Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 1 | /*********************************************************** |
Guido van Rossum | 524b588 | 1995-01-04 19:10:35 +0000 | [diff] [blame] | 2 | Copyright 1991-1995 by Stichting Mathematisch Centrum, Amsterdam, |
| 3 | The Netherlands. |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 4 | |
| 5 | All Rights Reserved |
| 6 | |
Guido van Rossum | d266eb4 | 1996-10-25 14:44:06 +0000 | [diff] [blame] | 7 | Permission to use, copy, modify, and distribute this software and its |
| 8 | documentation for any purpose and without fee is hereby granted, |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 9 | provided that the above copyright notice appear in all copies and that |
Guido van Rossum | d266eb4 | 1996-10-25 14:44:06 +0000 | [diff] [blame] | 10 | both that copyright notice and this permission notice appear in |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 11 | supporting documentation, and that the names of Stichting Mathematisch |
Guido van Rossum | d266eb4 | 1996-10-25 14:44:06 +0000 | [diff] [blame] | 12 | Centrum or CWI or Corporation for National Research Initiatives or |
| 13 | CNRI not be used in advertising or publicity pertaining to |
| 14 | distribution of the software without specific, written prior |
| 15 | permission. |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 16 | |
Guido van Rossum | d266eb4 | 1996-10-25 14:44:06 +0000 | [diff] [blame] | 17 | While CWI is the initial source for this software, a modified version |
| 18 | is made available by the Corporation for National Research Initiatives |
| 19 | (CNRI) at the Internet address ftp://ftp.python.org. |
| 20 | |
| 21 | STICHTING MATHEMATISCH CENTRUM AND CNRI DISCLAIM ALL WARRANTIES WITH |
| 22 | REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF |
| 23 | MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL STICHTING MATHEMATISCH |
| 24 | CENTRUM OR CNRI BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL |
| 25 | DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR |
| 26 | PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER |
| 27 | TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR |
| 28 | PERFORMANCE OF THIS SOFTWARE. |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 29 | |
| 30 | ******************************************************************/ |
| 31 | |
| 32 | /* strop module */ |
| 33 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 34 | static char strop_module__doc__[] = |
Guido van Rossum | 23e21e7 | 1997-12-29 19:57:36 +0000 | [diff] [blame] | 35 | "Common string manipulations, optimized for speed.\n\ |
| 36 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 37 | Always use \"import string\" rather than referencing\n\ |
Guido van Rossum | 23e21e7 | 1997-12-29 19:57:36 +0000 | [diff] [blame] | 38 | this module directly."; |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 39 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 40 | #include "Python.h" |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 41 | |
Guido van Rossum | 7b7c578 | 1997-03-14 04:13:56 +0000 | [diff] [blame] | 42 | #ifdef HAVE_LIMITS_H |
| 43 | #include <limits.h> |
| 44 | #else |
| 45 | #define INT_MAX 2147483647 |
| 46 | #endif |
| 47 | |
Guido van Rossum | d05eb8b | 1993-07-08 11:12:36 +0000 | [diff] [blame] | 48 | #include <ctype.h> |
Guido van Rossum | e22e644 | 1993-07-09 10:51:31 +0000 | [diff] [blame] | 49 | /* XXX This file assumes that the <ctype.h> is*() functions |
| 50 | XXX are defined for all 8-bit characters! */ |
Guido van Rossum | d05eb8b | 1993-07-08 11:12:36 +0000 | [diff] [blame] | 51 | |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 52 | /* The lstrip(), rstrip() and strip() functions are implemented |
| 53 | in do_strip(), which uses an additional parameter to indicate what |
| 54 | type of strip should occur. */ |
| 55 | |
| 56 | #define LEFTSTRIP 0 |
| 57 | #define RIGHTSTRIP 1 |
| 58 | #define BOTHSTRIP 2 |
| 59 | |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 60 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 61 | static PyObject * |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 62 | split_whitespace(s, len, maxsplit) |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 63 | char *s; |
Guido van Rossum | 009e79b | 1995-05-03 17:40:23 +0000 | [diff] [blame] | 64 | int len; |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 65 | int maxsplit; |
Guido van Rossum | 009e79b | 1995-05-03 17:40:23 +0000 | [diff] [blame] | 66 | { |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 67 | int i = 0, j, err; |
| 68 | int countsplit = 0; |
| 69 | PyObject* item; |
| 70 | PyObject *list = PyList_New(0); |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 71 | |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 72 | if (list == NULL) |
| 73 | return NULL; |
| 74 | |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 75 | while (i < len) { |
Guido van Rossum | 7f7f274 | 1995-02-10 17:01:56 +0000 | [diff] [blame] | 76 | while (i < len && isspace(Py_CHARMASK(s[i]))) { |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 77 | i = i+1; |
| 78 | } |
| 79 | j = i; |
Guido van Rossum | ee1813d | 1995-02-14 00:58:59 +0000 | [diff] [blame] | 80 | while (i < len && !isspace(Py_CHARMASK(s[i]))) { |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 81 | i = i+1; |
| 82 | } |
| 83 | if (j < i) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 84 | item = PyString_FromStringAndSize(s+j, (int)(i-j)); |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 85 | if (item == NULL) |
| 86 | goto finally; |
| 87 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 88 | err = PyList_Append(list, item); |
| 89 | Py_DECREF(item); |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 90 | if (err < 0) |
| 91 | goto finally; |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 92 | |
| 93 | countsplit++; |
Barry Warsaw | 93be92d | 1997-12-02 00:29:30 +0000 | [diff] [blame] | 94 | while (i < len && isspace(Py_CHARMASK(s[i]))) { |
| 95 | i = i+1; |
| 96 | } |
| 97 | if (maxsplit && (countsplit >= maxsplit) && i < len) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 98 | item = PyString_FromStringAndSize( |
| 99 | s+i, (int)(len - i)); |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 100 | if (item == NULL) |
| 101 | goto finally; |
| 102 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 103 | err = PyList_Append(list, item); |
| 104 | Py_DECREF(item); |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 105 | if (err < 0) |
| 106 | goto finally; |
| 107 | |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 108 | i = len; |
| 109 | } |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 110 | } |
| 111 | } |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 112 | return list; |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 113 | finally: |
| 114 | Py_DECREF(list); |
| 115 | return NULL; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 116 | } |
| 117 | |
| 118 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 119 | static char splitfields__doc__[] = |
| 120 | "split(str [,sep [,maxsplit]]) -> list of strings\n\ |
| 121 | splitfields(str [,sep [,maxsplit]]) -> list of strings\n\ |
| 122 | \n\ |
| 123 | Return a list of the words in the string s, using sep as the\n\ |
| 124 | delimiter string. If maxsplit is nonzero, splits into at most\n\ |
| 125 | maxsplit words If sep is not specified, any whitespace string\n\ |
| 126 | is a separator. Maxsplit defaults to 0.\n\ |
| 127 | \n\ |
| 128 | (split and splitfields are synonymous)"; |
| 129 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 130 | static PyObject * |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 131 | strop_splitfields(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 132 | PyObject *self; /* Not used */ |
| 133 | PyObject *args; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 134 | { |
