cpython/Python/getargs.c

1275 lines
28 KiB
C
Raw Normal View History

/* New getargs implementation */
/* XXX There are several unchecked sprintf or strcat calls in this file.
XXX The only way these can become a danger is if some C code in the
XXX Python source (or in an extension) uses ridiculously long names
XXX or ridiculously deep nesting in format strings. */
#include "Python.h"
1996-08-22 07:38:24 +08:00
#include <ctype.h>
#ifdef HAVE_LIMITS_H
#include <limits.h>
#endif
1996-08-22 07:38:24 +08:00
int PyArg_Parse(PyObject *, char *, ...);
int PyArg_ParseTuple(PyObject *, char *, ...);
int PyArg_VaParse(PyObject *, char *, va_list);
int PyArg_ParseTupleAndKeywords(PyObject *, PyObject *,
char *, char **, ...);
/* Forward */
static int vgetargs1(PyObject *, char *, va_list *, int);
static void seterror(int, char *, int *, char *, char *);
static char *convertitem(PyObject *, char **, va_list *, int *, char *);
static char *converttuple(PyObject *, char **, va_list *,
int *, char *, int);
static char *convertsimple(PyObject *, char **, va_list *, char *);
static char *convertsimple1(PyObject *, char **, va_list *);
static int vgetargskeywords(PyObject *, PyObject *,
char *, char **, va_list *);
static char *skipitem(char **, va_list *);
int PyArg_Parse(PyObject *args, char *format, ...)
{
int retval;
va_list va;
va_start(va, format);
retval = vgetargs1(args, format, &va, 1);
va_end(va);
return retval;
}
int PyArg_ParseTuple(PyObject *args, char *format, ...)
{
int retval;
va_list va;
va_start(va, format);
retval = vgetargs1(args, format, &va, 0);
va_end(va);
return retval;
}
int
PyArg_VaParse(PyObject *args, char *format, va_list va)
{
va_list lva;
#ifdef VA_LIST_IS_ARRAY
memcpy(lva, va, sizeof(va_list));
#else
lva = va;
#endif
return vgetargs1(args, format, &lva, 0);
}
static int
vgetargs1(PyObject *args, char *format, va_list *p_va, int compat)
{
char msgbuf[256];
int levels[32];
char *fname = NULL;
char *message = NULL;
int min = -1;
int max = 0;
int level = 0;
char *formatsave = format;
int i, len;
char *msg;
for (;;) {
int c = *format++;
if (c == '(' /* ')' */) {
if (level == 0)
max++;
level++;
}
else if (/* '(' */ c == ')') {
if (level == 0)
Py_FatalError(/* '(' */
"excess ')' in getargs format");
else
level--;
}
else if (c == '\0')
break;
else if (c == ':') {
fname = format;
break;
}
else if (c == ';') {
message = format;
break;
}
else if (level != 0)
; /* Pass */
else if (c == 'e')
; /* Pass */
else if (isalpha(c))
max++;
else if (c == '|')
min = max;
}
if (level != 0)
Py_FatalError(/* '(' */ "missing ')' in getargs format");
if (min < 0)
min = max;
format = formatsave;
if (compat) {
if (max == 0) {
if (args == NULL)
return 1;
sprintf(msgbuf, "%s requires no arguments",
fname==NULL ? "function" : fname);
PyErr_SetString(PyExc_TypeError, msgbuf);
return 0;
}
else if (min == 1 && max == 1) {
if (args == NULL) {
sprintf(msgbuf,
"%s requires at least one argument",
fname==NULL ? "function" : fname);
PyErr_SetString(PyExc_TypeError, msgbuf);
return 0;
}
msg = convertitem(args, &format, p_va, levels, msgbuf);
if (msg == NULL)
return 1;
seterror(levels[0], msg, levels+1, fname, message);
return 0;
}
else {
PyErr_SetString(PyExc_SystemError,
