git/parse-options.c

1339 lines
35 KiB
C
Raw Normal View History

#include "git-compat-util.h"
#include "parse-options.h"
#include "abspath.h"
#include "parse.h"
#include "gettext.h"
#include "strbuf.h"
#include "string-list.h"
#include "utf8.h"
static int disallow_abbreviated_options;
enum opt_parsed {
OPT_LONG = 0,
OPT_SHORT = 1<<0,
OPT_UNSET = 1<<1,
};
static void optbug(const struct option *opt, const char *reason)
{
if (opt->long_name && opt->short_name)
bug("switch '%c' (--%s) %s", opt->short_name,
opt->long_name, reason);
else if (opt->long_name)
bug("option '%s' %s", opt->long_name, reason);
else
bug("switch '%c' %s", opt->short_name, reason);
}
static const char *optname(const struct option *opt, enum opt_parsed flags)
{
static struct strbuf sb = STRBUF_INIT;
strbuf_reset(&sb);
if (flags & OPT_SHORT)
strbuf_addf(&sb, "switch `%c'", opt->short_name);
else if (flags & OPT_UNSET)
strbuf_addf(&sb, "option `no-%s'", opt->long_name);
else if (flags == OPT_LONG)
strbuf_addf(&sb, "option `%s'", opt->long_name);
else
BUG("optname() got unknown flags %d", flags);
return sb.buf;
}
static enum parse_opt_result get_arg(struct parse_opt_ctx_t *p,
const struct option *opt,
enum opt_parsed flags, const char **arg)
{
if (p->opt) {
*arg = p->opt;
p->opt = NULL;
} else if (p->argc == 1 && (opt->flags & PARSE_OPT_LASTARG_DEFAULT)) {
*arg = (const char *)opt->defval;
} else if (p->argc > 1) {
p->argc--;
*arg = *++p->argv;
} else
return error(_("%s requires a value"), optname(opt, flags));
return 0;
}
static char *fix_filename(const char *prefix, const char *file)
{
parse-options: consistently allocate memory in fix_filename() When handling OPT_FILENAME(), we have to stick the "prefix" (if any) in front of the filename to make up for the fact that Git has chdir()'d to the top of the repository. We can do this with prefix_filename(), but there are a few special cases we handle ourselves. Unfortunately the memory allocation is inconsistent here; if we do make it to prefix_filename(), we'll allocate a string which the caller must free to avoid a leak. But if we hit our special cases, we'll return the string as-is, and a caller which tries to free it will crash. So there's no way to win. Let's consistently allocate, so that callers can do the right thing. There are now three cases to care about in the function (and hence a three-armed if/else): 1. we got a NULL input (and should leave it as NULL, though arguably this is the sign of a bug; let's keep the status quo for now and we can pick at that scab later) 2. we hit a special case that means we leave the name intact; we should duplicate the string. This includes our special "-" matching. Prior to this patch, it also included empty prefixes and absolute filenames. But we can observe that prefix_filename() already handles these, so we don't need to detect them. 3. everything else goes to prefix_filename() I've dropped the "const" from the "char **file" parameter to indicate that we're allocating, though in practice it's not really important. This is all being shuffled through a void pointer via opt->value before it hits code which ever looks at the string. And it's even a bit weird, because we are really taking _in_ a const string and using the same out-parameter for a non-const string. A better function signature would be: static char *fix_filename(const char *prefix, const char *file); but that would mean the caller dereferences the double-pointer (and the NULL check is currently handled inside this function). So I took the path of least-change here. Note that we have to fix several callers in this commit, too, or we'll break the leak-checking tests. These are "new" leaks in the sense that they are now triggered by the test suite, but these spots have always been leaky when Git is run in a subdirectory of the repository. I fixed all of the cases that trigger with GIT_TEST_PASSING_SANITIZE_LEAK. There may be others in scripts that have other leaks, but we can fix them later along with those other leaks (and again, you _couldn't_ fix them before this patch, so this is the necessary first step). Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-03-04 18:31:22 +08:00
if (!file || !*file)
return NULL;
parse-options: consistently allocate memory in fix_filename() When handling OPT_FILENAME(), we have to stick the "prefix" (if any) in front of the filename to make up for the fact that Git has chdir()'d to the top of the repository. We can do this with prefix_filename(), but there are a few special cases we handle ourselves. Unfortunately the memory allocation is inconsistent here; if we do make it to prefix_filename(), we'll allocate a string which the caller must free to avoid a leak. But if we hit our special cases, we'll return the string as-is, and a caller which tries to free it will crash. So there's no way to win. Let's consistently allocate, so that callers can do the right thing. There are now three cases to care about in the function (and hence a three-armed if/else): 1. we got a NULL input (and should leave it as NULL, though arguably this is the sign of a bug; let's keep the status quo for now and we can pick at that scab later) 2. we hit a special case that means we leave the name intact; we should duplicate the string. This includes our special "-" matching. Prior to this patch, it also included empty prefixes and absolute filenames. But we can observe that prefix_filename() already handles these, so we don't need to detect them. 3. everything else goes to prefix_filename() I've dropped the "const" from the "char **file" parameter to indicate that we're allocating, though in practice it's not really important. This is all being shuffled through a void pointer via opt->value before it hits code which ever looks at the string. And it's even a bit weird, because we are really taking _in_ a const string and using the same out-parameter for a non-const string. A better function signature would be: static char *fix_filename(const char *prefix, const char *file); but that would mean the caller dereferences the double-pointer (and the NULL check is currently handled inside this function). So I took the path of least-change here. Note that we have to fix several callers in this commit, too, or we'll break the leak-checking tests. These are "new" leaks in the sense that they are now triggered by the test suite, but these spots have always been leaky when Git is run in a subdirectory of the repository. I fixed all of the cases that trigger with GIT_TEST_PASSING_SANITIZE_LEAK. There may be others in scripts that have other leaks, but we can fix them later along with those other leaks (and again, you _couldn't_ fix them before this patch, so this is the necessary first step). Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-03-04 18:31:22 +08:00
else
return prefix_filename_except_for_dash(prefix, file);
}
parse-options: make CMDMODE errors more precise Only a single PARSE_OPT_CMDMODE option can be specified for the same variable at the same time. This is enforced by get_value(), but the error messages are imprecise in three ways: 1. If a non-PARSE_OPT_CMDMODE option changes the value variable of a PARSE_OPT_CMDMODE option then an ominously vague message is shown: $ t/helper/test-tool parse-options --set23 --mode1 error: option `mode1' : incompatible with something else Worse: If the order of options is reversed then no error is reported at all: $ t/helper/test-tool parse-options --mode1 --set23 boolean: 0 integer: 23 magnitude: 0 timestamp: 0 string: (not set) abbrev: 7 verbose: -1 quiet: 0 dry run: no file: (not set) Fortunately this can currently only happen in the test helper; actual Git commands don't share the same variable for the value of options with and without the flag PARSE_OPT_CMDMODE. 2. If there are multiple options with the same value (synonyms), then the one that is defined first is shown rather than the one actually given on the command line, which is confusing: $ git am --resolved --quit error: option `quit' is incompatible with --continue 3. Arguments of PARSE_OPT_CMDMODE options are not handled by the parse-option machinery. This is left to the callback function. We currently only have a single affected option, --show-current-patch of git am. Errors for it can show an argument that was not actually given on the command line: $ git am --show-current-patch --show-current-patch=diff error: options '--show-current-patch=diff' and '--show-current-patch=raw' cannot be used together The options --show-current-patch and --show-current-patch=raw are synonyms, but the error accuses the user of input they did not actually made. Or it can awkwardly print a NULL pointer: $ git am --show-current-patch=diff --show-current-patch error: options '--show-current-patch=(null)' and '--show-current-patch=diff' cannot be used together The reasons for these shortcomings is that the current code checks incompatibility only when encountering a PARSE_OPT_CMDMODE option at the command line, and that it searches the previous incompatible option by value. Fix the first two points by checking all PARSE_OPT_CMDMODE variables after parsing each option and by storing all relevant details if their value changed. Do that whether or not the changing options has the flag PARSE_OPT_CMDMODE set. Report an incompatibility only if two options change the variable to different values and at least one of them is a PARSE_OPT_CMDMODE option. This changes the output of the first three examples above to: $ t/helper/test-tool parse-options --set23 --mode1 error: --mode1 is incompatible with --set23 $ t/helper/test-tool parse-options --mode1 --set23 error: --set23 is incompatible with --mode1 $ git am --resolved --quit error: --quit is incompatible with --resolved Store the argument of PARSE_OPT_CMDMODE options of type OPTION_CALLBACK as well to allow taking over the responsibility for compatibility checking from the callback function. The next patch will use this capability to fix the messages for git am --show-current-patch. Use a linked list for storing the PARSE_OPT_CMDMODE variables. This somewhat outdated data structure is simple and suffices, as the number of elements per command is currently only zero or one. We do support multiple different command modes variables per command, but I don't expect that we'd ever use a significant number of them. Once we do we can switch to a hashmap. Since we no longer need to search the conflicting option, the all_opts parameter of get_value() is no longer used. Remove it. Extend the tests to check for both conflicting option names, but don't insist on a particular order. Signed-off-by: René Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-10-28 19:53:01 +08:00
static enum parse_opt_result do_get_value(struct parse_opt_ctx_t *p,
const struct option *opt,
enum opt_parsed flags,
const char **argp)
{
const char *s, *arg;
const int unset = flags & OPT_UNSET;
int err;
if (unset && p->opt)
return error(_("%s takes no value"), optname(opt, flags));
if (unset && (opt->flags & PARSE_OPT_NONEG))
return error(_("%s isn't available"), optname(opt, flags));
if (!(flags & OPT_SHORT) && p->opt && (opt->flags & PARSE_OPT_NOARG))
return error(_("%s takes no value"), optname(opt, flags));
switch (opt->type) {
case OPTION_LOWLEVEL_CALLBACK:
return opt->ll_callback(p, opt, NULL, unset);
case OPTION_BIT:
if (unset)
*(int *)opt->value &= ~opt->defval;
else
*(int *)opt->value |= opt->defval;
return 0;
case OPTION_NEGBIT:
if (unset)
*(int *)opt->value |= opt->defval;
else
*(int *)opt->value &= ~opt->defval;
return 0;
case OPTION_BITOP:
if (unset)
BUG("BITOP can't have unset form");
*(int *)opt->value &= ~opt->extra;
*(int *)opt->value |= opt->defval;
return 0;
parse-options: deprecate OPT_BOOLEAN It is natural to expect that an option defined with OPT_BOOLEAN() could be used in this way: int option = -1; /* unspecified */ struct option options[] = { OPT_BOOLEAN(0, "option", &option, "set option"), OPT_END() }; parse_options(ac, av, prefix, options, usage, 0); if (option < 0) ... do the default thing ... else if (!option) ... --no-option was given ... else ... --option was given ... to easily tell three cases apart: - There is no mention of the `--option` on the command line; - The variable is positively set with `--option`; or - The variable is explicitly negated with `--no-option`. Unfortunately, this is not the case. OPT_BOOLEAN() increments the variable every time `--option` is given, and resets it to zero when `--no-option` is given. As a first step to remedy this, introduce a true boolean OPT_BOOL(), and rename OPT_BOOLEAN() to OPT_COUNTUP(). To help transitioning, OPT_BOOLEAN and OPTION_BOOLEAN are defined as deprecated synonyms to OPT_COUNTUP and OPTION_COUNTUP respectively. This is what db7244b (parse-options new features., 2007-11-07) from four years ago started by marking OPTION_BOOLEAN as "INCR would have been a better name". Some existing users do depend on the count-up semantics; for example, users of OPT__VERBOSE() could use it to raise the verbosity level with repeated use of `-v` on the command line, but they probably should be rewritten to use OPT__VERBOSITY() instead these days. I suspect that some users of OPT__FORCE() may also use it to implement different level of forcibleness but I didn't check. On top of this patch, here are the remaining clean-up tasks that other people can help: - Look at each hit in "git grep -e OPT_BOOLEAN"; trace all uses of the value that is set to the underlying variable, and if it can proven that the variable is only used as a boolean, replace it with OPT_BOOL(). If the caller does depend on the count-up semantics, replace it with OPT_COUNTUP() instead. - Same for OPTION_BOOLEAN; replace it with OPTION_SET_INT and arrange to set 1 to the variable for a true boolean, and otherwise replace it with OPTION_COUNTUP. - Look at each hit in "git grep -e OPT__VERBOSE -e OPT__QUIET" and see if they can be replaced with OPT__VERBOSITY(). I'll follow this message up with a separate patch as an example. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-09-28 07:56:49 +08:00
case OPTION_COUNTUP:
parse-options.c: make OPTION_COUNTUP respect "unspecified" values OPT_COUNTUP() merely increments the counter upon --option, and resets it to 0 upon --no-option, which means that there is no "unspecified" value with which a client can initialize the counter to determine whether or not --[no]-option was seen at all. Make OPT_COUNTUP() treat any negative number as an "unspecified" value to address this shortcoming. In particular, if a client initializes the counter to -1, then if it is still -1 after parse_options(), then neither --option nor --no-option was seen; if it is 0, then --no-option was seen last, and if it is 1 or greater, than --option was seen last. This change does not affect the behavior of existing clients because they all use the initial value of 0 (or more). Note that builtin/clean.c initializes the variable used with OPT__FORCE (which uses OPT_COUNTUP()) to a negative value, but it is set to either 0 or 1 by reading the configuration before the code calls parse_options(), i.e. as far as parse_options() is concerned, the initial value of the variable is not negative. To test this behavior, in test-parse-options.c, "verbose" is set to "unspecified" while quiet is set to 0 which will test the new behavior with all sets of values. Helped-by: Jeff King <peff@peff.net> Helped-by: Eric Sunshine <sunshine@sunshineco.com> Helped-by: Junio C Hamano <gitster@pobox.com> Signed-off-by: Pranit Bauva <pranit.bauva@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-05-05 17:50:00 +08:00
if (*(int *)opt->value < 0)
*(int *)opt->value = 0;
*(int *)opt->value = unset ? 0 : *(int *)opt->value + 1;
return 0;
case OPTION_SET_INT:
*(int *)opt->value = unset ? 0 : opt->defval;
return 0;
case OPTION_STRING:
if (unset)
*(const char **)opt->value = NULL;
else if (opt->flags & PARSE_OPT_OPTARG && !p->opt)
*(const char **)opt->value = (const char *)opt->defval;
else
return get_arg(p, opt, flags, (const char **)opt->value);
return 0;
case OPTION_FILENAME:
{
const char *value;
FREE_AND_NULL(*(char **)opt->value);
err = 0;
if (unset)
value = NULL;
else if (opt->flags & PARSE_OPT_OPTARG && !p->opt)
value = (const char *) opt->defval;
else
err = get_arg(p, opt, flags, &value);
if (!err)
*(char **)opt->value = fix_filename(p->prefix, value);
return err;
}
case OPTION_CALLBACK:
{
const char *p_arg = NULL;
int p_unset;
if (unset)
p_unset = 1;
else if (opt->flags & PARSE_OPT_NOARG)
p_unset = 0;
else if (opt->flags & PARSE_OPT_OPTARG && !p->opt)
p_unset = 0;
else if (get_arg(p, opt, flags, &arg))
return -1;
else {
p_unset = 0;
p_arg = arg;
}
