cpython/Python/marshal.c

1892 lines
50 KiB
C
Raw Normal View History

1991-06-05 04:42:30 +09:00
/* Write Python objects to files and read them back.
This is primarily intended for writing and reading compiled Python code,
even though dicts, lists, sets and frozensets, not commonly seen in
code objects, are supported.
Version 3 of this protocol properly supports circular links
and sharing. */
1991-06-05 04:42:30 +09:00
2006-03-02 07:49:13 +08:00
#define PY_SSIZE_T_CLEAN
#include "Python.h"
#include "pycore_call.h" // _PyObject_CallNoArgs()
#include "pycore_code.h" // _PyCode_New()
#include "pycore_hashtable.h" // _Py_hashtable_t
#include "marshal.h" // Py_MARSHAL_VERSION
1991-06-05 04:42:30 +09:00
/*[clinic input]
module marshal
[clinic start generated code]*/
/*[clinic end generated code: output=da39a3ee5e6b4b0d input=c982b7930dee17db]*/
#include "clinic/marshal.c.h"
/* High water mark to determine when the marshalled object is dangerously deep
* and risks coring the interpreter. When the object stack gets this deep,
* raise an exception instead of continuing.
* On Windows debug builds, reduce this value.
*
* BUG: https://bugs.python.org/issue33720
* On Windows PGO builds, the r_object function overallocates its stack and
* can cause a stack overflow. We reduce the maximum depth for all Windows
* releases to protect against this.
* #if defined(MS_WINDOWS) && defined(_DEBUG)
*/
#if defined(MS_WINDOWS)
#define MAX_MARSHAL_STACK_DEPTH 1000
#elif defined(__wasi__)
#define MAX_MARSHAL_STACK_DEPTH 1500
#else
Merged revisions 55407-55513 via svnmerge from svn+ssh://pythondev@svn.python.org/python/branches/p3yk ................ r55413 | fred.drake | 2007-05-17 12:30:10 -0700 (Thu, 17 May 2007) | 1 line fix argument name in documentation; match the implementation ................ r55430 | jack.diederich | 2007-05-18 06:39:59 -0700 (Fri, 18 May 2007) | 1 line Implements class decorators, PEP 3129. ................ r55432 | guido.van.rossum | 2007-05-18 08:09:41 -0700 (Fri, 18 May 2007) | 2 lines obsubmit. ................ r55434 | guido.van.rossum | 2007-05-18 09:39:10 -0700 (Fri, 18 May 2007) | 3 lines Fix bug in test_inspect. (I presume this is how it should be fixed; Jack Diedrich, please verify.) ................ r55460 | brett.cannon | 2007-05-20 00:31:57 -0700 (Sun, 20 May 2007) | 4 lines Remove the imageop module. With imgfile already removed in Python 3.0 and rgbimg gone in Python 2.6 the unit tests themselves were made worthless. Plus third-party libraries perform the same function much better. ................ r55469 | neal.norwitz | 2007-05-20 11:28:20 -0700 (Sun, 20 May 2007) | 118 lines Merged revisions 55324-55467 via svnmerge from svn+ssh://pythondev@svn.python.org/python/trunk ........ r55348 | georg.brandl | 2007-05-15 13:19:34 -0700 (Tue, 15 May 2007) | 4 lines HTML-escape the plain traceback in cgitb's HTML output, to prevent the traceback inadvertently or maliciously closing the comment and injecting HTML into the error page. ........ r55372 | neal.norwitz | 2007-05-15 21:33:50 -0700 (Tue, 15 May 2007) | 6 lines Port rev 55353 from Guido: Add what looks like a necessary call to PyErr_NoMemory() when PyMem_MALLOC() fails. Will backport. ........ r55377 | neal.norwitz | 2007-05-15 22:06:33 -0700 (Tue, 15 May 2007) | 1 line Mention removal of some directories for obsolete platforms ........ r55380 | brett.cannon | 2007-05-15 22:50:03 -0700 (Tue, 15 May 2007) | 2 lines Change the maintainer of the BeOS port. ........ r55383 | georg.brandl | 2007-05-16 06:44:18 -0700 (Wed, 16 May 2007) | 2 lines Bug #1719995: don't use deprecated method in sets example. ........ r55386 | neal.norwitz | 2007-05-16 13:05:11 -0700 (Wed, 16 May 2007) | 5 lines Fix bug in marshal where bad data would cause a segfault due to lack of an infinite recursion check. Contributed by Damien Miller at Google. ........ r55389 | brett.cannon | 2007-05-16 15:42:29 -0700 (Wed, 16 May 2007) | 6 lines Remove the gopherlib module. It has been raising a DeprecationWarning since Python 2.5. Also remove gopher support from urllib/urllib2. As both imported gopherlib the usage of the support would have raised a DeprecationWarning. ........ r55394 | raymond.hettinger | 2007-05-16 18:08:04 -0700 (Wed, 16 May 2007) | 1 line calendar.py gets no benefit from xrange() instead of range() ........ r55395 | brett.cannon | 2007-05-16 19:02:56 -0700 (Wed, 16 May 2007) | 3 lines Complete deprecation of BaseException.message. Some subclasses were directly accessing the message attribute instead of using the descriptor. ........ r55396 | neal.norwitz | 2007-05-16 23:11:36 -0700 (Wed, 16 May 2007) | 4 lines Reduce the max stack depth to see if this fixes the segfaults on Windows and some other boxes. If this is successful, this rev should be backported. I'm not sure how close to the limit we should push this. ........ r55397 | neal.norwitz | 2007-05-16 23:23:50 -0700 (Wed, 16 May 2007) | 4 lines Set the depth to something very small to try to determine if the crashes on Windows are really due to the stack size or possibly some other problem. ........ r55398 | neal.norwitz | 2007-05-17 00:04:46 -0700 (Thu, 17 May 2007) | 4 lines Last try for tweaking the max stack depth. 5000 was the original value, 4000 didn't work either. 1000 does work on Windows. If 2000 works, that will hopefully be a reasonable balance. ........ r55412 | fred.drake | 2007-05-17 12:29:58 -0700 (Thu, 17 May 2007) | 1 line fix argument name in documentation; match the implementation ........ r55427 | neal.norwitz | 2007-05-17 22:47:16 -0700 (Thu, 17 May 2007) | 1 line Verify neither dumps or loads overflow the stack and segfault. ........ r55446 | collin.winter | 2007-05-18 16:11:24 -0700 (Fri, 18 May 2007) | 1 line Backport PEP 3110's new 'except' syntax to 2.6. ........ r55448 | raymond.hettinger | 2007-05-18 18:11:16 -0700 (Fri, 18 May 2007) | 1 line Improvements to NamedTuple's implementation, tests, and documentation ........ r55449 | raymond.hettinger | 2007-05-18 18:50:11 -0700 (Fri, 18 May 2007) | 1 line Fix beginner mistake -- don't mix spaces and tabs. ........ r55450 | neal.norwitz | 2007-05-18 20:48:47 -0700 (Fri, 18 May 2007) | 1 line Clear data so random memory does not get freed. Will backport. ........ r55452 | neal.norwitz | 2007-05-18 21:34:55 -0700 (Fri, 18 May 2007) | 3 lines Whoops, need to pay attention to those test failures. Move the clear to *before* the first use, not after. ........ r55453 | neal.norwitz | 2007-05-18 21:35:52 -0700 (Fri, 18 May 2007) | 1 line Give some clue as to what happened if the test fails. ........ r55455 | georg.brandl | 2007-05-19 11:09:26 -0700 (Sat, 19 May 2007) | 2 lines Fix docstring for add_package in site.py. ........ r55458 | brett.cannon | 2007-05-20 00:09:50 -0700 (Sun, 20 May 2007) | 2 lines Remove the rgbimg module. It has been deprecated since Python 2.5. ........ r55465 | nick.coghlan | 2007-05-20 04:12:49 -0700 (Sun, 20 May 2007) | 1 line Fix typo in example (should be backported, but my maintenance branch is woefully out of date) ........ ................ r55472 | brett.cannon | 2007-05-20 12:06:18 -0700 (Sun, 20 May 2007) | 2 lines Remove imageop from the Windows build process. ................ r55486 | neal.norwitz | 2007-05-20 23:59:52 -0700 (Sun, 20 May 2007) | 1 line Remove callable() builtin ................ r55506 | neal.norwitz | 2007-05-22 00:43:29 -0700 (Tue, 22 May 2007) | 78 lines Merged revisions 55468-55505 via svnmerge from svn+ssh://pythondev@svn.python.org/python/trunk ........ r55468 | neal.norwitz | 2007-05-20 11:06:27 -0700 (Sun, 20 May 2007) | 1 line rotor is long gone. ........ r55470 | neal.norwitz | 2007-05-20 11:43:00 -0700 (Sun, 20 May 2007) | 1 line Update directories/files at the top-level. ........ r55471 | brett.cannon | 2007-05-20 12:05:06 -0700 (Sun, 20 May 2007) | 2 lines Try to remove rgbimg from Windows builds. ........ r55474 | brett.cannon | 2007-05-20 16:17:38 -0700 (Sun, 20 May 2007) | 4 lines Remove the macfs module. This led to the deprecation of macostools.touched(); it completely relied on macfs and is a no-op on OS X according to code comments. ........ r55476 | brett.cannon | 2007-05-20 16:56:18 -0700 (Sun, 20 May 2007) | 3 lines Move imgfile import to the global namespace to trigger an import error ASAP to prevent creation of a test file. ........ r55477 | brett.cannon | 2007-05-20 16:57:38 -0700 (Sun, 20 May 2007) | 3 lines Cause posixfile to raise a DeprecationWarning. Documented as deprecated since Ptyhon 1.5. ........ r55479 | andrew.kuchling | 2007-05-20 17:03:15 -0700 (Sun, 20 May 2007) | 1 line Note removed modules ........ r55481 | martin.v.loewis | 2007-05-20 21:35:47 -0700 (Sun, 20 May 2007) | 2 lines Add Alexandre Vassalotti. ........ r55482 | george.yoshida | 2007-05-20 21:41:21 -0700 (Sun, 20 May 2007) | 4 lines fix against r55474 [Remove the macfs module] Remove "libmacfs.tex" from Makefile.deps and mac/mac.tex. ........ r55487 | raymond.hettinger | 2007-05-21 01:13:35 -0700 (Mon, 21 May 2007) | 1 line Replace assertion with straight error-checking. ........ r55489 | raymond.hettinger | 2007-05-21 09:40:10 -0700 (Mon, 21 May 2007) | 1 line Allow all alphanumeric and underscores in type and field names. ........ r55490 | facundo.batista | 2007-05-21 10:32:32 -0700 (Mon, 21 May 2007) | 5 lines Added timeout support to HTTPSConnection, through the socket.create_connection function. Also added a small test for this, and updated NEWS file. ........ r55495 | georg.brandl | 2007-05-21 13:34:16 -0700 (Mon, 21 May 2007) | 2 lines Patch #1686487: you can now pass any mapping after '**' in function calls. ........ r55502 | neal.norwitz | 2007-05-21 23:03:36 -0700 (Mon, 21 May 2007) | 1 line Document new params to HTTPSConnection ........ r55504 | neal.norwitz | 2007-05-22 00:16:10 -0700 (Tue, 22 May 2007) | 1 line Stop using METH_OLDARGS ........ r55505 | neal.norwitz | 2007-05-22 00:16:44 -0700 (Tue, 22 May 2007) | 1 line Stop using METH_OLDARGS implicitly ........ ................
2007-05-23 02:11:13 +08:00
#define MAX_MARSHAL_STACK_DEPTH 2000
#endif
#define TYPE_NULL '0'
#define TYPE_NONE 'N'
#define TYPE_FALSE 'F'
#define TYPE_TRUE 'T'
#define TYPE_STOPITER 'S'
#define TYPE_ELLIPSIS '.'
#define TYPE_INT 'i'
/* TYPE_INT64 is not generated anymore.
Supported for backward compatibility only. */
#define TYPE_INT64 'I'
#define TYPE_FLOAT 'f'
#define TYPE_BINARY_FLOAT 'g'
#define TYPE_COMPLEX 'x'
#define TYPE_BINARY_COMPLEX 'y'
#define TYPE_LONG 'l'
#define TYPE_STRING 's'
#define TYPE_INTERNED 't'
#define TYPE_REF 'r'
#define TYPE_TUPLE '('
#define TYPE_LIST '['
#define TYPE_DICT '{'
#define TYPE_CODE 'c'
#define TYPE_UNICODE 'u'
#define TYPE_UNKNOWN '?'
#define TYPE_SET '<'
#define TYPE_FROZENSET '>'
#define FLAG_REF '\x80' /* with a type, add obj to index */
1991-06-05 04:42:30 +09:00
#define TYPE_ASCII 'a'
#define TYPE_ASCII_INTERNED 'A'
#define TYPE_SMALL_TUPLE ')'
#define TYPE_SHORT_ASCII 'z'
#define TYPE_SHORT_ASCII_INTERNED 'Z'
#define WFERR_OK 0
#define WFERR_UNMARSHALLABLE 1
#define WFERR_NESTEDTOODEEP 2
#define WFERR_NOMEMORY 3
typedef struct {
FILE *fp;
int error; /* see WFERR_* values */
int depth;
PyObject *str;
char *ptr;
const char *end;
char *buf;
_Py_hashtable_t *hashtable;
int version;
} WFILE;
1991-06-05 04:42:30 +09:00
#define w_byte(c, p) do { \
if ((p)->ptr != (p)->end || w_reserve((p), 1)) \
*(p)->ptr++ = (c); \
} while(0)
static void
w_flush(WFILE *p)
{
assert(p->fp != NULL);
fwrite(p->buf, 1, p->ptr - p->buf, p->fp);
p->ptr = p->buf;
}
static int
w_reserve(WFILE *p, Py_ssize_t needed)
{
Py_ssize_t pos, size, delta;
if (p->ptr == NULL)
return 0; /* An error already occurred */
if (p->fp != NULL) {
w_flush(p);
return needed <= p->end - p->ptr;
}
assert(p->str != NULL);
pos = p->ptr - p->buf;
size = PyBytes_GET_SIZE(p->str);
if (size > 16*1024*1024)
delta = (size >> 3); /* 12.5% overallocation */
else
delta = size + 1024;
delta = Py_MAX(delta, needed);
if (delta > PY_SSIZE_T_MAX - size) {
p->error = WFERR_NOMEMORY;
return 0;
}
size += delta;
if (_PyBytes_Resize(&p->str, size) != 0) {
p->end = p->ptr = p->buf = NULL;
return 0;
}
else {
p->buf = PyBytes_AS_STRING(p->str);
p->ptr = p->buf + pos;
p->end = p->buf + size;
return 1;
}
}
static void
w_string(const void *s, Py_ssize_t n, WFILE *p)
{
Py_ssize_t m;
if (!n || p->ptr == NULL)
return;
m = p->end - p->ptr;
if (p->fp != NULL) {
if (n <= m) {
memcpy(p->ptr, s, n);
p->ptr += n;
}
else {
w_flush(p);
fwrite(s, 1, n, p->fp);
}
}
else {
if (n <= m || w_reserve(p, n - m)) {
memcpy(p->ptr, s, n);
p->ptr += n;
}
}
}
static void
w_short(int x, WFILE *p)
1991-06-05 04:42:30 +09:00
{
w_byte((char)( x & 0xff), p);
w_byte((char)((x>> 8) & 0xff), p);
1991-06-05 04:42:30 +09:00
}
static void
w_long(long x, WFILE *p)
1991-06-05 04:42:30 +09:00
{
w_byte((char)( x & 0xff), p);
w_byte((char)((x>> 8) & 0xff), p);
w_byte((char)((x>>16) & 0xff), p);
w_byte((char)((x>>24) & 0xff), p);
1991-06-05 04:42:30 +09:00
}
#define SIZE32_MAX 0x7FFFFFFF
#if SIZEOF_SIZE_T > 4
# define W_SIZE(n, p) do { \
if ((n) > SIZE32_MAX) { \
(p)->depth--; \
(p)->error = WFERR_UNMARSHALLABLE; \
return; \
} \
w_long((long)(n), p); \
} while(0)
#else
# define W_SIZE w_long
#endif
static void
w_pstring(const void *s, Py_ssize_t n, WFILE *p)
{
W_SIZE(n, p);
w_string(s, n, p);
}
static void
w_short_pstring(const void *s, Py_ssize_t n, WFILE *p)
{
w_byte(Py_SAFE_DOWNCAST(n, Py_ssize_t, unsigned char), p);
w_string(s, n, p);
}
/* We assume that Python ints are stored internally in base some power of
2**15; for the sake of portability we'll always read and write them in base
exactly 2**15. */
#define PyLong_MARSHAL_SHIFT 15
#define PyLong_MARSHAL_BASE ((short)1 << PyLong_MARSHAL_SHIFT)
#define PyLong_MARSHAL_MASK (PyLong_MARSHAL_BASE - 1)
#if PyLong_SHIFT % PyLong_MARSHAL_SHIFT != 0
#error "PyLong_SHIFT must be a multiple of PyLong_MARSHAL_SHIFT"
#endif
#define PyLong_MARSHAL_RATIO (PyLong_SHIFT / PyLong_MARSHAL_SHIFT)
#define W_TYPE(t, p) do { \
w_byte((t) | flag, (p)); \
} while(0)
static void
w_PyLong(const PyLongObject *ob, char flag, WFILE *p)
{
Py_ssize_t i, j, n, l;
digit d;
W_TYPE(TYPE_LONG, p);
if (Py_SIZE(ob) == 0) {
w_long((long)0, p);
return;
}
/* set l to number of base PyLong_MARSHAL_BASE digits */
n = Py_ABS(Py_SIZE(ob));
l = (n-1) * PyLong_MARSHAL_RATIO;
d = ob->ob_digit[n-1];
assert(d != 0); /* a PyLong is always normalized */
do {
d >>= PyLong_MARSHAL_SHIFT;
l++;
} while (d != 0);
if (l > SIZE32_MAX) {
p->depth--;
p->error = WFERR_UNMARSHALLABLE;
return;
}
w_long((long)(Py_SIZE(ob) > 0 ? l : -l), p);
for (i=0; i < n-1; i++) {
d = ob->ob_digit[i];
for (j=0; j < PyLong_MARSHAL_RATIO; j++) {
w_short(d & PyLong_MARSHAL_MASK, p);
d >>= PyLong_MARSHAL_SHIFT;
}
assert (d == 0);
}
d = ob->ob_digit[n-1];
do {
w_short(d & PyLong_MARSHAL_MASK, p);
d >>= PyLong_MARSHAL_SHIFT;
} while (d != 0);
}
static void
w_float_bin(double v, WFILE *p)
{
char buf[8];
if (PyFloat_Pack8(v, buf, 1) < 0) {
p->error = WFERR_UNMARSHALLABLE;
return;
}
w_string(buf, 8, p);
}
static void
w_float_str(double v, WFILE *p)
{
char *buf = PyOS_double_to_string(v, 'g', 17, 0, NULL);
if (!buf) {
p->error = WFERR_NOMEMORY;
return;
}
w_short_pstring(buf, strlen(buf), p);
PyMem_Free(buf);
}
static int
w_ref(PyObject *v, char *flag, WFILE *p)
{
_Py_hashtable_entry_t *entry;
int w;
if (p->version < 3 || p->hashtable == NULL)
return 0; /* not writing object references */
/* If it has only one reference, it definitely isn't shared.