"old style getargs format uses new features");
return 0;
}
}
if (!PyTuple_Check(args)) {
PyErr_SetString(PyExc_SystemError,
"new style getargs format but argument is not a tuple");
return 0;
}
len = PyTuple_Size(args);
if (len < min || max < len) {
if (message == NULL) {
sprintf(msgbuf,
"%s requires %s %d argument%s; %d given",
fname==NULL ? "function" : fname,
min==max ? "exactly"
: len < min ? "at least" : "at most",
len < min ? min : max,
(len < min ? min : max) == 1 ? "" : "s",
len);
message = msgbuf;
}
PyErr_SetString(PyExc_TypeError, message);
return 0;
}
for (i = 0; i < len; i++) {
if (*format == '|')
format++;
msg = convertitem(PyTuple_GetItem(args, i), &format, p_va,
levels, msgbuf);
if (msg) {
seterror(i+1, msg, levels, fname, message);
return 0;
}
}
1998-04-11 06:27:42 +08:00
if (*format != '\0' && !isalpha((int)(*format)) &&
*format != '(' &&
*format != '|' && *format != ':' && *format != ';') {
PyErr_Format(PyExc_SystemError,
"bad format string: %.200s", formatsave);
return 0;
}
return 1;
}
static void
seterror(int iarg, char *msg, int *levels, char *fname, char *message)
{
char buf[256];
int i;
char *p = buf;
if (PyErr_Occurred())
return;
if (iarg == 0 && message == NULL)
message = msg;
else if (message == NULL) {
if (fname != NULL) {
sprintf(p, "%s, ", fname);
p += strlen(p);
}
sprintf(p, "argument %d", iarg);
i = 0;
p += strlen(p);
while (levels[i] > 0) {
sprintf(p, ", item %d", levels[i]-1);
p += strlen(p);
i++;
}
sprintf(p, ": expected %s found", msg);
message = buf;
}
PyErr_SetString(PyExc_TypeError, message);
}
/* Convert a tuple argument.
On entry, *p_format points to the character _after_ the opening '('.
On successful exit, *p_format points to the closing ')'.
If successful:
*p_format and *p_va are updated,
*levels and *msgbuf are untouched,
and NULL is returned.
If the argument is invalid:
*p_format is unchanged,
*p_va is undefined,
*levels is a 0-terminated list of item numbers,
*msgbuf contains an error message, whose format is:
"<typename1>, <typename2>", where:
<typename1> is the name of the expected type, and
<typename2> is the name of the actual type,
(so you can surround it by "expected ... found"),
and msgbuf is returned.
*/
static char *
converttuple(PyObject *arg, char **p_format, va_list *p_va, int *levels,
char *msgbuf, int toplevel)
{
int level = 0;
int n = 0;
char *format = *p_format;
int i;
for (;;) {
int c = *format++;
if (c == '(') {
if (level == 0)
n++;
level++;
}
else if (c == ')') {
if (level == 0)
break;
level--;
}
else if (c == ':' || c == ';' || c == '\0')
break;
else if (level == 0 && isalpha(c))
n++;
}
if (!PySequence_Check(arg)) {
levels[0] = 0;
sprintf(msgbuf,
toplevel ? "%d arguments, %s" : "%d-sequence, %s",
n, arg == Py_None ? "None" : arg->ob_type->tp_name);
return msgbuf;
}
if ((i = PySequence_Size(arg)) != n) {
levels[0] = 0;
sprintf(msgbuf,
toplevel ? "%d arguments, %d" : "%d-sequence, %d-sequence",
n, i);
return msgbuf;
}
format = *p_format;
for (i = 0; i < n; i++) {
char *msg;
PyObject *item;
item = PySequence_GetItem(arg, i);
msg = convertitem(item, &format, p_va, levels+1, msgbuf);
/* PySequence_GetItem calls tp->sq_item, which INCREFs */
Py_XDECREF(item);
if (msg != NULL) {
levels[0] = i+1;
return msg;
}
}
