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 | fd71b9e | 2000-06-30 23:50:40 +0000 | [diff] [blame] | 7 | Copyright (c) 2000, BeOpen.com. |
| 8 | Copyright (c) 1995-2000, Corporation for National Research Initiatives. |
| 9 | Copyright (c) 1990-1995, Stichting Mathematisch Centrum. |
| 10 | All rights reserved. |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 11 | |
Guido van Rossum | fd71b9e | 2000-06-30 23:50:40 +0000 | [diff] [blame] | 12 | See the file "Misc/COPYRIGHT" for information on usage and |
| 13 | redistribution of this file, and for a DISCLAIMER OF ALL WARRANTIES. |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 14 | ******************************************************************/ |
| 15 | |
| 16 | /* Regular expression objects */ |
Guido van Rossum | 1cab95c | 1992-01-19 16:31:57 +0000 | [diff] [blame] | 17 | /* This uses Tatu Ylonen's copyleft-free reimplementation of |
| 18 | GNU regular expressions */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 19 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 20 | #include "Python.h" |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 21 | |
Guido van Rossum | a376cc5 | 1996-12-05 23:43:35 +0000 | [diff] [blame] | 22 | #include <ctype.h> |
| 23 | |
Guido van Rossum | 1cab95c | 1992-01-19 16:31:57 +0000 | [diff] [blame] | 24 | #include "regexpr.h" |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 25 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 26 | static PyObject *RegexError; /* Exception */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 27 | |
| 28 | typedef struct { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 29 | PyObject_HEAD |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 30 | struct re_pattern_buffer re_patbuf; /* The compiled expression */ |
| 31 | struct re_registers re_regs; /* The registers from the last match */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 32 | char re_fastmap[256]; /* Storage for fastmap */ |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 33 | PyObject *re_translate; /* String object for translate table */ |
| 34 | PyObject *re_lastok; /* String object last matched/searched */ |
| 35 | PyObject *re_groupindex; /* Group name to index dictionary */ |
| 36 | PyObject *re_givenpat; /* Pattern with symbolic groups */ |
| 37 | PyObject *re_realpat; /* Pattern without symbolic groups */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 38 | } regexobject; |
| 39 | |
| 40 | /* Regex object methods */ |
| 41 | |
| 42 | static void |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 43 | reg_dealloc(regexobject *re) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 44 | { |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 45 | if (re->re_patbuf.buffer) |
Vladimir Marangozov | 9e3d73a | 2000-07-12 00:49:17 +0000 | [diff] [blame] | 46 | free(re->re_patbuf.buffer); |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 47 | Py_XDECREF(re->re_translate); |
| 48 | Py_XDECREF(re->re_lastok); |
| 49 | Py_XDECREF(re->re_groupindex); |
| 50 | Py_XDECREF(re->re_givenpat); |
| 51 | Py_XDECREF(re->re_realpat); |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 52 | PyObject_Del(re); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 53 | } |
| 54 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 55 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 56 | makeresult(struct re_registers *regs) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 57 | { |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 58 | PyObject *v; |
| 59 | int i; |
| 60 | static PyObject *filler = NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 61 | |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 62 | if (filler == NULL) { |
| 63 | filler = Py_BuildValue("(ii)", -1, -1); |
| 64 | if (filler == NULL) |
| 65 | return NULL; |
| 66 | } |
| 67 | v = PyTuple_New(RE_NREGS); |
| 68 | if (v == NULL) |
| 69 | return NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 70 | |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 71 | for (i = 0; i < RE_NREGS; i++) { |
| 72 | int lo = regs->start[i]; |
| 73 | int hi = regs->end[i]; |
| 74 | PyObject *w; |
| 75 | if (lo == -1 && hi == -1) { |
| 76 | w = filler; |
| 77 | Py_INCREF(w); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 78 | } |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 79 | else |
| 80 | w = Py_BuildValue("(ii)", lo, hi); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 81 | if (w == NULL || PyTuple_SetItem(v, i, w) < 0) { |
| 82 | Py_DECREF(v); |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 83 | return NULL; |
