Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 1 | /* |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 2 | XXX support range parameter on search |
| 3 | XXX support mstop parameter on search |
| 4 | */ |
| 5 | |
| 6 | /*********************************************************** |
Guido van Rossum | 524b588 | 1995-01-04 19:10:35 +0000 | [diff] [blame] | 7 | Copyright 1991-1995 by Stichting Mathematisch Centrum, Amsterdam, |
| 8 | The Netherlands. |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 9 | |
| 10 | All Rights Reserved |
| 11 | |
Guido van Rossum | d266eb4 | 1996-10-25 14:44:06 +0000 | [diff] [blame] | 12 | Permission to use, copy, modify, and distribute this software and its |
| 13 | documentation for any purpose and without fee is hereby granted, |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 14 | provided that the above copyright notice appear in all copies and that |
Guido van Rossum | d266eb4 | 1996-10-25 14:44:06 +0000 | [diff] [blame] | 15 | both that copyright notice and this permission notice appear in |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 16 | supporting documentation, and that the names of Stichting Mathematisch |
Guido van Rossum | d266eb4 | 1996-10-25 14:44:06 +0000 | [diff] [blame] | 17 | Centrum or CWI or Corporation for National Research Initiatives or |
| 18 | CNRI not be used in advertising or publicity pertaining to |
| 19 | distribution of the software without specific, written prior |
| 20 | permission. |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 21 | |
Guido van Rossum | d266eb4 | 1996-10-25 14:44:06 +0000 | [diff] [blame] | 22 | While CWI is the initial source for this software, a modified version |
| 23 | is made available by the Corporation for National Research Initiatives |
| 24 | (CNRI) at the Internet address ftp://ftp.python.org. |
| 25 | |
| 26 | STICHTING MATHEMATISCH CENTRUM AND CNRI DISCLAIM ALL WARRANTIES WITH |
| 27 | REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF |
| 28 | MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL STICHTING MATHEMATISCH |
| 29 | CENTRUM OR CNRI BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL |
| 30 | DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR |
| 31 | PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER |
| 32 | TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR |
| 33 | PERFORMANCE OF THIS SOFTWARE. |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 34 | |
| 35 | ******************************************************************/ |
| 36 | |
| 37 | /* Regular expression objects */ |
Guido van Rossum | 1cab95c | 1992-01-19 16:31:57 +0000 | [diff] [blame] | 38 | /* This uses Tatu Ylonen's copyleft-free reimplementation of |
| 39 | GNU regular expressions */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 40 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 41 | #include "Python.h" |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 42 | |
Guido van Rossum | a376cc5 | 1996-12-05 23:43:35 +0000 | [diff] [blame] | 43 | #include <ctype.h> |
| 44 | |
Guido van Rossum | 1cab95c | 1992-01-19 16:31:57 +0000 | [diff] [blame] | 45 | #include "regexpr.h" |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 46 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 47 | static PyObject *RegexError; /* Exception */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 48 | |
| 49 | typedef struct { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 50 | PyObject_HEAD |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 51 | struct re_pattern_buffer re_patbuf; /* The compiled expression */ |
| 52 | struct re_registers re_regs; /* The registers from the last match */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 53 | char re_fastmap[256]; /* Storage for fastmap */ |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 54 | PyObject *re_translate; /* String object for translate table */ |
| 55 | PyObject *re_lastok; /* String object last matched/searched */ |
| 56 | PyObject *re_groupindex; /* Group name to index dictionary */ |
| 57 | PyObject *re_givenpat; /* Pattern with symbolic groups */ |
| 58 | PyObject *re_realpat; /* Pattern without symbolic groups */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 59 | } regexobject; |
| 60 | |
| 61 | /* Regex object methods */ |
| 62 | |
| 63 | static void |
| 64 | reg_dealloc(re) |
| 65 | regexobject *re; |
| 66 | { |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 67 | if (re->re_patbuf.buffer) |
| 68 | PyMem_DEL(re->re_patbuf.buffer); |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 69 | Py_XDECREF(re->re_translate); |
| 70 | Py_XDECREF(re->re_lastok); |
| 71 | Py_XDECREF(re->re_groupindex); |
| 72 | Py_XDECREF(re->re_givenpat); |
| 73 | Py_XDECREF(re->re_realpat); |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 74 | PyObject_Del(re); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 75 | } |