*p_format = format;
return NULL;
}
/* Convert a single item. */
static char *
convertitem(PyObject *arg, char **p_format, va_list *p_va, int *levels,
char *msgbuf)
{
char *msg;
char *format = *p_format;
if (*format == '(' /* ')' */) {
format++;
msg = converttuple(arg, &format, p_va, levels, msgbuf, 0);
if (msg == NULL)
format++;
}
else {
msg = convertsimple(arg, &format, p_va, msgbuf);
if (msg != NULL)
levels[0] = 0;
}
if (msg == NULL)
*p_format = format;
return msg;
}
/* Convert a non-tuple argument. Adds to convertsimple1 functionality
by appending ", <actual argument type>" to error message. */
static char *
convertsimple(PyObject *arg, char **p_format, va_list *p_va, char *msgbuf)
{
char *msg = convertsimple1(arg, p_format, p_va);
if (msg != NULL) {
sprintf(msgbuf, "%.50s, %.50s", msg,
arg == Py_None ? "None" : arg->ob_type->tp_name);
msg = msgbuf;
}
return msg;
}
/* Internal API needed by convertsimple1(): */
extern
PyObject *_PyUnicode_AsDefaultEncodedString(PyObject *unicode,
const char *errors);
/* Convert a non-tuple argument. Return NULL if conversion went OK,
or a string representing the expected type if the conversion failed.
When failing, an exception may or may not have been raised.
Don't call if a tuple is expected. */
static char *
convertsimple1(PyObject *arg, char **p_format, va_list *p_va)
{
char *format = *p_format;
char c = *format++;
switch (c) {
case 'b': /* unsigned byte -- very short int */
{
char *p = va_arg(*p_va, char *);
long ival = PyInt_AsLong(arg);
if (ival == -1 && PyErr_Occurred())
return "integer<b>";
else if (ival < 0) {
PyErr_SetString(PyExc_OverflowError,
"unsigned byte integer is less than minimum");
return "integer<b>";
}
else if (ival > UCHAR_MAX) {
PyErr_SetString(PyExc_OverflowError,
"unsigned byte integer is greater than maximum");
return "integer<b>";
}
else
*p = (unsigned char) ival;
break;
}
case 'B': /* byte sized bitfield - both signed and unsigned values allowed */
{
char *p = va_arg(*p_va, char *);
long ival = PyInt_AsLong(arg);
if (ival == -1 && PyErr_Occurred())
return "integer<b>";
else if (ival < SCHAR_MIN) {
PyErr_SetString(PyExc_OverflowError,
"byte-sized integer bitfield is less than minimum");
return "integer<B>";
}
else if (ival > UCHAR_MAX) {
PyErr_SetString(PyExc_OverflowError,
"byte-sized integer bitfield is greater than maximum");
return "integer<B>";
}
else
*p = (unsigned char) ival;
break;
}
case 'h': /* signed short int */
{
short *p = va_arg(*p_va, short *);
long ival = PyInt_AsLong(arg);
if (ival == -1 && PyErr_Occurred())
return "integer<h>";
else if (ival < SHRT_MIN) {
PyErr_SetString(PyExc_OverflowError,
"signed short integer is less than minimum");
return "integer<h>";
}
else if (ival > SHRT_MAX) {
PyErr_SetString(PyExc_OverflowError,
"signed short integer is greater than maximum");
return "integer<h>";
}
else
1997-04-12 04:37:35 +08:00
*p = (short) ival;
break;
}
case 'H': /* short int sized bitfield, both signed and unsigned allowed */
{
unsigned short *p = va_arg(*p_va, unsigned short *);
long ival = PyInt_AsLong(arg);
if (ival == -1 && PyErr_Occurred())
return "integer<H>";
else if (ival < SHRT_MIN) {
PyErr_SetString(PyExc_OverflowError,
"short integer bitfield is less than minimum");