| 84 | } |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 85 | } |
| 86 | return v; |
| 87 | } |
| 88 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 89 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 90 | regobj_match(regexobject *re, PyObject *args) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 91 | { |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 92 | PyObject *argstring; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 93 | char *buffer; |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 94 | int size; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 95 | int offset = 0; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 96 | int result; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 97 | |
Guido van Rossum | 43713e5 | 2000-02-29 13:59:29 +0000 | [diff] [blame] | 98 | if (!PyArg_ParseTuple(args, "O|i:match", &argstring, &offset)) |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 99 | return NULL; |
Guido van Rossum | 7e48898 | 1998-10-08 02:25:24 +0000 | [diff] [blame] | 100 | if (!PyArg_Parse(argstring, "t#", &buffer, &size)) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 101 | return NULL; |
| 102 | |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 103 | if (offset < 0 || offset > size) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 104 | PyErr_SetString(RegexError, "match offset out of range"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 105 | return NULL; |
| 106 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 107 | Py_XDECREF(re->re_lastok); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 108 | re->re_lastok = NULL; |
Guido van Rossum | ed2554a | 1997-08-18 15:31:24 +0000 | [diff] [blame] | 109 | 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] | 110 | &re->re_regs); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 111 | if (result < -1) { |
Guido van Rossum | 95e8053 | 1997-08-13 22:34:14 +0000 | [diff] [blame] | 112 | /* Serious failure of some sort; if re_match didn't |
| 113 | set an exception, raise a generic error */ |
| 114 | if (!PyErr_Occurred()) |
| 115 | PyErr_SetString(RegexError, "match failure"); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 116 | return NULL; |
| 117 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 118 | if (result >= 0) { |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 119 | Py_INCREF(argstring); |
| 120 | re->re_lastok = argstring; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 121 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 122 | return PyInt_FromLong((long)result); /* Length of the match or -1 */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 123 | } |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 124 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 125 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 126 | regobj_search(regexobject *re, PyObject *args) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 127 | { |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 128 | PyObject *argstring; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 129 | char *buffer; |
| 130 | int size; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 131 | int offset = 0; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 132 | int range; |
| 133 | int result; |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 134 | |
Guido van Rossum | 43713e5 | 2000-02-29 13:59:29 +0000 | [diff] [blame] | 135 | if (!PyArg_ParseTuple(args, "O|i:search", &argstring, &offset)) |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 136 | return NULL; |
Guido van Rossum | 43713e5 | 2000-02-29 13:59:29 +0000 | [diff] [blame] | 137 | if (!PyArg_Parse(argstring, "t#:search", &buffer, &size)) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 138 | return NULL; |
| 139 | |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 140 | if (offset < 0 || offset > size) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 141 | PyErr_SetString(RegexError, "search offset out of range"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 142 | return NULL; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 143 | } |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 144 | /* NB: In Emacs 18.57, the documentation for re_search[_2] and |
| 145 | the implementation don't match: the documentation states that |
| 146 | |range| positions are tried, while the code tries |range|+1 |
| 147 | positions. It seems more productive to believe the code! */ |