| 76 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 77 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 78 | makeresult(regs) |
| 79 | struct re_registers *regs; |
| 80 | { |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 81 | PyObject *v; |
| 82 | int i; |
| 83 | static PyObject *filler = NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 84 | |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 85 | if (filler == NULL) { |
| 86 | filler = Py_BuildValue("(ii)", -1, -1); |
| 87 | if (filler == NULL) |
| 88 | return NULL; |
| 89 | } |
| 90 | v = PyTuple_New(RE_NREGS); |
| 91 | if (v == NULL) |
| 92 | return NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 93 | |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 94 | for (i = 0; i < RE_NREGS; i++) { |
| 95 | int lo = regs->start[i]; |
| 96 | int hi = regs->end[i]; |
| 97 | PyObject *w; |
| 98 | if (lo == -1 && hi == -1) { |
| 99 | w = filler; |
| 100 | Py_INCREF(w); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 101 | } |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 102 | else |
| 103 | w = Py_BuildValue("(ii)", lo, hi); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 104 | if (w == NULL || PyTuple_SetItem(v, i, w) < 0) { |
| 105 | Py_DECREF(v); |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 106 | return NULL; |
| 107 | } |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 108 | } |
| 109 | return v; |
| 110 | } |
| 111 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 112 | static PyObject * |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 113 | regobj_match(re, args) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 114 | regexobject *re; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 115 | PyObject *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 116 | { |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 117 | PyObject *argstring; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 118 | char *buffer; |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 119 | int size; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 120 | int offset = 0; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 121 | int result; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 122 | |
Guido van Rossum | 43713e5 | 2000-02-29 13:59:29 +0000 | [diff] [blame] | 123 | if (!PyArg_ParseTuple(args, "O|i:match", &argstring, &offset)) |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 124 | return NULL; |
Guido van Rossum | 7e48898 | 1998-10-08 02:25:24 +0000 | [diff] [blame] | 125 | if (!PyArg_Parse(argstring, "t#", &buffer, &size)) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 126 | return NULL; |
| 127 | |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 128 | if (offset < 0 || offset > size) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 129 | PyErr_SetString(RegexError, "match offset out of range"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 130 | return NULL; |
| 131 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 132 | Py_XDECREF(re->re_lastok); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 133 | re->re_lastok = NULL; |
Guido van Rossum | ed2554a | 1997-08-18 15:31:24 +0000 | [diff] [blame] | 134 | result = _Py_re_match(&re->re_patbuf, (unsigned char *)buffer, size, offset, |
Guido van Rossum | 0318bd6 | 1997-08-14 14:35:12 +0000 | [diff] [blame] | 135 | &re->re_regs); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 136 | if (result < -1) { |
Guido van Rossum | 95e8053 | 1997-08-13 22:34:14 +0000 | [diff] [blame] | 137 | /* Serious failure of some sort; if re_match didn't |
| 138 | set an exception, raise a generic error */ |
| 139 | if (!PyErr_Occurred()) |
| 140 | PyErr_SetString(RegexError, "match failure"); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 141 | return NULL; |
| 142 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 143 | if (result >= 0) { |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 144 | Py_INCREF(argstring); |
| 145 | re->re_lastok = argstring; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 146 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 147 | return PyInt_FromLong((long)result); /* Length of the match or -1 */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 148 | } |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 149 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 150 | static PyObject * |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 151 | regobj_search(re, args) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 152 | regexobject *re; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 153 | PyObject *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 154 | { |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 155 | PyObject *argstring; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 156 | char *buffer; |
| 157 | int size; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 158 | int offset = 0; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 159 | int range; |
