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 | |
| 12 | Permission to use, copy, modify, and distribute this software and its |
| 13 | documentation for any purpose and without fee is hereby granted, |
| 14 | provided that the above copyright notice appear in all copies and that |
| 15 | both that copyright notice and this permission notice appear in |
| 16 | supporting documentation, and that the names of Stichting Mathematisch |
| 17 | Centrum or CWI not be used in advertising or publicity pertaining to |
| 18 | distribution of the software without specific, written prior permission. |
| 19 | |
| 20 | STICHTING MATHEMATISCH CENTRUM DISCLAIMS ALL WARRANTIES WITH REGARD TO |
| 21 | THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND |
| 22 | FITNESS, IN NO EVENT SHALL STICHTING MATHEMATISCH CENTRUM BE LIABLE |
| 23 | FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES |
| 24 | WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN |
| 25 | ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT |
| 26 | OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. |
| 27 | |
| 28 | ******************************************************************/ |
| 29 | |
| 30 | /* Regular expression objects */ |
Guido van Rossum | 1cab95c | 1992-01-19 16:31:57 +0000 | [diff] [blame] | 31 | /* This uses Tatu Ylonen's copyleft-free reimplementation of |
| 32 | GNU regular expressions */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 33 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 34 | #include "Python.h" |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 35 | |
Guido van Rossum | 1cab95c | 1992-01-19 16:31:57 +0000 | [diff] [blame] | 36 | #include "regexpr.h" |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 37 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 38 | static PyObject *RegexError; /* Exception */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 39 | |
| 40 | typedef struct { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 41 | PyObject_HEAD |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 42 | struct re_pattern_buffer re_patbuf; /* The compiled expression */ |
| 43 | struct re_registers re_regs; /* The registers from the last match */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 44 | char re_fastmap[256]; /* Storage for fastmap */ |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 45 | PyObject *re_translate; /* String object for translate table */ |
| 46 | PyObject *re_lastok; /* String object last matched/searched */ |
| 47 | PyObject *re_groupindex; /* Group name to index dictionary */ |
| 48 | PyObject *re_givenpat; /* Pattern with symbolic groups */ |
| 49 | PyObject *re_realpat; /* Pattern without symbolic groups */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 50 | } regexobject; |
| 51 | |
| 52 | /* Regex object methods */ |
| 53 | |
| 54 | static void |
| 55 | reg_dealloc(re) |
| 56 | regexobject *re; |
| 57 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 58 | PyMem_XDEL(re->re_patbuf.buffer); |
| 59 | Py_XDECREF(re->re_translate); |
| 60 | Py_XDECREF(re->re_lastok); |
| 61 | Py_XDECREF(re->re_groupindex); |
| 62 | Py_XDECREF(re->re_givenpat); |
| 63 | Py_XDECREF(re->re_realpat); |
| 64 | PyMem_DEL(re); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 65 | } |
| 66 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 67 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 68 | makeresult(regs) |
| 69 | struct re_registers *regs; |
| 70 | { |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 71 | PyObject *v; |
| 72 | int i; |
| 73 | static PyObject *filler = NULL; |
| 74 | if (filler == NULL) { |
| 75 | filler = Py_BuildValue("(ii)", -1, -1); |
| 76 | if (filler == NULL) |
| 77 | return NULL; |
| 78 | } |
| 79 | v = PyTuple_New(RE_NREGS); |
| 80 | if (v == NULL) |
| 81 | return NULL; |
| 82 | for (i = 0; i < RE_NREGS; i++) { |
| 83 | int lo = regs->start[i]; |
| 84 | int hi = regs->end[i]; |
| 85 | PyObject *w; |
| 86 | if (lo == -1 && hi == -1) { |
| 87 | w = filler; |
| 88 | Py_INCREF(w); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 89 | } |
Guido van Rossum | c196202 | 1996-10-08 14:18:42 +0000 | [diff] [blame] | 90 | else |
| 91 | w = Py_BuildValue("(ii)", lo, hi); |
| 92 | if (w == NULL) { |
| 93 | Py_XDECREF(v); |
| 94 | return NULL; |
| 95 | } |
| 96 | PyTuple_SetItem(v, i, w); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 97 | } |
| 98 | return v; |
| 99 | } |
| 100 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 101 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 102 | reg_match(re, args) |
| 103 | regexobject *re; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 104 | PyObject *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 105 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 106 | PyObject *argstring; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 107 | char *buffer; |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 108 | int size; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 109 | int offset; |
| 110 | int result; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 111 | if (PyArg_Parse(args, "S", &argstring)) { |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 112 | offset = 0; |
| 113 | } |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 114 | else { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 115 | PyErr_Clear(); |