Guido van Rossum | 2d78590 | 1992-01-26 18:12:41 +0000 | [diff] [blame] | 148 | range = size - offset; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 149 | Py_XDECREF(re->re_lastok); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 150 | re->re_lastok = NULL; |
Guido van Rossum | ed2554a | 1997-08-18 15:31:24 +0000 | [diff] [blame] | 151 | 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] | 152 | &re->re_regs); |
| 153 | if (result < -1) { |
Guido van Rossum | 95e8053 | 1997-08-13 22:34:14 +0000 | [diff] [blame] | 154 | /* Serious failure of some sort; if re_match didn't |
| 155 | set an exception, raise a generic error */ |
| 156 | if (!PyErr_Occurred()) |
| 157 | PyErr_SetString(RegexError, "match failure"); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 158 | return NULL; |
| 159 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 160 | if (result >= 0) { |
Guido van Rossum | 4a807f5 | 1997-05-12 16:04:09 +0000 | [diff] [blame] | 161 | Py_INCREF(argstring); |
| 162 | re->re_lastok = argstring; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 163 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 164 | return PyInt_FromLong((long)result); /* Position of the match or -1 */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 165 | } |
| 166 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 167 | /* get the group from the regex where index can be a string (group name) or |
| 168 | an integer index [0 .. 99] |
| 169 | */ |
| 170 | static PyObject* |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 171 | group_from_index(regexobject *re, PyObject *index) |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 172 | { |
| 173 | int i, a, b; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 174 | char *v; |
| 175 | |
| 176 | if (PyString_Check(index)) |
| 177 | if (re->re_groupindex == NULL || |
| 178 | !(index = PyDict_GetItem(re->re_groupindex, index))) |
| 179 | { |
| 180 | PyErr_SetString(RegexError, |
| 181 | "group() group name doesn't exist"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 182 | return NULL; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 183 | } |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 184 | |
| 185 | i = PyInt_AsLong(index); |
| 186 | if (i == -1 && PyErr_Occurred()) |
| 187 | return NULL; |
| 188 | |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 189 | if (i < 0 || i >= RE_NREGS) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 190 | PyErr_SetString(RegexError, "group() index out of range"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 191 | return NULL; |
| 192 | } |
| 193 | if (re->re_lastok == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 194 | PyErr_SetString(RegexError, |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 195 | "group() only valid after successful match/search"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 196 | return NULL; |
| 197 | } |
| 198 | a = re->re_regs.start[i]; |
| 199 | b = re->re_regs.end[i]; |
| 200 | if (a < 0 || b < 0) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 201 | Py_INCREF(Py_None); |
| 202 | return Py_None; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 203 | } |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 204 | |
| 205 | if (!(v = PyString_AsString(re->re_lastok))) |
| 206 | return NULL; |
| 207 | |
| 208 | return PyString_FromStringAndSize(v+a, b-a); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 209 | } |
| 210 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 211 | |
| 212 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 213 | regobj_group(regexobject *re, PyObject *args) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 214 | { |
| 215 | int n = PyTuple_Size(args); |
| 216 | int i; |
| 217 | PyObject *res = NULL; |
| 218 | |
| 219 | if (n < 0) |
| 220 | return NULL; |
| 221 | if (n == 0) { |
| 222 | PyErr_SetString(PyExc_TypeError, "not enough arguments"); |
| 223 | return NULL; |
| 224 | } |
| 225 | if (n == 1) { |
| 226 | /* return value is a single string */ |
| 227 | PyObject *index = PyTuple_GetItem(args, 0); |
| 228 | if (!index) |
| 229 | return NULL; |
| 230 | |
| 231 | return group_from_index(re, index); |
| 232 | } |
| 233 | |
| 234 | /* return value is a tuple */ |
| 235 | if (!(res = PyTuple_New(n))) |
| 236 | return NULL; |
| 237 | |
| 238 | for (i = 0; i < n; i++) { |
| 239 | PyObject *index = PyTuple_GetItem(args, i); |
| 240 | PyObject *group = NULL; |
| 241 | |
| 242 | if (!index) |
| 243 | goto finally; |
| 244 | if (!(group = group_from_index(re, index))) |
| 245 | goto finally; |
| 246 | if (PyTuple_SetItem(res, i, group) < 0) |
| 247 | goto finally; |
| 248 | } |