| 160 | int result; |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 161 | |
Guido van Rossum | 43713e5 | 2000-02-29 13:59:29 +0000 | [diff] [blame] | 162 | if (!PyArg_ParseTuple(args, "O|i:search", &argstring, &offset)) |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 163 | return NULL; |
Guido van Rossum | 43713e5 | 2000-02-29 13:59:29 +0000 | [diff] [blame] | 164 | if (!PyArg_Parse(argstring, "t#:search", &buffer, &size)) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 165 | return NULL; |
| 166 | |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 167 | if (offset < 0 || offset > size) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 168 | PyErr_SetString(RegexError, "search offset out of range"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 169 | return NULL; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 170 | } |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 171 | /* NB: In Emacs 18.57, the documentation for re_search[_2] and |
| 172 | the implementation don't match: the documentation states that |
| 173 | |range| positions are tried, while the code tries |range|+1 |
| 174 | positions. It seems more productive to believe the code! */ |
Guido van Rossum | 2d78590 | 1992-01-26 18:12:41 +0000 | [diff] [blame] | 175 | range = size - offset; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 176 | Py_XDECREF(re->re_lastok); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 177 | re->re_lastok = NULL; |
Guido van Rossum | ed2554a | 1997-08-18 15:31:24 +0000 | [diff] [blame] | 178 | result = _Py_re_search(&re->re_patbuf, (unsigned char *)buffer, size, offset, range, |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 179 | &re->re_regs); |
| 180 | if (result < -1) { |
Guido van Rossum | 95e8053 | 1997-08-13 22:34:14 +0000 | [diff] [blame] | 181 | /* Serious failure of some sort; if re_match didn't |
| 182 | set an exception, raise a generic error */ |
| 183 | if (!PyErr_Occurred()) |
| 184 | PyErr_SetString(RegexError, "match failure"); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 185 | return NULL; |
| 186 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 187 | if (result >= 0) { |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 188 | Py_INCREF(argstring); |
| 189 | re->re_lastok = argstring; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 190 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 191 | return PyInt_FromLong((long)result); /* Position of the match or -1 */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 192 | } |
| 193 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 194 | /* get the group from the regex where index can be a string (group name) or |
| 195 | an integer index [0 .. 99] |
| 196 | */ |
| 197 | static PyObject* |
| 198 | group_from_index(re, index) |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 199 | regexobject *re; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 200 | PyObject *index; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 201 | { |
| 202 | int i, a, b; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 203 | char *v; |
| 204 | |
| 205 | if (PyString_Check(index)) |
| 206 | if (re->re_groupindex == NULL || |
| 207 | !(index = PyDict_GetItem(re->re_groupindex, index))) |
| 208 | { |
| 209 | PyErr_SetString(RegexError, |
| 210 | "group() group name doesn't exist"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 211 | return NULL; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 212 | } |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 213 | |
| 214 | i = PyInt_AsLong(index); |
| 215 | if (i == -1 && PyErr_Occurred()) |
| 216 | return NULL; |
| 217 | |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 218 | if (i < 0 || i >= RE_NREGS) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 219 | PyErr_SetString(RegexError, "group() index out of range"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 220 | return NULL; |
| 221 | } |
| 222 | if (re->re_lastok == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 223 | PyErr_SetString(RegexError, |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 224 | "group() only valid after successful match/search"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 225 | return NULL; |
| 226 | } |
| 227 | a = re->re_regs.start[i]; |
| 228 | b = re->re_regs.end[i]; |
| 229 | if (a < 0 || b < 0) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 230 | Py_INCREF(Py_None); |
| 231 | return Py_None; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 232 | } |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 233 | |
| 234 | if (!(v = PyString_AsString(re->re_lastok))) |
| 235 | return NULL; |
| 236 | |
| 237 | return PyString_FromStringAndSize(v+a, b-a); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 238 | } |
| 239 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 240 | |
| 241 | static PyObject * |