* But we use TYPE_REF always for interned string, to PYC file stable
* as possible.
*/
if (Py_REFCNT(v) == 1 &&
!(PyUnicode_CheckExact(v) && PyUnicode_CHECK_INTERNED(v))) {
return 0;
}
entry = _Py_hashtable_get_entry(p->hashtable, v);
if (entry != NULL) {
/* write the reference index to the stream */
w = (int)(uintptr_t)entry->value;
/* we don't store "long" indices in the dict */
assert(0 <= w && w <= 0x7fffffff);
w_byte(TYPE_REF, p);
w_long(w, p);
return 1;
} else {
size_t s = p->hashtable->nentries;
/* we don't support long indices */
if (s >= 0x7fffffff) {
PyErr_SetString(PyExc_ValueError, "too many objects");
goto err;
}
w = (int)s;
Py_INCREF(v);
if (_Py_hashtable_set(p->hashtable, v, (void *)(uintptr_t)w) < 0) {
Py_DECREF(v);
goto err;
}
*flag |= FLAG_REF;
return 0;
}
err:
p->error = WFERR_UNMARSHALLABLE;
return 1;
}
static void
w_complex_object(PyObject *v, char flag, WFILE *p);
static void
w_object(PyObject *v, WFILE *p)
1991-06-05 04:42:30 +09:00
{
char flag = '\0';
p->depth++;
if (p->depth > MAX_MARSHAL_STACK_DEPTH) {
p->error = WFERR_NESTEDTOODEEP;
}
else if (v == NULL) {
w_byte(TYPE_NULL, p);
}
else if (v == Py_None) {
w_byte(TYPE_NONE, p);
}
else if (v == PyExc_StopIteration) {
w_byte(TYPE_STOPITER, p);
}
else if (v == Py_Ellipsis) {
w_byte(TYPE_ELLIPSIS, p);
}
else if (v == Py_False) {
w_byte(TYPE_FALSE, p);
}
else if (v == Py_True) {
w_byte(TYPE_TRUE, p);
}
else if (!w_ref(v, &flag, p))
w_complex_object(v, flag, p);
p->depth--;
}
static void
w_complex_object(PyObject *v, char flag, WFILE *p)
{
Py_ssize_t i, n;
if (PyLong_CheckExact(v)) {
int overflow;
long x = PyLong_AsLongAndOverflow(v, &overflow);
if (overflow) {
w_PyLong((PyLongObject *)v, flag, p);
}
else {
#if SIZEOF_LONG > 4
long y = Py_ARITHMETIC_RIGHT_SHIFT(long, x, 31);
if (y && y != -1) {
/* Too large for TYPE_INT */
w_PyLong((PyLongObject*)v, flag, p);
}
else
#endif
{
W_TYPE(TYPE_INT, p);
w_long(x, p);
}
}
}
else if (PyFloat_CheckExact(v)) {
if (p->version > 1) {
W_TYPE(TYPE_BINARY_FLOAT, p);
w_float_bin(PyFloat_AS_DOUBLE(v), p);
}
else {
W_TYPE(TYPE_FLOAT, p);
w_float_str(PyFloat_AS_DOUBLE(v), p);
}
}
else if (PyComplex_CheckExact(v)) {
if (p->version > 1) {
W_TYPE(TYPE_BINARY_COMPLEX, p);
w_float_bin(PyComplex_RealAsDouble(v), p);
w_float_bin(PyComplex_ImagAsDouble(v), p);
}
else {
W_TYPE(TYPE_COMPLEX, p);
w_float_str(PyComplex_RealAsDouble(v), p);
w_float_str(PyComplex_ImagAsDouble(v), p);
}
}
else if (PyBytes_CheckExact(v)) {
W_TYPE(TYPE_STRING, p);
w_pstring(PyBytes_AS_STRING(v), PyBytes_GET_SIZE(v), p);
}
else if (PyUnicode_CheckExact(v)) {
if (p->version >= 4 && PyUnicode_IS_ASCII(v)) {
int is_short = PyUnicode_GET_LENGTH(v) < 256;
if (is_short) {
if (PyUnicode_CHECK_INTERNED(v))
W_TYPE(TYPE_SHORT_ASCII_INTERNED, p);
else
W_TYPE(TYPE_SHORT_ASCII, p);
w_short_pstring(PyUnicode_1BYTE_DATA(v),
PyUnicode_GET_LENGTH(v), p);
}
else {
if (PyUnicode_CHECK_INTERNED(v))
W_TYPE(TYPE_ASCII_INTERNED, p);
else
W_TYPE(TYPE_ASCII, p);
w_pstring(PyUnicode_1BYTE_DATA(v),
PyUnicode_GET_LENGTH(v), p);
}
}
else {
PyObject *utf8;
utf8 = PyUnicode_AsEncodedString(v, "utf8", "surrogatepass");
if (utf8 == NULL) {
p->depth--;
p->error = WFERR_UNMARSHALLABLE;
return;
}
if (p->version >= 3 && PyUnicode_CHECK_INTERNED(v))
W_TYPE(TYPE_INTERNED, p);
else
W_TYPE(TYPE_UNICODE, p);
w_pstring(PyBytes_AS_STRING(utf8), PyBytes_GET_SIZE(utf8), p);
Py_DECREF(utf8);
}
}
else if (PyTuple_CheckExact(v)) {
n = PyTuple_GET_SIZE(v);
if (p->version >= 4 && n < 256) {
W_TYPE(TYPE_SMALL_TUPLE, p);
w_byte((unsigned char)n, p);
}
else {
W_TYPE(TYPE_TUPLE, p);
W_SIZE(n, p);
}
for (i = 0; i < n; i++) {
w_object(PyTuple_GET_ITEM(v, i), p);
}
}
else if (PyList_CheckExact(v)) {
W_TYPE(TYPE_LIST, p);
n = PyList_GET_SIZE(v);
W_SIZE(n, p);
for (i = 0; i < n; i++) {
w_object(PyList_GET_ITEM(v, i), p);
}
}
else if (PyDict_CheckExact(v)) {
Py_ssize_t pos;
PyObject *key, *value;
W_TYPE(TYPE_DICT, p);
/* This one is NULL object terminated! */
pos = 0;
while (PyDict_Next(v, &pos, &key, &value)) {
w_object(key, p);
w_object(value, p);
}
w_object((PyObject *)NULL, p);
}
else if (PyAnySet_CheckExact(v)) {
PyObject *value;
Py_ssize_t pos = 0;
Py_hash_t hash;
if (PyFrozenSet_CheckExact(v))
W_TYPE(TYPE_FROZENSET, p);
else
W_TYPE(TYPE_SET, p);
n = PySet_GET_SIZE(v);
W_SIZE(n, p);
// bpo-37596: To support reproducible builds, sets and frozensets need
// to have their elements serialized in a consistent order (even when
// they have been scrambled by hash randomization). To ensure this, we
// use an order equivalent to sorted(v, key=marshal.dumps):
PyObject *pairs = PyList_New(n);
if (pairs == NULL) {
p->error = WFERR_NOMEMORY;
return;
}
Py_ssize_t i = 0;
while (_PySet_NextEntry(v, &pos, &value, &hash)) {
PyObject *dump = PyMarshal_WriteObjectToString(value, p->version);
if (dump == NULL) {
p->error = WFERR_UNMARSHALLABLE;
Py_DECREF(pairs);
return;
}
PyObject *pair = PyTuple_Pack(2, dump, value);
Py_DECREF(dump);
if (pair == NULL) {
p->error = WFERR_NOMEMORY;
Py_DECREF(pairs);
return;
}
PyList_SET_ITEM(pairs, i++, pair);
}
assert(i == n);
if (PyList_Sort(pairs)) {
p->error = WFERR_NOMEMORY;
Py_DECREF(pairs);
return;
}
for (Py_ssize_t i = 0; i < n; i++) {
PyObject *pair = PyList_GET_ITEM(pairs, i);
value = PyTuple_GET_ITEM(pair, 1);
w_object(value, p);
}