| 249 | return res; |
| 250 | |
| 251 | finally: |
| 252 | Py_DECREF(res); |
| 253 | return NULL; |
| 254 | } |
| 255 | |
| 256 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 257 | static struct PyMethodDef reg_methods[] = { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 258 | {"match", (PyCFunction)regobj_match, 1}, |
| 259 | {"search", (PyCFunction)regobj_search, 1}, |
| 260 | {"group", (PyCFunction)regobj_group, 1}, |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 261 | {NULL, NULL} /* sentinel */ |
| 262 | }; |
| 263 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 264 | |
| 265 | |
| 266 | static char* members[] = { |
| 267 | "last", "regs", "translate", |
| 268 | "groupindex", "realpat", "givenpat", |
| 269 | NULL |
| 270 | }; |
| 271 | |
| 272 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 273 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 274 | regobj_getattr(regexobject *re, char *name) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 275 | { |
Guido van Rossum | b824fc6 | 1992-01-01 14:52:16 +0000 | [diff] [blame] | 276 | if (strcmp(name, "regs") == 0) { |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 277 | if (re->re_lastok == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 278 | Py_INCREF(Py_None); |
| 279 | return Py_None; |
Guido van Rossum | b824fc6 | 1992-01-01 14:52:16 +0000 | [diff] [blame] | 280 | } |
| 281 | return makeresult(&re->re_regs); |
| 282 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 283 | if (strcmp(name, "last") == 0) { |
| 284 | if (re->re_lastok == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 285 | Py_INCREF(Py_None); |
| 286 | return Py_None; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 287 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 288 | Py_INCREF(re->re_lastok); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 289 | return re->re_lastok; |
| 290 | } |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 291 | if (strcmp(name, "translate") == 0) { |
| 292 | if (re->re_translate == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 293 | Py_INCREF(Py_None); |
| 294 | return Py_None; |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 295 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 296 | Py_INCREF(re->re_translate); |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 297 | return re->re_translate; |
| 298 | } |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 299 | if (strcmp(name, "groupindex") == 0) { |
| 300 | if (re->re_groupindex == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 301 | Py_INCREF(Py_None); |
| 302 | return Py_None; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 303 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 304 | Py_INCREF(re->re_groupindex); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 305 | return re->re_groupindex; |
| 306 | } |
| 307 | if (strcmp(name, "realpat") == 0) { |
| 308 | if (re->re_realpat == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 309 | Py_INCREF(Py_None); |
| 310 | return Py_None; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 311 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 312 | Py_INCREF(re->re_realpat); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 313 | return re->re_realpat; |
| 314 | } |
| 315 | if (strcmp(name, "givenpat") == 0) { |
| 316 | if (re->re_givenpat == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 317 | Py_INCREF(Py_None); |
| 318 | return Py_None; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 319 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 320 | Py_INCREF(re->re_givenpat); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 321 | return re->re_givenpat; |
| 322 | } |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 323 | if (strcmp(name, "__members__") == 0) { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 324 | int i = 0; |
| 325 | PyObject *list = NULL; |
| 326 | |
| 327 | /* okay, so it's unlikely this list will change that often. |
| 328 | still, it's easier to change it in just one place. |
| 329 | */ |
| 330 | while (members[i]) |
| 331 | i++; |
| 332 | if (!(list = PyList_New(i))) |
| 333 | return NULL; |
| 334 | |
| 335 | i = 0; |
| 336 | while (members[i]) { |
| 337 | PyObject* v = PyString_FromString(members[i]); |
| 338 | if (!v || PyList_SetItem(list, i, v) < 0) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 339 | Py_DECREF(list); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 340 | return NULL; |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 341 | } |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 342 | i++; |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 343 | } |
