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