parse-options: make CMDMODE errors more precise Only a single PARSE_OPT_CMDMODE option can be specified for the same variable at the same time. This is enforced by get_value(), but the error messages are imprecise in three ways: 1. If a non-PARSE_OPT_CMDMODE option changes the value variable of a PARSE_OPT_CMDMODE option then an ominously vague message is shown: $ t/helper/test-tool parse-options --set23 --mode1 error: option `mode1' : incompatible with something else Worse: If the order of options is reversed then no error is reported at all: $ t/helper/test-tool parse-options --mode1 --set23 boolean: 0 integer: 23 magnitude: 0 timestamp: 0 string: (not set) abbrev: 7 verbose: -1 quiet: 0 dry run: no file: (not set) Fortunately this can currently only happen in the test helper; actual Git commands don't share the same variable for the value of options with and without the flag PARSE_OPT_CMDMODE. 2. If there are multiple options with the same value (synonyms), then the one that is defined first is shown rather than the one actually given on the command line, which is confusing: $ git am --resolved --quit error: option `quit' is incompatible with --continue 3. Arguments of PARSE_OPT_CMDMODE options are not handled by the parse-option machinery. This is left to the callback function. We currently only have a single affected option, --show-current-patch of git am. Errors for it can show an argument that was not actually given on the command line: $ git am --show-current-patch --show-current-patch=diff error: options '--show-current-patch=diff' and '--show-current-patch=raw' cannot be used together The options --show-current-patch and --show-current-patch=raw are synonyms, but the error accuses the user of input they did not actually made. Or it can awkwardly print a NULL pointer: $ git am --show-current-patch=diff --show-current-patch error: options '--show-current-patch=(null)' and '--show-current-patch=diff' cannot be used together The reasons for these shortcomings is that the current code checks incompatibility only when encountering a PARSE_OPT_CMDMODE option at the command line, and that it searches the previous incompatible option by value. Fix the first two points by checking all PARSE_OPT_CMDMODE variables after parsing each option and by storing all relevant details if their value changed. Do that whether or not the changing options has the flag PARSE_OPT_CMDMODE set. Report an incompatibility only if two options change the variable to different values and at least one of them is a PARSE_OPT_CMDMODE option. This changes the output of the first three examples above to: $ t/helper/test-tool parse-options --set23 --mode1 error: --mode1 is incompatible with --set23 $ t/helper/test-tool parse-options --mode1 --set23 error: --set23 is incompatible with --mode1 $ git am --resolved --quit error: --quit is incompatible with --resolved Store the argument of PARSE_OPT_CMDMODE options of type OPTION_CALLBACK as well to allow taking over the responsibility for compatibility checking from the callback function. The next patch will use this capability to fix the messages for git am --show-current-patch. Use a linked list for storing the PARSE_OPT_CMDMODE variables. This somewhat outdated data structure is simple and suffices, as the number of elements per command is currently only zero or one. We do support multiple different command modes variables per command, but I don't expect that we'd ever use a significant number of them. Once we do we can switch to a hashmap. Since we no longer need to search the conflicting option, the all_opts parameter of get_value() is no longer used. Remove it. Extend the tests to check for both conflicting option names, but don't insist on a particular order. Signed-off-by: René Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-10-28 19:53:01 +08:00
if (opt->flags & PARSE_OPT_CMDMODE)
*argp = p_arg;
if (opt->callback)
return (*opt->callback)(opt, p_arg, p_unset) ? (-1) : 0;
else
return (*opt->ll_callback)(p, opt, p_arg, p_unset);
}
case OPTION_INTEGER:
if (unset) {
*(int *)opt->value = 0;
return 0;
}
if (opt->flags & PARSE_OPT_OPTARG && !p->opt) {
*(int *)opt->value = opt->defval;
return 0;
}
if (get_arg(p, opt, flags, &arg))
return -1;
if (!*arg)
return error(_("%s expects a numerical value"),
optname(opt, flags));
*(int *)opt->value = strtol(arg, (char **)&s, 10);
if (*s)
return error(_("%s expects a numerical value"),
optname(opt, flags));
return 0;
case OPTION_MAGNITUDE:
if (unset) {
*(unsigned long *)opt->value = 0;
return 0;
}
if (opt->flags & PARSE_OPT_OPTARG && !p->opt) {
*(unsigned long *)opt->value = opt->defval;
return 0;
}
if (get_arg(p, opt, flags, &arg))
return -1;
if (!git_parse_ulong(arg, opt->value))
return error(_("%s expects a non-negative integer value"
" with an optional k/m/g suffix"),
optname(opt, flags));
return 0;
default:
BUG("opt->type %d should not happen", opt->type);
}
}
parse-options: make CMDMODE errors more precise Only a single PARSE_OPT_CMDMODE option can be specified for the same variable at the same time. This is enforced by get_value(), but the error messages are imprecise in three ways: 1. If a non-PARSE_OPT_CMDMODE option changes the value variable of a PARSE_OPT_CMDMODE option then an ominously vague message is shown: $ t/helper/test-tool parse-options --set23 --mode1 error: option `mode1' : incompatible with something else Worse: If the order of options is reversed then no error is reported at all: $ t/helper/test-tool parse-options --mode1 --set23 boolean: 0 integer: 23 magnitude: 0 timestamp: 0 string: (not set) abbrev: 7 verbose: -1 quiet: 0 dry run: no file: (not set) Fortunately this can currently only happen in the test helper; actual Git commands don't share the same variable for the value of options with and without the flag PARSE_OPT_CMDMODE. 2. If there are multiple options with the same value (synonyms), then the one that is defined first is shown rather than the one actually given on the command line, which is confusing: $ git am --resolved --quit error: option `quit' is incompatible with --continue 3. Arguments of PARSE_OPT_CMDMODE options are not handled by the parse-option machinery. This is left to the callback function. We currently only have a single affected option, --show-current-patch of git am. Errors for it can show an argument that was not actually given on the command line: $ git am --show-current-patch --show-current-patch=diff error: options '--show-current-patch=diff' and '--show-current-patch=raw' cannot be used together The options --show-current-patch and --show-current-patch=raw are synonyms, but the error accuses the user of input they did not actually made. Or it can awkwardly print a NULL pointer: $ git am --show-current-patch=diff --show-current-patch error: options '--show-current-patch=(null)' and '--show-current-patch=diff' cannot be used together The reasons for these shortcomings is that the current code checks incompatibility only when encountering a PARSE_OPT_CMDMODE option at the command line, and that it searches the previous incompatible option by value. Fix the first two points by checking all PARSE_OPT_CMDMODE variables after parsing each option and by storing all relevant details if their value changed. Do that whether or not the changing options has the flag PARSE_OPT_CMDMODE set. Report an incompatibility only if two options change the variable to different values and at least one of them is a PARSE_OPT_CMDMODE option. This changes the output of the first three examples above to: $ t/helper/test-tool parse-options --set23 --mode1 error: --mode1 is incompatible with --set23 $ t/helper/test-tool parse-options --mode1 --set23 error: --set23 is incompatible with --mode1 $ git am --resolved --quit error: --quit is incompatible with --resolved Store the argument of PARSE_OPT_CMDMODE options of type OPTION_CALLBACK as well to allow taking over the responsibility for compatibility checking from the callback function. The next patch will use this capability to fix the messages for git am --show-current-patch. Use a linked list for storing the PARSE_OPT_CMDMODE variables. This somewhat outdated data structure is simple and suffices, as the number of elements per command is currently only zero or one. We do support multiple different command modes variables per command, but I don't expect that we'd ever use a significant number of them. Once we do we can switch to a hashmap. Since we no longer need to search the conflicting option, the all_opts parameter of get_value() is no longer used. Remove it. Extend the tests to check for both conflicting option names, but don't insist on a particular order. Signed-off-by: René Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-10-28 19:53:01 +08:00
struct parse_opt_cmdmode_list {
int value, *value_ptr;
const struct option *opt;
const char *arg;
enum opt_parsed flags;
struct parse_opt_cmdmode_list *next;
};
static void build_cmdmode_list(struct parse_opt_ctx_t *ctx,
const struct option *opts)
{
ctx->cmdmode_list = NULL;
for (; opts->type != OPTION_END; opts++) {
struct parse_opt_cmdmode_list *elem = ctx->cmdmode_list;
int *value_ptr = opts->value;
if (!(opts->flags & PARSE_OPT_CMDMODE) || !value_ptr)
continue;
while (elem && elem->value_ptr != value_ptr)
elem = elem->next;
if (elem)
continue;
CALLOC_ARRAY(elem, 1);
elem->value_ptr = value_ptr;
elem->value = *value_ptr;
elem->next = ctx->cmdmode_list;
ctx->cmdmode_list = elem;
}
}
static char *optnamearg(const struct option *opt, const char *arg,
enum opt_parsed flags)
{
if (flags & OPT_SHORT)
return xstrfmt("-%c%s", opt->short_name, arg ? arg : "");
return xstrfmt("--%s%s%s%s", flags & OPT_UNSET ? "no-" : "",
opt->long_name, arg ? "=" : "", arg ? arg : "");
}
static enum parse_opt_result get_value(struct parse_opt_ctx_t *p,
const struct option *opt,
enum opt_parsed flags)
{
const char *arg = NULL;
enum parse_opt_result result = do_get_value(p, opt, flags, &arg);
struct parse_opt_cmdmode_list *elem = p->cmdmode_list;
char *opt_name, *other_opt_name;
for (; elem; elem = elem->next) {
if (*elem->value_ptr == elem->value)
continue;
if (elem->opt &&
(elem->opt->flags | opt->flags) & PARSE_OPT_CMDMODE)
break;
elem->opt = opt;
elem->arg = arg;
elem->flags = flags;
elem->value = *elem->value_ptr;
}
if (result || !elem)
return result;
opt_name = optnamearg(opt, arg, flags);
other_opt_name = optnamearg(elem->opt, elem->arg, elem->flags);
error(_("options '%s' and '%s' cannot be used together"),
opt_name, other_opt_name);
parse-options: make CMDMODE errors more precise Only a single PARSE_OPT_CMDMODE option can be specified for the same variable at the same time. This is enforced by get_value(), but the error messages are imprecise in three ways: 1. If a non-PARSE_OPT_CMDMODE option changes the value variable of a PARSE_OPT_CMDMODE option then an ominously vague message is shown: $ t/helper/test-tool parse-options --set23 --mode1 error: option `mode1' : incompatible with something else Worse: If the order of options is reversed then no error is reported at all: $ t/helper/test-tool parse-options --mode1 --set23 boolean: 0 integer: 23 magnitude: 0 timestamp: 0 string: (not set) abbrev: 7 verbose: -1 quiet: 0 dry run: no file: (not set) Fortunately this can currently only happen in the test helper; actual Git commands don't share the same variable for the value of options with and without the flag PARSE_OPT_CMDMODE. 2. If there are multiple options with the same value (synonyms), then the one that is defined first is shown rather than the one actually given on the command line, which is confusing: $ git am --resolved --quit error: option `quit' is incompatible with --continue 3. Arguments of PARSE_OPT_CMDMODE options are not handled by the parse-option machinery. This is left to the callback function. We currently only have a single affected option, --show-current-patch of git am. Errors for it can show an argument that was not actually given on the command line: $ git am --show-current-patch --show-current-patch=diff error: options '--show-current-patch=diff' and '--show-current-patch=raw' cannot be used together The options --show-current-patch and --show-current-patch=raw are synonyms, but the error accuses the user of input they did not actually made. Or it can awkwardly print a NULL pointer: $ git am --show-current-patch=diff --show-current-patch error: options '--show-current-patch=(null)' and '--show-current-patch=diff' cannot be used together The reasons for these shortcomings is that the current code checks incompatibility only when encountering a PARSE_OPT_CMDMODE option at the command line, and that it searches the previous incompatible option by value. Fix the first two points by checking all PARSE_OPT_CMDMODE variables after parsing each option and by storing all relevant details if their value changed. Do that whether or not the changing options has the flag PARSE_OPT_CMDMODE set. Report an incompatibility only if two options change the variable to different values and at least one of them is a PARSE_OPT_CMDMODE option. This changes the output of the first three examples above to: $ t/helper/test-tool parse-options --set23 --mode1 error: --mode1 is incompatible with --set23 $ t/helper/test-tool parse-options --mode1 --set23 error: --set23 is incompatible with --mode1 $ git am --resolved --quit error: --quit is incompatible with --resolved Store the argument of PARSE_OPT_CMDMODE options of type OPTION_CALLBACK as well to allow taking over the responsibility for compatibility checking from the callback function. The next patch will use this capability to fix the messages for git am --show-current-patch. Use a linked list for storing the PARSE_OPT_CMDMODE variables. This somewhat outdated data structure is simple and suffices, as the number of elements per command is currently only zero or one. We do support multiple different command modes variables per command, but I don't expect that we'd ever use a significant number of them. Once we do we can switch to a hashmap. Since we no longer need to search the conflicting option, the all_opts parameter of get_value() is no longer used. Remove it. Extend the tests to check for both conflicting option names, but don't insist on a particular order. Signed-off-by: René Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-10-28 19:53:01 +08:00
free(opt_name);
free(other_opt_name);
return -1;
}
static enum parse_opt_result parse_short_opt(struct parse_opt_ctx_t *p,
const struct option *options)
{
const struct option *numopt = NULL;
for (; options->type != OPTION_END; options++) {
if (options->short_name == *p->opt) {
p->opt = p->opt[1] ? p->opt + 1 : NULL;
parse-options: make CMDMODE errors more precise Only a single PARSE_OPT_CMDMODE option can be specified for the same variable at the same time. This is enforced by get_value(), but the error messages are imprecise in three ways: 1. If a non-PARSE_OPT_CMDMODE option changes the value variable of a PARSE_OPT_CMDMODE option then an ominously vague message is shown: $ t/helper/test-tool parse-options --set23 --mode1 error: option `mode1' : incompatible with something else Worse: If the order of options is reversed then no error is reported at all: $ t/helper/test-tool parse-options --mode1 --set23 boolean: 0 integer: 23 magnitude: 0 timestamp: 0 string: (not set) abbrev: 7 verbose: -1 quiet: 0 dry run: no file: (not set) Fortunately this can currently only happen in the test helper; actual Git commands don't share the same variable for the value of options with and without the flag PARSE_OPT_CMDMODE. 2. If there are multiple options with the same value (synonyms), then the one that is defined first is shown rather than the one actually given on the command line, which is confusing: $ git am --resolved --quit error: option `quit' is incompatible with --continue 3. Arguments of PARSE_OPT_CMDMODE options are not handled by the parse-option machinery. This is left to the callback function. We currently only have a single affected option, --show-current-patch of git am. Errors for it can show an argument that was not actually given on the command line: $ git am --show-current-patch --show-current-patch=diff error: options '--show-current-patch=diff' and '--show-current-patch=raw' cannot be used together The options --show-current-patch and --show-current-patch=raw are synonyms, but the error accuses the user of input they did not actually made. Or it can awkwardly print a NULL pointer: $ git am --show-current-patch=diff --show-current-patch error: options '--show-current-patch=(null)' and '--show-current-patch=diff' cannot be used together The reasons for these shortcomings is that the current code checks incompatibility only when encountering a PARSE_OPT_CMDMODE option at the command line, and that it searches the previous incompatible option by value. Fix the first two points by checking all PARSE_OPT_CMDMODE variables after parsing each option and by storing all relevant details if their value changed. Do that whether or not the changing options has the flag PARSE_OPT_CMDMODE set. Report an incompatibility only if two options change the variable to different values and at least one of them is a PARSE_OPT_CMDMODE option. This changes the output of the first three examples above to: $ t/helper/test-tool parse-options --set23 --mode1 error: --mode1 is incompatible with --set23 $ t/helper/test-tool parse-options --mode1 --set23 error: --set23 is incompatible with --mode1 $ git am --resolved --quit error: --quit is incompatible with --resolved Store the argument of PARSE_OPT_CMDMODE options of type OPTION_CALLBACK as well to allow taking over the responsibility for compatibility checking from the callback function. The next patch will use this capability to fix the messages for git am --show-current-patch. Use a linked list for storing the PARSE_OPT_CMDMODE variables. This somewhat outdated data structure is simple and suffices, as the number of elements per command is currently only zero or one. We do support multiple different command modes variables per command, but I don't expect that we'd ever use a significant number of them. Once we do we can switch to a hashmap. Since we no longer need to search the conflicting option, the all_opts parameter of get_value() is no longer used. Remove it. Extend the tests to check for both conflicting option names, but don't insist on a particular order. Signed-off-by: René Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-10-28 19:53:01 +08:00
return get_value(p, options, OPT_SHORT);
}
/*
* Handle the numerical option later, explicit one-digit
* options take precedence over it.