| 344 | return list; |
| 345 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 346 | return Py_FindMethod(reg_methods, (PyObject *)re, name); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 347 | } |
| 348 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 349 | static PyTypeObject Regextype = { |
| 350 | PyObject_HEAD_INIT(&PyType_Type) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 351 | 0, /*ob_size*/ |
| 352 | "regex", /*tp_name*/ |
| 353 | sizeof(regexobject), /*tp_size*/ |
| 354 | 0, /*tp_itemsize*/ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 355 | /* methods */ |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 356 | (destructor)reg_dealloc, /*tp_dealloc*/ |
| 357 | 0, /*tp_print*/ |
| 358 | (getattrfunc)regobj_getattr, /*tp_getattr*/ |
| 359 | 0, /*tp_setattr*/ |
| 360 | 0, /*tp_compare*/ |
| 361 | 0, /*tp_repr*/ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 362 | }; |
| 363 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 364 | /* reference counting invariants: |
| 365 | pattern: borrowed |
| 366 | translate: borrowed |
| 367 | givenpat: borrowed |
| 368 | groupindex: transferred |
| 369 | */ |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 370 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 371 | newregexobject(PyObject *pattern, PyObject *translate, PyObject *givenpat, PyObject *groupindex) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 372 | { |
| 373 | regexobject *re; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 374 | char *pat; |
| 375 | int size; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 376 | |
Guido van Rossum | 7e48898 | 1998-10-08 02:25:24 +0000 | [diff] [blame] | 377 | if (!PyArg_Parse(pattern, "t#", &pat, &size)) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 378 | return NULL; |
| 379 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 380 | if (translate != NULL && PyString_Size(translate) != 256) { |
| 381 | PyErr_SetString(RegexError, |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 382 | "translation table must be 256 bytes"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 383 | return NULL; |
| 384 | } |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 385 | re = PyObject_New(regexobject, &Regextype); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 386 | if (re != NULL) { |
| 387 | char *error; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 388 | re->re_patbuf.buffer = NULL; |
| 389 | re->re_patbuf.allocated = 0; |
Guido van Rossum | ed2554a | 1997-08-18 15:31:24 +0000 | [diff] [blame] | 390 | re->re_patbuf.fastmap = (unsigned char *)re->re_fastmap; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 391 | if (translate) { |
Guido van Rossum | ed2554a | 1997-08-18 15:31:24 +0000 | [diff] [blame] | 392 | re->re_patbuf.translate = (unsigned char *)PyString_AsString(translate); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 393 | if (!re->re_patbuf.translate) |
| 394 | goto finally; |
| 395 | Py_INCREF(translate); |
| 396 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 397 | else |
| 398 | re->re_patbuf.translate = NULL; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 399 | re->re_translate = translate; |
| 400 | re->re_lastok = NULL; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 401 | re->re_groupindex = groupindex; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 402 | Py_INCREF(pattern); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 403 | re->re_realpat = pattern; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 404 | Py_INCREF(givenpat); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 405 | re->re_givenpat = givenpat; |
Guido van Rossum | d19c04a | 1997-09-03 00:47:36 +0000 | [diff] [blame] | 406 | 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] | 407 | if (error != NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 408 | PyErr_SetString(RegexError, error); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 409 | goto finally; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 410 | } |
| 411 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 412 | return (PyObject *)re; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 413 | finally: |
| 414 | Py_DECREF(re); |
| 415 | return NULL; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 416 | } |
| 417 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 418 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 419 | regex_compile(PyObject *self, PyObject *args) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 420 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 421 | PyObject *pat = NULL; |