| 242 | regobj_group(re, args) |
| 243 | regexobject *re; |
| 244 | PyObject *args; |
| 245 | { |
| 246 | int n = PyTuple_Size(args); |
| 247 | int i; |
| 248 | PyObject *res = NULL; |
| 249 | |
| 250 | if (n < 0) |
| 251 | return NULL; |
| 252 | if (n == 0) { |
| 253 | PyErr_SetString(PyExc_TypeError, "not enough arguments"); |
| 254 | return NULL; |
| 255 | } |
| 256 | if (n == 1) { |
| 257 | /* return value is a single string */ |
| 258 | PyObject *index = PyTuple_GetItem(args, 0); |
| 259 | if (!index) |
| 260 | return NULL; |
| 261 | |
| 262 | return group_from_index(re, index); |
| 263 | } |
| 264 | |
| 265 | /* return value is a tuple */ |
| 266 | if (!(res = PyTuple_New(n))) |
| 267 | return NULL; |
| 268 | |
| 269 | for (i = 0; i < n; i++) { |
| 270 | PyObject *index = PyTuple_GetItem(args, i); |
| 271 | PyObject *group = NULL; |
| 272 | |
| 273 | if (!index) |
| 274 | goto finally; |
| 275 | if (!(group = group_from_index(re, index))) |
| 276 | goto finally; |
| 277 | if (PyTuple_SetItem(res, i, group) < 0) |
| 278 | goto finally; |
| 279 | } |
| 280 | return res; |
| 281 | |
| 282 | finally: |
| 283 | Py_DECREF(res); |
| 284 | return NULL; |
| 285 | } |
| 286 | |
| 287 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 288 | static struct PyMethodDef reg_methods[] = { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 289 | {"match", (PyCFunction)regobj_match, 1}, |
| 290 | {"search", (PyCFunction)regobj_search, 1}, |
| 291 | {"group", (PyCFunction)regobj_group, 1}, |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 292 | {NULL, NULL} /* sentinel */ |
| 293 | }; |
| 294 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 295 | |
| 296 | |
| 297 | static char* members[] = { |
| 298 | "last", "regs", "translate", |
| 299 | "groupindex", "realpat", "givenpat", |
| 300 | NULL |
| 301 | }; |
| 302 | |
| 303 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 304 | static PyObject * |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 305 | regobj_getattr(re, name) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 306 | regexobject *re; |
| 307 | char *name; |
| 308 | { |
Guido van Rossum | b824fc6 | 1992-01-01 14:52:16 +0000 | [diff] [blame] | 309 | if (strcmp(name, "regs") == 0) { |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 310 | if (re->re_lastok == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 311 | Py_INCREF(Py_None); |
| 312 | return Py_None; |
Guido van Rossum | b824fc6 | 1992-01-01 14:52:16 +0000 | [diff] [blame] | 313 | } |
| 314 | return makeresult(&re->re_regs); |
| 315 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 316 | if (strcmp(name, "last") == 0) { |
| 317 | if (re->re_lastok == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 318 | Py_INCREF(Py_None); |
| 319 | return Py_None; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 320 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 321 | Py_INCREF(re->re_lastok); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 322 | return re->re_lastok; |
| 323 | } |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 324 | if (strcmp(name, "translate") == 0) { |
| 325 | if (re->re_translate == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 326 | Py_INCREF(Py_None); |
| 327 | return Py_None; |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 328 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 329 | Py_INCREF(re->re_translate); |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 330 | return re->re_translate; |
| 331 | } |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 332 | if (strcmp(name, "groupindex") == 0) { |
| 333 | if (re->re_groupindex == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 334 | Py_INCREF(Py_None); |
| 335 | return Py_None; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 336 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 337 | Py_INCREF(re->re_groupindex); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 338 | return re->re_groupindex; |
| 339 | } |
| 340 | if (strcmp(name, "realpat") == 0) { |
| 341 | if (re->re_realpat == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 342 | Py_INCREF(Py_None); |
| 343 | return Py_None; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 344 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 345 | Py_INCREF(re->re_realpat); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 346 | return re->re_realpat; |
| 347 | } |
| 348 | if (strcmp(name, "givenpat") == 0) { |
| 349 | if (re->re_givenpat == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 350 | Py_INCREF(Py_None); |
| 351 | return Py_None; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 352 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 353 | Py_INCREF(re->re_givenpat); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 354 | return re->re_givenpat; |
| 355 | } |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 356 | if (strcmp(name, "__members__") == 0) { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 357 | int i = 0; |