Guido van Rossum | 572d2d9 | 1993-11-05 10:14:49 +0000 | [diff] [blame] | 135 | int len, n, i, j, err; |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 136 | int splitcount, maxsplit; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 137 | char *s, *sub; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 138 | PyObject *list, *item; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 139 | |
Guido van Rossum | 009e79b | 1995-05-03 17:40:23 +0000 | [diff] [blame] | 140 | sub = NULL; |
| 141 | n = 0; |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 142 | splitcount = 0; |
| 143 | maxsplit = 0; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 144 | if (!PyArg_ParseTuple(args, "s#|z#i", &s, &len, &sub, &n, &maxsplit)) |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 145 | return NULL; |
Guido van Rossum | 009e79b | 1995-05-03 17:40:23 +0000 | [diff] [blame] | 146 | if (sub == NULL) |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 147 | return split_whitespace(s, len, maxsplit); |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 148 | if (n == 0) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 149 | PyErr_SetString(PyExc_ValueError, "empty separator"); |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 150 | return NULL; |
| 151 | } |
| 152 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 153 | list = PyList_New(0); |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 154 | if (list == NULL) |
| 155 | return NULL; |
| 156 | |
| 157 | i = j = 0; |
| 158 | while (i+n <= len) { |
Guido van Rossum | a0ca4c4 | 1996-10-04 13:39:37 +0000 | [diff] [blame] | 159 | if (s[i] == sub[0] && (n == 1 || memcmp(s+i, sub, n) == 0)) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 160 | item = PyString_FromStringAndSize(s+j, (int)(i-j)); |
Guido van Rossum | 572d2d9 | 1993-11-05 10:14:49 +0000 | [diff] [blame] | 161 | if (item == NULL) |
| 162 | goto fail; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 163 | err = PyList_Append(list, item); |
| 164 | Py_DECREF(item); |
Guido van Rossum | 572d2d9 | 1993-11-05 10:14:49 +0000 | [diff] [blame] | 165 | if (err < 0) |
| 166 | goto fail; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 167 | i = j = i + n; |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 168 | splitcount++; |
| 169 | if (maxsplit && (splitcount >= maxsplit)) |
| 170 | break; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 171 | } |
| 172 | else |
| 173 | i++; |
| 174 | } |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 175 | item = PyString_FromStringAndSize(s+j, (int)(len-j)); |
Guido van Rossum | 572d2d9 | 1993-11-05 10:14:49 +0000 | [diff] [blame] | 176 | if (item == NULL) |
| 177 | goto fail; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 178 | err = PyList_Append(list, item); |
| 179 | Py_DECREF(item); |
Guido van Rossum | 572d2d9 | 1993-11-05 10:14:49 +0000 | [diff] [blame] | 180 | if (err < 0) |
| 181 | goto fail; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 182 | |
| 183 | return list; |
Guido van Rossum | 572d2d9 | 1993-11-05 10:14:49 +0000 | [diff] [blame] | 184 | |
| 185 | fail: |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 186 | Py_DECREF(list); |
Guido van Rossum | 572d2d9 | 1993-11-05 10:14:49 +0000 | [diff] [blame] | 187 | return NULL; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 188 | } |
| 189 | |
| 190 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 191 | static char joinfields__doc__[] = |
| 192 | "join(list [,sep]) -> string\n\ |
| 193 | joinfields(list [,sep]) -> string\n\ |
| 194 | \n\ |
| 195 | Return a string composed of the words in list, with\n\ |
Guido van Rossum | 23e21e7 | 1997-12-29 19:57:36 +0000 | [diff] [blame] | 196 | intervening occurences of sep. Sep defaults to a single\n\ |
| 197 | space.\n\ |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 198 | \n\ |
Guido van Rossum | 23e21e7 | 1997-12-29 19:57:36 +0000 | [diff] [blame] | 199 | (join and joinfields are synonymous)"; |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 200 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 201 | static PyObject * |
Guido van Rossum | c89705d | 1992-11-26 08:54:07 +0000 | [diff] [blame] | 202 | strop_joinfields(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 203 | PyObject *self; /* Not used */ |
| 204 | PyObject *args; |
Guido van Rossum | c89705d | 1992-11-26 08:54:07 +0000 | [diff] [blame] | 205 | { |
Barry Warsaw | d4ff1b9 | 1997-01-06 22:48:32 +0000 | [diff] [blame] | 206 | PyObject *seq; |
| 207 | char *sep = NULL; |
| 208 | int seqlen, seplen = 0; |
| 209 | int i, reslen = 0, slen = 0, sz = 100; |
| 210 | PyObject *res = NULL; |
| 211 | char* p = NULL; |
| 212 | intargfunc getitemfunc; |
Guido van Rossum | c89705d | 1992-11-26 08:54:07 +0000 | [diff] [blame] | 213 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 214 | if (!PyArg_ParseTuple(args, "O|s#", &seq, &sep, &seplen)) |
Guido van Rossum | c89705d | 1992-11-26 08:54:07 +0000 | [diff] [blame] | 215 | return NULL; |
Guido van Rossum | 009e79b | 1995-05-03 17:40:23 +0000 | [diff] [blame] | 216 | if (sep == NULL) { |
| 217 | sep = " "; |
| 218 | seplen = 1; |
| 219 | } |
Barry Warsaw | d4ff1b9 | 1997-01-06 22:48:32 +0000 | [diff] [blame] | 220 | |
| 221 | seqlen = PySequence_Length(seq); |
| 222 | if (seqlen < 0 && PyErr_Occurred()) |
| 223 | return NULL; |
| 224 | |
| 225 | if (seqlen == 1) { |
| 226 | /* Optimization if there's only one item */ |
| 227 | PyObject *item = PySequence_GetItem(seq, 0); |
| 228 | if (item && !PyString_Check(item)) |
| 229 | PyErr_SetString(PyExc_TypeError, |
| 230 | "first argument must be sequence of strings"); |
| 231 | return item; |
| 232 | } |
| 233 | |
| 234 | if (!(res = PyString_FromStringAndSize((char*)NULL, sz))) |
| 235 | return NULL; |
| 236 | p = PyString_AsString(res); |
| 237 | |
| 238 | /* optimize for lists, since it's the most common case. all others |
| 239 | * (tuples and arbitrary sequences) just use the sequence abstract |
| 240 | * interface. |
Barry Warsaw | 04d2d15 | 1997-01-03 23:46:51 +0000 | [diff] [blame] | 241 | */ |
| 242 | if (PyList_Check(seq)) { |
Barry Warsaw | d4ff1b9 | 1997-01-06 22:48:32 +0000 | [diff] [blame] | 243 | for (i = 0; i < seqlen; i++) { |
| 244 | PyObject *item = PyList_GET_ITEM(seq, i); |
| 245 | if (!PyString_Check(item)) { |
| 246 | PyErr_SetString(PyExc_TypeError, |
| 247 | "first argument must be sequence of strings"); |
| 248 | Py_DECREF(res); |
| 249 | return NULL; |
| 250 | } |
| 251 | slen = PyString_GET_SIZE(item); |
| 252 | while (reslen + slen + seplen >= sz) { |
| 253 | if (_PyString_Resize(&res, sz * 2)) { |
| 254 | Py_DECREF(res); |
| 255 | return NULL; |
| 256 | } |
| 257 | sz *= 2; |
| 258 | p = PyString_AsString(res) + reslen; |
| 259 | } |
| 260 | if (i > 0) { |
| 261 | memcpy(p, sep, seplen); |
| 262 | p += seplen; |
| 263 | reslen += seplen; |
| 264 | } |
| 265 | memcpy(p, PyString_AS_STRING(item), slen); |
| 266 | p += slen; |
| 267 | reslen += slen; |
| 268 | } |
| 269 | if (_PyString_Resize(&res, reslen)) { |
| 270 | Py_DECREF(res); |
| 271 | res = NULL; |
| 272 | } |
| 273 | return res; |
Barry Warsaw | 04d2d15 | 1997-01-03 23:46:51 +0000 | [diff] [blame] | 274 | } |
Barry Warsaw | d4ff1b9 | 1997-01-06 22:48:32 +0000 | [diff] [blame] | 275 | else if (!PySequence_Check(seq)) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 276 | PyErr_SetString(PyExc_TypeError, |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 277 | "first argument must be a sequence"); |
Guido van Rossum | c89705d | 1992-11-26 08:54:07 +0000 | [diff] [blame] | 278 | return NULL; |
| 279 | } |
Barry Warsaw | d4ff1b9 | 1997-01-06 22:48:32 +0000 | [diff] [blame] | 280 | /* type safe */ |
| 281 | getitemfunc = seq->ob_type->tp_as_sequence->sq_item; |
Guido van Rossum | c89705d | 1992-11-26 08:54:07 +0000 | [diff] [blame] | 282 | for (i = 0; i < seqlen; i++) { |
