Guido van Rossum | f70e43a | 1991-02-19 12:39:46 +0000 | [diff] [blame] | 1 | /*********************************************************** |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 2 | Copyright 1991, 1992, 1993 by Stichting Mathematisch Centrum, |
| 3 | Amsterdam, The Netherlands. |
Guido van Rossum | f70e43a | 1991-02-19 12:39:46 +0000 | [diff] [blame] | 4 | |
| 5 | All Rights Reserved |
| 6 | |
| 7 | Permission to use, copy, modify, and distribute this software and its |
| 8 | documentation for any purpose and without fee is hereby granted, |
| 9 | provided that the above copyright notice appear in all copies and that |
| 10 | both that copyright notice and this permission notice appear in |
| 11 | supporting documentation, and that the names of Stichting Mathematisch |
| 12 | Centrum or CWI not be used in advertising or publicity pertaining to |
| 13 | distribution of the software without specific, written prior permission. |
| 14 | |
| 15 | STICHTING MATHEMATISCH CENTRUM DISCLAIMS ALL WARRANTIES WITH REGARD TO |
| 16 | THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND |
| 17 | FITNESS, IN NO EVENT SHALL STICHTING MATHEMATISCH CENTRUM BE LIABLE |
| 18 | FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES |
| 19 | WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN |
| 20 | ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT |
| 21 | OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. |
| 22 | |
| 23 | ******************************************************************/ |
| 24 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 25 | /* String object implementation */ |
| 26 | |
Guido van Rossum | 3f5da24 | 1990-12-20 15:06:42 +0000 | [diff] [blame] | 27 | #include "allobjects.h" |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 28 | |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 29 | #ifdef COUNT_ALLOCS |
| 30 | int null_strings, one_strings; |
| 31 | #endif |
| 32 | |
| 33 | #ifdef __STDC__ |
| 34 | #include <limits.h> |
| 35 | #else |
| 36 | #ifndef UCHAR_MAX |
| 37 | #define UCHAR_MAX 255 |
| 38 | #endif |
| 39 | #endif |
| 40 | |
| 41 | static stringobject *characters[UCHAR_MAX + 1]; |
| 42 | static stringobject *nullstring; |
| 43 | |
| 44 | /* |
| 45 | Newsizedstringobject() and newstringobject() try in certain cases |
| 46 | to share string objects. When the size of the string is zero, |
| 47 | these routines always return a pointer to the same string object; |
| 48 | when the size is one, they return a pointer to an already existing |
| 49 | object if the contents of the string is known. For |
| 50 | newstringobject() this is always the case, for |
| 51 | newsizedstringobject() this is the case when the first argument in |
| 52 | not NULL. |
| 53 | A common practice to allocate a string and then fill it in or |
| 54 | change it must be done carefully. It is only allowed to change the |
| 55 | contents of the string if the obect was gotten from |
| 56 | newsizedstringobject() with a NULL first argument, because in the |
| 57 | future these routines may try to do even more sharing of objects. |
| 58 | */ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 59 | object * |
| 60 | newsizedstringobject(str, size) |
| 61 | char *str; |
| 62 | int size; |
| 63 | { |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 64 | register stringobject *op; |
| 65 | if (size == 0 && (op = nullstring) != NULL) { |
| 66 | #ifdef COUNT_ALLOCS |
| 67 | null_strings++; |
| 68 | #endif |
| 69 | INCREF(op); |
| 70 | return op; |
| 71 | } |
| 72 | if (size == 1 && str != NULL && (op = characters[*str & UCHAR_MAX]) != NULL) { |
| 73 | #ifdef COUNT_ALLOCS |
| 74 | one_strings++; |
| 75 | #endif |
| 76 | INCREF(op); |
| 77 | return op; |
| 78 | } |
| 79 | op = (stringobject *) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 80 | malloc(sizeof(stringobject) + size * sizeof(char)); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 81 | if (op == NULL) |
| 82 | return err_nomem(); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 83 | op->ob_type = &Stringtype; |
| 84 | op->ob_size = size; |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 85 | #ifdef CACHE_HASH |
| 86 | op->ob_shash = -1; |
| 87 | #endif |
Sjoerd Mullender | a9c3c22 | 1993-10-11 12:54:31 +0000 | [diff] [blame] | 88 | NEWREF(op); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 89 | if (str != NULL) |
| 90 | memcpy(op->ob_sval, str, size); |
| 91 | op->ob_sval[size] = '\0'; |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 92 | if (size == 0) { |
| 93 | nullstring = op; |
| 94 | INCREF(op); |
| 95 | } else if (size == 1 && str != NULL) { |
| 96 | characters[*str & UCHAR_MAX] = op; |