| 358 | PyObject *list = NULL; |
| 359 | |
| 360 | /* okay, so it's unlikely this list will change that often. |
| 361 | still, it's easier to change it in just one place. |
| 362 | */ |
| 363 | while (members[i]) |
| 364 | i++; |
| 365 | if (!(list = PyList_New(i))) |
| 366 | return NULL; |
| 367 | |
| 368 | i = 0; |
| 369 | while (members[i]) { |
| 370 | PyObject* v = PyString_FromString(members[i]); |
| 371 | if (!v || PyList_SetItem(list, i, v) < 0) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 372 | Py_DECREF(list); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 373 | return NULL; |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 374 | } |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 375 | i++; |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 376 | } |
| 377 | return list; |
| 378 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 379 | return Py_FindMethod(reg_methods, (PyObject *)re, name); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 380 | } |
| 381 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 382 | static PyTypeObject Regextype = { |
| 383 | PyObject_HEAD_INIT(&PyType_Type) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 384 | 0, /*ob_size*/ |
| 385 | "regex", /*tp_name*/ |
| 386 | sizeof(regexobject), /*tp_size*/ |
| 387 | 0, /*tp_itemsize*/ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 388 | /* methods */ |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 389 | (destructor)reg_dealloc, /*tp_dealloc*/ |
| 390 | 0, /*tp_print*/ |
| 391 | (getattrfunc)regobj_getattr, /*tp_getattr*/ |
| 392 | 0, /*tp_setattr*/ |
| 393 | 0, /*tp_compare*/ |
| 394 | 0, /*tp_repr*/ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 395 | }; |
| 396 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 397 | /* reference counting invariants: |
| 398 | pattern: borrowed |
| 399 | translate: borrowed |
| 400 | givenpat: borrowed |
| 401 | groupindex: transferred |
| 402 | */ |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 403 | static PyObject * |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 404 | newregexobject(pattern, translate, givenpat, groupindex) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 405 | PyObject *pattern; |
| 406 | PyObject *translate; |
| 407 | PyObject *givenpat; |
| 408 | PyObject *groupindex; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 409 | { |
| 410 | regexobject *re; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 411 | char *pat; |
| 412 | int size; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 413 | |
Guido van Rossum | 7e48898 | 1998-10-08 02:25:24 +0000 | [diff] [blame] | 414 | if (!PyArg_Parse(pattern, "t#", &pat, &size)) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 415 | return NULL; |
| 416 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 417 | if (translate != NULL && PyString_Size(translate) != 256) { |
| 418 | PyErr_SetString(RegexError, |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 419 | "translation table must be 256 bytes"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 420 | return NULL; |
| 421 | } |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 422 | re = PyObject_New(regexobject, &Regextype); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 423 | if (re != NULL) { |
| 424 | char *error; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 425 | re->re_patbuf.buffer = NULL; |
| 426 | re->re_patbuf.allocated = 0; |
Guido van Rossum | ed2554a | 1997-08-18 15:31:24 +0000 | [diff] [blame] | 427 | re->re_patbuf.fastmap = (unsigned char *)re->re_fastmap; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 428 | if (translate) { |
Guido van Rossum | ed2554a | 1997-08-18 15:31:24 +0000 | [diff] [blame] | 429 | re->re_patbuf.translate = (unsigned char *)PyString_AsString(translate); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 430 | if (!re->re_patbuf.translate) |
| 431 | goto finally; |
| 432 | Py_INCREF(translate); |
| 433 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 434 | else |
| 435 | re->re_patbuf.translate = NULL; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 436 | re->re_translate = translate; |
| 437 | re->re_lastok = NULL; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 438 | re->re_groupindex = groupindex; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 439 | Py_INCREF(pattern); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 440 | re->re_realpat = pattern; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 441 | Py_INCREF(givenpat); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 442 | re->re_givenpat = givenpat; |
Guido van Rossum | d19c04a | 1997-09-03 00:47:36 +0000 | [diff] [blame] | 443 | error = _Py_re_compile_pattern((unsigned char *)pat, size, &re->re_patbuf); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 444 | if (error != NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 445 | PyErr_SetString(RegexError, error); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 446 | goto finally; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 447 | } |