Barry Warsaw | d4ff1b9 | 1997-01-06 22:48:32 +0000 | [diff] [blame] | 283 | PyObject *item = getitemfunc(seq, i); |
| 284 | if (!item || !PyString_Check(item)) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 285 | PyErr_SetString(PyExc_TypeError, |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 286 | "first argument must be sequence of strings"); |
Barry Warsaw | d4ff1b9 | 1997-01-06 22:48:32 +0000 | [diff] [blame] | 287 | Py_DECREF(res); |
| 288 | Py_XDECREF(item); |
Guido van Rossum | c89705d | 1992-11-26 08:54:07 +0000 | [diff] [blame] | 289 | return NULL; |
| 290 | } |
Barry Warsaw | d4ff1b9 | 1997-01-06 22:48:32 +0000 | [diff] [blame] | 291 | slen = PyString_GET_SIZE(item); |
| 292 | while (reslen + slen + seplen >= sz) { |
| 293 | if (_PyString_Resize(&res, sz * 2)) { |
| 294 | Py_DECREF(res); |
| 295 | Py_DECREF(item); |
| 296 | return NULL; |
| 297 | } |
| 298 | sz *= 2; |
| 299 | p = PyString_AsString(res) + reslen; |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 300 | } |
Guido van Rossum | c89705d | 1992-11-26 08:54:07 +0000 | [diff] [blame] | 301 | if (i > 0) { |
| 302 | memcpy(p, sep, seplen); |
| 303 | p += seplen; |
Barry Warsaw | d4ff1b9 | 1997-01-06 22:48:32 +0000 | [diff] [blame] | 304 | reslen += seplen; |
Guido van Rossum | c89705d | 1992-11-26 08:54:07 +0000 | [diff] [blame] | 305 | } |
Barry Warsaw | d4ff1b9 | 1997-01-06 22:48:32 +0000 | [diff] [blame] | 306 | memcpy(p, PyString_AS_STRING(item), slen); |
| 307 | p += slen; |
| 308 | reslen += slen; |
| 309 | Py_DECREF(item); |
Guido van Rossum | c89705d | 1992-11-26 08:54:07 +0000 | [diff] [blame] | 310 | } |
Barry Warsaw | d4ff1b9 | 1997-01-06 22:48:32 +0000 | [diff] [blame] | 311 | if (_PyString_Resize(&res, reslen)) { |
| 312 | Py_DECREF(res); |
| 313 | res = NULL; |
Guido van Rossum | c89705d | 1992-11-26 08:54:07 +0000 | [diff] [blame] | 314 | } |
| 315 | return res; |
| 316 | } |
| 317 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 318 | |
| 319 | static char find__doc__[] = |
| 320 | "find(s, sub [,start [,end]]) -> in\n\ |
| 321 | \n\ |
| 322 | Return the lowest index in s where substring sub is found,\n\ |
| 323 | such that sub is contained within s[start,end]. Optional\n\ |
| 324 | arguments start and end are interpreted as in slice notation.\n\ |
| 325 | \n\ |
| 326 | Return -1 on failure."; |
| 327 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 328 | static PyObject * |
Guido van Rossum | 5806a4f | 1994-08-17 13:15:46 +0000 | [diff] [blame] | 329 | strop_find(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 330 | PyObject *self; /* Not used */ |
| 331 | PyObject *args; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 332 | { |
| 333 | char *s, *sub; |
Guido van Rossum | 7b7c578 | 1997-03-14 04:13:56 +0000 | [diff] [blame] | 334 | int len, n, i = 0, last = INT_MAX; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 335 | |
Guido van Rossum | 7b7c578 | 1997-03-14 04:13:56 +0000 | [diff] [blame] | 336 | if (!PyArg_ParseTuple(args, "s#s#|ii", &s, &len, &sub, &n, &i, &last)) |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 337 | return NULL; |
| 338 | |
Guido van Rossum | 7b7c578 | 1997-03-14 04:13:56 +0000 | [diff] [blame] | 339 | if (last > len) |
| 340 | last = len; |
| 341 | if (last < 0) |
| 342 | last += len; |
| 343 | if (last < 0) |
| 344 | last = 0; |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 345 | if (i < 0) |
| 346 | i += len; |
| 347 | if (i < 0) |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 348 | i = 0; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 349 | |
| 350 | if (n == 0) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 351 | return PyInt_FromLong((long)i); |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 352 | |
Guido van Rossum | 7b7c578 | 1997-03-14 04:13:56 +0000 | [diff] [blame] | 353 | last -= n; |
| 354 | for (; i <= last; ++i) |
Guido van Rossum | ee9012f | 1993-10-26 15:23:55 +0000 | [diff] [blame] | 355 | if (s[i] == sub[0] && |
Guido van Rossum | a0ca4c4 | 1996-10-04 13:39:37 +0000 | [diff] [blame] | 356 | (n == 1 || memcmp(&s[i+1], &sub[1], n-1) == 0)) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 357 | return PyInt_FromLong((long)i); |
Guido van Rossum | ee9012f | 1993-10-26 15:23:55 +0000 | [diff] [blame] | 358 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 359 | return PyInt_FromLong(-1L); |
Guido van Rossum | ee9012f | 1993-10-26 15:23:55 +0000 | [diff] [blame] | 360 | } |
| 361 | |
| 362 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 363 | static char rfind__doc__[] = |
| 364 | "rfind(s, sub [,start [,end]]) -> int\n\ |
| 365 | \n\ |
| 366 | Return the highest index in s where substring sub is found,\n\ |
| 367 | such that sub is contained within s[start,end]. Optional\n\ |
| 368 | arguments start and end are interpreted as in slice notation.\n\ |
| 369 | \n\ |
| 370 | Return -1 on failure."; |
| 371 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 372 | static PyObject * |
Guido van Rossum | 5806a4f | 1994-08-17 13:15:46 +0000 | [diff] [blame] | 373 | strop_rfind(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 374 | PyObject *self; /* Not used */ |
| 375 | PyObject *args; |
Guido van Rossum | ee9012f | 1993-10-26 15:23:55 +0000 | [diff] [blame] | 376 | { |
| 377 | char *s, *sub; |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 378 | int len, n, j; |
Guido van Rossum | 7b7c578 | 1997-03-14 04:13:56 +0000 | [diff] [blame] | 379 | int i = 0, last = INT_MAX; |
Guido van Rossum | ee9012f | 1993-10-26 15:23:55 +0000 | [diff] [blame] | 380 | |
Guido van Rossum | 7b7c578 | 1997-03-14 04:13:56 +0000 | [diff] [blame] | 381 | if (!PyArg_ParseTuple(args, "s#s#|ii", &s, &len, &sub, &n, &i, &last)) |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 382 | return NULL; |
| 383 | |
Guido van Rossum | 7b7c578 | 1997-03-14 04:13:56 +0000 | [diff] [blame] | 384 | if (last > len) |
| 385 | last = len; |
| 386 | if (last < 0) |
| 387 | last += len; |
| 388 | if (last < 0) |
| 389 | last = 0; |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 390 | if (i < 0) |
| 391 | i += len; |
| 392 | if (i < 0) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 393 | i = 0; |
Guido van Rossum | ee9012f | 1993-10-26 15:23:55 +0000 | [diff] [blame] | 394 | |
| 395 | if (n == 0) |
Guido van Rossum | 7b7c578 | 1997-03-14 04:13:56 +0000 | [diff] [blame] | 396 | return PyInt_FromLong((long)last); |
Guido van Rossum | ee9012f | 1993-10-26 15:23:55 +0000 | [diff] [blame] | 397 | |
Guido van Rossum | 7b7c578 | 1997-03-14 04:13:56 +0000 | [diff] [blame] | 398 | for (j = last-n; j >= i; --j) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 399 | if (s[j] == sub[0] && |
Guido van Rossum | a0ca4c4 | 1996-10-04 13:39:37 +0000 | [diff] [blame] | 400 | (n == 1 || memcmp(&s[j+1], &sub[1], n-1) == 0)) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 401 | return PyInt_FromLong((long)j); |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 402 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 403 | return PyInt_FromLong(-1L); |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 404 | } |
| 405 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 406 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 407 | static PyObject * |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 408 | do_strip(args, striptype) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 409 | PyObject *args; |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 410 | int striptype; |
| 411 | { |
| 412 | char *s; |
| 413 | int len, i, j; |
| 414 | |
| 415 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 416 | if (!PyArg_Parse(args, "s#", &s, &len)) |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 417 | return NULL; |
| 418 | |
| 419 | i = 0; |
| 420 | if (striptype != RIGHTSTRIP) { |
| 421 | while (i < len && isspace(Py_CHARMASK(s[i]))) { |
| 422 | i++; |
| 423 | } |