return "integer<H>";
}
else if (ival > USHRT_MAX) {
PyErr_SetString(PyExc_OverflowError,
"short integer bitfield is greater than maximum");
return "integer<H>";
}
else
*p = (unsigned short) ival;
break;
}
case 'i': /* signed int */
{
int *p = va_arg(*p_va, int *);
long ival = PyInt_AsLong(arg);
if (ival == -1 && PyErr_Occurred())
return "integer<i>";
else if (ival > INT_MAX) {
PyErr_SetString(PyExc_OverflowError,
"signed integer is greater than maximum");
return "integer<i>";
}
else if (ival < INT_MIN) {
PyErr_SetString(PyExc_OverflowError,
"signed integer is less than minimum");
return "integer<i>";
}
else
*p = ival;
break;
}
case 'l': /* long int */
{
long *p = va_arg(*p_va, long *);
long ival = PyInt_AsLong(arg);
if (ival == -1 && PyErr_Occurred())
return "integer<l>";
else
*p = ival;
break;
}
#ifdef HAVE_LONG_LONG
case 'L': /* LONG_LONG */
{
LONG_LONG *p = va_arg( *p_va, LONG_LONG * );
LONG_LONG ival = PyLong_AsLongLong( arg );
if( ival == (LONG_LONG)-1 && PyErr_Occurred() ) {
return "long<L>";
} else {
*p = ival;
}
break;
}
#endif
case 'f': /* float */
{
float *p = va_arg(*p_va, float *);
double dval = PyFloat_AsDouble(arg);
if (PyErr_Occurred())
return "float<f>";
else
1997-04-12 04:37:35 +08:00
*p = (float) dval;
break;
}
case 'd': /* double */
{
double *p = va_arg(*p_va, double *);
double dval = PyFloat_AsDouble(arg);
if (PyErr_Occurred())
return "float<d>";
else
*p = dval;
break;
}
#ifndef WITHOUT_COMPLEX
1996-01-12 09:09:56 +08:00
case 'D': /* complex double */
{
Py_complex *p = va_arg(*p_va, Py_complex *);
Py_complex cval;
cval = PyComplex_AsCComplex(arg);
if (PyErr_Occurred())
1996-01-12 09:09:56 +08:00
return "complex<D>";
else
*p = cval;
break;
}
#endif /* WITHOUT_COMPLEX */
1996-01-12 09:09:56 +08:00
case 'c': /* char */
{
char *p = va_arg(*p_va, char *);
if (PyString_Check(arg) && PyString_Size(arg) == 1)
*p = PyString_AsString(arg)[0];
else
return "char";
break;
}
case 's': /* string */
{
if (*format == '#') { /* any buffer-like object */
void **p = (void **)va_arg(*p_va, char **);
PyBufferProcs *pb = arg->ob_type->tp_as_buffer;
int *q = va_arg(*p_va, int *);
int count;
if ( pb == NULL ||
pb->bf_getreadbuffer == NULL ||
pb->bf_getsegcount == NULL )
return "read-only buffer";
if ( (*pb->bf_getsegcount)(arg, NULL) != 1 )
return "single-segment read-only buffer";
if ( (count =
(*pb->bf_getreadbuffer)(arg, 0, p)) < 0 )
return "(unspecified)";
*q = count;
format++;
} else {
char **p = va_arg(*p_va, char **);
if (PyString_Check(arg))
*p = PyString_AS_STRING(arg);
else if (PyUnicode_Check(arg)) {
arg = _PyUnicode_AsDefaultEncodedString(
arg, NULL);
if (arg == NULL)
return "unicode conversion error";
*p = PyString_AS_STRING(arg);
}
else
return "string";
if ((int)strlen(*p) != PyString_Size(arg))
return "string without null bytes";
}
break;
}
case 'z': /* string, may be NULL (None) */
{
if (*format == '#') { /* any buffer-like object */
void **p = (void **)va_arg(*p_va, char **);
PyBufferProcs *pb = arg->ob_type->tp_as_buffer;
int *q = va_arg(*p_va, int *);
int count;
if (arg == Py_None) {
*p = 0;
*q = 0;
} else {
if ( pb == NULL ||
pb->bf_getreadbuffer == NULL ||
pb->bf_getsegcount == NULL )
return "read-only buffer";
if ( (*pb->bf_getsegcount)(arg, NULL) != 1 )