| 448 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 449 | return (PyObject *)re; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 450 | finally: |
| 451 | Py_DECREF(re); |
| 452 | return NULL; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 453 | } |
| 454 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 455 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 456 | regex_compile(self, args) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 457 | PyObject *self; |
| 458 | PyObject *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 459 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 460 | PyObject *pat = NULL; |
| 461 | PyObject *tran = NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 462 | |
Guido van Rossum | 43713e5 | 2000-02-29 13:59:29 +0000 | [diff] [blame] | 463 | if (!PyArg_ParseTuple(args, "S|S:compile", &pat, &tran)) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 464 | return NULL; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 465 | return newregexobject(pat, tran, pat, NULL); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 466 | } |
| 467 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 468 | static PyObject * |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 469 | symcomp(pattern, gdict) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 470 | PyObject *pattern; |
| 471 | PyObject *gdict; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 472 | { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 473 | char *opat, *oend, *o, *n, *g, *v; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 474 | int group_count = 0; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 475 | int sz; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 476 | int escaped = 0; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 477 | char name_buf[128]; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 478 | PyObject *npattern; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 479 | int require_escape = re_syntax & RE_NO_BK_PARENS ? 0 : 1; |
| 480 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 481 | if (!(opat = PyString_AsString(pattern))) |
| 482 | return NULL; |
| 483 | |
| 484 | if ((sz = PyString_Size(pattern)) < 0) |
| 485 | return NULL; |
| 486 | |
| 487 | oend = opat + sz; |
| 488 | o = opat; |
| 489 | |
Guido van Rossum | ab28c56 | 1996-06-11 18:33:14 +0000 | [diff] [blame] | 490 | if (oend == opat) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 491 | Py_INCREF(pattern); |
Guido van Rossum | ab28c56 | 1996-06-11 18:33:14 +0000 | [diff] [blame] | 492 | return pattern; |
| 493 | } |
| 494 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 495 | if (!(npattern = PyString_FromStringAndSize((char*)NULL, sz)) || |
| 496 | !(n = PyString_AsString(npattern))) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 497 | return NULL; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 498 | |
| 499 | while (o < oend) { |
| 500 | if (*o == '(' && escaped == require_escape) { |
| 501 | char *backtrack; |
| 502 | escaped = 0; |
| 503 | ++group_count; |
| 504 | *n++ = *o; |
| 505 | if (++o >= oend || *o != '<') |
| 506 | continue; |
| 507 | /* *o == '<' */ |
| 508 | if (o+1 < oend && *(o+1) == '>') |
| 509 | continue; |
| 510 | backtrack = o; |
| 511 | g = name_buf; |
| 512 | for (++o; o < oend;) { |
| 513 | if (*o == '>') { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 514 | PyObject *group_name = NULL; |
| 515 | PyObject *group_index = NULL; |
| 516 | *g++ = '\0'; |
| 517 | group_name = PyString_FromString(name_buf); |
| 518 | group_index = PyInt_FromLong(group_count); |
| 519 | if (group_name == NULL || |
| 520 | group_index == NULL || |
| 521 | PyDict_SetItem(gdict, group_name, |
| 522 | group_index) != 0) |
| 523 | { |
| 524 | Py_XDECREF(group_name); |
| 525 | Py_XDECREF(group_index); |
| 526 | Py_XDECREF(npattern); |
| 527 | return NULL; |
| 528 | } |
Barry Warsaw | 4bc9d39 | 1997-01-09 22:22:05 +0000 | [diff] [blame] | 529 | Py_DECREF(group_name); |
| 530 | Py_DECREF(group_index); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 531 | ++o; /* eat the '>' */ |
| 532 | break; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 533 | } |
Guido van Rossum | 7f7f274 | 1995-02-10 17:01:56 +0000 | [diff] [blame] | 534 | if (!isalnum(Py_CHARMASK(*o)) && *o != '_') { |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 535 | o = backtrack; |
| 536 | break; |
| 537 | } |
| 538 | *g++ = *o++; |
| 539 | } |
| 540 | } |
Guido van Rossum | 0cbaff4 | 1996-10-23 17:53:06 +0000 | [diff] [blame] | 541 | else if (*o == '[' && !escaped) { |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 542 | *n++ = *o; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 543 | ++o; /* eat the char following '[' */ |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 544 | *n++ = *o; |