| 116 | if (!PyArg_Parse(args, "(Si)", &argstring, &offset)) |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 117 | return NULL; |
| 118 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 119 | buffer = PyString_AsString(argstring); |
| 120 | size = PyString_Size(argstring); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 121 | if (offset < 0 || offset > size) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 122 | PyErr_SetString(RegexError, "match offset out of range"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 123 | return NULL; |
| 124 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 125 | Py_XDECREF(re->re_lastok); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 126 | re->re_lastok = NULL; |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 127 | result = re_match(&re->re_patbuf, buffer, size, offset, &re->re_regs); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 128 | if (result < -1) { |
| 129 | /* Failure like stack overflow */ |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 130 | PyErr_SetString(RegexError, "match failure"); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 131 | return NULL; |
| 132 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 133 | if (result >= 0) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 134 | Py_INCREF(argstring); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 135 | re->re_lastok = argstring; |
| 136 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 137 | return PyInt_FromLong((long)result); /* Length of the match or -1 */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 138 | } |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 139 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 140 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 141 | reg_search(re, args) |
| 142 | regexobject *re; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 143 | PyObject *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 144 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 145 | PyObject *argstring; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 146 | char *buffer; |
| 147 | int size; |
| 148 | int offset; |
| 149 | int range; |
| 150 | int result; |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 151 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 152 | if (PyArg_Parse(args, "S", &argstring)) { |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 153 | offset = 0; |
| 154 | } |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 155 | else { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 156 | PyErr_Clear(); |
| 157 | if (!PyArg_Parse(args, "(Si)", &argstring, &offset)) |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 158 | return NULL; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 159 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 160 | buffer = PyString_AsString(argstring); |
| 161 | size = PyString_Size(argstring); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 162 | if (offset < 0 || offset > size) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 163 | PyErr_SetString(RegexError, "search offset out of range"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 164 | return NULL; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 165 | } |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 166 | /* NB: In Emacs 18.57, the documentation for re_search[_2] and |
| 167 | the implementation don't match: the documentation states that |
| 168 | |range| positions are tried, while the code tries |range|+1 |
| 169 | positions. It seems more productive to believe the code! */ |
Guido van Rossum | 2d78590 | 1992-01-26 18:12:41 +0000 | [diff] [blame] | 170 | range = size - offset; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 171 | Py_XDECREF(re->re_lastok); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 172 | re->re_lastok = NULL; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 173 | result = re_search(&re->re_patbuf, buffer, size, offset, range, |
| 174 | &re->re_regs); |
| 175 | if (result < -1) { |
| 176 | /* Failure like stack overflow */ |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 177 | PyErr_SetString(RegexError, "match failure"); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 178 | return NULL; |
| 179 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 180 | if (result >= 0) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 181 | Py_INCREF(argstring); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 182 | re->re_lastok = argstring; |
| 183 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 184 | return PyInt_FromLong((long)result); /* Position of the match or -1 */ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 185 | } |
| 186 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 187 | static PyObject * |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 188 | reg_group(re, args) |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 189 | regexobject *re; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 190 | PyObject *args; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 191 | { |
| 192 | int i, a, b; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 193 | if (args != NULL && PyTuple_Check(args)) { |
| 194 | int n = PyTuple_Size(args); |
| 195 | PyObject *res = PyTuple_New(n); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 196 | if (res == NULL) |
| 197 | return NULL; |