*/
if (options->type == OPTION_NUMBER)
numopt = options;
}
if (numopt && isdigit(*p->opt)) {
size_t len = 1;
char *arg;
int rc;
while (isdigit(p->opt[len]))
len++;
arg = xmemdupz(p->opt, len);
p->opt = p->opt[len] ? p->opt + len : NULL;
if (numopt->callback)
rc = (*numopt->callback)(numopt, arg, 0) ? (-1) : 0;
else
rc = (*numopt->ll_callback)(p, numopt, arg, 0);
free(arg);
return rc;
}
return PARSE_OPT_UNKNOWN;
}
static int has_string(const char *it, const char **array)
{
while (*array)
if (!strcmp(it, *(array++)))
return 1;
return 0;
}
static int is_alias(struct parse_opt_ctx_t *ctx,
const struct option *one_opt,
const struct option *another_opt)
{
const char **group;
if (!ctx->alias_groups)
return 0;
if (!one_opt->long_name || !another_opt->long_name)
return 0;
for (group = ctx->alias_groups; *group; group += 3) {
/* it and other are from the same family? */
if (has_string(one_opt->long_name, group) &&
has_string(another_opt->long_name, group))
return 1;
}
return 0;
}
struct parsed_option {
const struct option *option;
enum opt_parsed flags;
};
static void register_abbrev(struct parse_opt_ctx_t *p,
const struct option *option, enum opt_parsed flags,
struct parsed_option *abbrev,
struct parsed_option *ambiguous)
{
if (p->flags & PARSE_OPT_KEEP_UNKNOWN_OPT)
return;
if (abbrev->option &&
!(abbrev->flags == flags && is_alias(p, abbrev->option, option))) {
/*
* If this is abbreviated, it is
* ambiguous. So when there is no
* exact match later, we need to
* error out.
*/
ambiguous->option = abbrev->option;
ambiguous->flags = abbrev->flags;
}
abbrev->option = option;
abbrev->flags = flags;
}
static enum parse_opt_result parse_long_opt(
struct parse_opt_ctx_t *p, const char *arg,
const struct option *options)
{
const char *arg_end = strchrnul(arg, '=');
const char *arg_start = arg;
enum opt_parsed flags = OPT_LONG;
int arg_starts_with_no_no = 0;
struct parsed_option abbrev = { .option = NULL, .flags = OPT_LONG };
struct parsed_option ambiguous = { .option = NULL, .flags = OPT_LONG };
if (skip_prefix(arg_start, "no-", &arg_start)) {
if (skip_prefix(arg_start, "no-", &arg_start))
arg_starts_with_no_no = 1;
else
flags |= OPT_UNSET;
}
for (; options->type != OPTION_END; options++) {
const char *rest, *long_name = options->long_name;
enum opt_parsed opt_flags = OPT_LONG;
int allow_unset = !(options->flags & PARSE_OPT_NONEG);
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
if (options->type == OPTION_SUBCOMMAND)
continue;
if (!long_name)
continue;
if (skip_prefix(long_name, "no-", &long_name))
opt_flags |= OPT_UNSET;
else if (arg_starts_with_no_no)
continue;
if (((flags ^ opt_flags) & OPT_UNSET) && !allow_unset)
continue;
if (skip_prefix(arg_start, long_name, &rest)) {
if (*rest == '=')
p->opt = rest + 1;
else if (*rest)
continue;
return get_value(p, options, flags ^ opt_flags);
}
/* abbreviated? */
if (!strncmp(long_name, arg_start, arg_end - arg_start))
register_abbrev(p, options, flags ^ opt_flags,
&abbrev, &ambiguous);
/* negated and abbreviated very much? */
if (allow_unset && starts_with("no-", arg))
register_abbrev(p, options, OPT_UNSET ^ opt_flags,
&abbrev, &ambiguous);
}
if (disallow_abbreviated_options && (ambiguous.option || abbrev.option))
die("disallowed abbreviated or ambiguous option '%.*s'",
(int)(arg_end - arg), arg);
if (ambiguous.option) {
error(_("ambiguous option: %s "
"(could be --%s%s or --%s%s)"),
arg,
(ambiguous.flags & OPT_UNSET) ? "no-" : "",
ambiguous.option->long_name,
(abbrev.flags & OPT_UNSET) ? "no-" : "",
abbrev.option->long_name);
return PARSE_OPT_HELP;
}
if (abbrev.option) {
if (*arg_end)
p->opt = arg_end + 1;
return get_value(p, abbrev.option, abbrev.flags);
}
return PARSE_OPT_UNKNOWN;
}
static enum parse_opt_result parse_nodash_opt(struct parse_opt_ctx_t *p,
const char *arg,
const struct option *options)
{
for (; options->type != OPTION_END; options++) {
if (!(options->flags & PARSE_OPT_NODASH))
continue;
if (options->short_name == arg[0] && arg[1] == '\0')
parse-options: make CMDMODE errors more precise Only a single PARSE_OPT_CMDMODE option can be specified for the same variable at the same time. This is enforced by get_value(), but the error messages are imprecise in three ways: 1. If a non-PARSE_OPT_CMDMODE option changes the value variable of a PARSE_OPT_CMDMODE option then an ominously vague message is shown: $ t/helper/test-tool parse-options --set23 --mode1 error: option `mode1' : incompatible with something else Worse: If the order of options is reversed then no error is reported at all: $ t/helper/test-tool parse-options --mode1 --set23 boolean: 0 integer: 23 magnitude: 0 timestamp: 0 string: (not set) abbrev: 7 verbose: -1 quiet: 0 dry run: no file: (not set) Fortunately this can currently only happen in the test helper; actual Git commands don't share the same variable for the value of options with and without the flag PARSE_OPT_CMDMODE. 2. If there are multiple options with the same value (synonyms), then the one that is defined first is shown rather than the one actually given on the command line, which is confusing: $ git am --resolved --quit error: option `quit' is incompatible with --continue 3. Arguments of PARSE_OPT_CMDMODE options are not handled by the parse-option machinery. This is left to the callback function. We currently only have a single affected option, --show-current-patch of git am. Errors for it can show an argument that was not actually given on the command line: $ git am --show-current-patch --show-current-patch=diff error: options '--show-current-patch=diff' and '--show-current-patch=raw' cannot be used together The options --show-current-patch and --show-current-patch=raw are synonyms, but the error accuses the user of input they did not actually made. Or it can awkwardly print a NULL pointer: $ git am --show-current-patch=diff --show-current-patch error: options '--show-current-patch=(null)' and '--show-current-patch=diff' cannot be used together The reasons for these shortcomings is that the current code checks incompatibility only when encountering a PARSE_OPT_CMDMODE option at the command line, and that it searches the previous incompatible option by value. Fix the first two points by checking all PARSE_OPT_CMDMODE variables after parsing each option and by storing all relevant details if their value changed. Do that whether or not the changing options has the flag PARSE_OPT_CMDMODE set. Report an incompatibility only if two options change the variable to different values and at least one of them is a PARSE_OPT_CMDMODE option. This changes the output of the first three examples above to: $ t/helper/test-tool parse-options --set23 --mode1 error: --mode1 is incompatible with --set23 $ t/helper/test-tool parse-options --mode1 --set23 error: --set23 is incompatible with --mode1 $ git am --resolved --quit error: --quit is incompatible with --resolved Store the argument of PARSE_OPT_CMDMODE options of type OPTION_CALLBACK as well to allow taking over the responsibility for compatibility checking from the callback function. The next patch will use this capability to fix the messages for git am --show-current-patch. Use a linked list for storing the PARSE_OPT_CMDMODE variables. This somewhat outdated data structure is simple and suffices, as the number of elements per command is currently only zero or one. We do support multiple different command modes variables per command, but I don't expect that we'd ever use a significant number of them. Once we do we can switch to a hashmap. Since we no longer need to search the conflicting option, the all_opts parameter of get_value() is no longer used. Remove it. Extend the tests to check for both conflicting option names, but don't insist on a particular order. Signed-off-by: René Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-10-28 19:53:01 +08:00
return get_value(p, options, OPT_SHORT);
}
return PARSE_OPT_ERROR;
}
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
static enum parse_opt_result parse_subcommand(const char *arg,
const struct option *options)
{
for (; options->type != OPTION_END; options++)
if (options->type == OPTION_SUBCOMMAND &&
!strcmp(options->long_name, arg)) {
*(parse_opt_subcommand_fn **)options->value = options->subcommand_fn;
return PARSE_OPT_SUBCOMMAND;
}
return PARSE_OPT_UNKNOWN;
}
static void check_typos(const char *arg, const struct option *options)
{
if (strlen(arg) < 3)
return;
if (starts_with(arg, "no-")) {
error(_("did you mean `--%s` (with two dashes)?"), arg);
exit(129);
}
for (; options->type != OPTION_END; options++) {
if (!options->long_name)
continue;
if (starts_with(options->long_name, arg)) {
error(_("did you mean `--%s` (with two dashes)?"), arg);
exit(129);
}
}
}
static void parse_options_check(const struct option *opts)
{
char short_opts[128];
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
void *subcommand_value = NULL;
memset(short_opts, '\0', sizeof(short_opts));
for (; opts->type != OPTION_END; opts++) {
if ((opts->flags & PARSE_OPT_LASTARG_DEFAULT) &&
(opts->flags & PARSE_OPT_OPTARG))
optbug(opts, "uses incompatible flags "
"LASTARG_DEFAULT and OPTARG");
if (opts->short_name) {
if (0x7F <= opts->short_name)
optbug(opts, "invalid short name");
else if (short_opts[opts->short_name]++)
optbug(opts, "short name already used");
}
if (opts->flags & PARSE_OPT_NODASH &&
((opts->flags & PARSE_OPT_OPTARG) ||
!(opts->flags & PARSE_OPT_NOARG) ||
!(opts->flags & PARSE_OPT_NONEG) ||
opts->long_name))
optbug(opts, "uses feature "
"not supported for dashless options");
if (opts->type == OPTION_SET_INT && !opts->defval &&
opts->long_name && !(opts->flags & PARSE_OPT_NONEG))
optbug(opts, "OPTION_SET_INT 0 should not be negatable");
switch (opts->type) {
parse-options: deprecate OPT_BOOLEAN It is natural to expect that an option defined with OPT_BOOLEAN() could be used in this way: int option = -1; /* unspecified */ struct option options[] = { OPT_BOOLEAN(0, "option", &option, "set option"), OPT_END() }; parse_options(ac, av, prefix, options, usage, 0); if (option < 0) ... do the default thing ... else if (!option) ... --no-option was given ... else ... --option was given ... to easily tell three cases apart: - There is no mention of the `--option` on the command line; - The variable is positively set with `--option`; or - The variable is explicitly negated with `--no-option`. Unfortunately, this is not the case. OPT_BOOLEAN() increments the variable every time `--option` is given, and resets it to zero when `--no-option` is given. As a first step to remedy this, introduce a true boolean OPT_BOOL(), and rename OPT_BOOLEAN() to OPT_COUNTUP(). To help transitioning, OPT_BOOLEAN and OPTION_BOOLEAN are defined as deprecated synonyms to OPT_COUNTUP and OPTION_COUNTUP respectively. This is what db7244b (parse-options new features., 2007-11-07) from four years ago started by marking OPTION_BOOLEAN as "INCR would have been a better name". Some existing users do depend on the count-up semantics; for example, users of OPT__VERBOSE() could use it to raise the verbosity level with repeated use of `-v` on the command line, but they probably should be rewritten to use OPT__VERBOSITY() instead these days. I suspect that some users of OPT__FORCE() may also use it to implement different level of forcibleness but I didn't check. On top of this patch, here are the remaining clean-up tasks that other people can help: - Look at each hit in "git grep -e OPT_BOOLEAN"; trace all uses of the value that is set to the underlying variable, and if it can proven that the variable is only used as a boolean, replace it with OPT_BOOL(). If the caller does depend on the count-up semantics, replace it with OPT_COUNTUP() instead. - Same for OPTION_BOOLEAN; replace it with OPTION_SET_INT and arrange to set 1 to the variable for a true boolean, and otherwise replace it with OPTION_COUNTUP. - Look at each hit in "git grep -e OPT__VERBOSE -e OPT__QUIET" and see if they can be replaced with OPT__VERBOSITY(). I'll follow this message up with a separate patch as an example. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-09-28 07:56:49 +08:00
case OPTION_COUNTUP:
case OPTION_BIT:
case OPTION_NEGBIT:
case OPTION_SET_INT:
case OPTION_NUMBER:
if ((opts->flags & PARSE_OPT_OPTARG) ||
!(opts->flags & PARSE_OPT_NOARG))
optbug(opts, "should not accept an argument");
break;
case OPTION_CALLBACK:
if (!opts->callback && !opts->ll_callback)
optbug(opts, "OPTION_CALLBACK needs one callback");
else if (opts->callback && opts->ll_callback)
optbug(opts, "OPTION_CALLBACK can't have two callbacks");
break;
case OPTION_LOWLEVEL_CALLBACK:
if (!opts->ll_callback)
optbug(opts, "OPTION_LOWLEVEL_CALLBACK needs a callback");
if (opts->callback)
optbug(opts, "OPTION_LOWLEVEL_CALLBACK needs no high level callback");
break;
case OPTION_ALIAS:
optbug(opts, "OPT_ALIAS() should not remain at this point. "
"Are you using parse_options_step() directly?\n"
"That case is not supported yet.");
break;
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
case OPTION_SUBCOMMAND:
if (!opts->value || !opts->subcommand_fn)
optbug(opts, "OPTION_SUBCOMMAND needs a value and a subcommand function");
if (!subcommand_value)
subcommand_value = opts->value;
else if (subcommand_value != opts->value)
optbug(opts, "all OPTION_SUBCOMMANDs need the same value");
break;
default:
; /* ok. (usually accepts an argument) */
}
if (opts->argh &&
strcspn(opts->argh, " _") != strlen(opts->argh))
optbug(opts, "multi-word argh should use dash to separate words");
}
BUG_if_bug("invalid 'struct option'");
}
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
static int has_subcommands(const struct option *options)
{
for (; options->type != OPTION_END; options++)
if (options->type == OPTION_SUBCOMMAND)
return 1;
return 0;
}
static void parse_options_start_1(struct parse_opt_ctx_t *ctx,
int argc, const char **argv, const char *prefix,
const struct option *options,
enum parse_opt_flags flags)
{
ctx->argc = argc;
ctx->argv = argv;
if (!(flags & PARSE_OPT_ONE_SHOT)) {
ctx->argc--;
ctx->argv++;
}
ctx->total = ctx->argc;
ctx->out = argv;
ctx->prefix = prefix;
ctx->cpidx = ((flags & PARSE_OPT_KEEP_ARGV0) != 0);
ctx->flags = flags;
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
ctx->has_subcommands = has_subcommands(options);
if (!ctx->has_subcommands && (flags & PARSE_OPT_SUBCOMMAND_OPTIONAL))
BUG("Using PARSE_OPT_SUBCOMMAND_OPTIONAL without subcommands");
if (ctx->has_subcommands) {
if (flags & PARSE_OPT_STOP_AT_NON_OPTION)
BUG("subcommands are incompatible with PARSE_OPT_STOP_AT_NON_OPTION");
if (!(flags & PARSE_OPT_SUBCOMMAND_OPTIONAL)) {
if (flags & PARSE_OPT_KEEP_UNKNOWN_OPT)
BUG("subcommands are incompatible with PARSE_OPT_KEEP_UNKNOWN_OPT unless in combination with PARSE_OPT_SUBCOMMAND_OPTIONAL");
if (flags & PARSE_OPT_KEEP_DASHDASH)
BUG("subcommands are incompatible with PARSE_OPT_KEEP_DASHDASH unless in combination with PARSE_OPT_SUBCOMMAND_OPTIONAL");
}
}
if ((flags & PARSE_OPT_KEEP_UNKNOWN_OPT) &&
(flags & PARSE_OPT_STOP_AT_NON_OPTION) &&
!(flags & PARSE_OPT_ONE_SHOT))
BUG("STOP_AT_NON_OPTION and KEEP_UNKNOWN don't go together");
if ((flags & PARSE_OPT_ONE_SHOT) &&
(flags & PARSE_OPT_KEEP_ARGV0))
BUG("Can't keep argv0 if you don't have it");
parse_options_check(options);
