1 /* -*- c-file-style: "python"; indent-tabs-mode: nil; -*-
3 Python wrapper for Samba tdb pack/unpack functions
4 Copyright (C) Martin Pool 2002
7 NOTE PYTHON STYLE GUIDE
8 http://www.python.org/peps/pep-0007.html
11 This program is free software; you can redistribute it and/or modify
12 it under the terms of the GNU General Public License as published by
13 the Free Software Foundation; either version 2 of the License, or
14 (at your option) any later version.
16 This program is distributed in the hope that it will be useful,
17 but WITHOUT ANY WARRANTY; without even the implied warranty of
18 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 GNU General Public License for more details.
21 You should have received a copy of the GNU General Public License
22 along with this program; if not, write to the Free Software
23 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
30 static PyObject * pytdbpack_number(char ch, PyObject *val_iter, PyObject *packed_list);
31 static PyObject * pytdbpack_str_850(PyObject *val_iter, PyObject *packed_list);
32 static PyObject * pytdbpack_buffer(PyObject *val_iter, PyObject *packed_list);
34 static PyObject *pytdbunpack_item(char, char **pbuf, int *plen, PyObject *);
36 static PyObject *pytdbpack_data(const char *format_str,
40 static void pack_le_uint32(unsigned long val_long, unsigned char *pbuf);
43 static PyObject *pytdbpack_bad_type(char ch,
47 static const char * pytdbpack_docstring =
48 "Convert between Python values and Samba binary encodings.
50 This module is conceptually similar to the standard 'struct' module, but it
51 uses both a different binary format and a different description string.
53 Samba's encoding is based on that used inside DCE-RPC and SMB: a
54 little-endian, unpadded, non-self-describing binary format. It is intended
55 that these functions be as similar as possible to the routines in Samba's
56 tdb/tdbutil module, with appropriate adjustments for Python datatypes.
58 Python strings are used to specify the format of data to be packed or
61 Strings are always stored in codepage 850. Unicode objects are translated
62 to cp850; plain strings are assumed to be in latin-1 and are also
65 This may be a problem in the future if it is different to the Samba codepage.
66 It might be better to have the caller do the conversion, but that would conflict
67 with existing CMI code.
69 tdbpack format strings:
71 'f': NULL-terminated string in codepage 850
75 'd': 4 byte little-endian unsigned number
77 'w': 2 byte little-endian unsigned number
79 'P': \"Pointer\" value -- in the subset of DCERPC used by Samba, this is
80 really just an \"exists\" or \"does not exist\" flag. The boolean
81 value of the Python object is used.
83 'B': 4-byte LE length, followed by that many bytes of binary data.
84 Corresponds to a Python integer giving the length, followed by a byte
85 string of the appropriate length.
87 '$': Special flag indicating that the preceding format code should be
88 repeated while data remains. This is only supported for unpacking.
90 Every code corresponds to a single Python object, except 'B' which
91 corresponds to two values (length and contents), and '$', which produces
92 however many make sense.
96 static char const pytdbpack_doc[] =
97 "pack(format, values) -> buffer
98 Pack Python objects into Samba binary format according to format string.
101 format -- string of tdbpack format characters
102 values -- sequence of value objects corresponding 1:1 to format characters
105 buffer -- string containing packed data
108 IndexError -- if there are too few values for the format
109 ValueError -- if any of the format characters is illegal
110 TypeError -- if the format is not a string, or values is not a sequence,
111 or any of the values is of the wrong type for the corresponding
115 For historical reasons, it is not an error to pass more values than are consumed
120 static char const pytdbunpack_doc[] =
121 "unpack(format, buffer) -> (values, rest)
122 Unpack Samba binary data according to format string.
125 format -- string of tdbpack characters
126 buffer -- string of packed binary data
130 values -- sequence of values corresponding 1:1 to format characters
131 rest -- string containing data that was not decoded, or '' if the
132 whole string was consumed
135 IndexError -- if there is insufficient data in the buffer for the
136 format (or if the data is corrupt and contains a variable-length
137 field extending past the end)
138 ValueError -- if any of the format characters is illegal
141 Because unconsumed data is returned, you can feed it back in to the
142 unpacker to extract further fields. Alternatively, if you wish to modify
143 some fields near the start of the data, you may be able to save time by
144 only unpacking and repacking the necessary part.