| 424 | } |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 425 | |
| 426 | j = len; |
| 427 | if (striptype != LEFTSTRIP) { |
| 428 | do { |
| 429 | j--; |
| 430 | } while (j >= i && isspace(Py_CHARMASK(s[j]))); |
| 431 | j++; |
| 432 | } |
| 433 | |
| 434 | if (i == 0 && j == len) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 435 | Py_INCREF(args); |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 436 | return args; |
| 437 | } |
| 438 | else |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 439 | return PyString_FromStringAndSize(s+i, j-i); |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 440 | } |
| 441 | |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 442 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 443 | static char strip__doc__[] = |
| 444 | "strip(s) -> string\n\ |
| 445 | \n\ |
| 446 | Return a copy of the string s with leading and trailing\n\ |
| 447 | whitespace removed."; |
| 448 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 449 | static PyObject * |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 450 | strop_strip(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 451 | PyObject *self; /* Not used */ |
| 452 | PyObject *args; |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 453 | { |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 454 | return do_strip(args, BOTHSTRIP); |
| 455 | } |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 456 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 457 | |
| 458 | static char lstrip__doc__[] = |
| 459 | "lstrip(s) -> string\n\ |
| 460 | \n\ |
| 461 | Return a copy of the string s with leading whitespace removed."; |
| 462 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 463 | static PyObject * |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 464 | strop_lstrip(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 465 | PyObject *self; /* Not used */ |
| 466 | PyObject *args; |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 467 | { |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 468 | return do_strip(args, LEFTSTRIP); |
| 469 | } |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 470 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 471 | |
| 472 | static char rstrip__doc__[] = |
| 473 | "rstrip(s) -> string\n\ |
| 474 | \n\ |
| 475 | Return a copy of the string s with trailing whitespace removed."; |
| 476 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 477 | static PyObject * |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 478 | strop_rstrip(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 479 | PyObject *self; /* Not used */ |
| 480 | PyObject *args; |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 481 | { |
Guido van Rossum | 7999a5c | 1996-08-08 19:16:15 +0000 | [diff] [blame] | 482 | return do_strip(args, RIGHTSTRIP); |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 483 | } |
| 484 | |
| 485 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 486 | static char lower__doc__[] = |
| 487 | "lower(s) -> string\n\ |
| 488 | \n\ |
| 489 | Return a copy of the string s converted to lowercase."; |
| 490 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 491 | static PyObject * |
Barry Warsaw | 04d2d15 | 1997-01-03 23:46:51 +0000 | [diff] [blame] | 492 | strop_lower(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 493 | PyObject *self; /* Not used */ |
| 494 | PyObject *args; |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 495 | { |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame] | 496 | char *s, *s_new; |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 497 | int i, n; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 498 | PyObject *new; |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 499 | int changed; |
| 500 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 501 | if (!PyArg_Parse(args, "s#", &s, &n)) |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 502 | return NULL; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 503 | new = PyString_FromStringAndSize(NULL, n); |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 504 | if (new == NULL) |
| 505 | return NULL; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 506 | s_new = PyString_AsString(new); |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 507 | changed = 0; |
| 508 | for (i = 0; i < n; i++) { |
Guido van Rossum | 7f7f274 | 1995-02-10 17:01:56 +0000 | [diff] [blame] | 509 | int c = Py_CHARMASK(*s++); |
Barry Warsaw | 04d2d15 | 1997-01-03 23:46:51 +0000 | [diff] [blame] | 510 | if (isupper(c)) { |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 511 | changed = 1; |
Barry Warsaw | 04d2d15 | 1997-01-03 23:46:51 +0000 | [diff] [blame] | 512 | *s_new = tolower(c); |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame] | 513 | } else |
| 514 | *s_new = c; |
| 515 | s_new++; |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 516 | } |
| 517 | if (!changed) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 518 | Py_DECREF(new); |
| 519 | Py_INCREF(args); |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 520 | return args; |
| 521 | } |
| 522 | return new; |
| 523 | } |
| 524 | |
| 525 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 526 | static char upper__doc__[] = |
| 527 | "upper(s) -> string\n\ |
| 528 | \n\ |
| 529 | Return a copy of the string s converted to uppercase."; |
| 530 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 531 | static PyObject * |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 532 | strop_upper(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 533 | PyObject *self; /* Not used */ |
| 534 | PyObject *args; |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 535 | { |
Barry Warsaw | 04d2d15 | 1997-01-03 23:46:51 +0000 | [diff] [blame] | 536 | char *s, *s_new; |
| 537 | int i, n; |
| 538 | PyObject *new; |
| 539 | int changed; |
| 540 | |
| 541 | if (!PyArg_Parse(args, "s#", &s, &n)) |
| 542 | return NULL; |
| 543 | new = PyString_FromStringAndSize(NULL, n); |
| 544 | if (new == NULL) |
| 545 | return NULL; |
| 546 | s_new = PyString_AsString(new); |
| 547 | changed = 0; |
| 548 | for (i = 0; i < n; i++) { |
| 549 | int c = Py_CHARMASK(*s++); |
| 550 | if (islower(c)) { |
| 551 | changed = 1; |
| 552 | *s_new = toupper(c); |
| 553 | } else |
| 554 | *s_new = c; |
| 555 | s_new++; |
| 556 | } |
| 557 | if (!changed) { |
| 558 | Py_DECREF(new); |
| 559 | Py_INCREF(args); |
| 560 | return args; |
| 561 | } |
| 562 | return new; |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 563 | } |
| 564 | |
| 565 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 566 | static char capitalize__doc__[] = |
| 567 | "capitalize(s) -> string\n\ |
| 568 | \n\ |
| 569 | Return a copy of the string s with only its first character\n\ |
| 570 | capitalized."; |
| 571 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 572 | static PyObject * |
Guido van Rossum | 2745753 | 1996-06-12 04:24:52 +0000 | [diff] [blame] | 573 | strop_capitalize(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 574 | PyObject *self; /* Not used */ |
| 575 | PyObject *args; |
Guido van Rossum | 2745753 | 1996-06-12 04:24:52 +0000 | [diff] [blame] | 576 | { |
| 577 | char *s, *s_new; |
| 578 | int i, n; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 579 | PyObject *new; |
Guido van Rossum | 2745753 | 1996-06-12 04:24:52 +0000 | [diff] [blame] | 580 | int changed; |
| 581 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 582 | if (!PyArg_Parse(args, "s#", &s, &n)) |
Guido van Rossum | 2745753 | 1996-06-12 04:24:52 +0000 | [diff] [blame] | 583 | return NULL; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 584 | new = PyString_FromStringAndSize(NULL, n); |
Guido van Rossum | 2745753 | 1996-06-12 04:24:52 +0000 | [diff] [blame] | 585 | if (new == NULL) |
| 586 | return NULL; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 587 | s_new = PyString_AsString(new); |