| 97 | INCREF(op); |
| 98 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 99 | return (object *) op; |
| 100 | } |
| 101 | |
| 102 | object * |
| 103 | newstringobject(str) |
| 104 | char *str; |
| 105 | { |
| 106 | register unsigned int size = strlen(str); |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 107 | register stringobject *op; |
| 108 | if (size == 0 && (op = nullstring) != NULL) { |
| 109 | #ifdef COUNT_ALLOCS |
| 110 | null_strings++; |
| 111 | #endif |
| 112 | INCREF(op); |
| 113 | return op; |
| 114 | } |
| 115 | if (size == 1 && (op = characters[*str & UCHAR_MAX]) != NULL) { |
| 116 | #ifdef COUNT_ALLOCS |
| 117 | one_strings++; |
| 118 | #endif |
| 119 | INCREF(op); |
| 120 | return op; |
| 121 | } |
| 122 | op = (stringobject *) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 123 | malloc(sizeof(stringobject) + size * sizeof(char)); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 124 | if (op == NULL) |
| 125 | return err_nomem(); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 126 | op->ob_type = &Stringtype; |
| 127 | op->ob_size = size; |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 128 | #ifdef CACHE_HASH |
| 129 | op->ob_shash = -1; |
| 130 | #endif |
Sjoerd Mullender | a9c3c22 | 1993-10-11 12:54:31 +0000 | [diff] [blame] | 131 | NEWREF(op); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 132 | strcpy(op->ob_sval, str); |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 133 | if (size == 0) { |
| 134 | nullstring = op; |
| 135 | INCREF(op); |
| 136 | } else if (size == 1) { |
| 137 | characters[*str & UCHAR_MAX] = op; |
| 138 | INCREF(op); |
| 139 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 140 | return (object *) op; |
| 141 | } |
| 142 | |
Guido van Rossum | 234f942 | 1993-06-17 12:35:49 +0000 | [diff] [blame] | 143 | static void |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 144 | string_dealloc(op) |
Guido van Rossum | 719f5fa | 1992-03-27 17:31:02 +0000 | [diff] [blame] | 145 | object *op; |
| 146 | { |
| 147 | DEL(op); |
| 148 | } |
| 149 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 150 | unsigned int |
| 151 | getstringsize(op) |
| 152 | register object *op; |
| 153 | { |
| 154 | if (!is_stringobject(op)) { |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 155 | err_badcall(); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 156 | return -1; |
| 157 | } |
| 158 | return ((stringobject *)op) -> ob_size; |
| 159 | } |
| 160 | |
| 161 | /*const*/ char * |
| 162 | getstringvalue(op) |
| 163 | register object *op; |
| 164 | { |
| 165 | if (!is_stringobject(op)) { |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 166 | err_badcall(); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 167 | return NULL; |
| 168 | } |
| 169 | return ((stringobject *)op) -> ob_sval; |
| 170 | } |
| 171 | |
| 172 | /* Methods */ |
| 173 | |
Guido van Rossum | bcaa31c | 1991-06-07 22:58:57 +0000 | [diff] [blame] | 174 | static int |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 175 | string_print(op, fp, flags) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 176 | stringobject *op; |
| 177 | FILE *fp; |
| 178 | int flags; |
| 179 | { |
| 180 | int i; |
| 181 | char c; |
Guido van Rossum | bcaa31c | 1991-06-07 22:58:57 +0000 | [diff] [blame] | 182 | /* XXX Ought to check for interrupts when writing long strings */ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 183 | if (flags & PRINT_RAW) { |
| 184 | fwrite(op->ob_sval, 1, (int) op->ob_size, fp); |
Guido van Rossum | bcaa31c | 1991-06-07 22:58:57 +0000 | [diff] [blame] | 185 | return 0; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 186 | } |
| 187 | fprintf(fp, "'"); |
| 188 | for (i = 0; i < op->ob_size; i++) { |
| 189 | c = op->ob_sval[i]; |
| 190 | if (c == '\'' || c == '\\') |
| 191 | fprintf(fp, "\\%c", c); |
| 192 | else if (c < ' ' || c >= 0177) |
| 193 | fprintf(fp, "\\%03o", c&0377); |
| 194 | else |
| 195 | putc(c, fp); |
| 196 | } |
| 197 | fprintf(fp, "'"); |
Guido van Rossum | bcaa31c | 1991-06-07 22:58:57 +0000 | [diff] [blame] | 198 | return 0; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 199 | } |
| 200 | |
| 201 | static object * |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 202 | string_repr(op) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 203 | register stringobject *op; |
| 204 | { |
| 205 | /* XXX overflow? */ |
| 206 | int newsize = 2 + 4 * op->ob_size * sizeof(char); |
| 207 | object *v = newsizedstringobject((char *)NULL, newsize); |
| 208 | if (v == NULL) { |
Guido van Rossum | bcaa31c | 1991-06-07 22:58:57 +0000 | [diff] [blame] | 209 | return NULL; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 210 | } |