| 422 | PyObject *tran = NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 423 | |
Guido van Rossum | 43713e5 | 2000-02-29 13:59:29 +0000 | [diff] [blame] | 424 | if (!PyArg_ParseTuple(args, "S|S:compile", &pat, &tran)) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 425 | return NULL; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 426 | return newregexobject(pat, tran, pat, NULL); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 427 | } |
| 428 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 429 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 430 | symcomp(PyObject *pattern, PyObject *gdict) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 431 | { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 432 | char *opat, *oend, *o, *n, *g, *v; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 433 | int group_count = 0; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 434 | int sz; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 435 | int escaped = 0; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 436 | char name_buf[128]; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 437 | PyObject *npattern; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 438 | int require_escape = re_syntax & RE_NO_BK_PARENS ? 0 : 1; |
| 439 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 440 | if (!(opat = PyString_AsString(pattern))) |
| 441 | return NULL; |
| 442 | |
| 443 | if ((sz = PyString_Size(pattern)) < 0) |
| 444 | return NULL; |
| 445 | |
| 446 | oend = opat + sz; |
| 447 | o = opat; |
| 448 | |
Guido van Rossum | ab28c56 | 1996-06-11 18:33:14 +0000 | [diff] [blame] | 449 | if (oend == opat) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 450 | Py_INCREF(pattern); |
Guido van Rossum | ab28c56 | 1996-06-11 18:33:14 +0000 | [diff] [blame] | 451 | return pattern; |
| 452 | } |
| 453 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 454 | if (!(npattern = PyString_FromStringAndSize((char*)NULL, sz)) || |
| 455 | !(n = PyString_AsString(npattern))) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 456 | return NULL; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 457 | |
| 458 | while (o < oend) { |
| 459 | if (*o == '(' && escaped == require_escape) { |
| 460 | char *backtrack; |
| 461 | escaped = 0; |
| 462 | ++group_count; |
| 463 | *n++ = *o; |
| 464 | if (++o >= oend || *o != '<') |
| 465 | continue; |
| 466 | /* *o == '<' */ |
| 467 | if (o+1 < oend && *(o+1) == '>') |
| 468 | continue; |
| 469 | backtrack = o; |
| 470 | g = name_buf; |
| 471 | for (++o; o < oend;) { |
| 472 | if (*o == '>') { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 473 | PyObject *group_name = NULL; |
| 474 | PyObject *group_index = NULL; |
| 475 | *g++ = '\0'; |
| 476 | group_name = PyString_FromString(name_buf); |
| 477 | group_index = PyInt_FromLong(group_count); |
| 478 | if (group_name == NULL || |
| 479 | group_index == NULL || |
| 480 | PyDict_SetItem(gdict, group_name, |
| 481 | group_index) != 0) |
| 482 | { |
| 483 | Py_XDECREF(group_name); |
| 484 | Py_XDECREF(group_index); |
| 485 | Py_XDECREF(npattern); |
| 486 | return NULL; |
| 487 | } |
Barry Warsaw | 4bc9d39 | 1997-01-09 22:22:05 +0000 | [diff] [blame] | 488 | Py_DECREF(group_name); |
| 489 | Py_DECREF(group_index); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 490 | ++o; /* eat the '>' */ |
| 491 | break; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 492 | } |
Guido van Rossum | 7f7f274 | 1995-02-10 17:01:56 +0000 | [diff] [blame] | 493 | if (!isalnum(Py_CHARMASK(*o)) && *o != '_') { |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 494 | o = backtrack; |
| 495 | break; |
| 496 | } |
| 497 | *g++ = *o++; |
| 498 | } |
| 499 | } |
Guido van Rossum | 0cbaff4 | 1996-10-23 17:53:06 +0000 | [diff] [blame] | 500 | else if (*o == '[' && !escaped) { |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 501 | *n++ = *o; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 502 | ++o; /* eat the char following '[' */ |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 503 | *n++ = *o; |
| 504 | while (o < oend && *o != ']') { |
| 505 | ++o; |
| 506 | *n++ = *o; |
| 507 | } |
| 508 | if (o < oend) |
| 509 | ++o; |
| 510 | } |
| 511 | else if (*o == '\\') { |
| 512 | escaped = 1; |
| 513 | *n++ = *o; |
| 514 | ++o; |
| 515 | } |
| 516 | else { |
| 517 | escaped = 0; |
| 518 | *n++ = *o; |
| 519 | ++o; |
| 520 | } |
| 521 | } |