Guido van Rossum | 2745753 | 1996-06-12 04:24:52 +0000 | [diff] [blame] | 588 | changed = 0; |
Guido van Rossum | 529c963 | 1996-06-17 16:59:33 +0000 | [diff] [blame] | 589 | if (0 < n) { |
Guido van Rossum | 2745753 | 1996-06-12 04:24:52 +0000 | [diff] [blame] | 590 | int c = Py_CHARMASK(*s++); |
| 591 | if (islower(c)) { |
| 592 | changed = 1; |
| 593 | *s_new = toupper(c); |
| 594 | } else |
| 595 | *s_new = c; |
| 596 | s_new++; |
| 597 | } |
| 598 | for (i = 1; i < n; i++) { |
| 599 | int c = Py_CHARMASK(*s++); |
| 600 | if (isupper(c)) { |
| 601 | changed = 1; |
| 602 | *s_new = tolower(c); |
| 603 | } else |
| 604 | *s_new = c; |
| 605 | s_new++; |
| 606 | } |
| 607 | if (!changed) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 608 | Py_DECREF(new); |
| 609 | Py_INCREF(args); |
Guido van Rossum | 2745753 | 1996-06-12 04:24:52 +0000 | [diff] [blame] | 610 | return args; |
| 611 | } |
| 612 | return new; |
| 613 | } |
| 614 | |
| 615 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 616 | static char swapcase__doc__[] = |
| 617 | "swapcase(s) -> string\n\ |
| 618 | \n\ |
| 619 | Return a copy of the string s with upper case characters\n\ |
| 620 | converted to lowercase and vice versa."; |
| 621 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 622 | static PyObject * |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 623 | strop_swapcase(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 624 | PyObject *self; /* Not used */ |
| 625 | PyObject *args; |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 626 | { |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame] | 627 | char *s, *s_new; |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 628 | int i, n; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 629 | PyObject *new; |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 630 | int changed; |
| 631 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 632 | if (!PyArg_Parse(args, "s#", &s, &n)) |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 633 | return NULL; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 634 | new = PyString_FromStringAndSize(NULL, n); |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 635 | if (new == NULL) |
| 636 | return NULL; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 637 | s_new = PyString_AsString(new); |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 638 | changed = 0; |
| 639 | for (i = 0; i < n; i++) { |
Guido van Rossum | 7f7f274 | 1995-02-10 17:01:56 +0000 | [diff] [blame] | 640 | int c = Py_CHARMASK(*s++); |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 641 | if (islower(c)) { |
| 642 | changed = 1; |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame] | 643 | *s_new = toupper(c); |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 644 | } |
| 645 | else if (isupper(c)) { |
| 646 | changed = 1; |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame] | 647 | *s_new = tolower(c); |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 648 | } |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame] | 649 | else |
| 650 | *s_new = c; |
| 651 | s_new++; |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 652 | } |
| 653 | if (!changed) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 654 | Py_DECREF(new); |
| 655 | Py_INCREF(args); |
Guido van Rossum | 5c85062 | 1992-09-11 23:55:51 +0000 | [diff] [blame] | 656 | return args; |
| 657 | } |
| 658 | return new; |
| 659 | } |
| 660 | |
| 661 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 662 | static char atoi__doc__[] = |
| 663 | "atoi(s [,base]) -> int\n\ |
| 664 | \n\ |
| 665 | Return the integer represented by the string s in the given\n\ |
| 666 | base, which defaults to 10. The string s must consist of one\n\ |
| 667 | or more digits, possibly preceded by a sign. If base is 0, it\n\ |
| 668 | is chosen from the leading characters of s, 0 for octal, 0x or\n\ |
| 669 | 0X for hexadecimal. If base is 16, a preceding 0x or 0X is\n\ |
| 670 | accepted."; |
| 671 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 672 | static PyObject * |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 673 | strop_atoi(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 674 | PyObject *self; /* Not used */ |
| 675 | PyObject *args; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 676 | { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 677 | extern long PyOS_strtol Py_PROTO((const char *, char **, int)); |
| 678 | extern unsigned long |
| 679 | PyOS_strtoul Py_PROTO((const char *, char **, int)); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 680 | char *s, *end; |
| 681 | int base = 10; |
| 682 | long x; |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 683 | char buffer[256]; /* For errors */ |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 684 | |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 685 | if (!PyArg_ParseTuple(args, "s|i", &s, &base)) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 686 | return NULL; |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 687 | |
| 688 | if ((base != 0 && base < 2) || base > 36) { |
| 689 | PyErr_SetString(PyExc_ValueError, "invalid base for atoi()"); |
| 690 | return NULL; |
| 691 | } |
| 692 | |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 693 | while (*s && isspace(Py_CHARMASK(*s))) |
| 694 | s++; |
Guido van Rossum | 171191e | 1996-08-21 20:02:25 +0000 | [diff] [blame] | 695 | if (s[0] == '\0') { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 696 | PyErr_SetString(PyExc_ValueError, "empty string for atoi()"); |
Guido van Rossum | 171191e | 1996-08-21 20:02:25 +0000 | [diff] [blame] | 697 | return NULL; |
| 698 | } |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 699 | errno = 0; |
| 700 | if (base == 0 && s[0] == '0') |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 701 | x = (long) PyOS_strtoul(s, &end, base); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 702 | else |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 703 | x = PyOS_strtol(s, &end, base); |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 704 | while (*end && isspace(Py_CHARMASK(*end))) |
| 705 | end++; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 706 | if (*end != '\0') { |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 707 | sprintf(buffer, "invalid literal for atoi(): %.200s", s); |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 708 | PyErr_SetString(PyExc_ValueError, buffer); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 709 | return NULL; |
| 710 | } |
| 711 | else if (errno != 0) { |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 712 | sprintf(buffer, "atoi() literal too large: %.200s", s); |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 713 | PyErr_SetString(PyExc_ValueError, buffer); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 714 | return NULL; |
| 715 | } |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 716 | return PyInt_FromLong(x); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 717 | } |
| 718 | |
| 719 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 720 | static char atol__doc__[] = |
| 721 | "atol(s [,base]) -> long\n\ |
| 722 | \n\ |
| 723 | Return the long integer represented by the string s in the\n\ |
| 724 | given base, which defaults to 10. The string s must consist\n\ |
| 725 | of one or more digits, possibly preceded by a sign. If base\n\ |
| 726 | is 0, it is chosen from the leading characters of s, 0 for\n\ |
| 727 | octal, 0x or 0X for hexadecimal. If base is 16, a preceding\n\ |
| 728 | 0x or 0X is accepted. A trailing L or l is not accepted,\n\ |
| 729 | unless base is 0."; |
| 730 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 731 | static PyObject * |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 732 | strop_atol(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 733 | PyObject *self; /* Not used */ |