parse-options: make CMDMODE errors more precise Only a single PARSE_OPT_CMDMODE option can be specified for the same variable at the same time. This is enforced by get_value(), but the error messages are imprecise in three ways: 1. If a non-PARSE_OPT_CMDMODE option changes the value variable of a PARSE_OPT_CMDMODE option then an ominously vague message is shown: $ t/helper/test-tool parse-options --set23 --mode1 error: option `mode1' : incompatible with something else Worse: If the order of options is reversed then no error is reported at all: $ t/helper/test-tool parse-options --mode1 --set23 boolean: 0 integer: 23 magnitude: 0 timestamp: 0 string: (not set) abbrev: 7 verbose: -1 quiet: 0 dry run: no file: (not set) Fortunately this can currently only happen in the test helper; actual Git commands don't share the same variable for the value of options with and without the flag PARSE_OPT_CMDMODE. 2. If there are multiple options with the same value (synonyms), then the one that is defined first is shown rather than the one actually given on the command line, which is confusing: $ git am --resolved --quit error: option `quit' is incompatible with --continue 3. Arguments of PARSE_OPT_CMDMODE options are not handled by the parse-option machinery. This is left to the callback function. We currently only have a single affected option, --show-current-patch of git am. Errors for it can show an argument that was not actually given on the command line: $ git am --show-current-patch --show-current-patch=diff error: options '--show-current-patch=diff' and '--show-current-patch=raw' cannot be used together The options --show-current-patch and --show-current-patch=raw are synonyms, but the error accuses the user of input they did not actually made. Or it can awkwardly print a NULL pointer: $ git am --show-current-patch=diff --show-current-patch error: options '--show-current-patch=(null)' and '--show-current-patch=diff' cannot be used together The reasons for these shortcomings is that the current code checks incompatibility only when encountering a PARSE_OPT_CMDMODE option at the command line, and that it searches the previous incompatible option by value. Fix the first two points by checking all PARSE_OPT_CMDMODE variables after parsing each option and by storing all relevant details if their value changed. Do that whether or not the changing options has the flag PARSE_OPT_CMDMODE set. Report an incompatibility only if two options change the variable to different values and at least one of them is a PARSE_OPT_CMDMODE option. This changes the output of the first three examples above to: $ t/helper/test-tool parse-options --set23 --mode1 error: --mode1 is incompatible with --set23 $ t/helper/test-tool parse-options --mode1 --set23 error: --set23 is incompatible with --mode1 $ git am --resolved --quit error: --quit is incompatible with --resolved Store the argument of PARSE_OPT_CMDMODE options of type OPTION_CALLBACK as well to allow taking over the responsibility for compatibility checking from the callback function. The next patch will use this capability to fix the messages for git am --show-current-patch. Use a linked list for storing the PARSE_OPT_CMDMODE variables. This somewhat outdated data structure is simple and suffices, as the number of elements per command is currently only zero or one. We do support multiple different command modes variables per command, but I don't expect that we'd ever use a significant number of them. Once we do we can switch to a hashmap. Since we no longer need to search the conflicting option, the all_opts parameter of get_value() is no longer used. Remove it. Extend the tests to check for both conflicting option names, but don't insist on a particular order. Signed-off-by: René Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-10-28 19:53:01 +08:00
build_cmdmode_list(ctx, options);
}
void parse_options_start(struct parse_opt_ctx_t *ctx,
int argc, const char **argv, const char *prefix,
const struct option *options,
enum parse_opt_flags flags)
{
memset(ctx, 0, sizeof(*ctx));
parse_options_start_1(ctx, argc, argv, prefix, options, flags);
}
static void show_negated_gitcomp(const struct option *opts, int show_all,
int nr_noopts)
completion: collapse extra --no-.. options The commands that make use of --git-completion-helper feature could now produce a lot of --no-xxx options that a command can take. This in many case could nearly double the amount of completable options, using more screen estate and also harder to search for the wanted option. This patch attempts to mitigate that by collapsing extra --no- options, the ones that are added by --git-completion-helper and not in original struct option arrays. The "--no-..." option will be displayed in this case to hint about more options, e.g. > ~/w/git $ git clone -- --bare --origin= --branch= --progress --checkout --quiet --config= --recurse-submodules --depth= --reference= --dissociate --reference-if-able= --filter= --separate-git-dir= --hardlinks --shallow-exclude= --ipv4 --shallow-since= --ipv6 --shallow-submodules --jobs= --shared --local --single-branch --mirror --tags --no-... --template= --no-checkout --upload-pack= --no-hardlinks --verbose --no-tags and when you complete it with --no-<tab>, all negative options will be presented: > ~/w/git $ git clone --no- --no-bare --no-quiet --no-branch --no-recurse-submodules --no-checkout --no-reference --no-config --no-reference-if-able --no-depth --no-separate-git-dir --no-dissociate --no-shallow-exclude --no-filter --no-shallow-since --no-hardlinks --no-shallow-submodules --no-ipv4 --no-shared --no-ipv6 --no-single-branch --no-jobs --no-tags --no-local --no-template --no-mirror --no-upload-pack --no-origin --no-verbose --no-progress Corner case: to make sure that people will never accidentally complete the fake option "--no-..." there must be one real --no- in the first complete listing even if it's not from the original struct option. PS. This could could be made simpler with ";&" to fall through from "--no-*" block and share the code but ";&" is not available on bash-3 (i.e. Mac) Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-06-06 17:41:39 +08:00
{
int printed_dashdash = 0;
for (; opts->type != OPTION_END; opts++) {
int has_unset_form = 0;
const char *name;
if (!opts->long_name)
continue;
if (!show_all &&
(opts->flags & (PARSE_OPT_HIDDEN | PARSE_OPT_NOCOMPLETE)))
completion: collapse extra --no-.. options The commands that make use of --git-completion-helper feature could now produce a lot of --no-xxx options that a command can take. This in many case could nearly double the amount of completable options, using more screen estate and also harder to search for the wanted option. This patch attempts to mitigate that by collapsing extra --no- options, the ones that are added by --git-completion-helper and not in original struct option arrays. The "--no-..." option will be displayed in this case to hint about more options, e.g. > ~/w/git $ git clone -- --bare --origin= --branch= --progress --checkout --quiet --config= --recurse-submodules --depth= --reference= --dissociate --reference-if-able= --filter= --separate-git-dir= --hardlinks --shallow-exclude= --ipv4 --shallow-since= --ipv6 --shallow-submodules --jobs= --shared --local --single-branch --mirror --tags --no-... --template= --no-checkout --upload-pack= --no-hardlinks --verbose --no-tags and when you complete it with --no-<tab>, all negative options will be presented: > ~/w/git $ git clone --no- --no-bare --no-quiet --no-branch --no-recurse-submodules --no-checkout --no-reference --no-config --no-reference-if-able --no-depth --no-separate-git-dir --no-dissociate --no-shallow-exclude --no-filter --no-shallow-since --no-hardlinks --no-shallow-submodules --no-ipv4 --no-shared --no-ipv6 --no-single-branch --no-jobs --no-tags --no-local --no-template --no-mirror --no-upload-pack --no-origin --no-verbose --no-progress Corner case: to make sure that people will never accidentally complete the fake option "--no-..." there must be one real --no- in the first complete listing even if it's not from the original struct option. PS. This could could be made simpler with ";&" to fall through from "--no-*" block and share the code but ";&" is not available on bash-3 (i.e. Mac) Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-06-06 17:41:39 +08:00
continue;
if (opts->flags & PARSE_OPT_NONEG)
continue;
switch (opts->type) {
case OPTION_STRING:
case OPTION_FILENAME:
case OPTION_INTEGER:
case OPTION_MAGNITUDE:
case OPTION_CALLBACK:
case OPTION_BIT:
case OPTION_NEGBIT:
case OPTION_COUNTUP:
case OPTION_SET_INT:
has_unset_form = 1;
break;
default:
break;
}
if (!has_unset_form)
continue;
if (skip_prefix(opts->long_name, "no-", &name)) {
if (nr_noopts < 0)
printf(" --%s", name);
} else if (nr_noopts >= 0) {
if (nr_noopts && !printed_dashdash) {
printf(" --");
printed_dashdash = 1;
}
printf(" --no-%s", opts->long_name);
nr_noopts++;
}
}
}
static int show_gitcomp(const struct option *opts, int show_all)
{
completion: collapse extra --no-.. options The commands that make use of --git-completion-helper feature could now produce a lot of --no-xxx options that a command can take. This in many case could nearly double the amount of completable options, using more screen estate and also harder to search for the wanted option. This patch attempts to mitigate that by collapsing extra --no- options, the ones that are added by --git-completion-helper and not in original struct option arrays. The "--no-..." option will be displayed in this case to hint about more options, e.g. > ~/w/git $ git clone -- --bare --origin= --branch= --progress --checkout --quiet --config= --recurse-submodules --depth= --reference= --dissociate --reference-if-able= --filter= --separate-git-dir= --hardlinks --shallow-exclude= --ipv4 --shallow-since= --ipv6 --shallow-submodules --jobs= --shared --local --single-branch --mirror --tags --no-... --template= --no-checkout --upload-pack= --no-hardlinks --verbose --no-tags and when you complete it with --no-<tab>, all negative options will be presented: > ~/w/git $ git clone --no- --no-bare --no-quiet --no-branch --no-recurse-submodules --no-checkout --no-reference --no-config --no-reference-if-able --no-depth --no-separate-git-dir --no-dissociate --no-shallow-exclude --no-filter --no-shallow-since --no-hardlinks --no-shallow-submodules --no-ipv4 --no-shared --no-ipv6 --no-single-branch --no-jobs --no-tags --no-local --no-template --no-mirror --no-upload-pack --no-origin --no-verbose --no-progress Corner case: to make sure that people will never accidentally complete the fake option "--no-..." there must be one real --no- in the first complete listing even if it's not from the original struct option. PS. This could could be made simpler with ";&" to fall through from "--no-*" block and share the code but ";&" is not available on bash-3 (i.e. Mac) Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-06-06 17:41:39 +08:00
const struct option *original_opts = opts;
int nr_noopts = 0;
for (; opts->type != OPTION_END; opts++) {
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
const char *prefix = "--";
const char *suffix = "";
if (!opts->long_name)
continue;
if (!show_all &&
parse-options: don't complete option aliases by default Since 'OPT_ALIAS' was created in 5c387428f1 (parse-options: don't emit "ambiguous option" for aliases, 2019-04-29), 'git clone --git-completion-helper', which is used by the Bash completion script to list options accepted by clone (via '__gitcomp_builtin'), lists both '--recurse-submodules' and its alias '--recursive', which was not the case before since '--recursive' had the PARSE_OPT_HIDDEN flag set, and options with this flag are skipped by 'parse-options.c::show_gitcomp', which implements 'git <cmd> --git-completion-helper'. This means that typing 'git clone --recurs<TAB>' will yield both '--recurse-submodules' and '--recursive', which is not ideal since both do the same thing, and so the completion should directly complete the canonical option. At the point where 'show_gitcomp' is called in 'parse_options_step', 'preprocess_options' was already called in 'parse_options', so any aliases are now copies of the original options with a modified help text indicating they are aliases. Helpfully, since 64cc539fd2 (parse-options: don't leak alias help messages, 2021-03-21) these copies have the PARSE_OPT_FROM_ALIAS flag set, so check that flag early in 'show_gitcomp' and do not print them, unless the user explicitely requested that *all* completion be shown (by setting 'GIT_COMPLETION_SHOW_ALL'). After all, if we want to encourage the use of '--recurse-submodules' over '--recursive', we'd better just suggest the former. The only other options alias is 'log' and friends' '--mailmap', which is an alias for '--use-mailmap', but the Bash completion helpers for these commands do not use '__gitcomp_builtin', and thus are unnaffected by this change. Test the new behaviour in t9902-completion.sh. As a side effect, this also tests the correct behaviour of GIT_COMPLETION_SHOW_ALL, which was not tested before. Note that since '__gitcomp_builtin' caches the options it shows, we need to re-source the completion script to clear that cache for the second test. Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-07-16 09:55:43 +08:00
(opts->flags & (PARSE_OPT_HIDDEN | PARSE_OPT_NOCOMPLETE | PARSE_OPT_FROM_ALIAS)))
continue;
switch (opts->type) {
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
case OPTION_SUBCOMMAND:
prefix = "";
break;
case OPTION_GROUP:
continue;
case OPTION_STRING:
case OPTION_FILENAME:
case OPTION_INTEGER:
case OPTION_MAGNITUDE:
case OPTION_CALLBACK:
if (opts->flags & PARSE_OPT_NOARG)
break;
if (opts->flags & PARSE_OPT_OPTARG)
break;
if (opts->flags & PARSE_OPT_LASTARG_DEFAULT)
break;
suffix = "=";
break;
default:
break;
}
if (opts->flags & PARSE_OPT_COMP_ARG)
suffix = "=";
completion: collapse extra --no-.. options The commands that make use of --git-completion-helper feature could now produce a lot of --no-xxx options that a command can take. This in many case could nearly double the amount of completable options, using more screen estate and also harder to search for the wanted option. This patch attempts to mitigate that by collapsing extra --no- options, the ones that are added by --git-completion-helper and not in original struct option arrays. The "--no-..." option will be displayed in this case to hint about more options, e.g. > ~/w/git $ git clone -- --bare --origin= --branch= --progress --checkout --quiet --config= --recurse-submodules --depth= --reference= --dissociate --reference-if-able= --filter= --separate-git-dir= --hardlinks --shallow-exclude= --ipv4 --shallow-since= --ipv6 --shallow-submodules --jobs= --shared --local --single-branch --mirror --tags --no-... --template= --no-checkout --upload-pack= --no-hardlinks --verbose --no-tags and when you complete it with --no-<tab>, all negative options will be presented: > ~/w/git $ git clone --no- --no-bare --no-quiet --no-branch --no-recurse-submodules --no-checkout --no-reference --no-config --no-reference-if-able --no-depth --no-separate-git-dir --no-dissociate --no-shallow-exclude --no-filter --no-shallow-since --no-hardlinks --no-shallow-submodules --no-ipv4 --no-shared --no-ipv6 --no-single-branch --no-jobs --no-tags --no-local --no-template --no-mirror --no-upload-pack --no-origin --no-verbose --no-progress Corner case: to make sure that people will never accidentally complete the fake option "--no-..." there must be one real --no- in the first complete listing even if it's not from the original struct option. PS. This could could be made simpler with ";&" to fall through from "--no-*" block and share the code but ";&" is not available on bash-3 (i.e. Mac) Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-06-06 17:41:39 +08:00
if (starts_with(opts->long_name, "no-"))
nr_noopts++;
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
printf("%s%s%s%s", opts == original_opts ? "" : " ",
prefix, opts->long_name, suffix);
}
show_negated_gitcomp(original_opts, show_all, -1);
show_negated_gitcomp(original_opts, show_all, nr_noopts);
fputc('\n', stdout);
return PARSE_OPT_COMPLETE;
}
/*
* Scan and may produce a new option[] array, which should be used
* instead of the original 'options'.