return "single-segment read-only buffer";
if ( (count = (*pb->bf_getreadbuffer)
(arg, 0, p)) < 0 )
return "(unspecified)";
*q = count;
}
format++;
} else {
char **p = va_arg(*p_va, char **);
if (arg == Py_None)
*p = 0;
else if (PyString_Check(arg))
*p = PyString_AsString(arg);
else if (PyUnicode_Check(arg)) {
arg = _PyUnicode_AsDefaultEncodedString(
arg, NULL);
if (arg == NULL)
return "unicode conversion error";
*p = PyString_AS_STRING(arg);
}
else
return "None or string";
if (*format == '#') {
int *q = va_arg(*p_va, int *);
if (arg == Py_None)
*q = 0;
else
*q = PyString_Size(arg);
format++;
}
else if (*p != NULL &&
(int)strlen(*p) != PyString_Size(arg))
return "None or string without null bytes";
}
break;
}
case 'e': /* encoded string */
{
char **buffer;
const char *encoding;
PyObject *u, *s;
int size;
/* Get 'e' parameter: the encoding name */
encoding = (const char *)va_arg(*p_va, const char *);
if (encoding == NULL)
encoding = PyUnicode_GetDefaultEncoding();
/* Get 's' parameter: the output buffer to use */
if (*format != 's')
2000-03-28 10:00:29 +08:00
return "(unknown parser marker combination)";
buffer = (char **)va_arg(*p_va, char **);
format++;
if (buffer == NULL)
return "(buffer is NULL)";
/* Convert object to Unicode */
u = PyUnicode_FromObject(arg);
if (u == NULL)
return "string, unicode or text buffer";
/* Encode object; use default error handling */
s = PyUnicode_AsEncodedString(u,
encoding,
NULL);
Py_DECREF(u);
if (s == NULL)
return "(encoding failed)";
if (!PyString_Check(s)) {
Py_DECREF(s);
return "(encoder failed to return a string)";
}
size = PyString_GET_SIZE(s);
/* Write output; output is guaranteed to be
0-terminated */
if (*format == '#') {
/* Using buffer length parameter '#':
- if *buffer is NULL, a new buffer
of the needed size is allocated and
the data copied into it; *buffer is
updated to point to the new buffer;
the caller is responsible for
PyMem_Free()ing it after usage
- if *buffer is not NULL, the data
is copied to *buffer; *buffer_len
has to be set to the size of the
buffer on input; buffer overflow is
signalled with an error; buffer has
to provide enough room for the
encoded string plus the trailing
0-byte
- in both cases, *buffer_len is
updated to the size of the buffer
/excluding/ the trailing 0-byte
*/
int *buffer_len = va_arg(*p_va, int *);
format++;
if (buffer_len == NULL)
return "(buffer_len is NULL)";
if (*buffer == NULL) {
*buffer = PyMem_NEW(char, size + 1);
if (*buffer == NULL) {
Py_DECREF(s);
return "(memory error)";
}
} else {
if (size + 1 > *buffer_len) {
Py_DECREF(s);
return "(buffer overflow)";
}
}
memcpy(*buffer,
PyString_AS_STRING(s),
size + 1);
*buffer_len = size;
} else {
/* Using a 0-terminated buffer:
- the encoded string has to be
0-terminated for this variant to
work; if it is not, an error raised
- a new buffer of the needed size
is allocated and the data copied
into it; *buffer is updated to
point to the new buffer; the caller
is responsible for PyMem_Free()ing it
after usage
*/
2000-03-28 10:00:29 +08:00
if ((int)strlen(PyString_AS_STRING(s)) != size)
return "(encoded string without "\
"NULL bytes)";
*buffer = PyMem_NEW(char, size + 1);
if (*buffer == NULL) {
Py_DECREF(s);
return "(memory error)";