148 const char *pytdb_string_encoding = "cp850";
152 * Pack objects to bytes.
154 * All objects are first individually encoded onto a list, and then the list
155 * of strings is concatenated. This is faster than concatenating strings,
156 * and reasonably simple to code.
159 pytdbpack(PyObject *self,
163 PyObject *val_seq, *val_iter = NULL,
164 *packed_list = NULL, *packed_str = NULL,
167 /* TODO: Test passing wrong types or too many arguments */
168 if (!PyArg_ParseTuple(args, "sO", &format_str, &val_seq))
171 if (!(val_iter = PyObject_GetIter(val_seq)))
174 /* Create list to hold strings until we're done, then join them all. */
175 if (!(packed_list = PyList_New(0)))
178 if (!pytdbpack_data(format_str, val_iter, packed_list))
181 /* this function is not officially documented but it works */
182 if (!(empty_str = PyString_InternFromString("")))
185 packed_str = _PyString_Join(empty_str, packed_list);
188 Py_XDECREF(empty_str);
189 Py_XDECREF(val_iter);
190 Py_XDECREF(packed_list);
197 Pack data according to FORMAT_STR from the elements of VAL_SEQ into
200 The string has already been checked out, so we know that VAL_SEQ is large
201 enough to hold the packed data, and that there are enough value items.
202 (However, their types may not have been thoroughly checked yet.)
204 In addition, val_seq is a Python Fast sequence.
206 Returns NULL for error (with exception set), or None.
209 pytdbpack_data(const char *format_str,
211 PyObject *packed_list)
213 int format_i, val_i = 0;
215 for (format_i = 0, val_i = 0; format_str[format_i]; format_i++) {
216 char ch = format_str[format_i];
219 /* dispatch to the appropriate packer for this type,
220 which should pull things off the iterator, and
221 append them to the packed_list */
225 if (!(packed_list = pytdbpack_number(ch, val_iter, packed_list)))
231 if (!(packed_list = pytdbpack_str_850(val_iter, packed_list)))
236 if (!(packed_list = pytdbpack_buffer(val_iter, packed_list)))
241 PyErr_Format(PyExc_ValueError,
242 "%s: format character '%c' is not supported",
253 pytdbpack_number(char ch, PyObject *val_iter, PyObject *packed_list)
255 unsigned long val_long;
256 PyObject *val_obj = NULL, *long_obj = NULL, *result_obj = NULL;
257 PyObject *new_list = NULL;
258 unsigned char pack_buf[4];
260 if (!(val_obj = PyIter_Next(val_iter)))
263 if (!(long_obj = PyNumber_Long(val_obj))) {
264 pytdbpack_bad_type(ch, "Number", val_obj);
268 val_long = PyLong_AsUnsignedLong(long_obj);
269 pack_le_uint32(val_long, pack_buf);
271 /* pack as 32-bit; if just packing a 'w' 16-bit word then only take
272 the first two bytes. */
274 if (!(result_obj = PyString_FromStringAndSize(pack_buf, ch == 'w' ? 2 : 4)))
277 if (PyList_Append(packed_list, result_obj) != -1)
278 new_list = packed_list;
282 Py_XDECREF(long_obj);
283 Py_XDECREF(result_obj);
290 * Take one string from the iterator val_iter, convert it to 8-bit CP850, and
293 * If the input is neither a string nor Unicode, an exception is raised.
295 * If the input is Unicode, then it is converted to CP850.
297 * If the input is a String, then it is converted to Unicode using the default
298 * decoding method, and then converted to CP850. This in effect gives
299 * conversion from latin-1 (currently the PSA's default) to CP850, without
300 * needing a custom translation table.
302 * I hope this approach avoids being too fragile w.r.t. being passed either
303 * Unicode or String objects.