| 198 | for (i = 0; i < n; i++) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 199 | PyObject *v = reg_group(re, PyTuple_GetItem(args, i)); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 200 | if (v == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 201 | Py_DECREF(res); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 202 | return NULL; |
| 203 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 204 | PyTuple_SetItem(res, i, v); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 205 | } |
| 206 | return res; |
| 207 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 208 | if (!PyArg_Parse(args, "i", &i)) { |
| 209 | PyObject *n; |
| 210 | PyErr_Clear(); |
| 211 | if (!PyArg_Parse(args, "S", &n)) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 212 | return NULL; |
| 213 | else { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 214 | PyObject *index; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 215 | if (re->re_groupindex == NULL) |
| 216 | index = NULL; |
| 217 | else |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 218 | index = PyDict_GetItem(re->re_groupindex, n); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 219 | if (index == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 220 | PyErr_SetString(RegexError, "group() group name doesn't exist"); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 221 | return NULL; |
| 222 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 223 | i = PyInt_AsLong(index); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 224 | } |
| 225 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 226 | if (i < 0 || i >= RE_NREGS) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 227 | PyErr_SetString(RegexError, "group() index out of range"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 228 | return NULL; |
| 229 | } |
| 230 | if (re->re_lastok == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 231 | PyErr_SetString(RegexError, |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 232 | "group() only valid after successful match/search"); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 233 | return NULL; |
| 234 | } |
| 235 | a = re->re_regs.start[i]; |
| 236 | b = re->re_regs.end[i]; |
| 237 | if (a < 0 || b < 0) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 238 | Py_INCREF(Py_None); |
| 239 | return Py_None; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 240 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 241 | return PyString_FromStringAndSize(PyString_AsString(re->re_lastok)+a, b-a); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 242 | } |
| 243 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 244 | static struct PyMethodDef reg_methods[] = { |
| 245 | {"match", (PyCFunction)reg_match}, |
| 246 | {"search", (PyCFunction)reg_search}, |
| 247 | {"group", (PyCFunction)reg_group}, |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 248 | {NULL, NULL} /* sentinel */ |
| 249 | }; |
| 250 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 251 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 252 | reg_getattr(re, name) |
| 253 | regexobject *re; |
| 254 | char *name; |
| 255 | { |
Guido van Rossum | b824fc6 | 1992-01-01 14:52:16 +0000 | [diff] [blame] | 256 | if (strcmp(name, "regs") == 0) { |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 257 | if (re->re_lastok == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 258 | Py_INCREF(Py_None); |
| 259 | return Py_None; |
Guido van Rossum | b824fc6 | 1992-01-01 14:52:16 +0000 | [diff] [blame] | 260 | } |
| 261 | return makeresult(&re->re_regs); |
| 262 | } |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 263 | if (strcmp(name, "last") == 0) { |
| 264 | if (re->re_lastok == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 265 | Py_INCREF(Py_None); |
| 266 | return Py_None; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 267 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 268 | Py_INCREF(re->re_lastok); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 269 | return re->re_lastok; |
| 270 | } |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 271 | if (strcmp(name, "translate") == 0) { |
| 272 | if (re->re_translate == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 273 | Py_INCREF(Py_None); |
| 274 | return Py_None; |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 275 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 276 | Py_INCREF(re->re_translate); |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 277 | return re->re_translate; |
| 278 | } |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 279 | if (strcmp(name, "groupindex") == 0) { |
| 280 | if (re->re_groupindex == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 281 | Py_INCREF(Py_None); |
| 282 | return Py_None; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 283 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 284 | Py_INCREF(re->re_groupindex); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 285 | return re->re_groupindex; |
| 286 | } |
| 287 | if (strcmp(name, "realpat") == 0) { |
| 288 | if (re->re_realpat == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 289 | Py_INCREF(Py_None); |
| 290 | return Py_None; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 291 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 292 | Py_INCREF(re->re_realpat); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 293 | return re->re_realpat; |
| 294 | } |
| 295 | if (strcmp(name, "givenpat") == 0) { |