}
memcpy(*buffer,
PyString_AS_STRING(s),
size + 1);
}
Py_DECREF(s);
break;
}
case 'u': /* raw unicode buffer (Py_UNICODE *) */
{
if (*format == '#') { /* any buffer-like object */
void **p = (void **)va_arg(*p_va, char **);
PyBufferProcs *pb = arg->ob_type->tp_as_buffer;
int *q = va_arg(*p_va, int *);
int count;
if ( pb == NULL ||
pb->bf_getreadbuffer == NULL ||
pb->bf_getsegcount == NULL )
return "read-only buffer";
if ( (*pb->bf_getsegcount)(arg, NULL) != 1 )
return "single-segment read-only buffer";
if ( (count =
(*pb->bf_getreadbuffer)(arg, 0, p)) < 0 )
return "(unspecified)";
/* buffer interface returns bytes, we want
length in characters */
*q = count/(sizeof(Py_UNICODE));
format++;
} else {
Py_UNICODE **p = va_arg(*p_va, Py_UNICODE **);
if (PyUnicode_Check(arg))
*p = PyUnicode_AS_UNICODE(arg);
else
return "unicode";
}
break;
}
case 'S': /* string object */
{
PyObject **p = va_arg(*p_va, PyObject **);
if (PyString_Check(arg))
*p = arg;
else
return "string";
break;
}
case 'U': /* Unicode object */
{
PyObject **p = va_arg(*p_va, PyObject **);
if (PyUnicode_Check(arg))
*p = arg;
else
return "unicode";
break;
}
case 'O': /* object */
{
PyTypeObject *type;
PyObject **p;
if (*format == '!') {
type = va_arg(*p_va, PyTypeObject*);
p = va_arg(*p_va, PyObject **);
format++;
if (arg->ob_type == type)
*p = arg;
else
return type->tp_name;
}
else if (*format == '?') {
inquiry pred = va_arg(*p_va, inquiry);
p = va_arg(*p_va, PyObject **);
format++;
if ((*pred)(arg))
*p = arg;
else
return "(unspecified)";
}
else if (*format == '&') {
typedef int (*converter)(PyObject *, void *);
converter convert = va_arg(*p_va, converter);
void *addr = va_arg(*p_va, void *);
format++;
if (! (*convert)(arg, addr))
return "(unspecified)";
}
else {
p = va_arg(*p_va, PyObject **);
*p = arg;
}
break;
}
case 'w': /* memory buffer, read-write access */
{
void **p = va_arg(*p_va, void **);
PyBufferProcs *pb = arg->ob_type->tp_as_buffer;
int count;
if ( pb == NULL || pb->bf_getwritebuffer == NULL ||
pb->bf_getsegcount == NULL )
return "read-write buffer";
if ( (*pb->bf_getsegcount)(arg, NULL) != 1 )
return "single-segment read-write buffer";
if ( (count = pb->bf_getwritebuffer(arg, 0, p)) < 0 )
return "(unspecified)";
if (*format == '#') {
int *q = va_arg(*p_va, int *);
*q = count;
format++;
}
break;
}
case 't': /* 8-bit character buffer, read-only access */
{
const char **p = va_arg(*p_va, const char **);
PyBufferProcs *pb = arg->ob_type->tp_as_buffer;
int count;
if ( *format++ != '#' )
return "invalid use of 't' format character";
if ( !PyType_HasFeature(
arg->ob_type,
Py_TPFLAGS_HAVE_GETCHARBUFFER) ||
pb == NULL ||
pb->bf_getcharbuffer == NULL ||
pb->bf_getsegcount == NULL )
return "read-only character buffer";
if ( (*pb->bf_getsegcount)(arg, NULL) != 1 )
return "single-segment read-only buffer";
if ( (count = pb->bf_getcharbuffer(arg, 0, p)) < 0 )
return "(unspecified)";
*va_arg(*p_va, int *) = count;
break;
}
default:
return "impossible<bad format char>";
}
*p_format = format;
return NULL;
}
/* Support for keyword arguments donated by
Geoff Philbrick <philbric@delphi.hks.com> */
int PyArg_ParseTupleAndKeywords(PyObject *args,
PyObject *keywords,
char *format,
char **kwlist, ...)