306 pytdbpack_str_850(PyObject *val_iter, PyObject *packed_list)
308 PyObject *val_obj = NULL;
309 PyObject *unicode_obj = NULL;
310 PyObject *cp850_str = NULL;
311 PyObject *nul_str = NULL;
312 PyObject *new_list = NULL;
314 if (!(val_obj = PyIter_Next(val_iter)))
317 if (PyUnicode_Check(val_obj)) {
318 unicode_obj = val_obj;
322 if (!(unicode_obj = PyString_AsDecodedObject(val_obj, NULL, NULL)))
328 if (!(cp850_str = PyUnicode_AsEncodedString(unicode_obj, pytdb_string_encoding, NULL)))
332 /* this is constant and often-used; hold it forever */
333 if (!(nul_str = PyString_FromStringAndSize("", 1)))
336 if ((PyList_Append(packed_list, cp850_str) != -1)
337 && (PyList_Append(packed_list, nul_str) != -1))
338 new_list = packed_list;
341 Py_XDECREF(unicode_obj);
342 Py_XDECREF(cp850_str);
349 * Pack (LENGTH, BUFFER) pair onto the list.
351 * The buffer must already be a String, not Unicode, because it contains 8-bit
352 * untranslated data. In some cases it will actually be UTF_16_LE data.
355 pytdbpack_buffer(PyObject *val_iter, PyObject *packed_list)
358 PyObject *new_list = NULL;
360 /* pull off integer and stick onto list */
361 if (!(packed_list = pytdbpack_number('d', val_iter, packed_list)))
364 /* this assumes that the string is the right length; the old code did the same. */
365 if (!(val_obj = PyIter_Next(val_iter)))
368 if (!PyString_Check(val_obj)) {
369 pytdbpack_bad_type('B', "String", val_obj);
373 if (PyList_Append(packed_list, val_obj) != -1)
374 new_list = packed_list;
382 static PyObject *pytdbpack_bad_type(char ch,
383 const char *expected,
386 PyObject *r = PyObject_Repr(val_obj);
389 PyErr_Format(PyExc_TypeError,
390 "tdbpack: format '%c' requires %s, not %s",
391 ch, expected, PyString_AS_STRING(r));
398 XXX: glib and Samba have quicker macro for doing the endianness conversions,
399 but I don't know of one in plain libc, and it's probably not a big deal. I
400 realize this is kind of dumb because we'll almost always be on x86, but
401 being safe is important.
403 static void pack_le_uint32(unsigned long val_long, unsigned char *pbuf)
405 pbuf[0] = val_long & 0xff;
406 pbuf[1] = (val_long >> 8) & 0xff;
407 pbuf[2] = (val_long >> 16) & 0xff;
408 pbuf[3] = (val_long >> 24) & 0xff;
412 static void pack_bytes(long len, const char *from,
413 unsigned char **pbuf)
415 memcpy(*pbuf, from, len);
422 pytdbunpack(PyObject *self,
425 char *format_str, *packed_str, *ppacked;
426 PyObject *val_list = NULL, *ret_tuple = NULL;
427 PyObject *rest_string = NULL;
428 int format_len, packed_len;
429 char last_format = '#'; /* invalid */
433 if (!PyArg_ParseTuple(args, "ss#", &format_str, &packed_str, &packed_len))
436 format_len = strlen(format_str);
438 /* Allocate list to hold results. Initially empty, and we append
439 results as we go along. */
440 val_list = PyList_New(0);
443 ret_tuple = PyTuple_New(2);
447 /* For every object, unpack. */
448 for (ppacked = packed_str, i = 0; i < format_len && format_str[i] != '$'; i++) {
449 last_format = format_str[i];
450 /* packed_len is reduced in place */
451 if (!pytdbunpack_item(format_str[i], &ppacked, &packed_len, val_list))
455 /* If the last character was '$', keep going until out of space */
456 if (format_str[i] == '$') {
458 PyErr_Format(PyExc_ValueError,