| 734 | PyObject *args; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 735 | { |
| 736 | char *s, *end; |
| 737 | int base = 10; |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 738 | PyObject *x; |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 739 | char buffer[256]; /* For errors */ |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 740 | |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 741 | if (!PyArg_ParseTuple(args, "s|i", &s, &base)) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 742 | return NULL; |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 743 | |
| 744 | if ((base != 0 && base < 2) || base > 36) { |
| 745 | PyErr_SetString(PyExc_ValueError, "invalid base for atol()"); |
| 746 | return NULL; |
| 747 | } |
| 748 | |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 749 | while (*s && isspace(Py_CHARMASK(*s))) |
| 750 | s++; |
Guido van Rossum | 171191e | 1996-08-21 20:02:25 +0000 | [diff] [blame] | 751 | if (s[0] == '\0') { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 752 | PyErr_SetString(PyExc_ValueError, "empty string for atol()"); |
Guido van Rossum | 171191e | 1996-08-21 20:02:25 +0000 | [diff] [blame] | 753 | return NULL; |
| 754 | } |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 755 | x = PyLong_FromString(s, &end, base); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 756 | if (x == NULL) |
| 757 | return NULL; |
| 758 | if (base == 0 && (*end == 'l' || *end == 'L')) |
| 759 | end++; |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 760 | while (*end && isspace(Py_CHARMASK(*end))) |
| 761 | end++; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 762 | if (*end != '\0') { |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 763 | sprintf(buffer, "invalid literal for atol(): %.200s", s); |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 764 | PyErr_SetString(PyExc_ValueError, buffer); |
| 765 | Py_DECREF(x); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 766 | return NULL; |
| 767 | } |
| 768 | return x; |
| 769 | } |
| 770 | |
| 771 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 772 | static char atof__doc__[] = |
| 773 | "atof(s) -> float\n\ |
| 774 | \n\ |
| 775 | Return the floating point number represented by the string s."; |
| 776 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 777 | static PyObject * |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 778 | strop_atof(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 779 | PyObject *self; /* Not used */ |
| 780 | PyObject *args; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 781 | { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 782 | extern double strtod Py_PROTO((const char *, char **)); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 783 | char *s, *end; |
| 784 | double x; |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 785 | char buffer[256]; /* For errors */ |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 786 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 787 | if (!PyArg_Parse(args, "s", &s)) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 788 | return NULL; |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 789 | while (*s && isspace(Py_CHARMASK(*s))) |
| 790 | s++; |
Guido van Rossum | 171191e | 1996-08-21 20:02:25 +0000 | [diff] [blame] | 791 | if (s[0] == '\0') { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 792 | PyErr_SetString(PyExc_ValueError, "empty string for atof()"); |
Guido van Rossum | 171191e | 1996-08-21 20:02:25 +0000 | [diff] [blame] | 793 | return NULL; |
| 794 | } |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 795 | errno = 0; |
Guido van Rossum | 52fa3a6 | 1997-02-14 22:59:58 +0000 | [diff] [blame] | 796 | PyFPE_START_PROTECT("strop_atof", return 0) |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 797 | x = strtod(s, &end); |
Guido van Rossum | 7b7c578 | 1997-03-14 04:13:56 +0000 | [diff] [blame] | 798 | PyFPE_END_PROTECT(x) |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 799 | while (*end && isspace(Py_CHARMASK(*end))) |
| 800 | end++; |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 801 | if (*end != '\0') { |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 802 | sprintf(buffer, "invalid literal for atof(): %.200s", s); |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 803 | PyErr_SetString(PyExc_ValueError, buffer); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 804 | return NULL; |
| 805 | } |
| 806 | else if (errno != 0) { |
Guido van Rossum | c35f933 | 1996-09-11 23:30:42 +0000 | [diff] [blame] | 807 | sprintf(buffer, "atof() literal too large: %.200s", s); |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 808 | PyErr_SetString(PyExc_ValueError, buffer); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 809 | return NULL; |
| 810 | } |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 811 | return PyFloat_FromDouble(x); |
Guido van Rossum | b6775db | 1994-08-01 11:34:53 +0000 | [diff] [blame] | 812 | } |
| 813 | |
| 814 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 815 | static char maketrans__doc__[] = |
| 816 | "maketrans(frm, to) -> string\n\ |
| 817 | \n\ |
| 818 | Return a translation table (a string of 256 bytes long)\n\ |
| 819 | suitable for use in string.translate. The strings frm and to\n\ |
| 820 | must be of the same length."; |
| 821 | |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 822 | static PyObject * |
| 823 | strop_maketrans(self, args) |
| 824 | PyObject *self; /* Not used */ |
| 825 | PyObject *args; |
| 826 | { |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 827 | unsigned char *c, *from=NULL, *to=NULL; |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 828 | int i, fromlen=0, tolen=0; |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 829 | PyObject *result; |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 830 | |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 831 | if (!PyArg_ParseTuple(args, "s#s#", &from, &fromlen, &to, &tolen)) |
| 832 | return NULL; |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 833 | |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 834 | if (fromlen != tolen) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 835 | PyErr_SetString(PyExc_ValueError, |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 836 | "maketrans arguments must have same length"); |
| 837 | return NULL; |
| 838 | } |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 839 | |
| 840 | result = PyString_FromStringAndSize((char *)NULL, 256); |
| 841 | if (result == NULL) |
| 842 | return NULL; |
| 843 | c = (unsigned char *) PyString_AS_STRING((PyStringObject *)result); |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 844 | for (i = 0; i < 256; i++) |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 845 | c[i]=(unsigned char)i; |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 846 | for (i = 0; i < fromlen; i++) |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 847 | c[from[i]]=to[i]; |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 848 | |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 849 | return result; |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 850 | } |
| 851 | |
| 852 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 853 | static char translate__doc__[] = |
| 854 | "translate(s,table [,deletechars]) -> string\n\ |
| 855 | \n\ |
| 856 | Return a copy of the string s, where all characters occurring\n\ |
| 857 | in the optional argument deletechars are removed, and the\n\ |
| 858 | remaining characters have been mapped through the given\n\ |
| 859 | translation table, which must be a string of length 256."; |
| 860 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 861 | static PyObject * |
Guido van Rossum | a3127e8 | 1995-09-13 17:39:06 +0000 | [diff] [blame] | 862 | strop_translate(self, args) |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 863 | PyObject *self; |