*
* Right now this is only used to preprocess and substitute
* OPTION_ALIAS.
parse-options: don't leak alias help messages preprocess_options() allocates new strings for help messages for OPTION_ALIAS. Therefore we also need to clean those help messages up when freeing the returned options. First introduced in: 7c280589cf (parse-options: teach "git cmd -h" to show alias as alias, 2020-03-16) The preprocessed options themselves no longer contain any indication that a given option is/was an alias - therefore we add a new flag to indicate former aliases. (An alternative approach would be to look back at the original options to determine which options are aliases - but that seems like a fragile approach. Or we could even look at the alias_groups list - which might be less fragile, but would be slower as it requires nested looping.) As far as I can tell, parse_options() is only ever used once per command, and the help messages are small - hence this leak has very little impact. This leak was found while running t0001. LSAN output can be found below: Direct leak of 65 byte(s) in 1 object(s) allocated from: #0 0x49a859 in realloc /home/abuild/rpmbuild/BUILD/llvm-11.0.0.src/build/../projects/compiler-rt/lib/asan/asan_malloc_linux.cpp:164:3 #1 0x9aae36 in xrealloc /home/ahunt/oss-fuzz/git/wrapper.c:126:8 #2 0x939d8d in strbuf_grow /home/ahunt/oss-fuzz/git/strbuf.c:98:2 #3 0x93b936 in strbuf_vaddf /home/ahunt/oss-fuzz/git/strbuf.c:392:3 #4 0x93b7ff in strbuf_addf /home/ahunt/oss-fuzz/git/strbuf.c:333:2 #5 0x86747e in preprocess_options /home/ahunt/oss-fuzz/git/parse-options.c:666:3 #6 0x866ed2 in parse_options /home/ahunt/oss-fuzz/git/parse-options.c:847:17 #7 0x51c4a7 in cmd_clone /home/ahunt/oss-fuzz/git/builtin/clone.c:989:9 #8 0x4cd60d in run_builtin /home/ahunt/oss-fuzz/git/git.c:453:11 #9 0x4cb2da in handle_builtin /home/ahunt/oss-fuzz/git/git.c:704:3 #10 0x4ccc37 in run_argv /home/ahunt/oss-fuzz/git/git.c:771:4 #11 0x4cac29 in cmd_main /home/ahunt/oss-fuzz/git/git.c:902:19 #12 0x69c9fe in main /home/ahunt/oss-fuzz/git/common-main.c:52:11 #13 0x7fdac42d4349 in __libc_start_main (/lib64/libc.so.6+0x24349) Signed-off-by: Andrzej Hunt <ajrhunt@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-03-22 00:58:36 +08:00
*
* The returned options should be freed using free_preprocessed_options.
*/
static struct option *preprocess_options(struct parse_opt_ctx_t *ctx,
const struct option *options)
{
struct option *newopt;
int i, nr, alias;
int nr_aliases = 0;
for (nr = 0; options[nr].type != OPTION_END; nr++) {
if (options[nr].type == OPTION_ALIAS)
nr_aliases++;
}
if (!nr_aliases)
return NULL;
DUP_ARRAY(newopt, options, nr + 1);
/* each alias has two string pointers and NULL */
CALLOC_ARRAY(ctx->alias_groups, 3 * (nr_aliases + 1));
for (alias = 0, i = 0; i < nr; i++) {
int short_name;
const char *long_name;
const char *source;
struct strbuf help = STRBUF_INIT;
int j;
if (newopt[i].type != OPTION_ALIAS)
continue;
short_name = newopt[i].short_name;
long_name = newopt[i].long_name;
source = newopt[i].value;
if (!long_name)
BUG("An alias must have long option name");
strbuf_addf(&help, _("alias of --%s"), source);
for (j = 0; j < nr; j++) {
const char *name = options[j].long_name;
if (!name || strcmp(name, source))
continue;
if (options[j].type == OPTION_ALIAS)
BUG("No please. Nested aliases are not supported.");
memcpy(newopt + i, options + j, sizeof(*newopt));
newopt[i].short_name = short_name;
newopt[i].long_name = long_name;
newopt[i].help = strbuf_detach(&help, NULL);
parse-options: don't leak alias help messages preprocess_options() allocates new strings for help messages for OPTION_ALIAS. Therefore we also need to clean those help messages up when freeing the returned options. First introduced in: 7c280589cf (parse-options: teach "git cmd -h" to show alias as alias, 2020-03-16) The preprocessed options themselves no longer contain any indication that a given option is/was an alias - therefore we add a new flag to indicate former aliases. (An alternative approach would be to look back at the original options to determine which options are aliases - but that seems like a fragile approach. Or we could even look at the alias_groups list - which might be less fragile, but would be slower as it requires nested looping.) As far as I can tell, parse_options() is only ever used once per command, and the help messages are small - hence this leak has very little impact. This leak was found while running t0001. LSAN output can be found below: Direct leak of 65 byte(s) in 1 object(s) allocated from: #0 0x49a859 in realloc /home/abuild/rpmbuild/BUILD/llvm-11.0.0.src/build/../projects/compiler-rt/lib/asan/asan_malloc_linux.cpp:164:3 #1 0x9aae36 in xrealloc /home/ahunt/oss-fuzz/git/wrapper.c:126:8 #2 0x939d8d in strbuf_grow /home/ahunt/oss-fuzz/git/strbuf.c:98:2 #3 0x93b936 in strbuf_vaddf /home/ahunt/oss-fuzz/git/strbuf.c:392:3 #4 0x93b7ff in strbuf_addf /home/ahunt/oss-fuzz/git/strbuf.c:333:2 #5 0x86747e in preprocess_options /home/ahunt/oss-fuzz/git/parse-options.c:666:3 #6 0x866ed2 in parse_options /home/ahunt/oss-fuzz/git/parse-options.c:847:17 #7 0x51c4a7 in cmd_clone /home/ahunt/oss-fuzz/git/builtin/clone.c:989:9 #8 0x4cd60d in run_builtin /home/ahunt/oss-fuzz/git/git.c:453:11 #9 0x4cb2da in handle_builtin /home/ahunt/oss-fuzz/git/git.c:704:3 #10 0x4ccc37 in run_argv /home/ahunt/oss-fuzz/git/git.c:771:4 #11 0x4cac29 in cmd_main /home/ahunt/oss-fuzz/git/git.c:902:19 #12 0x69c9fe in main /home/ahunt/oss-fuzz/git/common-main.c:52:11 #13 0x7fdac42d4349 in __libc_start_main (/lib64/libc.so.6+0x24349) Signed-off-by: Andrzej Hunt <ajrhunt@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-03-22 00:58:36 +08:00
newopt[i].flags |= PARSE_OPT_FROM_ALIAS;
break;
}
if (j == nr)
BUG("could not find source option '%s' of alias '%s'",
source, newopt[i].long_name);
ctx->alias_groups[alias * 3 + 0] = newopt[i].long_name;
ctx->alias_groups[alias * 3 + 1] = options[j].long_name;
ctx->alias_groups[alias * 3 + 2] = NULL;
alias++;
}
return newopt;
}
parse-options: don't leak alias help messages preprocess_options() allocates new strings for help messages for OPTION_ALIAS. Therefore we also need to clean those help messages up when freeing the returned options. First introduced in: 7c280589cf (parse-options: teach "git cmd -h" to show alias as alias, 2020-03-16) The preprocessed options themselves no longer contain any indication that a given option is/was an alias - therefore we add a new flag to indicate former aliases. (An alternative approach would be to look back at the original options to determine which options are aliases - but that seems like a fragile approach. Or we could even look at the alias_groups list - which might be less fragile, but would be slower as it requires nested looping.) As far as I can tell, parse_options() is only ever used once per command, and the help messages are small - hence this leak has very little impact. This leak was found while running t0001. LSAN output can be found below: Direct leak of 65 byte(s) in 1 object(s) allocated from: #0 0x49a859 in realloc /home/abuild/rpmbuild/BUILD/llvm-11.0.0.src/build/../projects/compiler-rt/lib/asan/asan_malloc_linux.cpp:164:3 #1 0x9aae36 in xrealloc /home/ahunt/oss-fuzz/git/wrapper.c:126:8 #2 0x939d8d in strbuf_grow /home/ahunt/oss-fuzz/git/strbuf.c:98:2 #3 0x93b936 in strbuf_vaddf /home/ahunt/oss-fuzz/git/strbuf.c:392:3 #4 0x93b7ff in strbuf_addf /home/ahunt/oss-fuzz/git/strbuf.c:333:2 #5 0x86747e in preprocess_options /home/ahunt/oss-fuzz/git/parse-options.c:666:3 #6 0x866ed2 in parse_options /home/ahunt/oss-fuzz/git/parse-options.c:847:17 #7 0x51c4a7 in cmd_clone /home/ahunt/oss-fuzz/git/builtin/clone.c:989:9 #8 0x4cd60d in run_builtin /home/ahunt/oss-fuzz/git/git.c:453:11 #9 0x4cb2da in handle_builtin /home/ahunt/oss-fuzz/git/git.c:704:3 #10 0x4ccc37 in run_argv /home/ahunt/oss-fuzz/git/git.c:771:4 #11 0x4cac29 in cmd_main /home/ahunt/oss-fuzz/git/git.c:902:19 #12 0x69c9fe in main /home/ahunt/oss-fuzz/git/common-main.c:52:11 #13 0x7fdac42d4349 in __libc_start_main (/lib64/libc.so.6+0x24349) Signed-off-by: Andrzej Hunt <ajrhunt@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-03-22 00:58:36 +08:00
static void free_preprocessed_options(struct option *options)
{
int i;
if (!options)
return;
for (i = 0; options[i].type != OPTION_END; i++) {
if (options[i].flags & PARSE_OPT_FROM_ALIAS)
free((void *)options[i].help);
}
free(options);
}
static enum parse_opt_result usage_with_options_internal(struct parse_opt_ctx_t *,
const char * const *,
const struct option *,
int, int);
enum parse_opt_result parse_options_step(struct parse_opt_ctx_t *ctx,
const struct option *options,
const char * const usagestr[])
{
int internal_help = !(ctx->flags & PARSE_OPT_NO_INTERNAL_HELP);
/* we must reset ->opt, unknown short option leave it dangling */
ctx->opt = NULL;
for (; ctx->argc; ctx->argc--, ctx->argv++) {
const char *arg = ctx->argv[0];
if (ctx->flags & PARSE_OPT_ONE_SHOT &&
ctx->argc != ctx->total)
break;
if (*arg != '-' || !arg[1]) {
if (parse_nodash_opt(ctx, arg, options) == 0)
continue;
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
if (!ctx->has_subcommands) {
if (ctx->flags & PARSE_OPT_STOP_AT_NON_OPTION)
return PARSE_OPT_NON_OPTION;
ctx->out[ctx->cpidx++] = ctx->argv[0];
continue;
}
switch (parse_subcommand(arg, options)) {
case PARSE_OPT_SUBCOMMAND:
return PARSE_OPT_SUBCOMMAND;
case PARSE_OPT_UNKNOWN:
if (ctx->flags & PARSE_OPT_SUBCOMMAND_OPTIONAL)
/*
* arg is neither a short or long
* option nor a subcommand. Since
* this command has a default
* operation mode, we have to treat
* this arg and all remaining args
* as args meant to that default
* operation mode.
* So we are done parsing.