| 296 | if (re->re_givenpat == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 297 | Py_INCREF(Py_None); |
| 298 | return Py_None; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 299 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 300 | Py_INCREF(re->re_givenpat); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 301 | return re->re_givenpat; |
| 302 | } |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 303 | if (strcmp(name, "__members__") == 0) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 304 | PyObject *list = PyList_New(6); |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 305 | if (list) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 306 | PyList_SetItem(list, 0, PyString_FromString("last")); |
| 307 | PyList_SetItem(list, 1, PyString_FromString("regs")); |
| 308 | PyList_SetItem(list, 2, PyString_FromString("translate")); |
| 309 | PyList_SetItem(list, 3, PyString_FromString("groupindex")); |
| 310 | PyList_SetItem(list, 4, PyString_FromString("realpat")); |
| 311 | PyList_SetItem(list, 5, PyString_FromString("givenpat")); |
| 312 | if (PyErr_Occurred()) { |
| 313 | Py_DECREF(list); |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 314 | list = NULL; |
| 315 | } |
| 316 | } |
| 317 | return list; |
| 318 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 319 | return Py_FindMethod(reg_methods, (PyObject *)re, name); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 320 | } |
| 321 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 322 | static PyTypeObject Regextype = { |
| 323 | PyObject_HEAD_INIT(&PyType_Type) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 324 | 0, /*ob_size*/ |
| 325 | "regex", /*tp_name*/ |
| 326 | sizeof(regexobject), /*tp_size*/ |
| 327 | 0, /*tp_itemsize*/ |
| 328 | /* methods */ |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 329 | (destructor)reg_dealloc, /*tp_dealloc*/ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 330 | 0, /*tp_print*/ |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 331 | (getattrfunc)reg_getattr, /*tp_getattr*/ |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 332 | 0, /*tp_setattr*/ |
| 333 | 0, /*tp_compare*/ |
| 334 | 0, /*tp_repr*/ |
| 335 | }; |
| 336 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 337 | static PyObject * |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 338 | newregexobject(pattern, translate, givenpat, groupindex) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 339 | PyObject *pattern; |
| 340 | PyObject *translate; |
| 341 | PyObject *givenpat; |
| 342 | PyObject *groupindex; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 343 | { |
| 344 | regexobject *re; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 345 | char *pat = PyString_AsString(pattern); |
| 346 | int size = PyString_Size(pattern); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 347 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 348 | if (translate != NULL && PyString_Size(translate) != 256) { |
| 349 | PyErr_SetString(RegexError, |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 350 | "translation table must be 256 bytes"); |
| 351 | return NULL; |
| 352 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 353 | re = PyObject_NEW(regexobject, &Regextype); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 354 | if (re != NULL) { |
| 355 | char *error; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 356 | re->re_patbuf.buffer = NULL; |
| 357 | re->re_patbuf.allocated = 0; |
| 358 | re->re_patbuf.fastmap = re->re_fastmap; |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 359 | if (translate) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 360 | re->re_patbuf.translate = PyString_AsString(translate); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 361 | else |
| 362 | re->re_patbuf.translate = NULL; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 363 | Py_XINCREF(translate); |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 364 | re->re_translate = translate; |
| 365 | re->re_lastok = NULL; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 366 | re->re_groupindex = groupindex; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 367 | Py_INCREF(pattern); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 368 | re->re_realpat = pattern; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 369 | Py_INCREF(givenpat); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 370 | re->re_givenpat = givenpat; |
Guido van Rossum | d577c0c | 1992-01-27 16:46:19 +0000 | [diff] [blame] | 371 | error = re_compile_pattern(pat, size, &re->re_patbuf); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 372 | if (error != NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 373 | PyErr_SetString(RegexError, error); |
| 374 | Py_DECREF(re); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 375 | re = NULL; |
| 376 | } |
| 377 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 378 | return (PyObject *)re; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 379 | } |
| 380 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 381 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 382 | regex_compile(self, args) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 383 | PyObject *self; |
| 384 | PyObject *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 385 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 386 | PyObject *pat = NULL; |
| 387 | PyObject *tran = NULL; |
| 388 | if (!PyArg_Parse(args, "S", &pat)) { |