| 864 | PyObject *args; |
Guido van Rossum | a3127e8 | 1995-09-13 17:39:06 +0000 | [diff] [blame] | 865 | { |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 866 | register char *input, *table, *output; |
| 867 | register int i, c, changed = 0; |
| 868 | PyObject *input_obj; |
| 869 | char *table1, *output_start, *del_table=NULL; |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 870 | int inlen, tablen, dellen = 0; |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 871 | PyObject *result; |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 872 | int trans_table[256]; |
Guido van Rossum | a3127e8 | 1995-09-13 17:39:06 +0000 | [diff] [blame] | 873 | |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 874 | if (!PyArg_ParseTuple(args, "Ss#|s#", &input_obj, |
| 875 | &table1, &tablen, &del_table, &dellen)) |
Guido van Rossum | a3127e8 | 1995-09-13 17:39:06 +0000 | [diff] [blame] | 876 | return NULL; |
| 877 | if (tablen != 256) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 878 | PyErr_SetString(PyExc_ValueError, |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 879 | "translation table must be 256 characters long"); |
Guido van Rossum | a3127e8 | 1995-09-13 17:39:06 +0000 | [diff] [blame] | 880 | return NULL; |
| 881 | } |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 882 | |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 883 | table = table1; |
| 884 | inlen = PyString_Size(input_obj); |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 885 | result = PyString_FromStringAndSize((char *)NULL, inlen); |
Guido van Rossum | a3127e8 | 1995-09-13 17:39:06 +0000 | [diff] [blame] | 886 | if (result == NULL) |
| 887 | return NULL; |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 888 | output_start = output = PyString_AsString(result); |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 889 | input = PyString_AsString(input_obj); |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 890 | |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 891 | if (dellen == 0) { |
| 892 | /* If no deletions are required, use faster code */ |
| 893 | for (i = inlen; --i >= 0; ) { |
| 894 | c = Py_CHARMASK(*input++); |
| 895 | if (Py_CHARMASK((*output++ = table[c])) != c) |
| 896 | changed = 1; |
Guido van Rossum | ed7253c | 1996-07-23 18:12:39 +0000 | [diff] [blame] | 897 | } |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 898 | if (changed) |
| 899 | return result; |
| 900 | Py_DECREF(result); |
| 901 | Py_INCREF(input_obj); |
| 902 | return input_obj; |
Guido van Rossum | a3127e8 | 1995-09-13 17:39:06 +0000 | [diff] [blame] | 903 | } |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 904 | |
| 905 | for (i = 0; i < 256; i++) |
| 906 | trans_table[i] = Py_CHARMASK(table[i]); |
| 907 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 908 | for (i = 0; i < dellen; i++) |
Guido van Rossum | 1ed5e57 | 1997-04-29 21:34:16 +0000 | [diff] [blame] | 909 | trans_table[(int) Py_CHARMASK(del_table[i])] = -1; |
Guido van Rossum | e0548b8 | 1997-01-06 16:50:09 +0000 | [diff] [blame] | 910 | |
| 911 | for (i = inlen; --i >= 0; ) { |
| 912 | c = Py_CHARMASK(*input++); |
| 913 | if (trans_table[c] != -1) |
| 914 | if (Py_CHARMASK(*output++ = (char)trans_table[c]) == c) |
| 915 | continue; |
| 916 | changed = 1; |
| 917 | } |
| 918 | if (!changed) { |
| 919 | Py_DECREF(result); |
| 920 | Py_INCREF(input_obj); |
| 921 | return input_obj; |
| 922 | } |
| 923 | /* Fix the size of the resulting string */ |
| 924 | if (inlen > 0 &&_PyString_Resize(&result, output-output_start)) |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 925 | return NULL; |
Guido van Rossum | a3127e8 | 1995-09-13 17:39:06 +0000 | [diff] [blame] | 926 | return result; |
| 927 | } |
| 928 | |
| 929 | |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 930 | /* What follows is used for implementing replace(). Perry Stoll. */ |
| 931 | |
| 932 | /* |
| 933 | mymemfind |
| 934 | |
| 935 | strstr replacement for arbitrary blocks of memory. |
| 936 | |
| 937 | Locates the first occurance in the memory pointed to by MEM of the |
| 938 | contents of memory pointed to by PAT. Returns the index into MEM if |
| 939 | found, or -1 if not found. If len of PAT is greater than length of |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 940 | MEM, the function returns -1. |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 941 | */ |
| 942 | static int mymemfind(mem, len, pat, pat_len) |
| 943 | char *mem; |
| 944 | int len; |
| 945 | char *pat; |
| 946 | int pat_len; |
| 947 | { |
| 948 | register int ii; |
| 949 | |
| 950 | /* pattern can not occur in the last pat_len-1 chars */ |
| 951 | len -= pat_len; |
| 952 | |
| 953 | for (ii = 0; ii <= len; ii++) { |
| 954 | if (mem[ii] == pat[0] && |
| 955 | (pat_len == 1 || |
| 956 | memcmp(&mem[ii+1], &pat[1], pat_len-1) == 0)) { |
| 957 | return ii; |
| 958 | } |
| 959 | } |
| 960 | return -1; |
| 961 | } |
| 962 | |
| 963 | /* |
| 964 | mymemcnt |
| 965 | |
| 966 | Return the number of distinct times PAT is found in MEM. |
| 967 | meaning mem=1111 and pat==11 returns 2. |
| 968 | mem=11111 and pat==11 also return 2. |
| 969 | */ |
| 970 | static int mymemcnt(mem, len, pat, pat_len) |
| 971 | char *mem; |
| 972 | int len; |
| 973 | char *pat; |
| 974 | int pat_len; |
| 975 | { |
| 976 | register int offset = 0; |
| 977 | int nfound = 0; |
| 978 | |
| 979 | while (len >= 0) { |
| 980 | offset = mymemfind(mem, len, pat, pat_len); |
| 981 | if (offset == -1) |
| 982 | break; |
| 983 | mem += offset + pat_len; |
| 984 | len -= offset + pat_len; |
| 985 | nfound++; |
| 986 | } |
| 987 | return nfound; |
| 988 | } |
| 989 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 990 | /* |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 991 | mymemreplace |
| 992 | |
| 993 | Return a string in which all occurences of PAT in memory STR are |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 994 | replaced with SUB. |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 995 | |
| 996 | If length of PAT is less than length of STR or there are no occurences |
| 997 | of PAT in STR, then the original string is returned. Otherwise, a new |
| 998 | string is allocated here and returned. |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 999 | |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 1000 | on return, out_len is: |
| 1001 | the length of output string, or |
| 1002 | -1 if the input string is returned, or |
| 1003 | unchanged if an error occurs (no memory). |
| 1004 | |
| 1005 | return value is: |
| 1006 | the new string allocated locally, or |
| 1007 | NULL if an error occurred. |
| 1008 | */ |
Barry Warsaw | f577c08 | 1997-11-29 00:10:07 +0000 | [diff] [blame] | 1009 | static char *mymemreplace(str, len, pat, pat_len, sub, sub_len, count, out_len) |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 1010 | char *str; |
| 1011 | int len; /* input string */ |
| 1012 | char *pat; |
| 1013 | int pat_len; /* pattern string to find */ |
| 1014 | char *sub; |
| 1015 | int sub_len; /* substitution string */ |
Barry Warsaw | f577c08 | 1997-11-29 00:10:07 +0000 | [diff] [blame] | 1016 | int count; /* number of replacements, 0 == all */ |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 1017 | int *out_len; |
| 1018 | |
| 1019 | { |
| 1020 | char *out_s; |
| 1021 | char *new_s; |
| 1022 | int nfound, offset, new_len; |
| 1023 | |
| 1024 | if (len == 0 || pat_len > len) |
| 1025 | goto return_same; |
| 1026 | |
| 1027 | /* find length of output string */ |
| 1028 | nfound = mymemcnt(str, len, pat, pat_len); |
Barry Warsaw | f577c08 | 1997-11-29 00:10:07 +0000 | [diff] [blame] | 1029 | if (count > 0) |
| 1030 | nfound = nfound > count ? count : nfound; |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 1031 | if (nfound == 0) |
| 1032 | goto return_same; |
| 1033 | new_len = len + nfound*(sub_len - pat_len); |