{
int retval;
va_list va;
va_start(va, kwlist);
retval = vgetargskeywords(args, keywords, format, kwlist, &va);
va_end(va);
return retval;
}
static int
vgetargskeywords(PyObject *args, PyObject *keywords, char *format,
char **kwlist, va_list *p_va)
{
char msgbuf[256];
int levels[32];
char *fname = NULL;
char *message = NULL;
int min = -1;
int max = 0;
char *formatsave = format;
int i, len, tplen, kwlen;
char *msg, *ks, **p;
int nkwds, pos, match, converted;
PyObject *key, *value;
/* nested tuples cannot be parsed when using keyword arguments */
for (;;) {
int c = *format++;
if (c == '(') {
PyErr_SetString(PyExc_SystemError,
"tuple found in format when using keyword arguments");
return 0;
}
else if (c == '\0')
break;
else if (c == ':') {
fname = format;
break;
}
else if (c == ';') {
message = format;
break;
}
else if (isalpha(c))
max++;
else if (c == '|')
min = max;
}
if (min < 0)
min = max;
format = formatsave;
if (!PyTuple_Check(args)) {
PyErr_SetString(PyExc_SystemError,
"new style getargs format but argument is not a tuple");
return 0;
}
tplen = PyTuple_Size(args);
/* do a cursory check of the keywords just to see how many we got */
if (keywords) {
if (!PyDict_Check(keywords)) {
PyErr_SetString(PyExc_SystemError,
"non-dictionary object received when keyword dictionary expected");
return 0;
}
kwlen = PyDict_Size(keywords);
}
else {
kwlen = 0;
}
/* make sure there are no duplicate values for an argument;
its not clear when to use the term "keyword argument vs.
keyword parameter in messages */
if (keywords) {
for (i = 0; i < tplen; i++) {
if (PyMapping_HasKeyString(keywords, kwlist[i])) {
sprintf(msgbuf,
"keyword parameter %s redefined",
kwlist[i]);
PyErr_SetString(PyExc_TypeError, msgbuf);
return 0;
}
}
}
PyErr_Clear(); /* I'm not which Py functions set the error string */
/* required arguments missing from args can be supplied by keyword
arguments */
len = tplen;
if (keywords && tplen < min) {
for (i = tplen; i < min; i++) {
if (PyMapping_HasKeyString(keywords, kwlist[i])) {
len++;
}
}
}
PyErr_Clear();
/* make sure we got an acceptable number of arguments; the message
is a little confusing with keywords since keyword arguments
which are supplied, but don't match the required arguments
are not included in the "%d given" part of the message */
if (len < min || max < len) {
if (message == NULL) {
sprintf(msgbuf,
"%s requires %s %d argument%s; %d given",
fname==NULL ? "function" : fname,
min==max ? "exactly"
: len < min ? "at least" : "at most",
len < min ? min : max,
(len < min ? min : max) == 1 ? "" : "s",
len);
message = msgbuf;
}
PyErr_SetString(PyExc_TypeError, message);
return 0;
}
for (i = 0; i < tplen; i++) {
if (*format == '|')
format++;
msg = convertitem(PyTuple_GetItem(args, i), &format, p_va,
levels, msgbuf);
if (msg) {
seterror(i+1, msg, levels, fname, message);
return 0;
}
}
/* handle no keyword parameters in call */
if (!keywords) return 1;
/* make sure the number of keywords in the keyword list matches the
number of items in the format string */
nkwds = 0;
p = kwlist;
for (;;) {
if (!*(p++)) break;
nkwds++;
}
if (nkwds != max) {
PyErr_SetString(PyExc_SystemError,
"number of items in format string and keyword list do not match");
return 0;
}
/* convert the keyword arguments; this uses the format