| 389 | PyErr_Clear(); |
| 390 | if (!PyArg_Parse(args, "(SS)", &pat, &tran)) |
Guido van Rossum | 36d330b | 1993-02-21 20:12:16 +0000 | [diff] [blame] | 391 | return NULL; |
| 392 | } |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 393 | return newregexobject(pat, tran, pat, NULL); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 394 | } |
| 395 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 396 | static PyObject * |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 397 | symcomp(pattern, gdict) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 398 | PyObject *pattern; |
| 399 | PyObject *gdict; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 400 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 401 | char *opat = PyString_AsString(pattern); |
| 402 | char *oend = opat + PyString_Size(pattern); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 403 | int group_count = 0; |
| 404 | int escaped = 0; |
| 405 | char *o = opat; |
| 406 | char *n; |
| 407 | char name_buf[128]; |
| 408 | char *g; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 409 | PyObject *npattern; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 410 | int require_escape = re_syntax & RE_NO_BK_PARENS ? 0 : 1; |
| 411 | |
Guido van Rossum | ab28c56 | 1996-06-11 18:33:14 +0000 | [diff] [blame] | 412 | if (oend == opat) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 413 | Py_INCREF(pattern); |
Guido van Rossum | ab28c56 | 1996-06-11 18:33:14 +0000 | [diff] [blame] | 414 | return pattern; |
| 415 | } |
| 416 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 417 | npattern = PyString_FromStringAndSize((char*)NULL, PyString_Size(pattern)); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 418 | if (npattern == NULL) |
| 419 | return NULL; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 420 | n = PyString_AsString(npattern); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 421 | |
| 422 | while (o < oend) { |
| 423 | if (*o == '(' && escaped == require_escape) { |
| 424 | char *backtrack; |
| 425 | escaped = 0; |
| 426 | ++group_count; |
| 427 | *n++ = *o; |
| 428 | if (++o >= oend || *o != '<') |
| 429 | continue; |
| 430 | /* *o == '<' */ |
| 431 | if (o+1 < oend && *(o+1) == '>') |
| 432 | continue; |
| 433 | backtrack = o; |
| 434 | g = name_buf; |
| 435 | for (++o; o < oend;) { |
| 436 | if (*o == '>') { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 437 | PyObject *group_name = NULL; |
| 438 | PyObject *group_index = NULL; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 439 | *g++ = '\0'; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 440 | group_name = PyString_FromString(name_buf); |
| 441 | group_index = PyInt_FromLong(group_count); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 442 | if (group_name == NULL || group_index == NULL |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 443 | || PyDict_SetItem(gdict, group_name, group_index) != 0) { |
| 444 | Py_XDECREF(group_name); |
| 445 | Py_XDECREF(group_index); |
| 446 | Py_XDECREF(npattern); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 447 | return NULL; |
| 448 | } |
| 449 | ++o; /* eat the '>' */ |
| 450 | break; |
| 451 | } |
Guido van Rossum | 7f7f274 | 1995-02-10 17:01:56 +0000 | [diff] [blame] | 452 | if (!isalnum(Py_CHARMASK(*o)) && *o != '_') { |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 453 | o = backtrack; |
| 454 | break; |
| 455 | } |
| 456 | *g++ = *o++; |
| 457 | } |
| 458 | } |
Guido van Rossum | 0cbaff4 | 1996-10-23 17:53:06 +0000 | [diff] [blame] | 459 | else if (*o == '[' && !escaped) { |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 460 | *n++ = *o; |
| 461 | ++o; /* eat the char following '[' */ |
| 462 | *n++ = *o; |
| 463 | while (o < oend && *o != ']') { |
| 464 | ++o; |
| 465 | *n++ = *o; |
| 466 | } |
| 467 | if (o < oend) |
| 468 | ++o; |
| 469 | } |
| 470 | else if (*o == '\\') { |
| 471 | escaped = 1; |
| 472 | *n++ = *o; |
| 473 | ++o; |
| 474 | } |
| 475 | else { |
| 476 | escaped = 0; |
| 477 | *n++ = *o; |
| 478 | ++o; |
| 479 | } |
| 480 | } |
| 481 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 482 | if (_PyString_Resize(&npattern, n - PyString_AsString(npattern)) == 0) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 483 | return npattern; |
| 484 | else { |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 485 | return NULL; |
| 486 | } |
| 487 | |
| 488 | } |
| 489 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 490 | static PyObject * |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 491 | regex_symcomp(self, args) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 492 | PyObject *self; |
| 493 | PyObject *args; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 494 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 495 | PyObject *pattern; |
| 496 | PyObject *tran = NULL; |
| 497 | PyObject *gdict = NULL; |
| 498 | PyObject *npattern; |
| 499 | if (!PyArg_Parse(args, "S", &pattern)) { |
| 500 | PyErr_Clear(); |
| 501 | if (!PyArg_Parse(args, "(SS)", &pattern, &tran)) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 502 | return NULL; |
| 503 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 504 | gdict = PyDict_New(); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 505 | if (gdict == NULL |
| 506 | || (npattern = symcomp(pattern, gdict)) == NULL) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 507 | Py_DECREF(gdict); |