459 "%s: '$' may not be first character in format",
463 while (packed_len > 0)
464 if (!pytdbunpack_item(last_format, &ppacked, &packed_len, val_list))
468 /* save leftovers for next time */
469 rest_string = PyString_FromStringAndSize(ppacked, packed_len);
473 /* return (values, rest) tuple; give up references to them */
474 PyTuple_SET_ITEM(ret_tuple, 0, val_list);
476 PyTuple_SET_ITEM(ret_tuple, 1, rest_string);
481 /* handle failure: deallocate anything. XDECREF forms handle NULL
482 pointers for objects that haven't been allocated yet. */
483 Py_XDECREF(val_list);
484 Py_XDECREF(ret_tuple);
485 Py_XDECREF(rest_string);
491 pytdbunpack_err_too_short(void)
493 PyErr_Format(PyExc_IndexError,
494 __FUNCTION__ ": data too short for unpack format");
499 pytdbunpack_uint32(char **pbuf, int *plen)
505 pytdbunpack_err_too_short();
510 v = b[0] | b[1]<<8 | b[2]<<16 | b[3]<<24;
515 return PyLong_FromUnsignedLong(v);
519 static PyObject *pytdbunpack_int16(char **pbuf, int *plen)
525 pytdbunpack_err_too_short();
535 return PyInt_FromLong(v);
540 pytdbunpack_string(char **pbuf, int *plen)
543 char *nul_ptr, *start;
547 nul_ptr = memchr(start, '\0', *plen);
549 pytdbunpack_err_too_short();
553 len = nul_ptr - start;
555 *pbuf += len + 1; /* skip \0 */
558 return PyString_Decode(start, len, pytdb_string_encoding, NULL);
563 pytdbunpack_buffer(char **pbuf, int *plen, PyObject *val_list)
565 /* first get 32-bit len */
568 unsigned char *start;
569 PyObject *str_obj = NULL, *len_obj = NULL;
572 pytdbunpack_err_too_short();
577 slen = b[0] | b[1]<<8 | b[2]<<16 | b[3]<<24;
579 if (slen < 0) { /* surely you jest */
580 PyErr_Format(PyExc_ValueError,
581 __FUNCTION__ ": buffer seems to have negative length");
590 PyErr_Format(PyExc_IndexError,
591 __FUNCTION__ ": not enough data to unpack buffer: "
592 "need %d bytes, have %d",
600 if (!(len_obj = PyInt_FromLong(slen)))
603 if (PyList_Append(val_list, len_obj) == -1)
606 if (!(str_obj = PyString_FromStringAndSize(start, slen)))
609 if (PyList_Append(val_list, str_obj) == -1)
615 Py_XDECREF(len_obj); /* handles NULL */
621 /* Unpack a single field from packed data, according to format character CH.
622 Remaining data is at *PBUF, of *PLEN.
624 *PBUF is advanced, and *PLEN reduced to reflect the amount of data that has
627 Returns a reference to None, or NULL for failure.
629 static PyObject *pytdbunpack_item(char ch,
636 if (ch == 'w') { /* 16-bit int */
637 result = pytdbunpack_int16(pbuf, plen);
639 else if (ch == 'd' || ch == 'p') { /* 32-bit int */
640 /* pointers can just come through as integers */
641 result = pytdbunpack_uint32(pbuf, plen);
643 else if (ch == 'f' || ch == 'P') { /* nul-term string */
644 result = pytdbunpack_string(pbuf, plen);
646 else if (ch == 'B') { /* length, buffer */
647 return pytdbunpack_buffer(pbuf, plen, val_list);
650 PyErr_Format(PyExc_ValueError,
651 __FUNCTION__ ": format character '%c' is not supported",
660 if (PyList_Append(val_list, result) == -1)
671 static PyMethodDef pytdbpack_methods[] = {
672 { "pack", pytdbpack, METH_VARARGS, (char *) pytdbpack_doc },
673 { "unpack", pytdbunpack, METH_VARARGS, (char *) pytdbunpack_doc },
679 Py_InitModule3("tdbpack", pytdbpack_methods,
680 (char *) pytdbpack_docstring);