string where it was left after processing args */
converted = 0;
for (i = tplen; i < nkwds; i++) {
PyObject *item;
if (*format == '|')
format++;
1996-12-06 07:27:02 +08:00
item = PyMapping_GetItemString(keywords, kwlist[i]);
if (item != NULL) {
msg = convertitem(item, &format, p_va, levels, msgbuf);
if (msg) {
seterror(i+1, msg, levels, fname, message);
return 0;
}
converted++;
}
else {
PyErr_Clear();
msg = skipitem(&format, p_va);
if (msg) {
seterror(i+1, msg, levels, fname, message);
return 0;
}
}
}
/* make sure there are no extraneous keyword arguments */
pos = 0;
if (converted < kwlen) {
while (PyDict_Next(keywords, &pos, &key, &value)) {
match = 0;
ks = PyString_AsString(key);
for (i = 0; i < nkwds; i++) {
if (!strcmp(ks, kwlist[i])) {
match = 1;
break;
}
}
if (!match) {
sprintf(msgbuf,
"%s is an invalid keyword argument for this function",
ks);
PyErr_SetString(PyExc_TypeError, msgbuf);
return 0;
}
}
}
return 1;
}
static char *
skipitem(char **p_format, va_list *p_va)
{
char *format = *p_format;
char c = *format++;
switch (c) {
case 'b': /* byte -- very short int */
case 'B': /* byte as bitfield */
{
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, char *);
break;
}
case 'h': /* short int */
{
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, short *);
break;
}
case 'H': /* short int as bitfield */
{
(void) va_arg(*p_va, unsigned short *);
break;
}
case 'i': /* int */
{
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, int *);
break;
}
case 'l': /* long int */
{
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, long *);
break;
}
#ifdef HAVE_LONG_LONG
case 'L': /* LONG_LONG int */
{
(void) va_arg(*p_va, LONG_LONG *);
break;
}
#endif
case 'f': /* float */
{
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, float *);
break;
}
case 'd': /* double */
{
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, double *);
break;
}
#ifndef WITHOUT_COMPLEX
case 'D': /* complex double */
{
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, Py_complex *);
break;
}
#endif /* WITHOUT_COMPLEX */
case 'c': /* char */
{
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, char *);
break;
}
case 's': /* string */
{
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, char **);
if (*format == '#') {
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, int *);
format++;
}
break;
}
case 'z': /* string */
{
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, char **);
if (*format == '#') {
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, int *);
format++;
}
break;
}
case 'S': /* string object */
{
(void) va_arg(*p_va, PyObject **);
break;
}
case 'O': /* object */
{
if (*format == '!') {
format++;
(void) va_arg(*p_va, PyTypeObject*);
(void) va_arg(*p_va, PyObject **);
}
#if 0
/* I don't know what this is for */
else if (*format == '?') {
inquiry pred = va_arg(*p_va, inquiry);
format++;
if ((*pred)(arg)) {
(void) va_arg(*p_va, PyObject **);
}
}
#endif
else if (*format == '&') {
typedef int (*converter)(PyObject *, void *);
1996-12-06 07:27:02 +08:00
(void) va_arg(*p_va, converter);
(void) va_arg(*p_va, void *);
format++;
}
else {
(void) va_arg(*p_va, PyObject **);
}
break;
}
default:
return "impossible<bad format char>";
}
*p_format = format;
return NULL;
}