Py_DECREF(pairs);
}
else if (PyCode_Check(v)) {
PyCodeObject *co = (PyCodeObject *)v;
PyObject *co_code = _PyCode_GetCode(co);
if (co_code == NULL) {
p->error = WFERR_NOMEMORY;
return;
}
W_TYPE(TYPE_CODE, p);
w_long(co->co_argcount, p);
w_long(co->co_posonlyargcount, p);
w_long(co->co_kwonlyargcount, p);
w_long(co->co_stacksize, p);
w_long(co->co_flags, p);
w_object(co_code, p);
w_object(co->co_consts, p);
w_object(co->co_names, p);
w_object(co->co_localsplusnames, p);
w_object(co->co_localspluskinds, p);
w_object(co->co_filename, p);
w_object(co->co_name, p);
w_object(co->co_qualname, p);
w_long(co->co_firstlineno, p);
w_object(co->co_linetable, p);
w_object(co->co_exceptiontable, p);
Py_DECREF(co_code);
}
else if (PyObject_CheckBuffer(v)) {
/* Write unknown bytes-like objects as a bytes object */
Py_buffer view;
if (PyObject_GetBuffer(v, &view, PyBUF_SIMPLE) != 0) {
w_byte(TYPE_UNKNOWN, p);
p->depth--;
p->error = WFERR_UNMARSHALLABLE;
return;
}
W_TYPE(TYPE_STRING, p);
w_pstring(view.buf, view.len, p);
PyBuffer_Release(&view);
}
else {
W_TYPE(TYPE_UNKNOWN, p);
p->error = WFERR_UNMARSHALLABLE;
}
1991-06-05 04:42:30 +09:00
}
static void
w_decref_entry(void *key)
{
PyObject *entry_key = (PyObject *)key;
Py_XDECREF(entry_key);
}
static int
w_init_refs(WFILE *wf, int version)
{
if (version >= 3) {
wf->hashtable = _Py_hashtable_new_full(_Py_hashtable_hash_ptr,
_Py_hashtable_compare_direct,
w_decref_entry, NULL, NULL);
if (wf->hashtable == NULL) {
PyErr_NoMemory();
return -1;
}
}
return 0;
}
static void
w_clear_refs(WFILE *wf)
{
if (wf->hashtable != NULL) {
_Py_hashtable_destroy(wf->hashtable);
}
}
/* version currently has no effect for writing ints. */
void
PyMarshal_WriteLongToFile(long x, FILE *fp, int version)
{
char buf[4];
WFILE wf;
memset(&wf, 0, sizeof(wf));
wf.fp = fp;
wf.ptr = wf.buf = buf;
wf.end = wf.ptr + sizeof(buf);
wf.error = WFERR_OK;
wf.version = version;
w_long(x, &wf);
w_flush(&wf);
}
1991-06-05 04:42:30 +09:00
void
PyMarshal_WriteObjectToFile(PyObject *x, FILE *fp, int version)
{
char buf[BUFSIZ];
WFILE wf;
if (PySys_Audit("marshal.dumps", "Oi", x, version) < 0) {
return; /* caller must check PyErr_Occurred() */
}
memset(&wf, 0, sizeof(wf));
wf.fp = fp;
wf.ptr = wf.buf = buf;
wf.end = wf.ptr + sizeof(buf);
wf.error = WFERR_OK;
wf.version = version;
if (w_init_refs(&wf, version)) {
return; /* caller must check PyErr_Occurred() */
}
w_object(x, &wf);
w_clear_refs(&wf);
w_flush(&wf);
}
typedef struct {
FILE *fp;
int depth;
PyObject *readable; /* Stream-like object being read from */
const char *ptr;
const char *end;
char *buf;
Py_ssize_t buf_size;
PyObject *refs; /* a list */
} RFILE;
static const char *
r_string(Py_ssize_t n, RFILE *p)
{
Py_ssize_t read = -1;
if (p->ptr != NULL) {
/* Fast path for loads() */
const char *res = p->ptr;
Py_ssize_t left = p->end - p->ptr;
if (left < n) {
PyErr_SetString(PyExc_EOFError,
"marshal data too short");
return NULL;
}
p->ptr += n;
return res;
}
if (p->buf == NULL) {
p->buf = PyMem_Malloc(n);
if (p->buf == NULL) {
PyErr_NoMemory();
return NULL;
}
p->buf_size = n;
}
else if (p->buf_size < n) {
char *tmp = PyMem_Realloc(p->buf, n);
if (tmp == NULL) {
PyErr_NoMemory();
return NULL;
}
p->buf = tmp;
p->buf_size = n;
}
if (!p->readable) {
assert(p->fp != NULL);
read = fread(p->buf, 1, n, p->fp);
}
else {
PyObject *res, *mview;
Py_buffer buf;
if (PyBuffer_FillInfo(&buf, NULL, p->buf, n, 0, PyBUF_CONTIG) == -1)
return NULL;
mview = PyMemoryView_FromBuffer(&buf);
if (mview == NULL)
return NULL;
bpo-46541: Replace core use of _Py_IDENTIFIER() with statically initialized global objects. (gh-30928) We're no longer using _Py_IDENTIFIER() (or _Py_static_string()) in any core CPython code. It is still used in a number of non-builtin stdlib modules. The replacement is: PyUnicodeObject (not pointer) fields under _PyRuntimeState, statically initialized as part of _PyRuntime. A new _Py_GET_GLOBAL_IDENTIFIER() macro facilitates lookup of the fields (along with _Py_GET_GLOBAL_STRING() for non-identifier strings). https://bugs.python.org/issue46541#msg411799 explains the rationale for this change. The core of the change is in: * (new) Include/internal/pycore_global_strings.h - the declarations for the global strings, along with the macros * Include/internal/pycore_runtime_init.h - added the static initializers for the global strings * Include/internal/pycore_global_objects.h - where the struct in pycore_global_strings.h is hooked into _PyRuntimeState * Tools/scripts/generate_global_objects.py - added generation of the global string declarations and static initializers I've also added a --check flag to generate_global_objects.py (along with make check-global-objects) to check for unused global strings. That check is added to the PR CI config. The remainder of this change updates the core code to use _Py_GET_GLOBAL_IDENTIFIER() instead of _Py_IDENTIFIER() and the related _Py*Id functions (likewise for _Py_GET_GLOBAL_STRING() instead of _Py_static_string()). This includes adding a few functions where there wasn't already an alternative to _Py*Id(), replacing the _Py_Identifier * parameter with PyObject *. The following are not changed (yet): * stop using _Py_IDENTIFIER() in the stdlib modules * (maybe) get rid of _Py_IDENTIFIER(), etc. entirely -- this may not be doable as at least one package on PyPI using this (private) API * (maybe) intern the strings during runtime init https://bugs.python.org/issue46541