| 508 | Py_DECREF(pattern); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 509 | return NULL; |
| 510 | } |
| 511 | return newregexobject(npattern, tran, pattern, gdict); |
| 512 | } |
| 513 | |
| 514 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 515 | static PyObject *cache_pat; |
| 516 | static PyObject *cache_prog; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 517 | |
| 518 | static int |
| 519 | update_cache(pat) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 520 | PyObject *pat; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 521 | { |
| 522 | if (pat != cache_pat) { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 523 | Py_XDECREF(cache_pat); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 524 | cache_pat = NULL; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 525 | Py_XDECREF(cache_prog); |
| 526 | cache_prog = regex_compile((PyObject *)NULL, pat); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 527 | if (cache_prog == NULL) |
| 528 | return -1; |
| 529 | cache_pat = pat; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 530 | Py_INCREF(cache_pat); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 531 | } |
| 532 | return 0; |
| 533 | } |
| 534 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 535 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 536 | regex_match(self, args) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 537 | PyObject *self; |
| 538 | PyObject *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 539 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 540 | PyObject *pat, *string; |
| 541 | if (!PyArg_Parse(args, "(SS)", &pat, &string)) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 542 | return NULL; |
| 543 | if (update_cache(pat) < 0) |
| 544 | return NULL; |
| 545 | return reg_match((regexobject *)cache_prog, string); |
| 546 | } |
| 547 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 548 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 549 | regex_search(self, args) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 550 | PyObject *self; |
| 551 | PyObject *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 552 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 553 | PyObject *pat, *string; |
| 554 | if (!PyArg_Parse(args, "(SS)", &pat, &string)) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 555 | return NULL; |
| 556 | if (update_cache(pat) < 0) |
| 557 | return NULL; |
| 558 | return reg_search((regexobject *)cache_prog, string); |
| 559 | } |
| 560 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 561 | static PyObject * |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 562 | regex_set_syntax(self, args) |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 563 | PyObject *self, *args; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 564 | { |
| 565 | int syntax; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 566 | if (!PyArg_Parse(args, "i", &syntax)) |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 567 | return NULL; |
| 568 | syntax = re_set_syntax(syntax); |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 569 | return PyInt_FromLong((long)syntax); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 570 | } |
| 571 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 572 | static struct PyMethodDef regex_global_methods[] = { |
Guido van Rossum | 295d171 | 1995-02-19 15:55:19 +0000 | [diff] [blame] | 573 | {"compile", regex_compile, 0}, |
| 574 | {"symcomp", regex_symcomp, 0}, |
| 575 | {"match", regex_match, 0}, |
| 576 | {"search", regex_search, 0}, |
| 577 | {"set_syntax", regex_set_syntax, 0}, |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 578 | {NULL, NULL} /* sentinel */ |
| 579 | }; |
| 580 | |
Guido van Rossum | 8f3032d | 1996-08-19 22:03:12 +0000 | [diff] [blame] | 581 | void |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 582 | initregex() |
| 583 | { |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 584 | PyObject *m, *d, *v; |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 585 | |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 586 | m = Py_InitModule("regex", regex_global_methods); |
| 587 | d = PyModule_GetDict(m); |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 588 | |
| 589 | /* Initialize regex.error exception */ |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 590 | RegexError = PyString_FromString("regex.error"); |
| 591 | if (RegexError == NULL || PyDict_SetItemString(d, "error", RegexError) != 0) |
| 592 | Py_FatalError("can't define regex.error"); |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 593 | |
| 594 | /* Initialize regex.casefold constant */ |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 595 | v = PyString_FromStringAndSize((char *)NULL, 256); |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 596 | if (v != NULL) { |
| 597 | int i; |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 598 | char *s = PyString_AsString(v); |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 599 | for (i = 0; i < 256; i++) { |
| 600 | if (isupper(i)) |
| 601 | s[i] = tolower(i); |
| 602 | else |
| 603 | s[i] = i; |
| 604 | } |
Guido van Rossum | dfe8ad9 | 1996-07-24 00:51:20 +0000 | [diff] [blame] | 605 | PyDict_SetItemString(d, "casefold", v); |
| 606 | Py_DECREF(v); |
Guido van Rossum | ccd5bad | 1993-02-23 13:42:39 +0000 | [diff] [blame] | 607 | } |
Guido van Rossum | 6f4c43d | 1991-12-30 01:42:57 +0000 | [diff] [blame] | 608 | } |