*/
return PARSE_OPT_DONE;
error(_("unknown subcommand: `%s'"), arg);
usage_with_options(usagestr, options);
case PARSE_OPT_COMPLETE:
case PARSE_OPT_HELP:
case PARSE_OPT_ERROR:
case PARSE_OPT_DONE:
case PARSE_OPT_NON_OPTION:
/* Impossible. */
BUG("parse_subcommand() cannot return these");
}
}
/* lone -h asks for help */
if (internal_help && ctx->total == 1 && !strcmp(arg + 1, "h"))
goto show_usage;
/*
* lone --git-completion-helper and --git-completion-helper-all
* are asked by git-completion.bash
*/
if (ctx->total == 1 && !strcmp(arg, "--git-completion-helper"))
return show_gitcomp(options, 0);
if (ctx->total == 1 && !strcmp(arg, "--git-completion-helper-all"))
return show_gitcomp(options, 1);
if (arg[1] != '-') {
ctx->opt = arg + 1;
switch (parse_short_opt(ctx, options)) {
case PARSE_OPT_ERROR:
return PARSE_OPT_ERROR;
case PARSE_OPT_UNKNOWN:
if (ctx->opt)
check_typos(arg + 1, options);
if (internal_help && *ctx->opt == 'h')
goto show_usage;
goto unknown;
case PARSE_OPT_NON_OPTION:
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
case PARSE_OPT_SUBCOMMAND:
case PARSE_OPT_HELP:
case PARSE_OPT_COMPLETE:
BUG("parse_short_opt() cannot return these");
case PARSE_OPT_DONE:
break;
}
if (ctx->opt)
check_typos(arg + 1, options);
while (ctx->opt) {
switch (parse_short_opt(ctx, options)) {
case PARSE_OPT_ERROR:
return PARSE_OPT_ERROR;
case PARSE_OPT_UNKNOWN:
if (internal_help && *ctx->opt == 'h')
goto show_usage;
/* fake a short option thing to hide the fact that we may have
* started to parse aggregated stuff
*
* This is leaky, too bad.
*/
ctx->argv[0] = xstrdup(ctx->opt - 1);
*(char *)ctx->argv[0] = '-';
goto unknown;
case PARSE_OPT_NON_OPTION:
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
case PARSE_OPT_SUBCOMMAND:
case PARSE_OPT_COMPLETE:
case PARSE_OPT_HELP:
BUG("parse_short_opt() cannot return these");
case PARSE_OPT_DONE:
break;
}
}
continue;
}
parse-options: decouple "--end-of-options" and "--" When we added generic end-of-options support in 51b4594b40 (parse-options: allow --end-of-options as a synonym for "--", 2019-08-06), we made them true synonyms. They both stop option parsing, and they are both returned in the resulting argv if the KEEP_DASHDASH flag is used. The hope was that this would work for all callers: - most generic callers would not pass KEEP_DASHDASH, and so would just do the right thing (stop parsing there) without needing to know anything more. - callers with KEEP_DASHDASH were generally going to rely on setup_revisions(), which knew to handle --end-of-options specially But that turned out miss quite a few cases that pass KEEP_DASHDASH but do their own manual parsing. For example, "git reset", "git checkout", and so on want pass KEEP_DASHDASH so they can support: git reset $revs -- $paths but of course aren't going to actually do a traversal, so they don't call setup_revisions(). And those cases currently get confused by --end-of-options being left in place, like: $ git reset --end-of-options HEAD fatal: option '--end-of-options' must come before non-option arguments We could teach each of these callers to handle the leftover option explicitly. But let's try to be a bit more clever and see if we can solve it centrally in parse-options.c. The bogus assumption here is that KEEP_DASHDASH tells us the caller wants to see --end-of-options in the result. But really, the callers which need to know that --end-of-options was reached are those that may potentially parse more options from argv. In other words, those that pass the KEEP_UNKNOWN_OPT flag. If such a caller is aware of --end-of-options (e.g., because they call setup_revisions() with the result), then this will continue to do the right thing, treating anything after --end-of-options as a non-option. And if the caller is not aware of --end-of-options, they are better off keeping it intact, because either: 1. They are just passing the options along to somebody else anyway, in which case that somebody would need to know about the --end-of-options marker. 2. They are going to parse the remainder themselves, at which point choking on --end-of-options is much better than having it silently removed. The point is to avoid option injection from untrusted command line arguments, and bailing is better than quietly treating the untrusted argument as an option. This fixes bugs with --end-of-options across several commands, but I've focused on two in particular here: - t7102 confirms that "git reset --end-of-options --foo" now works. This checks two things. One, that we no longer barf on "--end-of-options" itself (which previously we did, even if the rev was something vanilla like "HEAD" instead of "--foo"). And two, that we correctly treat "--foo" as a revision rather than an option. This fix applies to any other cases which pass KEEP_DASHDASH but not KEEP_UNKNOWN_OPT, like "git checkout", "git check-attr", "git grep", etc, which would previously choke on "--end-of-options". - t9350 shows the opposite case: fast-export passed KEEP_UNKNOWN_OPT but not KEEP_DASHDASH, but then passed the result on to setup_revisions(). So it never saw --end-of-options, and would erroneously parse "fast-export --end-of-options --foo" as having a "--foo" option. This is now fixed. Note that this does shut the door for callers which want to know if we hit end-of-options, but don't otherwise need to keep unknown opts. The obvious thing here is feeding it to the DWIM verify_filename() machinery. And indeed, this is a problem even for commands which do understand --end-of-options already. For example, without this patch, you get: $ git log --end-of-options --foo fatal: option '--foo' must come before non-option arguments because we refuse to accept "--foo" as a filename (because it starts with a dash) even though we could know that we saw end-of-options. The verify_filename() function simply doesn't accept this extra information. So that is the status quo, and this patch doubles down further on that. Commands like "git reset" have the same problem, but they won't even know that parse-options saw --end-of-options! So even if we fixed verify_filename(), they wouldn't have anything to pass to it. But in practice I don't think this is a big deal. If you are being careful enough to use --end-of-options, then you should also be using "--" to disambiguate and avoid the DWIM behavior in the first place. In other words, doing: git log --end-of-options --this-is-a-rev -- --this-is-a-path works correctly, and will continue to do so. And likewise, with this patch now: git reset --end-of-options --this-is-a-rev -- --this-is-a-path will work, as well. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-12-07 06:21:45 +08:00
if (!arg[2] /* "--" */) {
if (!(ctx->flags & PARSE_OPT_KEEP_DASHDASH)) {
ctx->argc--;
ctx->argv++;
}
break;
parse-options: decouple "--end-of-options" and "--" When we added generic end-of-options support in 51b4594b40 (parse-options: allow --end-of-options as a synonym for "--", 2019-08-06), we made them true synonyms. They both stop option parsing, and they are both returned in the resulting argv if the KEEP_DASHDASH flag is used. The hope was that this would work for all callers: - most generic callers would not pass KEEP_DASHDASH, and so would just do the right thing (stop parsing there) without needing to know anything more. - callers with KEEP_DASHDASH were generally going to rely on setup_revisions(), which knew to handle --end-of-options specially But that turned out miss quite a few cases that pass KEEP_DASHDASH but do their own manual parsing. For example, "git reset", "git checkout", and so on want pass KEEP_DASHDASH so they can support: git reset $revs -- $paths but of course aren't going to actually do a traversal, so they don't call setup_revisions(). And those cases currently get confused by --end-of-options being left in place, like: $ git reset --end-of-options HEAD fatal: option '--end-of-options' must come before non-option arguments We could teach each of these callers to handle the leftover option explicitly. But let's try to be a bit more clever and see if we can solve it centrally in parse-options.c. The bogus assumption here is that KEEP_DASHDASH tells us the caller wants to see --end-of-options in the result. But really, the callers which need to know that --end-of-options was reached are those that may potentially parse more options from argv. In other words, those that pass the KEEP_UNKNOWN_OPT flag. If such a caller is aware of --end-of-options (e.g., because they call setup_revisions() with the result), then this will continue to do the right thing, treating anything after --end-of-options as a non-option. And if the caller is not aware of --end-of-options, they are better off keeping it intact, because either: 1. They are just passing the options along to somebody else anyway, in which case that somebody would need to know about the --end-of-options marker. 2. They are going to parse the remainder themselves, at which point choking on --end-of-options is much better than having it silently removed. The point is to avoid option injection from untrusted command line arguments, and bailing is better than quietly treating the untrusted argument as an option. This fixes bugs with --end-of-options across several commands, but I've focused on two in particular here: - t7102 confirms that "git reset --end-of-options --foo" now works. This checks two things. One, that we no longer barf on "--end-of-options" itself (which previously we did, even if the rev was something vanilla like "HEAD" instead of "--foo"). And two, that we correctly treat "--foo" as a revision rather than an option. This fix applies to any other cases which pass KEEP_DASHDASH but not KEEP_UNKNOWN_OPT, like "git checkout", "git check-attr", "git grep", etc, which would previously choke on "--end-of-options". - t9350 shows the opposite case: fast-export passed KEEP_UNKNOWN_OPT but not KEEP_DASHDASH, but then passed the result on to setup_revisions(). So it never saw --end-of-options, and would erroneously parse "fast-export --end-of-options --foo" as having a "--foo" option. This is now fixed. Note that this does shut the door for callers which want to know if we hit end-of-options, but don't otherwise need to keep unknown opts. The obvious thing here is feeding it to the DWIM verify_filename() machinery. And indeed, this is a problem even for commands which do understand --end-of-options already. For example, without this patch, you get: $ git log --end-of-options --foo fatal: option '--foo' must come before non-option arguments because we refuse to accept "--foo" as a filename (because it starts with a dash) even though we could know that we saw end-of-options. The verify_filename() function simply doesn't accept this extra information. So that is the status quo, and this patch doubles down further on that. Commands like "git reset" have the same problem, but they won't even know that parse-options saw --end-of-options! So even if we fixed verify_filename(), they wouldn't have anything to pass to it. But in practice I don't think this is a big deal. If you are being careful enough to use --end-of-options, then you should also be using "--" to disambiguate and avoid the DWIM behavior in the first place. In other words, doing: git log --end-of-options --this-is-a-rev -- --this-is-a-path works correctly, and will continue to do so. And likewise, with this patch now: git reset --end-of-options --this-is-a-rev -- --this-is-a-path will work, as well. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-12-07 06:21:45 +08:00
} else if (!strcmp(arg + 2, "end-of-options")) {
if (!(ctx->flags & PARSE_OPT_KEEP_UNKNOWN_OPT)) {
ctx->argc--;
ctx->argv++;
}
break;
}
if (internal_help && !strcmp(arg + 2, "help-all"))
return usage_with_options_internal(ctx, usagestr, options, 1, 0);
if (internal_help && !strcmp(arg + 2, "help"))
goto show_usage;
switch (parse_long_opt(ctx, arg + 2, options)) {
case PARSE_OPT_ERROR:
return PARSE_OPT_ERROR;
case PARSE_OPT_UNKNOWN:
goto unknown;
case PARSE_OPT_HELP:
goto show_usage;
case PARSE_OPT_NON_OPTION:
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
case PARSE_OPT_SUBCOMMAND:
case PARSE_OPT_COMPLETE:
BUG("parse_long_opt() cannot return these");
case PARSE_OPT_DONE:
break;
}
continue;
unknown:
if (ctx->flags & PARSE_OPT_ONE_SHOT)
break;
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
if (ctx->has_subcommands &&
(ctx->flags & PARSE_OPT_SUBCOMMAND_OPTIONAL) &&
(ctx->flags & PARSE_OPT_KEEP_UNKNOWN_OPT)) {
/*
* Found an unknown option given to a command with
* subcommands that has a default operation mode:
* we treat this option and all remaining args as
* arguments meant to that default operation mode.
* So we are done parsing.