2022-02-09 04:39:07 +08:00
res = _PyObject_CallMethod(p->readable, &_Py_ID(readinto), "N", mview);
if (res != NULL) {
read = PyNumber_AsSsize_t(res, PyExc_ValueError);
Py_DECREF(res);
}
}
if (read != n) {
if (!PyErr_Occurred()) {
if (read > n)
PyErr_Format(PyExc_ValueError,
"read() returned too much data: "
"%zd bytes requested, %zd returned",
n, read);
else
PyErr_SetString(PyExc_EOFError,
"EOF read where not expected");
}
return NULL;
}
return p->buf;
}
static int
r_byte(RFILE *p)
{
int c = EOF;
if (p->ptr != NULL) {
if (p->ptr < p->end)
c = (unsigned char) *p->ptr++;
return c;
}
if (!p->readable) {
assert(p->fp);
c = getc(p->fp);
}
else {
const char *ptr = r_string(1, p);
if (ptr != NULL)
c = *(const unsigned char *) ptr;
}
return c;
}
static int
r_short(RFILE *p)
1991-06-05 04:42:30 +09:00
{
short x = -1;
const unsigned char *buffer;
buffer = (const unsigned char *) r_string(2, p);
if (buffer != NULL) {
x = buffer[0];
x |= buffer[1] << 8;
/* Sign-extension, in case short greater than 16 bits */
x |= -(x & 0x8000);
}
return x;
1991-06-05 04:42:30 +09:00
}
static long
r_long(RFILE *p)
1991-06-05 04:42:30 +09:00
{
long x = -1;
const unsigned char *buffer;
buffer = (const unsigned char *) r_string(4, p);
if (buffer != NULL) {
x = buffer[0];
x |= (long)buffer[1] << 8;
x |= (long)buffer[2] << 16;
x |= (long)buffer[3] << 24;
#if SIZEOF_LONG > 4
/* Sign extension for 64-bit machines */
x |= -(x & 0x80000000L);
#endif
}
return x;
}
/* r_long64 deals with the TYPE_INT64 code. */
static PyObject *
r_long64(RFILE *p)
{
const unsigned char *buffer = (const unsigned char *) r_string(8, p);
if (buffer == NULL) {
return NULL;
}
return _PyLong_FromByteArray(buffer, 8,
1 /* little endian */,
1 /* signed */);
}
static PyObject *
r_PyLong(RFILE *p)
{
PyLongObject *ob;
long n, size, i;
int j, md, shorts_in_top_digit;
digit d;
n = r_long(p);
if (PyErr_Occurred())
return NULL;
if (n == 0)
return (PyObject *)_PyLong_New(0);
if (n < -SIZE32_MAX || n > SIZE32_MAX) {
PyErr_SetString(PyExc_ValueError,
"bad marshal data (long size out of range)");
return NULL;
}
size = 1 + (Py_ABS(n) - 1) / PyLong_MARSHAL_RATIO;
shorts_in_top_digit = 1 + (Py_ABS(n) - 1) % PyLong_MARSHAL_RATIO;
ob = _PyLong_New(size);
if (ob == NULL)
return NULL;
Py_SET_SIZE(ob, n > 0 ? size : -size);
for (i = 0; i < size-1; i++) {
d = 0;
for (j=0; j < PyLong_MARSHAL_RATIO; j++) {
md = r_short(p);
if (PyErr_Occurred()) {
Py_DECREF(ob);
return NULL;
}
if (md < 0 || md > PyLong_MARSHAL_BASE)
goto bad_digit;
d += (digit)md << j*PyLong_MARSHAL_SHIFT;
}
ob->ob_digit[i] = d;
}
d = 0;
for (j=0; j < shorts_in_top_digit; j++) {
md = r_short(p);
if (PyErr_Occurred()) {
Py_DECREF(ob);
return NULL;
}
if (md < 0 || md > PyLong_MARSHAL_BASE)
goto bad_digit;
/* topmost marshal digit should be nonzero */
if (md == 0 && j == shorts_in_top_digit - 1) {
Py_DECREF(ob);
PyErr_SetString(PyExc_ValueError,
"bad marshal data (unnormalized long data)");
return NULL;
}
d += (digit)md << j*PyLong_MARSHAL_SHIFT;
}
if (PyErr_Occurred()) {
Py_DECREF(ob);
return NULL;
}
/* top digit should be nonzero, else the resulting PyLong won't be
normalized */
ob->ob_digit[size-1] = d;
return (PyObject *)ob;
bad_digit:
Py_DECREF(ob);
PyErr_SetString(PyExc_ValueError,
"bad marshal data (digit out of range in long)");
return NULL;
}
static double
r_float_bin(RFILE *p)
{
const char *buf = r_string(8, p);
if (buf == NULL)
return -1;
return PyFloat_Unpack8(buf, 1);
}
/* Issue #33720: Disable inlining for reducing the C stack consumption
on PGO builds. */
Py_NO_INLINE static double
r_float_str(RFILE *p)
{
int n;
char buf[256];
const char *ptr;
n = r_byte(p);
if (n == EOF) {
PyErr_SetString(PyExc_EOFError,
"EOF read where object expected");
return -1;
}
ptr = r_string(n, p);
if (ptr == NULL) {
return -1;
}
memcpy(buf, ptr, n);
buf[n] = '\0';
return PyOS_string_to_double(buf, NULL, NULL);
}
/* allocate the reflist index for a new object. Return -1 on failure */
static Py_ssize_t
r_ref_reserve(int flag, RFILE *p)
{
if (flag) { /* currently only FLAG_REF is defined */
Py_ssize_t idx = PyList_GET_SIZE(p->refs);
if (idx >= 0x7ffffffe) {
PyErr_SetString(PyExc_ValueError, "bad marshal data (index list too large)");
return -1;
}
if (PyList_Append(p->refs, Py_None) < 0)
return -1;
return idx;
} else
return 0;
}
/* insert the new object 'o' to the reflist at previously
* allocated index 'idx'.
* 'o' can be NULL, in which case nothing is done.
* if 'o' was non-NULL, and the function succeeds, 'o' is returned.
* if 'o' was non-NULL, and the function fails, 'o' is released and
* NULL returned. This simplifies error checking at the call site since
* a single test for NULL for the function result is enough.
*/
static PyObject *
r_ref_insert(PyObject *o, Py_ssize_t idx, int flag, RFILE *p)
{
if (o != NULL && flag) { /* currently only FLAG_REF is defined */
PyObject *tmp = PyList_GET_ITEM(p->refs, idx);
PyList_SET_ITEM(p->refs, idx, Py_NewRef(o));
Py_DECREF(tmp);
}
return o;
}
/* combination of both above, used when an object can be
* created whenever it is seen in the file, as opposed to
* after having loaded its sub-objects.