| 522 | |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 523 | if (!(v = PyString_AsString(npattern))) { |
| 524 | Py_DECREF(npattern); |
| 525 | return NULL; |
| 526 | } |
| 527 | /* _PyString_Resize() decrements npattern on failure */ |
| 528 | if (_PyString_Resize(&npattern, n - v) == 0) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 529 | return npattern; |
| 530 | else { |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 531 | return NULL; |
| 532 | } |
| 533 | |
| 534 | } |
| 535 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 536 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 537 | regex_symcomp(PyObject *self, PyObject *args) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 538 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 539 | PyObject *pattern; |
| 540 | PyObject *tran = NULL; |
| 541 | PyObject *gdict = NULL; |
| 542 | PyObject *npattern; |
Barry Warsaw | 4bc9d39 | 1997-01-09 22:22:05 +0000 | [diff] [blame] | 543 | PyObject *retval = NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 544 | |
Guido van Rossum | 43713e5 | 2000-02-29 13:59:29 +0000 | [diff] [blame] | 545 | if (!PyArg_ParseTuple(args, "S|S:symcomp", &pattern, &tran)) |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 546 | return NULL; |
| 547 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 548 | gdict = PyDict_New(); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 549 | if (gdict == NULL || (npattern = symcomp(pattern, gdict)) == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 550 | Py_DECREF(gdict); |
| 551 | Py_DECREF(pattern); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 552 | return NULL; |
| 553 | } |
Barry Warsaw | 4bc9d39 | 1997-01-09 22:22:05 +0000 | [diff] [blame] | 554 | retval = newregexobject(npattern, tran, pattern, gdict); |
| 555 | Py_DECREF(npattern); |
| 556 | return retval; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 557 | } |
| 558 | |
| 559 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 560 | static PyObject *cache_pat; |
| 561 | static PyObject *cache_prog; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 562 | |
| 563 | static int |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 564 | update_cache(PyObject *pat) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 565 | { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 566 | PyObject *tuple = Py_BuildValue("(O)", pat); |
| 567 | int status = 0; |
| 568 | |
| 569 | if (!tuple) |
| 570 | return -1; |
| 571 | |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 572 | if (pat != cache_pat) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 573 | Py_XDECREF(cache_pat); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 574 | cache_pat = NULL; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 575 | Py_XDECREF(cache_prog); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 576 | cache_prog = regex_compile((PyObject *)NULL, tuple); |
| 577 | if (cache_prog == NULL) { |
| 578 | status = -1; |
| 579 | goto finally; |
| 580 | } |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 581 | cache_pat = pat; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 582 | Py_INCREF(cache_pat); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 583 | } |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 584 | finally: |
| 585 | Py_DECREF(tuple); |
| 586 | return status; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 587 | } |
| 588 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 589 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 590 | regex_match(PyObject *self, PyObject *args) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 591 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 592 | PyObject *pat, *string; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 593 | PyObject *tuple, *v; |
| 594 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 595 | if (!PyArg_Parse(args, "(SS)", &pat, &string)) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 596 | return NULL; |
| 597 | if (update_cache(pat) < 0) |
| 598 | return NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 599 | |
| 600 | if (!(tuple = Py_BuildValue("(S)", string))) |
| 601 | return NULL; |
| 602 | v = regobj_match((regexobject *)cache_prog, tuple); |
| 603 | Py_DECREF(tuple); |
| 604 | return v; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 605 | } |
| 606 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 607 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 608 | regex_search(PyObject *self, PyObject *args) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 609 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 610 | PyObject *pat, *string; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 611 | PyObject *tuple, *v; |