| 545 | while (o < oend && *o != ']') { |
| 546 | ++o; |
| 547 | *n++ = *o; |
| 548 | } |
| 549 | if (o < oend) |
| 550 | ++o; |
| 551 | } |
| 552 | else if (*o == '\\') { |
| 553 | escaped = 1; |
| 554 | *n++ = *o; |
| 555 | ++o; |
| 556 | } |
| 557 | else { |
| 558 | escaped = 0; |
| 559 | *n++ = *o; |
| 560 | ++o; |
| 561 | } |
| 562 | } |
| 563 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 564 | if (!(v = PyString_AsString(npattern))) { |
| 565 | Py_DECREF(npattern); |
| 566 | return NULL; |
| 567 | } |
| 568 | /* _PyString_Resize() decrements npattern on failure */ |
| 569 | if (_PyString_Resize(&npattern, n - v) == 0) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 570 | return npattern; |
| 571 | else { |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 572 | return NULL; |
| 573 | } |
| 574 | |
| 575 | } |
| 576 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 577 | static PyObject * |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 578 | regex_symcomp(self, args) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 579 | PyObject *self; |
| 580 | PyObject *args; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 581 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 582 | PyObject *pattern; |
| 583 | PyObject *tran = NULL; |
| 584 | PyObject *gdict = NULL; |
| 585 | PyObject *npattern; |
Barry Warsaw | 4bc9d39 | 1997-01-09 22:22:05 +0000 | [diff] [blame] | 586 | PyObject *retval = NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 587 | |
Guido van Rossum | 43713e5 | 2000-02-29 13:59:29 +0000 | [diff] [blame] | 588 | if (!PyArg_ParseTuple(args, "S|S:symcomp", &pattern, &tran)) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 589 | return NULL; |
| 590 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 591 | gdict = PyDict_New(); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 592 | if (gdict == NULL || (npattern = symcomp(pattern, gdict)) == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 593 | Py_DECREF(gdict); |
| 594 | Py_DECREF(pattern); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 595 | return NULL; |
| 596 | } |
Barry Warsaw | 4bc9d39 | 1997-01-09 22:22:05 +0000 | [diff] [blame] | 597 | retval = newregexobject(npattern, tran, pattern, gdict); |
| 598 | Py_DECREF(npattern); |
| 599 | return retval; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 600 | } |
| 601 | |
| 602 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 603 | static PyObject *cache_pat; |
| 604 | static PyObject *cache_prog; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 605 | |
| 606 | static int |
| 607 | update_cache(pat) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 608 | PyObject *pat; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 609 | { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 610 | PyObject *tuple = Py_BuildValue("(O)", pat); |
| 611 | int status = 0; |
| 612 | |
| 613 | if (!tuple) |
| 614 | return -1; |
| 615 | |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 616 | if (pat != cache_pat) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 617 | Py_XDECREF(cache_pat); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 618 | cache_pat = NULL; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 619 | Py_XDECREF(cache_prog); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 620 | cache_prog = regex_compile((PyObject *)NULL, tuple); |
| 621 | if (cache_prog == NULL) { |
| 622 | status = -1; |
| 623 | goto finally; |
| 624 | } |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 625 | cache_pat = pat; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 626 | Py_INCREF(cache_pat); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 627 | } |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 628 | finally: |
| 629 | Py_DECREF(tuple); |
| 630 | return status; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 631 | } |
| 632 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 633 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 634 | regex_match(self, args) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 635 | PyObject *self; |
| 636 | PyObject *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 637 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 638 | PyObject *pat, *string; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 639 | PyObject *tuple, *v; |
| 640 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 641 | if (!PyArg_Parse(args, "(SS)", &pat, &string)) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 642 | return NULL; |
| 643 | if (update_cache(pat) < 0) |
| 644 | return NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 645 | |
| 646 | if (!(tuple = Py_BuildValue("(S)", string))) |
| 647 | return NULL; |
| 648 | v = regobj_match((regexobject *)cache_prog, tuple); |
| 649 | Py_DECREF(tuple); |