*/
return PARSE_OPT_DONE;
}
if (!(ctx->flags & PARSE_OPT_KEEP_UNKNOWN_OPT))
return PARSE_OPT_UNKNOWN;
ctx->out[ctx->cpidx++] = ctx->argv[0];
ctx->opt = NULL;
}
return PARSE_OPT_DONE;
show_usage:
return usage_with_options_internal(ctx, usagestr, options, 0, 0);
}
int parse_options_end(struct parse_opt_ctx_t *ctx)
{
if (ctx->flags & PARSE_OPT_ONE_SHOT)
return ctx->total - ctx->argc;
MOVE_ARRAY(ctx->out + ctx->cpidx, ctx->argv, ctx->argc);
ctx->out[ctx->cpidx + ctx->argc] = NULL;
return ctx->cpidx + ctx->argc;
}
int parse_options(int argc, const char **argv,
const char *prefix,
const struct option *options,
const char * const usagestr[],
enum parse_opt_flags flags)
{
struct parse_opt_ctx_t ctx;
struct option *real_options;
disallow_abbreviated_options =
git_env_bool("GIT_TEST_DISALLOW_ABBREVIATED_OPTIONS", 0);
memset(&ctx, 0, sizeof(ctx));
real_options = preprocess_options(&ctx, options);
if (real_options)
options = real_options;
parse_options_start_1(&ctx, argc, argv, prefix, options, flags);
switch (parse_options_step(&ctx, options, usagestr)) {
case PARSE_OPT_HELP:
case PARSE_OPT_ERROR:
exit(129);
case PARSE_OPT_COMPLETE:
exit(0);
case PARSE_OPT_NON_OPTION:
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
case PARSE_OPT_SUBCOMMAND:
break;
case PARSE_OPT_DONE:
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
if (ctx.has_subcommands &&
!(flags & PARSE_OPT_SUBCOMMAND_OPTIONAL)) {
error(_("need a subcommand"));
usage_with_options(usagestr, options);
}
break;
case PARSE_OPT_UNKNOWN:
if (ctx.argv[0][1] == '-') {
error(_("unknown option `%s'"), ctx.argv[0] + 2);
} else if (isascii(*ctx.opt)) {
error(_("unknown switch `%c'"), *ctx.opt);
} else {
error(_("unknown non-ascii option in string: `%s'"),
ctx.argv[0]);
}
usage_with_options(usagestr, options);
}
MacOS: precompose_argv_prefix() The following sequence leads to a "BUG" assertion running under MacOS: DIR=git-test-restore-p Adiarnfd=$(printf 'A\314\210') DIRNAME=xx${Adiarnfd}yy mkdir $DIR && cd $DIR && git init && mkdir $DIRNAME && cd $DIRNAME && echo "Initial" >file && git add file && echo "One more line" >>file && echo y | git restore -p . Initialized empty Git repository in /tmp/git-test-restore-p/.git/ BUG: pathspec.c:495: error initializing pathspec_item Cannot close git diff-index --cached --numstat [snip] The command `git restore` is run from a directory inside a Git repo. Git needs to split the $CWD into 2 parts: The path to the repo and "the rest", if any. "The rest" becomes a "prefix" later used inside the pathspec code. As an example, "/path/to/repo/dir-inside-repå" would determine "/path/to/repo" as the root of the repo, the place where the configuration file .git/config is found. The rest becomes the prefix ("dir-inside-repå"), from where the pathspec machinery expands the ".", more about this later. If there is a decomposed form, (making the decomposing visible like this), "dir-inside-rep°a" doesn't match "dir-inside-repå". Git commands need to: (a) read the configuration variable "core.precomposeunicode" (b) precocompose argv[] (c) precompose the prefix, if there was any The first commit, 76759c7dff53 "git on Mac OS and precomposed unicode" addressed (a) and (b). The call to precompose_argv() was added into parse-options.c, because that seemed to be a good place when the patch was written. Commands that don't use parse-options need to do (a) and (b) themselfs. The commands `diff-files`, `diff-index`, `diff-tree` and `diff` learned (a) and (b) in commit 90a78b83e0b8 "diff: run arguments through precompose_argv" Branch names (or refs in general) using decomposed code points resulting in decomposed file names had been fixed in commit 8e712ef6fc97 "Honor core.precomposeUnicode in more places" The bug report from above shows 2 things: - more commands need to handle precomposed unicode - (c) should be implemented for all commands using pathspecs Solution: precompose_argv() now handles the prefix (if needed), and is renamed into precompose_argv_prefix(). Inside this function the config variable core.precomposeunicode is read into the global variable precomposed_unicode, as before. This reading is skipped if precomposed_unicode had been read before. The original patch for preocomposed unicode, 76759c7dff53, placed precompose_argv() into parse-options.c Now add it into git.c::run_builtin() as well. Existing precompose calls in diff-files.c and others may become redundant, and if we audit the callflows that reach these places to make sure that they can never be reached without going through the new call added to run_builtin(), we might be able to remove these existing ones. But in this commit, we do not bother to do so and leave these precompose callsites as they are. Because precompose() is idempotent and can be called on an already precomposed string safely, this is safer than removing existing calls without fully vetting the callflows. There is certainly room for cleanups - this change intends to be a bug fix. Cleanups needs more tests in e.g. t/t3910-mac-os-precompose.sh, and should be done in future commits. [1] git-bugreport-2021-01-06-1209.txt (git can't deal with special characters) [2] https://lore.kernel.org/git/A102844A-9501-4A86-854D-E3B387D378AA@icloud.com/ Reported-by: Daniel Troger <random_n0body@icloud.com> Helped-By: Philippe Blain <levraiphilippeblain@gmail.com> Signed-off-by: Torsten Bögershausen <tboegi@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-02-04 00:28:23 +08:00
precompose_argv_prefix(argc, argv, NULL);
parse-options: don't leak alias help messages preprocess_options() allocates new strings for help messages for OPTION_ALIAS. Therefore we also need to clean those help messages up when freeing the returned options. First introduced in: 7c280589cf (parse-options: teach "git cmd -h" to show alias as alias, 2020-03-16) The preprocessed options themselves no longer contain any indication that a given option is/was an alias - therefore we add a new flag to indicate former aliases. (An alternative approach would be to look back at the original options to determine which options are aliases - but that seems like a fragile approach. Or we could even look at the alias_groups list - which might be less fragile, but would be slower as it requires nested looping.) As far as I can tell, parse_options() is only ever used once per command, and the help messages are small - hence this leak has very little impact. This leak was found while running t0001. LSAN output can be found below: Direct leak of 65 byte(s) in 1 object(s) allocated from: #0 0x49a859 in realloc /home/abuild/rpmbuild/BUILD/llvm-11.0.0.src/build/../projects/compiler-rt/lib/asan/asan_malloc_linux.cpp:164:3 #1 0x9aae36 in xrealloc /home/ahunt/oss-fuzz/git/wrapper.c:126:8 #2 0x939d8d in strbuf_grow /home/ahunt/oss-fuzz/git/strbuf.c:98:2 #3 0x93b936 in strbuf_vaddf /home/ahunt/oss-fuzz/git/strbuf.c:392:3 #4 0x93b7ff in strbuf_addf /home/ahunt/oss-fuzz/git/strbuf.c:333:2 #5 0x86747e in preprocess_options /home/ahunt/oss-fuzz/git/parse-options.c:666:3 #6 0x866ed2 in parse_options /home/ahunt/oss-fuzz/git/parse-options.c:847:17 #7 0x51c4a7 in cmd_clone /home/ahunt/oss-fuzz/git/builtin/clone.c:989:9 #8 0x4cd60d in run_builtin /home/ahunt/oss-fuzz/git/git.c:453:11 #9 0x4cb2da in handle_builtin /home/ahunt/oss-fuzz/git/git.c:704:3 #10 0x4ccc37 in run_argv /home/ahunt/oss-fuzz/git/git.c:771:4 #11 0x4cac29 in cmd_main /home/ahunt/oss-fuzz/git/git.c:902:19 #12 0x69c9fe in main /home/ahunt/oss-fuzz/git/common-main.c:52:11 #13 0x7fdac42d4349 in __libc_start_main (/lib64/libc.so.6+0x24349) Signed-off-by: Andrzej Hunt <ajrhunt@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-03-22 00:58:36 +08:00
free_preprocessed_options(real_options);
free(ctx.alias_groups);
parse-options: make CMDMODE errors more precise Only a single PARSE_OPT_CMDMODE option can be specified for the same variable at the same time. This is enforced by get_value(), but the error messages are imprecise in three ways: 1. If a non-PARSE_OPT_CMDMODE option changes the value variable of a PARSE_OPT_CMDMODE option then an ominously vague message is shown: $ t/helper/test-tool parse-options --set23 --mode1 error: option `mode1' : incompatible with something else Worse: If the order of options is reversed then no error is reported at all: $ t/helper/test-tool parse-options --mode1 --set23 boolean: 0 integer: 23 magnitude: 0 timestamp: 0 string: (not set) abbrev: 7 verbose: -1 quiet: 0 dry run: no file: (not set) Fortunately this can currently only happen in the test helper; actual Git commands don't share the same variable for the value of options with and without the flag PARSE_OPT_CMDMODE. 2. If there are multiple options with the same value (synonyms), then the one that is defined first is shown rather than the one actually given on the command line, which is confusing: $ git am --resolved --quit error: option `quit' is incompatible with --continue 3. Arguments of PARSE_OPT_CMDMODE options are not handled by the parse-option machinery. This is left to the callback function. We currently only have a single affected option, --show-current-patch of git am. Errors for it can show an argument that was not actually given on the command line: $ git am --show-current-patch --show-current-patch=diff error: options '--show-current-patch=diff' and '--show-current-patch=raw' cannot be used together The options --show-current-patch and --show-current-patch=raw are synonyms, but the error accuses the user of input they did not actually made. Or it can awkwardly print a NULL pointer: $ git am --show-current-patch=diff --show-current-patch error: options '--show-current-patch=(null)' and '--show-current-patch=diff' cannot be used together The reasons for these shortcomings is that the current code checks incompatibility only when encountering a PARSE_OPT_CMDMODE option at the command line, and that it searches the previous incompatible option by value. Fix the first two points by checking all PARSE_OPT_CMDMODE variables after parsing each option and by storing all relevant details if their value changed. Do that whether or not the changing options has the flag PARSE_OPT_CMDMODE set. Report an incompatibility only if two options change the variable to different values and at least one of them is a PARSE_OPT_CMDMODE option. This changes the output of the first three examples above to: $ t/helper/test-tool parse-options --set23 --mode1 error: --mode1 is incompatible with --set23 $ t/helper/test-tool parse-options --mode1 --set23 error: --set23 is incompatible with --mode1 $ git am --resolved --quit error: --quit is incompatible with --resolved Store the argument of PARSE_OPT_CMDMODE options of type OPTION_CALLBACK as well to allow taking over the responsibility for compatibility checking from the callback function. The next patch will use this capability to fix the messages for git am --show-current-patch. Use a linked list for storing the PARSE_OPT_CMDMODE variables. This somewhat outdated data structure is simple and suffices, as the number of elements per command is currently only zero or one. We do support multiple different command modes variables per command, but I don't expect that we'd ever use a significant number of them. Once we do we can switch to a hashmap. Since we no longer need to search the conflicting option, the all_opts parameter of get_value() is no longer used. Remove it. Extend the tests to check for both conflicting option names, but don't insist on a particular order. Signed-off-by: René Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-10-28 19:53:01 +08:00
for (struct parse_opt_cmdmode_list *elem = ctx.cmdmode_list; elem;) {
struct parse_opt_cmdmode_list *next = elem->next;
free(elem);
elem = next;
}
return parse_options_end(&ctx);
}
static int usage_argh(const struct option *opts, FILE *outfile)
{
const char *s;
int literal = (opts->flags & PARSE_OPT_LITERAL_ARGHELP) ||
!opts->argh || !!strpbrk(opts->argh, "()<>[]|");
if (opts->flags & PARSE_OPT_OPTARG)
if (opts->long_name)
s = literal ? "[=%s]" : "[=<%s>]";
else
s = literal ? "[%s]" : "[<%s>]";
else
s = literal ? " %s" : " <%s>";
return utf8_fprintf(outfile, s, opts->argh ? _(opts->argh) : _("..."));
}
static int usage_indent(FILE *outfile)
{
return fprintf(outfile, " ");
}
#define USAGE_OPTS_WIDTH 26
static void usage_padding(FILE *outfile, size_t pos)
{
if (pos < USAGE_OPTS_WIDTH)
fprintf(outfile, "%*s", USAGE_OPTS_WIDTH - (int)pos, "");
else
fprintf(outfile, "\n%*s", USAGE_OPTS_WIDTH, "");
}
static const struct option *find_option_by_long_name(const struct option *opts,
const char *long_name)
{
for (; opts->type != OPTION_END; opts++) {
if (opts->long_name && !strcmp(opts->long_name, long_name))
return opts;
}
return NULL;
}
static enum parse_opt_result usage_with_options_internal(struct parse_opt_ctx_t *ctx,
const char * const *usagestr,
const struct option *opts,
int full, int err)
{
const struct option *all_opts = opts;
FILE *outfile = err ? stderr : stdout;
int need_newline;
parse-options: properly align continued usage output Some commands such as "git stash" emit continued options output with e.g. "git stash -h", because usage_with_options_internal() prefixes with its own whitespace the resulting output wasn't properly aligned. Let's account for the added whitespace, which properly aligns the output. The "git stash" command has usage output with a N_() translation that legitimately stretches across multiple lines; N_("git stash [push [-p|--patch] [-k|--[no-]keep-index] [-q|--quiet]\n" " [-u|--include-untracked] [-a|--all] [-m|--message <message>]\n" [...] We'd like to have that output aligned with the length of the initial "git stash " output, but since usage_with_options_internal() adds its own whitespace prefixing we fell short, before this change we'd emit: $ git stash -h usage: git stash list [<options>] or: git stash show [<options>] [<stash>] [...] or: git stash [push [-p|--patch] [-k|--[no-]keep-index] [-q|--quiet] [-u|--include-untracked] [-a|--all] [-m|--message <message>] [...] Now we'll properly emit aligned output. I.e. the last four lines above will instead be (a whitespace-only change to the above): [...] or: git stash [push [-p|--patch] [-k|--[no-]keep-index] [-q|--quiet] [-u|--include-untracked] [-a|--all] [-m|--message <message>] [...] We could also go for an approach where we have the caller support no padding of their own, i.e. (same as the first example, except for the padding on the second line): N_("git stash [push [-p|--patch] [-k|--[no-]keep-index] [-q|--quiet]\n" "[-u|--include-untracked] [-a|--all] [-m|--message <message>]\n" [...] But to do that we'll need to find the length of "git stash". We can discover that from the "cmd" in the "struct cmd_struct", but there might be cases with sub-commands or "git" itself taking arguments that would make that non-trivial. Even if it were I still think this approach is better, because this way we'll get the same legible alignment in the C code. The fact that usage_with_options_internal() is adding its own prefix padding is an implementation detail that callers shouldn't need to worry about. Implementation notes: We could skip the string_list_split() with a strchr(str, '\n') check, but we'd then need to duplicate our state machine for strings that do and don't contain a "\n". It's simpler to just always split into a "struct string_list", even though the common case is that that "struct string_list" will contain only one element. This is not performance-sensitive code. This change is relatively more complex since I've accounted for making it future-proof for RTL translation support. Later in usage_with_options_internal() we have some existing padding code dating back to d7a38c54a6c (parse-options: be able to generate usages automatically, 2007-10-15) which isn't RTL-safe, but that code would be easy to fix. Let's not introduce new RTL translation problems here. Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-09-21 21:30:11 +08:00
const char *usage_prefix = _("usage: %s");
/*
* The translation could be anything, but we can count on
* msgfmt(1)'s --check option to have asserted that "%s" is in
* the translation. So compute the length of the "usage: "
* part. We are assuming that the translator wasn't overly
* clever and used e.g. "%1$s" instead of "%s", there's only
* one "%s" in "usage_prefix" above, so there's no reason to
* do so even with a RTL language.
*/
size_t usage_len = strlen(usage_prefix) - strlen("%s");
/*
* TRANSLATORS: the colon here should align with the
* one in "usage: %s" translation.
*/
const char *or_prefix = _(" or: %s");
/*
* TRANSLATORS: You should only need to translate this format
* string if your language is a RTL language (e.g. Arabic,
* Hebrew etc.), not if it's a LTR language (e.g. German,
* Russian, Chinese etc.).
*
* When a translated usage string has an embedded "\n" it's
* because options have wrapped to the next line. The line
* after the "\n" will then be padded to align with the
* command name, such as N_("git cmd [opt]\n<8
* spaces>[opt2]"), where the 8 spaces are the same length as
* "git cmd ".
*
* This format string prints out that already-translated
* line. The "%*s" is whitespace padding to account for the
* padding at the start of the line that we add in this
* function. The "%s" is a line in the (hopefully already
* translated) N_() usage string, which contained embedded
* newlines before we split it up.