*/
static PyObject *
r_ref(PyObject *o, int flag, RFILE *p)
{
assert(flag & FLAG_REF);
if (o == NULL)
return NULL;
if (PyList_Append(p->refs, o) < 0) {
Py_DECREF(o); /* release the new object */
return NULL;
}
return o;
}
static PyObject *
r_object(RFILE *p)
1991-06-05 04:42:30 +09:00
{
/* NULL is a valid return value, it does not necessarily means that
an exception is set. */
PyObject *v, *v2;
2013-03-20 13:40:07 +08:00
Py_ssize_t idx = 0;
long i, n;
int type, code = r_byte(p);
int flag, is_interned = 0;
PyObject *retval = NULL;
if (code == EOF) {
PyErr_SetString(PyExc_EOFError,
"EOF read where object expected");
return NULL;
}
p->depth++;
if (p->depth > MAX_MARSHAL_STACK_DEPTH) {
p->depth--;
PyErr_SetString(PyExc_ValueError, "recursion limit exceeded");
return NULL;
}
flag = code & FLAG_REF;
type = code & ~FLAG_REF;
#define R_REF(O) do{\
if (flag) \
O = r_ref(O, flag, p);\
} while (0)
switch (type) {
case TYPE_NULL:
break;
case TYPE_NONE:
retval = Py_NewRef(Py_None);
break;
case TYPE_STOPITER:
retval = Py_NewRef(PyExc_StopIteration);
break;
case TYPE_ELLIPSIS:
retval = Py_NewRef(Py_Ellipsis);
break;
case TYPE_FALSE:
retval = Py_NewRef(Py_False);
break;
case TYPE_TRUE:
retval = Py_NewRef(Py_True);
break;
case TYPE_INT:
n = r_long(p);
retval = PyErr_Occurred() ? NULL : PyLong_FromLong(n);
R_REF(retval);
break;
case TYPE_INT64:
retval = r_long64(p);
R_REF(retval);
break;
case TYPE_LONG:
retval = r_PyLong(p);
R_REF(retval);
break;
case TYPE_FLOAT:
{
double x = r_float_str(p);
if (x == -1.0 && PyErr_Occurred())
break;
retval = PyFloat_FromDouble(x);
R_REF(retval);
break;
}
case TYPE_BINARY_FLOAT:
{
double x = r_float_bin(p);
if (x == -1.0 && PyErr_Occurred())
break;
retval = PyFloat_FromDouble(x);
R_REF(retval);
break;
}
case TYPE_COMPLEX:
{
Py_complex c;
c.real = r_float_str(p);
if (c.real == -1.0 && PyErr_Occurred())
break;
c.imag = r_float_str(p);
if (c.imag == -1.0 && PyErr_Occurred())
break;
retval = PyComplex_FromCComplex(c);
R_REF(retval);
break;
}
case TYPE_BINARY_COMPLEX:
{
Py_complex c;
c.real = r_float_bin(p);
if (c.real == -1.0 && PyErr_Occurred())
break;
c.imag = r_float_bin(p);
if (c.imag == -1.0 && PyErr_Occurred())
break;
retval = PyComplex_FromCComplex(c);
R_REF(retval);
break;
}
case TYPE_STRING:
{
const char *ptr;
n = r_long(p);
if (PyErr_Occurred())
break;
if (n < 0 || n > SIZE32_MAX) {
PyErr_SetString(PyExc_ValueError, "bad marshal data (bytes object size out of range)");
break;
}
v = PyBytes_FromStringAndSize((char *)NULL, n);
if (v == NULL)
break;
ptr = r_string(n, p);
if (ptr == NULL) {
Py_DECREF(v);
break;
}
memcpy(PyBytes_AS_STRING(v), ptr, n);
retval = v;
R_REF(retval);
break;
}
case TYPE_ASCII_INTERNED:
is_interned = 1;
/* fall through */
case TYPE_ASCII:
n = r_long(p);
if (PyErr_Occurred())
break;
if (n < 0 || n > SIZE32_MAX) {
PyErr_SetString(PyExc_ValueError, "bad marshal data (string size out of range)");
break;
}
goto _read_ascii;
case TYPE_SHORT_ASCII_INTERNED:
is_interned = 1;
/* fall through */
case TYPE_SHORT_ASCII:
n = r_byte(p);
if (n == EOF) {
PyErr_SetString(PyExc_EOFError,
"EOF read where object expected");
break;
}
_read_ascii:
{
const char *ptr;
ptr = r_string(n, p);
if (ptr == NULL)
break;
v = PyUnicode_FromKindAndData(PyUnicode_1BYTE_KIND, ptr, n);
if (v == NULL)
break;
if (is_interned)
PyUnicode_InternInPlace(&v);
retval = v;
R_REF(retval);
break;
}
case TYPE_INTERNED:
is_interned = 1;
/* fall through */
case TYPE_UNICODE:
{
const char *buffer;
n = r_long(p);
if (PyErr_Occurred())
break;
if (n < 0 || n > SIZE32_MAX) {
PyErr_SetString(PyExc_ValueError, "bad marshal data (string size out of range)");
break;
}
if (n != 0) {
buffer = r_string(n, p);
if (buffer == NULL)
break;
v = PyUnicode_DecodeUTF8(buffer, n, "surrogatepass");
}
else {
v = PyUnicode_New(0, 0);
}
if (v == NULL)
break;
if (is_interned)
PyUnicode_InternInPlace(&v);
retval = v;
R_REF(retval);
break;
}
case TYPE_SMALL_TUPLE:
n = (unsigned char) r_byte(p);
if (PyErr_Occurred())
break;
goto _read_tuple;
case TYPE_TUPLE:
n = r_long(p);
if (PyErr_Occurred())
break;
if (n < 0 || n > SIZE32_MAX) {
PyErr_SetString(PyExc_ValueError, "bad marshal data (tuple size out of range)");
break;
}
_read_tuple:
v = PyTuple_New(n);
R_REF(v);
if (v == NULL)
break;
for (i = 0; i < n; i++) {
v2 = r_object(p);
if ( v2 == NULL ) {
if (!PyErr_Occurred())
PyErr_SetString(PyExc_TypeError,
"NULL object in marshal data for tuple");
Py_DECREF(v);
v = NULL;
break;
}
PyTuple_SET_ITEM(v, i, v2);
}
retval = v;
break;
case TYPE_LIST:
n = r_long(p);
if (PyErr_Occurred())
break;
if (n < 0 || n > SIZE32_MAX) {
PyErr_SetString(PyExc_ValueError, "bad marshal data (list size out of range)");
break;
}
v = PyList_New(n);
R_REF(v);
if (v == NULL)
break;
for (i = 0; i < n; i++) {
v2 = r_object(p);
if ( v2 == NULL ) {
if (!PyErr_Occurred())
PyErr_SetString(PyExc_TypeError,
"NULL object in marshal data for list");
Py_DECREF(v);
v = NULL;
break;
}
PyList_SET_ITEM(v, i, v2);
}
retval = v;
break;
case TYPE_DICT:
v = PyDict_New();
R_REF(v);
if (v == NULL)
break;
for (;;) {
PyObject *key, *val;
key = r_object(p);
if (key == NULL)
break;
val = r_object(p);
if (val == NULL) {
Py_DECREF(key);
break;
}
if (PyDict_SetItem(v, key, val) < 0) {
Py_DECREF(key);
Py_DECREF(val);
break;
}
Py_DECREF(key);
Py_DECREF(val);
}
if (PyErr_Occurred()) {
Py_DECREF(v);
v = NULL;
}
retval = v;
break;
case TYPE_SET:
case TYPE_FROZENSET:
n = r_long(p);
if (PyErr_Occurred())
break;
if (n < 0 || n > SIZE32_MAX) {
PyErr_SetString(PyExc_ValueError, "bad marshal data (set size out of range)");
break;
}
if (n == 0 && type == TYPE_FROZENSET) {
/* call frozenset() to get the empty frozenset singleton */
v = _PyObject_CallNoArgs((PyObject*)&PyFrozenSet_Type);
if (v == NULL)
break;
R_REF(v);
retval = v;
}
else {
v = (type == TYPE_SET) ? PySet_New(NULL) : PyFrozenSet_New(NULL);
if (type == TYPE_SET) {
R_REF(v);
} else {
/* must use delayed registration of frozensets because they must
* be init with a refcount of 1
*/
idx = r_ref_reserve(flag, p);
if (idx < 0)
Py_CLEAR(v); /* signal error */
}
if (v == NULL)
break;
for (i = 0; i < n; i++) {
v2 = r_object(p);
if ( v2 == NULL ) {
if (!PyErr_Occurred())
PyErr_SetString(PyExc_TypeError,
"NULL object in marshal data for set");
Py_DECREF(v);
v = NULL;
break;
}
if (PySet_Add(v, v2) == -1) {
Py_DECREF(v);
Py_DECREF(v2);
v = NULL;
break;
}
Py_DECREF(v2);
}
if (type != TYPE_SET)
v = r_ref_insert(v, idx, flag, p);
retval = v;
}
break;
case TYPE_CODE:
{
int argcount;
int posonlyargcount;
int kwonlyargcount;
int stacksize;
int flags;
PyObject *code = NULL;
PyObject *consts = NULL;
PyObject *names = NULL;
PyObject *localsplusnames = NULL;
PyObject *localspluskinds = NULL;
PyObject *filename = NULL;
PyObject *name = NULL;
PyObject *qualname = NULL;
int firstlineno;
PyObject* linetable = NULL;
PyObject *exceptiontable = NULL;
idx = r_ref_reserve(flag, p);
if (idx < 0)
break;
v = NULL;
/* XXX ignore long->int overflows for now */
argcount = (int)r_long(p);
if (PyErr_Occurred())
goto code_error;
posonlyargcount = (int)r_long(p);
if (PyErr_Occurred()) {
goto code_error;
}
kwonlyargcount = (int)r_long(p);
if (PyErr_Occurred())
goto code_error;
stacksize = (int)r_long(p);
if (PyErr_Occurred())
goto code_error;
flags = (int)r_long(p);
if (PyErr_Occurred())
goto code_error;
code = r_object(p);
if (code == NULL)
goto code_error;
consts = r_object(p);
if (consts == NULL)
goto code_error;
names = r_object(p);
if (names == NULL)
goto code_error;
localsplusnames = r_object(p);
if (localsplusnames == NULL)
goto code_error;
localspluskinds = r_object(p);
if (localspluskinds == NULL)
goto code_error;
filename = r_object(p);
if (filename == NULL)
goto code_error;
name = r_object(p);
if (name == NULL)
goto code_error;
qualname = r_object(p);
if (qualname == NULL)
goto code_error;
firstlineno = (int)r_long(p);
if (firstlineno == -1 && PyErr_Occurred())
break;
linetable = r_object(p);
if (linetable == NULL)
goto code_error;
exceptiontable = r_object(p);
if (exceptiontable == NULL)
goto code_error;
struct _PyCodeConstructor con = {
.filename = filename,
.name = name,
.qualname = qualname,
.flags = flags,
.code = code,
.firstlineno = firstlineno,
.linetable = linetable,
.consts = consts,
.names = names,
.localsplusnames = localsplusnames,
.localspluskinds = localspluskinds,
.argcount = argcount,
.posonlyargcount = posonlyargcount,
.kwonlyargcount = kwonlyargcount,
.stacksize = stacksize,
.exceptiontable = exceptiontable,
};
if (_PyCode_Validate(&con) < 0) {
goto code_error;
}
v = (PyObject *)_PyCode_New(&con);
if (v == NULL) {
goto code_error;
}
v = r_ref_insert(v, idx, flag, p);
code_error:
Py_XDECREF(code);
Py_XDECREF(consts);
Py_XDECREF(names);
Py_XDECREF(localsplusnames);
Py_XDECREF(localspluskinds);
Py_XDECREF(filename);
Py_XDECREF(name);
Py_XDECREF(qualname);
Py_XDECREF(linetable);
Py_XDECREF(exceptiontable);
}
retval = v;
break;
case TYPE_REF:
n = r_long(p);
if (n < 0 || n >= PyList_GET_SIZE(p->refs)) {
if (n == -1 && PyErr_Occurred())
break;
PyErr_SetString(PyExc_ValueError, "bad marshal data (invalid reference)");
break;
}
v = PyList_GET_ITEM(p->refs, n);
if (v == Py_None) {
PyErr_SetString(PyExc_ValueError, "bad marshal data (invalid reference)");
break;
}
retval = Py_NewRef(v);
break;
default:
/* Bogus data got written, which isn't ideal.