| 650 | return v; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 651 | } |
| 652 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 653 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 654 | regex_search(self, args) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 655 | PyObject *self; |
| 656 | PyObject *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 657 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 658 | PyObject *pat, *string; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 659 | PyObject *tuple, *v; |
| 660 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 661 | if (!PyArg_Parse(args, "(SS)", &pat, &string)) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 662 | return NULL; |
| 663 | if (update_cache(pat) < 0) |
| 664 | return NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 665 | |
| 666 | if (!(tuple = Py_BuildValue("(S)", string))) |
| 667 | return NULL; |
| 668 | v = regobj_search((regexobject *)cache_prog, tuple); |
| 669 | Py_DECREF(tuple); |
| 670 | return v; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 671 | } |
| 672 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 673 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 674 | regex_set_syntax(self, args) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 675 | PyObject *self; |
| 676 | PyObject *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 677 | { |
| 678 | int syntax; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 679 | if (!PyArg_Parse(args, "i", &syntax)) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 680 | return NULL; |
| 681 | syntax = re_set_syntax(syntax); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 682 | /* wipe the global pattern cache */ |
| 683 | Py_XDECREF(cache_pat); |
| 684 | cache_pat = NULL; |
| 685 | Py_XDECREF(cache_prog); |
| 686 | cache_prog = NULL; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 687 | return PyInt_FromLong((long)syntax); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 688 | } |
| 689 | |
Barry Warsaw | 909d7c3 | 1997-02-18 18:48:50 +0000 | [diff] [blame] | 690 | static PyObject * |
| 691 | regex_get_syntax(self, args) |
| 692 | PyObject *self; |
| 693 | PyObject *args; |
| 694 | { |
| 695 | if (!PyArg_Parse(args, "")) |
| 696 | return NULL; |
| 697 | return PyInt_FromLong((long)re_syntax); |
| 698 | } |
| 699 | |
| 700 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 701 | static struct PyMethodDef regex_global_methods[] = { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 702 | {"compile", regex_compile, 1}, |
| 703 | {"symcomp", regex_symcomp, 1}, |
Guido van Rossum | 295d171 | 1995-02-19 15:55:19 +0000 | [diff] [blame] | 704 | {"match", regex_match, 0}, |
| 705 | {"search", regex_search, 0}, |
| 706 | {"set_syntax", regex_set_syntax, 0}, |
Barry Warsaw | 909d7c3 | 1997-02-18 18:48:50 +0000 | [diff] [blame] | 707 | {"get_syntax", regex_get_syntax, 0}, |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 708 | {NULL, NULL} /* sentinel */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 709 | }; |
| 710 | |
Guido van Rossum | 3886bb6 | 1998-12-04 18:50:17 +0000 | [diff] [blame] | 711 | DL_EXPORT(void) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 712 | initregex() |
| 713 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 714 | PyObject *m, *d, *v; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 715 | int i; |
| 716 | char *s; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 717 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 718 | m = Py_InitModule("regex", regex_global_methods); |
| 719 | d = PyModule_GetDict(m); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 720 | |
| 721 | /* Initialize regex.error exception */ |
Guido van Rossum | 0cb96de | 1997-10-01 04:29:29 +0000 | [diff] [blame] | 722 | v = RegexError = PyErr_NewException("regex.error", NULL, NULL); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 723 | if (v == NULL || PyDict_SetItemString(d, "error", v) != 0) |
| 724 | goto finally; |
| 725 | |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 726 | /* Initialize regex.casefold constant */ |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 727 | if (!(v = PyString_FromStringAndSize((char *)NULL, 256))) |
| 728 | goto finally; |
| 729 | |
| 730 | if (!(s = PyString_AsString(v))) |
| 731 | goto finally; |
| 732 | |
| 733 | for (i = 0; i < 256; i++) { |
| 734 | if (isupper(i)) |
| 735 | s[i] = tolower(i); |
| 736 | else |
| 737 | s[i] = i; |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 738 | } |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 739 | if (PyDict_SetItemString(d, "casefold", v) < 0) |
| 740 | goto finally; |
| 741 | Py_DECREF(v); |
| 742 | |
| 743 | if (!PyErr_Occurred()) |
| 744 | return; |
| 745 | finally: |
Guido van Rossum | 0cb96de | 1997-10-01 04:29:29 +0000 | [diff] [blame] | 746 | /* Nothing */ ; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 747 | } |