*/
const char *usage_continued = _("%*s%s");
const char *prefix = usage_prefix;
int saw_empty_line = 0;
if (!usagestr)
return PARSE_OPT_HELP;
if (!err && ctx && ctx->flags & PARSE_OPT_SHELL_EVAL)
fprintf(outfile, "cat <<\\EOF\n");
while (*usagestr) {
parse-options: properly align continued usage output Some commands such as "git stash" emit continued options output with e.g. "git stash -h", because usage_with_options_internal() prefixes with its own whitespace the resulting output wasn't properly aligned. Let's account for the added whitespace, which properly aligns the output. The "git stash" command has usage output with a N_() translation that legitimately stretches across multiple lines; N_("git stash [push [-p|--patch] [-k|--[no-]keep-index] [-q|--quiet]\n" " [-u|--include-untracked] [-a|--all] [-m|--message <message>]\n" [...] We'd like to have that output aligned with the length of the initial "git stash " output, but since usage_with_options_internal() adds its own whitespace prefixing we fell short, before this change we'd emit: $ git stash -h usage: git stash list [<options>] or: git stash show [<options>] [<stash>] [...] or: git stash [push [-p|--patch] [-k|--[no-]keep-index] [-q|--quiet] [-u|--include-untracked] [-a|--all] [-m|--message <message>] [...] Now we'll properly emit aligned output. I.e. the last four lines above will instead be (a whitespace-only change to the above): [...] or: git stash [push [-p|--patch] [-k|--[no-]keep-index] [-q|--quiet] [-u|--include-untracked] [-a|--all] [-m|--message <message>] [...] We could also go for an approach where we have the caller support no padding of their own, i.e. (same as the first example, except for the padding on the second line): N_("git stash [push [-p|--patch] [-k|--[no-]keep-index] [-q|--quiet]\n" "[-u|--include-untracked] [-a|--all] [-m|--message <message>]\n" [...] But to do that we'll need to find the length of "git stash". We can discover that from the "cmd" in the "struct cmd_struct", but there might be cases with sub-commands or "git" itself taking arguments that would make that non-trivial. Even if it were I still think this approach is better, because this way we'll get the same legible alignment in the C code. The fact that usage_with_options_internal() is adding its own prefix padding is an implementation detail that callers shouldn't need to worry about. Implementation notes: We could skip the string_list_split() with a strchr(str, '\n') check, but we'd then need to duplicate our state machine for strings that do and don't contain a "\n". It's simpler to just always split into a "struct string_list", even though the common case is that that "struct string_list" will contain only one element. This is not performance-sensitive code. This change is relatively more complex since I've accounted for making it future-proof for RTL translation support. Later in usage_with_options_internal() we have some existing padding code dating back to d7a38c54a6c (parse-options: be able to generate usages automatically, 2007-10-15) which isn't RTL-safe, but that code would be easy to fix. Let's not introduce new RTL translation problems here. Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-09-21 21:30:11 +08:00
const char *str = _(*usagestr++);
struct string_list list = STRING_LIST_INIT_DUP;
unsigned int j;
if (!saw_empty_line && !*str)
saw_empty_line = 1;
string_list_split(&list, str, '\n', -1);
for (j = 0; j < list.nr; j++) {
const char *line = list.items[j].string;
if (saw_empty_line && *line)
fprintf_ln(outfile, _(" %s"), line);
else if (saw_empty_line)
fputc('\n', outfile);
else if (!j)
fprintf_ln(outfile, prefix, line);
else
fprintf_ln(outfile, usage_continued,
(int)usage_len, "", line);
}
string_list_clear(&list, 0);
prefix = or_prefix;
}
need_newline = 1;
for (; opts->type != OPTION_END; opts++) {
size_t pos;
const char *cp, *np;
const char *positive_name = NULL;
parse-options: add support for parsing subcommands Several Git commands have subcommands to implement mutually exclusive "operation modes", and they usually parse their subcommand argument with a bunch of if-else if statements. Teach parse-options to handle subcommands as well, which will result in shorter and simpler code with consistent error handling and error messages on unknown or missing subcommand, and it will also make possible for our Bash completion script to handle subcommands programmatically. The approach is guided by the following observations: - Most subcommands [1] are implemented in dedicated functions, and most of those functions [2] either have a signature matching the 'int cmd_foo(int argc, const char **argc, const char *prefix)' signature of builtin commands or can be trivially converted to that signature, because they miss only that last prefix parameter or have no parameters at all. - Subcommand arguments only have long form, and they have no double dash prefix, no negated form, and no description, and they don't take any arguments, and can't be abbreviated. - There must be exactly one subcommand among the arguments, or zero if the command has a default operation mode. - All arguments following the subcommand are considered to be arguments of the subcommand, and, conversely, arguments meant for the subcommand may not preceed the subcommand. So in the end subcommand declaration and parsing would look something like this: parse_opt_subcommand_fn *fn = NULL; struct option builtin_commit_graph_options[] = { OPT_STRING(0, "object-dir", &opts.obj_dir, N_("dir"), N_("the object directory to store the graph")), OPT_SUBCOMMAND("verify", &fn, graph_verify), OPT_SUBCOMMAND("write", &fn, graph_write), OPT_END(), }; argc = parse_options(argc, argv, prefix, options, builtin_commit_graph_usage, 0); return fn(argc, argv, prefix); Here each OPT_SUBCOMMAND specifies the name of the subcommand and the function implementing it, and the address of the same 'fn' subcommand function pointer. parse_options() then processes the arguments until it finds the first argument matching one of the subcommands, sets 'fn' to the function associated with that subcommand, and returns, leaving the rest of the arguments unprocessed. If none of the listed subcommands is found among the arguments, parse_options() will show usage and abort. If a command has a default operation mode, 'fn' should be initialized to the function implementing that mode, and parse_options() should be invoked with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag. In this case parse_options() won't error out when not finding any subcommands, but will return leaving 'fn' unchanged. Note that if that default operation mode has any --options, then the PARSE_OPT_KEEP_UNKNOWN_OPT flag is necessary as well (otherwise parse_options() would error out upon seeing the unknown option meant to the default operation mode). Some thoughts about the implementation: - The same pointer to 'fn' must be specified as 'value' for each OPT_SUBCOMMAND, because there can be only one set of mutually exclusive subcommands; parse_options() will BUG() otherwise. There are other ways to tell parse_options() where to put the function associated with the subcommand given on the command line, but I didn't like them: - Change parse_options()'s signature by adding a pointer to subcommand function to be set to the function associated with the given subcommand, affecting all callsites, even those that don't have subcommands. - Introduce a specific parse_options_and_subcommand() variant with that extra funcion parameter. - I decided against automatically calling the subcommand function from within parse_options(), because: - There are commands that have to perform additional actions after option parsing but before calling the function implementing the specified subcommand. - The return code of the subcommand is usually the return code of the git command, but preserving the return code of the automatically called subcommand function would have made the API awkward. - Also add a OPT_SUBCOMMAND_F() variant to allow specifying an option flag: we have two subcommands that are purposefully excluded from completion ('git remote rm' and 'git stash save'), so they'll have to be specified with the PARSE_OPT_NOCOMPLETE flag. - Some of the 'parse_opt_flags' don't make sense with subcommands, and using them is probably just an oversight or misunderstanding. Therefore parse_options() will BUG() when invoked with any of the following flags while the options array contains at least one OPT_SUBCOMMAND: - PARSE_OPT_KEEP_DASHDASH: parse_options() stops parsing arguments when encountering a "--" argument, so it doesn't make sense to expect and keep one before a subcommand, because it would prevent the parsing of the subcommand. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, because the double dash might be meaningful for the command's default operation mode, e.g. to disambiguate refs and pathspecs. - PARSE_OPT_STOP_AT_NON_OPTION: As its name suggests, this flag tells parse_options() to stop as soon as it encouners a non-option argument, but subcommands are by definition not options... so how could they be parsed, then?! - PARSE_OPT_KEEP_UNKNOWN: This flag can be used to collect any unknown --options and then pass them to a different command or subsystem. Surely if a command has subcommands, then this functionality should rather be delegated to one of those subcommands, and not performed by the command itself. However, this flag is allowed in combination with the PARSE_OPT_SUBCOMMAND_OPTIONAL flag, making possible to pass --options to the default operation mode. - If the command with subcommands has a default operation mode, then all arguments to the command must preceed the arguments of the subcommand. AFAICT we don't have any commands where this makes a difference, because in those commands either only the command accepts any arguments ('notes' and 'remote'), or only the default subcommand ('reflog' and 'stash'), but never both. - The 'argv' array passed to subcommand functions currently starts with the name of the subcommand. Keep this behavior. AFAICT no subcommand functions depend on the actual content of 'argv[0]', but the parse_options() call handling their options expects that the options start at argv[1]. - To support handling subcommands programmatically in our Bash completion script, 'git cmd --git-completion-helper' will now list both subcommands and regular --options, if any. This means that the completion script will have to separate subcommands (i.e. words without a double dash prefix) from --options on its own, but that's rather easy to do, and it's not much work either, because the number of subcommands a command might have is rather low, and those commands accept only a single --option or none at all. An alternative would be to introduce a separate option that lists only subcommands, but then the completion script would need not one but two git invocations and command substitutions for commands with subcommands. Note that this change doesn't affect the behavior of our Bash completion script, because when completing the --option of a command with subcommands, e.g. for 'git notes --<TAB>', then all subcommands will be filtered out anyway, as none of them will match the word to be completed starting with that double dash prefix. [1] Except 'git rerere', because many of its subcommands are implemented in the bodies of the if-else if statements parsing the command's subcommand argument. [2] Except 'credential', 'credential-store' and 'fsmonitor--daemon', because some of the functions implementing their subcommands take special parameters. Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-08-20 00:04:00 +08:00
if (opts->type == OPTION_SUBCOMMAND)
continue;
if (opts->type == OPTION_GROUP) {
fputc('\n', outfile);
need_newline = 0;
if (*opts->help)
fprintf(outfile, "%s\n", _(opts->help));
continue;
}
if (!full && (opts->flags & PARSE_OPT_HIDDEN))
continue;
if (need_newline) {
fputc('\n', outfile);
need_newline = 0;
}
pos = usage_indent(outfile);
if (opts->short_name) {
if (opts->flags & PARSE_OPT_NODASH)
pos += fprintf(outfile, "%c", opts->short_name);
else
pos += fprintf(outfile, "-%c", opts->short_name);
}
if (opts->long_name && opts->short_name)
pos += fprintf(outfile, ", ");
if (opts->long_name) {
const char *long_name = opts->long_name;
if ((opts->flags & PARSE_OPT_NONEG) ||
skip_prefix(long_name, "no-", &positive_name))
pos += fprintf(outfile, "--%s", long_name);
else
pos += fprintf(outfile, "--[no-]%s", long_name);
}
if (opts->type == OPTION_NUMBER)
pos += utf8_fprintf(outfile, _("-NUM"));
if ((opts->flags & PARSE_OPT_LITERAL_ARGHELP) ||
!(opts->flags & PARSE_OPT_NOARG))
pos += usage_argh(opts, outfile);
if (opts->type == OPTION_ALIAS) {
usage_padding(outfile, pos);
fprintf_ln(outfile, _("alias of --%s"),
(const char *)opts->value);
continue;
}
for (cp = opts->help ? _(opts->help) : ""; *cp; cp = np) {
np = strchrnul(cp, '\n');
if (*np)
np++;
usage_padding(outfile, pos);
fwrite(cp, 1, np - cp, outfile);
pos = 0;
}
fputc('\n', outfile);
if (positive_name) {
if (find_option_by_long_name(all_opts, positive_name))
continue;
pos = usage_indent(outfile);
pos += fprintf(outfile, "--%s", positive_name);
usage_padding(outfile, pos);
fprintf_ln(outfile, _("opposite of --no-%s"),
positive_name);
}
}
fputc('\n', outfile);
if (!err && ctx && ctx->flags & PARSE_OPT_SHELL_EVAL)
fputs("EOF\n", outfile);
return PARSE_OPT_HELP;
}
Fix sparse warnings Fix warnings from 'make check'. - These files don't include 'builtin.h' causing sparse to complain that cmd_* isn't declared: builtin/clone.c:364, builtin/fetch-pack.c:797, builtin/fmt-merge-msg.c:34, builtin/hash-object.c:78, builtin/merge-index.c:69, builtin/merge-recursive.c:22 builtin/merge-tree.c:341, builtin/mktag.c:156, builtin/notes.c:426 builtin/notes.c:822, builtin/pack-redundant.c:596, builtin/pack-refs.c:10, builtin/patch-id.c:60, builtin/patch-id.c:149, builtin/remote.c:1512, builtin/remote-ext.c:240, builtin/remote-fd.c:53, builtin/reset.c:236, builtin/send-pack.c:384, builtin/unpack-file.c:25, builtin/var.c:75 - These files have symbols which should be marked static since they're only file scope: submodule.c:12, diff.c:631, replace_object.c:92, submodule.c:13, submodule.c:14, trace.c:78, transport.c:195, transport-helper.c:79, unpack-trees.c:19, url.c:3, url.c:18, url.c:104, url.c:117, url.c:123, url.c:129, url.c:136, thread-utils.c:21, thread-utils.c:48 - These files redeclare symbols to be different types: builtin/index-pack.c:210, parse-options.c:564, parse-options.c:571, usage.c:49, usage.c:58, usage.c:63, usage.c:72 - These files use a literal integer 0 when they really should use a NULL pointer: daemon.c:663, fast-import.c:2942, imap-send.c:1072, notes-merge.c:362 While we're in the area, clean up some unused #includes in builtin files (mostly exec_cmd.h). Signed-off-by: Stephen Boyd <bebarino@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-03-22 15:51:05 +08:00
void NORETURN usage_with_options(const char * const *usagestr,
const struct option *opts)
{
usage_with_options_internal(NULL, usagestr, opts, 0, 1);
exit(129);
}
Fix sparse warnings Fix warnings from 'make check'. - These files don't include 'builtin.h' causing sparse to complain that cmd_* isn't declared: builtin/clone.c:364, builtin/fetch-pack.c:797, builtin/fmt-merge-msg.c:34, builtin/hash-object.c:78, builtin/merge-index.c:69, builtin/merge-recursive.c:22 builtin/merge-tree.c:341, builtin/mktag.c:156, builtin/notes.c:426 builtin/notes.c:822, builtin/pack-redundant.c:596, builtin/pack-refs.c:10, builtin/patch-id.c:60, builtin/patch-id.c:149, builtin/remote.c:1512, builtin/remote-ext.c:240, builtin/remote-fd.c:53, builtin/reset.c:236, builtin/send-pack.c:384, builtin/unpack-file.c:25, builtin/var.c:75 - These files have symbols which should be marked static since they're only file scope: submodule.c:12, diff.c:631, replace_object.c:92, submodule.c:13, submodule.c:14, trace.c:78, transport.c:195, transport-helper.c:79, unpack-trees.c:19, url.c:3, url.c:18, url.c:104, url.c:117, url.c:123, url.c:129, url.c:136, thread-utils.c:21, thread-utils.c:48 - These files redeclare symbols to be different types: builtin/index-pack.c:210, parse-options.c:564, parse-options.c:571, usage.c:49, usage.c:58, usage.c:63, usage.c:72 - These files use a literal integer 0 when they really should use a NULL pointer: daemon.c:663, fast-import.c:2942, imap-send.c:1072, notes-merge.c:362 While we're in the area, clean up some unused #includes in builtin files (mostly exec_cmd.h). Signed-off-by: Stephen Boyd <bebarino@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-03-22 15:51:05 +08:00
void NORETURN usage_msg_opt(const char *msg,
const char * const *usagestr,
const struct option *options)
{
die_message("%s\n", msg); /* The extra \n is intentional */
usage_with_options(usagestr, options);
}
void NORETURN usage_msg_optf(const char * const fmt,
const char * const *usagestr,
const struct option *options, ...)
{
struct strbuf msg = STRBUF_INIT;
va_list ap;
va_start(ap, options);
strbuf_vaddf(&msg, fmt, ap);
va_end(ap);
usage_msg_opt(msg.buf, usagestr, options);
}
void die_for_incompatible_opt4(int opt1, const char *opt1_name,
int opt2, const char *opt2_name,
int opt3, const char *opt3_name,
int opt4, const char *opt4_name)
{
int count = 0;
const char *options[4];
if (opt1)
options[count++] = opt1_name;
if (opt2)
options[count++] = opt2_name;
if (opt3)
options[count++] = opt3_name;
if (opt4)
options[count++] = opt4_name;
switch (count) {
case 4:
die(_("options '%s', '%s', '%s', and '%s' cannot be used together"),
opt1_name, opt2_name, opt3_name, opt4_name);
break;
case 3:
die(_("options '%s', '%s', and '%s' cannot be used together"),
options[0], options[1], options[2]);
break;
case 2:
die(_("options '%s' and '%s' cannot be used together"),
options[0], options[1]);
break;
default:
break;
}
}