This will let you keep working and recover. */
PyErr_SetString(PyExc_ValueError, "bad marshal data (unknown type code)");
break;
}
p->depth--;
return retval;
1991-06-05 04:42:30 +09:00
}
2004-06-14 04:31:49 +08:00
static PyObject *
2004-03-26 23:09:27 +08:00
read_object(RFILE *p)
{
PyObject *v;
if (PyErr_Occurred()) {
fprintf(stderr, "XXX readobject called with exception set\n");
return NULL;
}
if (p->ptr && p->end) {
if (PySys_Audit("marshal.loads", "y#", p->ptr, (Py_ssize_t)(p->end - p->ptr)) < 0) {
return NULL;
}
} else if (p->fp || p->readable) {
if (PySys_Audit("marshal.load", NULL) < 0) {
return NULL;
}
}
v = r_object(p);
if (v == NULL && !PyErr_Occurred())
PyErr_SetString(PyExc_TypeError, "NULL object in marshal data for object");
return v;
2004-03-26 23:09:27 +08:00
}
int
PyMarshal_ReadShortFromFile(FILE *fp)
{
RFILE rf;
int res;
assert(fp);
rf.readable = NULL;
rf.fp = fp;
rf.end = rf.ptr = NULL;
rf.buf = NULL;
res = r_short(&rf);
if (rf.buf != NULL)
PyMem_Free(rf.buf);
return res;
}
long
PyMarshal_ReadLongFromFile(FILE *fp)
{
RFILE rf;
long res;
rf.fp = fp;
rf.readable = NULL;
rf.ptr = rf.end = NULL;
rf.buf = NULL;
res = r_long(&rf);
if (rf.buf != NULL)
PyMem_Free(rf.buf);
return res;
}
/* Return size of file in bytes; < 0 if unknown or INT_MAX if too big */
static off_t
getfilesize(FILE *fp)
{
struct _Py_stat_struct st;
if (_Py_fstat_noraise(fileno(fp), &st) != 0)
return -1;
#if SIZEOF_OFF_T == 4
else if (st.st_size >= INT_MAX)
return (off_t)INT_MAX;
#endif
else
return (off_t)st.st_size;
}
/* If we can get the size of the file up-front, and it's reasonably small,
* read it in one gulp and delegate to ...FromString() instead. Much quicker
* than reading a byte at a time from file; speeds .pyc imports.
* CAUTION: since this may read the entire remainder of the file, don't
* call it unless you know you're done with the file.
*/
PyObject *
PyMarshal_ReadLastObjectFromFile(FILE *fp)
{
/* REASONABLE_FILE_LIMIT is by defn something big enough for Tkinter.pyc. */
#define REASONABLE_FILE_LIMIT (1L << 18)
off_t filesize;
filesize = getfilesize(fp);
if (filesize > 0 && filesize <= REASONABLE_FILE_LIMIT) {
char* pBuf = (char *)PyMem_Malloc(filesize);
if (pBuf != NULL) {
size_t n = fread(pBuf, 1, (size_t)filesize, fp);
PyObject* v = PyMarshal_ReadObjectFromString(pBuf, n);
PyMem_Free(pBuf);
return v;
}
}
/* We don't have fstat, or we do but the file is larger than
* REASONABLE_FILE_LIMIT or malloc failed -- read a byte at a time.
*/
return PyMarshal_ReadObjectFromFile(fp);
#undef REASONABLE_FILE_LIMIT
}
PyObject *
PyMarshal_ReadObjectFromFile(FILE *fp)
{
RFILE rf;
PyObject *result;
rf.fp = fp;
rf.readable = NULL;
rf.depth = 0;
rf.ptr = rf.end = NULL;
rf.buf = NULL;
rf.refs = PyList_New(0);
if (rf.refs == NULL)
return NULL;
result = read_object(&rf);
Py_DECREF(rf.refs);
if (rf.buf != NULL)
PyMem_Free(rf.buf);
return result;
}
PyObject *
PyMarshal_ReadObjectFromString(const char *str, Py_ssize_t len)
{
RFILE rf;
PyObject *result;
rf.fp = NULL;
rf.readable = NULL;
rf.ptr = str;
rf.end = str + len;
rf.buf = NULL;
rf.depth = 0;
rf.refs = PyList_New(0);
if (rf.refs == NULL)
return NULL;
result = read_object(&rf);
Py_DECREF(rf.refs);
if (rf.buf != NULL)
PyMem_Free(rf.buf);
return result;
}
PyObject *
PyMarshal_WriteObjectToString(PyObject *x, int version)
{
WFILE wf;
if (PySys_Audit("marshal.dumps", "Oi", x, version) < 0) {
return NULL;
}
memset(&wf, 0, sizeof(wf));
wf.str = PyBytes_FromStringAndSize((char *)NULL, 50);
if (wf.str == NULL)
return NULL;
wf.ptr = wf.buf = PyBytes_AS_STRING(wf.str);
wf.end = wf.ptr + PyBytes_GET_SIZE(wf.str);
wf.error = WFERR_OK;
wf.version = version;
if (w_init_refs(&wf, version)) {
Py_DECREF(wf.str);
return NULL;
}
w_object(x, &wf);
w_clear_refs(&wf);
if (wf.str != NULL) {
const char *base = PyBytes_AS_STRING(wf.str);
if (_PyBytes_Resize(&wf.str, (Py_ssize_t)(wf.ptr - base)) < 0)
return NULL;
}
if (wf.error != WFERR_OK) {
Py_XDECREF(wf.str);
if (wf.error == WFERR_NOMEMORY)
PyErr_NoMemory();
else
PyErr_SetString(PyExc_ValueError,
(wf.error==WFERR_UNMARSHALLABLE)?"unmarshallable object"
:"object too deeply nested to marshal");
return NULL;
}
2012-03-03 01:22:23 +08:00
return wf.str;
}
1991-06-07 22:58:22 +09:00
/* And an interface for Python programs... */
/*[clinic input]
marshal.dump
value: object
Must be a supported type.
file: object
Must be a writeable binary file.
version: int(c_default="Py_MARSHAL_VERSION") = version
Indicates the data format that dump should use.
/
Write the value on the open file.
If the value has (or contains an object that has) an unsupported type, a
ValueError exception is raised - but garbage data will also be written
to the file. The object will not be properly read back by load().