| 612 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 613 | if (!PyArg_Parse(args, "(SS)", &pat, &string)) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 614 | return NULL; |
| 615 | if (update_cache(pat) < 0) |
| 616 | return NULL; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 617 | |
| 618 | if (!(tuple = Py_BuildValue("(S)", string))) |
| 619 | return NULL; |
| 620 | v = regobj_search((regexobject *)cache_prog, tuple); |
| 621 | Py_DECREF(tuple); |
| 622 | return v; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 623 | } |
| 624 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 625 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 626 | regex_set_syntax(PyObject *self, PyObject *args) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 627 | { |
| 628 | int syntax; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 629 | if (!PyArg_Parse(args, "i", &syntax)) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 630 | return NULL; |
| 631 | syntax = re_set_syntax(syntax); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 632 | /* wipe the global pattern cache */ |
| 633 | Py_XDECREF(cache_pat); |
| 634 | cache_pat = NULL; |
| 635 | Py_XDECREF(cache_prog); |
| 636 | cache_prog = NULL; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 637 | return PyInt_FromLong((long)syntax); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 638 | } |
| 639 | |
Barry Warsaw | 909d7c3 | 1997-02-18 18:48:50 +0000 | [diff] [blame] | 640 | static PyObject * |
Peter Schneider-Kamp | 7d0c71a | 2000-07-10 13:05:29 +0000 | [diff] [blame] | 641 | regex_get_syntax(PyObject *self, PyObject *args) |
Barry Warsaw | 909d7c3 | 1997-02-18 18:48:50 +0000 | [diff] [blame] | 642 | { |
| 643 | if (!PyArg_Parse(args, "")) |
| 644 | return NULL; |
| 645 | return PyInt_FromLong((long)re_syntax); |
| 646 | } |
| 647 | |
| 648 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 649 | static struct PyMethodDef regex_global_methods[] = { |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 650 | {"compile", regex_compile, 1}, |
| 651 | {"symcomp", regex_symcomp, 1}, |
Guido van Rossum | 295d171 | 1995-02-19 15:55:19 +0000 | [diff] [blame] | 652 | {"match", regex_match, 0}, |
| 653 | {"search", regex_search, 0}, |
| 654 | {"set_syntax", regex_set_syntax, 0}, |
Barry Warsaw | 909d7c3 | 1997-02-18 18:48:50 +0000 | [diff] [blame] | 655 | {"get_syntax", regex_get_syntax, 0}, |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 656 | {NULL, NULL} /* sentinel */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 657 | }; |
| 658 | |
Guido van Rossum | 3886bb6 | 1998-12-04 18:50:17 +0000 | [diff] [blame] | 659 | DL_EXPORT(void) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 660 | initregex() |
| 661 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 662 | PyObject *m, *d, *v; |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 663 | int i; |
| 664 | char *s; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 665 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 666 | m = Py_InitModule("regex", regex_global_methods); |
| 667 | d = PyModule_GetDict(m); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 668 | |
| 669 | /* Initialize regex.error exception */ |
Guido van Rossum | 0cb96de | 1997-10-01 04:29:29 +0000 | [diff] [blame] | 670 | v = RegexError = PyErr_NewException("regex.error", NULL, NULL); |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 671 | if (v == NULL || PyDict_SetItemString(d, "error", v) != 0) |
| 672 | goto finally; |
| 673 | |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 674 | /* Initialize regex.casefold constant */ |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 675 | if (!(v = PyString_FromStringAndSize((char *)NULL, 256))) |
| 676 | goto finally; |
| 677 | |
| 678 | if (!(s = PyString_AsString(v))) |
| 679 | goto finally; |
| 680 | |
| 681 | for (i = 0; i < 256; i++) { |
| 682 | if (isupper(i)) |
| 683 | s[i] = tolower(i); |
| 684 | else |
| 685 | s[i] = i; |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 686 | } |
Barry Warsaw | c357325 | 1996-12-20 21:56:07 +0000 | [diff] [blame] | 687 | if (PyDict_SetItemString(d, "casefold", v) < 0) |
| 688 | goto finally; |
| 689 | Py_DECREF(v); |
| 690 | |
| 691 | if (!PyErr_Occurred()) |
| 692 | return; |
| 693 | finally: |
Guido van Rossum | 0cb96de | 1997-10-01 04:29:29 +0000 | [diff] [blame] | 694 | /* Nothing */ ; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 695 | } |