| 1034 | |
| 1035 | new_s = (char *)malloc(new_len); |
| 1036 | if (new_s == NULL) return NULL; |
| 1037 | |
| 1038 | *out_len = new_len; |
| 1039 | out_s = new_s; |
| 1040 | |
| 1041 | while (len > 0) { |
| 1042 | /* find index of next instance of pattern */ |
| 1043 | offset = mymemfind(str, len, pat, pat_len); |
| 1044 | /* if not found, break out of loop */ |
| 1045 | if (offset == -1) break; |
| 1046 | |
| 1047 | /* copy non matching part of input string */ |
| 1048 | memcpy(new_s, str, offset); /* copy part of str before pat */ |
| 1049 | str += offset + pat_len; /* move str past pattern */ |
| 1050 | len -= offset + pat_len; /* reduce length of str remaining */ |
| 1051 | |
| 1052 | /* copy substitute into the output string */ |
| 1053 | new_s += offset; /* move new_s to dest for sub string */ |
| 1054 | memcpy(new_s, sub, sub_len); /* copy substring into new_s */ |
| 1055 | new_s += sub_len; /* offset new_s past sub string */ |
Barry Warsaw | f577c08 | 1997-11-29 00:10:07 +0000 | [diff] [blame] | 1056 | |
| 1057 | /* break when we've done count replacements */ |
| 1058 | if (--count == 0) break; |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 1059 | } |
| 1060 | /* copy any remaining values into output string */ |
| 1061 | if (len > 0) |
| 1062 | memcpy(new_s, str, len); |
| 1063 | return out_s; |
| 1064 | |
| 1065 | return_same: |
| 1066 | *out_len = -1; |
| 1067 | return str; |
| 1068 | } |
| 1069 | |
| 1070 | |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 1071 | static char replace__doc__[] = |
| 1072 | "replace (str, old, new[, maxsplit]) -> string\n\ |
| 1073 | \n\ |
| 1074 | Return a copy of string str with all occurrences of substring\n\ |
| 1075 | old replaced by new. If the optional argument maxsplit is\n\ |
| 1076 | given, only the first maxsplit occurrences are replaced."; |
| 1077 | |
| 1078 | static PyObject * |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 1079 | strop_replace(self, args) |
| 1080 | PyObject *self; /* Not used */ |
| 1081 | PyObject *args; |
| 1082 | { |
| 1083 | char *str, *pat,*sub,*new_s; |
| 1084 | int len,pat_len,sub_len,out_len; |
Barry Warsaw | f577c08 | 1997-11-29 00:10:07 +0000 | [diff] [blame] | 1085 | int count = 0; |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 1086 | PyObject *new; |
| 1087 | |
Barry Warsaw | f577c08 | 1997-11-29 00:10:07 +0000 | [diff] [blame] | 1088 | if (!PyArg_ParseTuple(args, "s#s#s#|i", |
| 1089 | &str, &len, &pat, &pat_len, &sub, &sub_len, |
| 1090 | &count)) |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 1091 | return NULL; |
Barry Warsaw | f577c08 | 1997-11-29 00:10:07 +0000 | [diff] [blame] | 1092 | new_s = mymemreplace(str,len,pat,pat_len,sub,sub_len,count,&out_len); |
Guido van Rossum | 101923b | 1997-04-02 06:11:18 +0000 | [diff] [blame] | 1093 | if (new_s == NULL) { |
| 1094 | PyErr_NoMemory(); |
| 1095 | return NULL; |
| 1096 | } |
| 1097 | if (out_len == -1) { |
| 1098 | /* we're returning another reference to the input string */ |
| 1099 | new = PyTuple_GetItem(args, 0); |
| 1100 | Py_XINCREF(new); |
| 1101 | } |
| 1102 | else { |
| 1103 | new = PyString_FromStringAndSize(new_s, out_len); |
| 1104 | free(new_s); |
| 1105 | } |
| 1106 | return new; |
| 1107 | } |
| 1108 | |
| 1109 | |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 1110 | /* List of functions defined in the module */ |
| 1111 | |
Barry Warsaw | e8fc29c | 1997-01-03 22:45:34 +0000 | [diff] [blame] | 1112 | static PyMethodDef |
| 1113 | strop_methods[] = { |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 1114 | {"atof", strop_atof, 1, atof__doc__}, |
| 1115 | {"atoi", strop_atoi, 1, atoi__doc__}, |
| 1116 | {"atol", strop_atol, 1, atol__doc__}, |
| 1117 | {"capitalize", strop_capitalize, 1, capitalize__doc__}, |
| 1118 | {"find", strop_find, 1, find__doc__}, |
| 1119 | {"join", strop_joinfields, 1, joinfields__doc__}, |
| 1120 | {"joinfields", strop_joinfields, 1, joinfields__doc__}, |
| 1121 | {"lstrip", strop_lstrip, 1, lstrip__doc__}, |
| 1122 | {"lower", strop_lower, 1, lower__doc__}, |
| 1123 | {"maketrans", strop_maketrans, 1, maketrans__doc__}, |
| 1124 | {"replace", strop_replace, 1, replace__doc__}, |
| 1125 | {"rfind", strop_rfind, 1, rfind__doc__}, |
| 1126 | {"rstrip", strop_rstrip, 1,rstrip__doc__}, |
| 1127 | {"split", strop_splitfields, 1, splitfields__doc__}, |
| 1128 | {"splitfields", strop_splitfields, 1, splitfields__doc__}, |
| 1129 | {"strip", strop_strip, 1, strip__doc__}, |
| 1130 | {"swapcase", strop_swapcase, 1, swapcase__doc__}, |
| 1131 | {"translate", strop_translate, 1, translate__doc__}, |
| 1132 | {"upper", strop_upper, 1, upper__doc__}, |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 1133 | {NULL, NULL} /* sentinel */ |
| 1134 | }; |
| 1135 | |
| 1136 | |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 1137 | void |
| 1138 | initstrop() |
| 1139 | { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 1140 | PyObject *m, *d, *s; |
Guido van Rossum | d05eb8b | 1993-07-08 11:12:36 +0000 | [diff] [blame] | 1141 | char buf[256]; |
| 1142 | int c, n; |
Guido van Rossum | 983c930 | 1997-12-29 19:52:29 +0000 | [diff] [blame] | 1143 | m = Py_InitModule4("strop", strop_methods, strop_module__doc__, |
| 1144 | (PyObject*)NULL, PYTHON_API_VERSION); |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 1145 | d = PyModule_GetDict(m); |
Guido van Rossum | e22e644 | 1993-07-09 10:51:31 +0000 | [diff] [blame] | 1146 | |
| 1147 | /* Create 'whitespace' object */ |
Guido van Rossum | d05eb8b | 1993-07-08 11:12:36 +0000 | [diff] [blame] | 1148 | n = 0; |
Guido van Rossum | 7f7f274 | 1995-02-10 17:01:56 +0000 | [diff] [blame] | 1149 | for (c = 0; c < 256; c++) { |
Guido van Rossum | d05eb8b | 1993-07-08 11:12:36 +0000 | [diff] [blame] | 1150 | if (isspace(c)) |
| 1151 | buf[n++] = c; |
| 1152 | } |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 1153 | s = PyString_FromStringAndSize(buf, n); |
Guido van Rossum | e22e644 | 1993-07-09 10:51:31 +0000 | [diff] [blame] | 1154 | if (s) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 1155 | PyDict_SetItemString(d, "whitespace", s); |
| 1156 | Py_DECREF(s); |
Guido van Rossum | e22e644 | 1993-07-09 10:51:31 +0000 | [diff] [blame] | 1157 | } |
| 1158 | /* Create 'lowercase' object */ |
| 1159 | n = 0; |
Guido van Rossum | 7f7f274 | 1995-02-10 17:01:56 +0000 | [diff] [blame] | 1160 | for (c = 0; c < 256; c++) { |
Guido van Rossum | e22e644 | 1993-07-09 10:51:31 +0000 | [diff] [blame] | 1161 | if (islower(c)) |
| 1162 | buf[n++] = c; |
| 1163 | } |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 1164 | s = PyString_FromStringAndSize(buf, n); |
Guido van Rossum | e22e644 | 1993-07-09 10:51:31 +0000 | [diff] [blame] | 1165 | if (s) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 1166 | PyDict_SetItemString(d, "lowercase", s); |
| 1167 | Py_DECREF(s); |
Guido van Rossum | e22e644 | 1993-07-09 10:51:31 +0000 | [diff] [blame] | 1168 | } |
| 1169 | |
| 1170 | /* Create 'uppercase' object */ |
| 1171 | n = 0; |
Guido van Rossum | 7f7f274 | 1995-02-10 17:01:56 +0000 | [diff] [blame] | 1172 | for (c = 0; c < 256; c++) { |
Guido van Rossum | e22e644 | 1993-07-09 10:51:31 +0000 | [diff] [blame] | 1173 | if (isupper(c)) |
| 1174 | buf[n++] = c; |
| 1175 | } |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 1176 | s = PyString_FromStringAndSize(buf, n); |
Guido van Rossum | e22e644 | 1993-07-09 10:51:31 +0000 | [diff] [blame] | 1177 | if (s) { |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 1178 | PyDict_SetItemString(d, "uppercase", s); |
| 1179 | Py_DECREF(s); |
Guido van Rossum | e22e644 | 1993-07-09 10:51:31 +0000 | [diff] [blame] | 1180 | } |
| 1181 | |
Barry Warsaw | f525601 | 1996-12-09 18:35:56 +0000 | [diff] [blame] | 1182 | if (PyErr_Occurred()) |
| 1183 | Py_FatalError("can't initialize module strop"); |
Guido van Rossum | e270b43 | 1992-09-03 20:21:07 +0000 | [diff] [blame] | 1184 | } |