[clinic start generated code]*/
1991-06-05 04:42:30 +09:00
static PyObject *
marshal_dump_impl(PyObject *module, PyObject *value, PyObject *file,
int version)
/*[clinic end generated code: output=aaee62c7028a7cb2 input=6c7a3c23c6fef556]*/
1991-06-05 04:42:30 +09:00
{
/* XXX Quick hack -- need to do this differently */
PyObject *s;
PyObject *res;
s = PyMarshal_WriteObjectToString(value, version);
if (s == NULL)
return NULL;
bpo-46541: Replace core use of _Py_IDENTIFIER() with statically initialized global objects. (gh-30928) We're no longer using _Py_IDENTIFIER() (or _Py_static_string()) in any core CPython code. It is still used in a number of non-builtin stdlib modules. The replacement is: PyUnicodeObject (not pointer) fields under _PyRuntimeState, statically initialized as part of _PyRuntime. A new _Py_GET_GLOBAL_IDENTIFIER() macro facilitates lookup of the fields (along with _Py_GET_GLOBAL_STRING() for non-identifier strings). https://bugs.python.org/issue46541#msg411799 explains the rationale for this change. The core of the change is in: * (new) Include/internal/pycore_global_strings.h - the declarations for the global strings, along with the macros * Include/internal/pycore_runtime_init.h - added the static initializers for the global strings * Include/internal/pycore_global_objects.h - where the struct in pycore_global_strings.h is hooked into _PyRuntimeState * Tools/scripts/generate_global_objects.py - added generation of the global string declarations and static initializers I've also added a --check flag to generate_global_objects.py (along with make check-global-objects) to check for unused global strings. That check is added to the PR CI config. The remainder of this change updates the core code to use _Py_GET_GLOBAL_IDENTIFIER() instead of _Py_IDENTIFIER() and the related _Py*Id functions (likewise for _Py_GET_GLOBAL_STRING() instead of _Py_static_string()). This includes adding a few functions where there wasn't already an alternative to _Py*Id(), replacing the _Py_Identifier * parameter with PyObject *. The following are not changed (yet): * stop using _Py_IDENTIFIER() in the stdlib modules * (maybe) get rid of _Py_IDENTIFIER(), etc. entirely -- this may not be doable as at least one package on PyPI using this (private) API * (maybe) intern the strings during runtime init https://bugs.python.org/issue46541
2022-02-09 04:39:07 +08:00
res = _PyObject_CallMethodOneArg(file, &_Py_ID(write), s);
Py_DECREF(s);
return res;
1991-06-05 04:42:30 +09:00
}
/*[clinic input]
marshal.load
file: object
Must be readable binary file.
/
Read one value from the open file and return it.
If no valid value is read (e.g. because the data has a different Python
version's incompatible marshal format), raise EOFError, ValueError or
TypeError.
Note: If an object containing an unsupported type was marshalled with
dump(), load() will substitute None for the unmarshallable type.
[clinic start generated code]*/
static PyObject *
marshal_load(PyObject *module, PyObject *file)
/*[clinic end generated code: output=f8e5c33233566344 input=c85c2b594cd8124a]*/
1991-06-05 04:42:30 +09:00
{
PyObject *data, *result;
RFILE rf;
/*
* Make a call to the read method, but read zero bytes.
* This is to ensure that the object passed in at least
* has a read method which returns bytes.
* This can be removed if we guarantee good error handling
* for r_string()
*/
bpo-46541: Replace core use of _Py_IDENTIFIER() with statically initialized global objects. (gh-30928) We're no longer using _Py_IDENTIFIER() (or _Py_static_string()) in any core CPython code. It is still used in a number of non-builtin stdlib modules. The replacement is: PyUnicodeObject (not pointer) fields under _PyRuntimeState, statically initialized as part of _PyRuntime. A new _Py_GET_GLOBAL_IDENTIFIER() macro facilitates lookup of the fields (along with _Py_GET_GLOBAL_STRING() for non-identifier strings). https://bugs.python.org/issue46541#msg411799 explains the rationale for this change. The core of the change is in: * (new) Include/internal/pycore_global_strings.h - the declarations for the global strings, along with the macros * Include/internal/pycore_runtime_init.h - added the static initializers for the global strings * Include/internal/pycore_global_objects.h - where the struct in pycore_global_strings.h is hooked into _PyRuntimeState * Tools/scripts/generate_global_objects.py - added generation of the global string declarations and static initializers I've also added a --check flag to generate_global_objects.py (along with make check-global-objects) to check for unused global strings. That check is added to the PR CI config. The remainder of this change updates the core code to use _Py_GET_GLOBAL_IDENTIFIER() instead of _Py_IDENTIFIER() and the related _Py*Id functions (likewise for _Py_GET_GLOBAL_STRING() instead of _Py_static_string()). This includes adding a few functions where there wasn't already an alternative to _Py*Id(), replacing the _Py_Identifier * parameter with PyObject *. The following are not changed (yet): * stop using _Py_IDENTIFIER() in the stdlib modules * (maybe) get rid of _Py_IDENTIFIER(), etc. entirely -- this may not be doable as at least one package on PyPI using this (private) API * (maybe) intern the strings during runtime init https://bugs.python.org/issue46541
2022-02-09 04:39:07 +08:00
data = _PyObject_CallMethod(file, &_Py_ID(read), "i", 0);
if (data == NULL)
return NULL;
if (!PyBytes_Check(data)) {
PyErr_Format(PyExc_TypeError,
"file.read() returned not bytes but %.100s",
Py_TYPE(data)->tp_name);
result = NULL;
}
else {
rf.depth = 0;
rf.fp = NULL;
rf.readable = file;
rf.ptr = rf.end = NULL;
rf.buf = NULL;
if ((rf.refs = PyList_New(0)) != NULL) {
result = read_object(&rf);
Py_DECREF(rf.refs);
if (rf.buf != NULL)
PyMem_Free(rf.buf);
} else
result = NULL;
}
Py_DECREF(data);
return result;
}
/*[clinic input]
marshal.dumps
value: object
Must be a supported type.
version: int(c_default="Py_MARSHAL_VERSION") = version
Indicates the data format that dumps should use.
/
Return the bytes object that would be written to a file by dump(value, file).
Raise a ValueError exception if value has (or contains an object that has) an
unsupported type.
[clinic start generated code]*/
static PyObject *
marshal_dumps_impl(PyObject *module, PyObject *value, int version)
/*[clinic end generated code: output=9c200f98d7256cad input=a2139ea8608e9b27]*/
{
return PyMarshal_WriteObjectToString(value, version);
}
/*[clinic input]
marshal.loads
bytes: Py_buffer
/
Convert the bytes-like object to a value.
If no valid value is found, raise EOFError, ValueError or TypeError. Extra
bytes in the input are ignored.
[clinic start generated code]*/
static PyObject *
marshal_loads_impl(PyObject *module, Py_buffer *bytes)
/*[clinic end generated code: output=9fc65985c93d1bb1 input=6f426518459c8495]*/
{
RFILE rf;
char *s = bytes->buf;
Py_ssize_t n = bytes->len;
PyObject* result;
rf.fp = NULL;
rf.readable = NULL;
rf.ptr = s;
rf.end = s + n;
rf.depth = 0;
if ((rf.refs = PyList_New(0)) == NULL)
return NULL;
result = read_object(&rf);
Py_DECREF(rf.refs);
return result;
1991-06-05 04:42:30 +09:00
}
static PyMethodDef marshal_methods[] = {
MARSHAL_DUMP_METHODDEF
MARSHAL_LOAD_METHODDEF
MARSHAL_DUMPS_METHODDEF
MARSHAL_LOADS_METHODDEF
{NULL, NULL} /* sentinel */
1991-06-05 04:42:30 +09:00
};
PyDoc_STRVAR(module_doc,
"This module contains functions that can read and write Python values in\n\
a binary format. The format is specific to Python, but independent of\n\
machine architecture issues.\n\
\n\
Not all Python object types are supported; in general, only objects\n\
whose value is independent from a particular invocation of Python can be\n\
written and read by this module. The following types are supported:\n\
None, integers, floating point numbers, strings, bytes, bytearrays,\n\
tuples, lists, sets, dictionaries, and code objects, where it\n\
should be understood that tuples, lists and dictionaries are only\n\
supported as long as the values contained therein are themselves\n\
supported; and recursive lists and dictionaries should not be written\n\
(they will cause infinite loops).\n\
\n\
Variables:\n\
\n\
version -- indicates the format that the module uses. Version 0 is the\n\
historical format, version 1 shares interned strings and version 2\n\
uses a binary format for floating point numbers.\n\
Version 3 shares common object references (New in version 3.4).\n\
\n\
Functions:\n\
\n\
dump() -- write value to a file\n\
load() -- read value from a file\n\
dumps() -- marshal value as a bytes object\n\
loads() -- read value from a bytes-like object");
static int
marshal_module_exec(PyObject *mod)
{
if (PyModule_AddIntConstant(mod, "version", Py_MARSHAL_VERSION) < 0) {
return -1;
}
return 0;
}
static PyModuleDef_Slot marshalmodule_slots[] = {
{Py_mod_exec, marshal_module_exec},
{0, NULL}
};
static struct PyModuleDef marshalmodule = {
PyModuleDef_HEAD_INIT,
.m_name = "marshal",
.m_doc = module_doc,
.m_methods = marshal_methods,
.m_slots = marshalmodule_slots,
};
PyMODINIT_FUNC
PyMarshal_Init(void)
1991-06-05 04:42:30 +09:00
{
return PyModuleDef_Init(&marshalmodule);
1991-06-05 04:42:30 +09:00
}