| 211 | else { |
| 212 | register int i; |
| 213 | register char c; |
| 214 | register char *p; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 215 | p = ((stringobject *)v)->ob_sval; |
| 216 | *p++ = '\''; |
| 217 | for (i = 0; i < op->ob_size; i++) { |
| 218 | c = op->ob_sval[i]; |
| 219 | if (c == '\'' || c == '\\') |
| 220 | *p++ = '\\', *p++ = c; |
| 221 | else if (c < ' ' || c >= 0177) { |
| 222 | sprintf(p, "\\%03o", c&0377); |
| 223 | while (*p != '\0') |
| 224 | p++; |
| 225 | |
| 226 | } |
| 227 | else |
| 228 | *p++ = c; |
| 229 | } |
| 230 | *p++ = '\''; |
| 231 | *p = '\0'; |
| 232 | resizestring(&v, (int) (p - ((stringobject *)v)->ob_sval)); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 233 | return v; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 234 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 235 | } |
| 236 | |
| 237 | static int |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 238 | string_length(a) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 239 | stringobject *a; |
| 240 | { |
| 241 | return a->ob_size; |
| 242 | } |
| 243 | |
| 244 | static object * |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 245 | string_concat(a, bb) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 246 | register stringobject *a; |
| 247 | register object *bb; |
| 248 | { |
| 249 | register unsigned int size; |
| 250 | register stringobject *op; |
| 251 | if (!is_stringobject(bb)) { |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 252 | err_badarg(); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 253 | return NULL; |
| 254 | } |
| 255 | #define b ((stringobject *)bb) |
| 256 | /* Optimize cases with empty left or right operand */ |
| 257 | if (a->ob_size == 0) { |
| 258 | INCREF(bb); |
| 259 | return bb; |
| 260 | } |
| 261 | if (b->ob_size == 0) { |
| 262 | INCREF(a); |
| 263 | return (object *)a; |
| 264 | } |
| 265 | size = a->ob_size + b->ob_size; |
| 266 | op = (stringobject *) |
| 267 | malloc(sizeof(stringobject) + size * sizeof(char)); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 268 | if (op == NULL) |
| 269 | return err_nomem(); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 270 | op->ob_type = &Stringtype; |
| 271 | op->ob_size = size; |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 272 | #ifdef CACHE_HASH |
| 273 | op->ob_shash = -1; |
| 274 | #endif |
Sjoerd Mullender | a9c3c22 | 1993-10-11 12:54:31 +0000 | [diff] [blame] | 275 | NEWREF(op); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 276 | memcpy(op->ob_sval, a->ob_sval, (int) a->ob_size); |
| 277 | memcpy(op->ob_sval + a->ob_size, b->ob_sval, (int) b->ob_size); |
| 278 | op->ob_sval[size] = '\0'; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 279 | return (object *) op; |
| 280 | #undef b |
| 281 | } |
| 282 | |
| 283 | static object * |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 284 | string_repeat(a, n) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 285 | register stringobject *a; |
| 286 | register int n; |
| 287 | { |
| 288 | register int i; |
| 289 | register unsigned int size; |
| 290 | register stringobject *op; |
| 291 | if (n < 0) |
| 292 | n = 0; |
| 293 | size = a->ob_size * n; |
| 294 | if (size == a->ob_size) { |
| 295 | INCREF(a); |
| 296 | return (object *)a; |
| 297 | } |
| 298 | op = (stringobject *) |
| 299 | malloc(sizeof(stringobject) + size * sizeof(char)); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 300 | if (op == NULL) |
| 301 | return err_nomem(); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 302 | op->ob_type = &Stringtype; |
| 303 | op->ob_size = size; |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 304 | #ifdef CACHE_HASH |
| 305 | op->ob_shash = -1; |
| 306 | #endif |
Sjoerd Mullender | a9c3c22 | 1993-10-11 12:54:31 +0000 | [diff] [blame] | 307 | NEWREF(op); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 308 | for (i = 0; i < size; i += a->ob_size) |
| 309 | memcpy(op->ob_sval+i, a->ob_sval, (int) a->ob_size); |
| 310 | op->ob_sval[size] = '\0'; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 311 | return (object *) op; |
| 312 | } |
| 313 | |
| 314 | /* String slice a[i:j] consists of characters a[i] ... a[j-1] */ |
| 315 | |
| 316 | static object * |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 317 | string_slice(a, i, j) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 318 | register stringobject *a; |
| 319 | register int i, j; /* May be negative! */ |
| 320 | { |
| 321 | if (i < 0) |
| 322 | i = 0; |
| 323 | if (j < 0) |
| 324 | j = 0; /* Avoid signed/unsigned bug in next line */ |
| 325 | if (j > a->ob_size) |
| 326 | j = a->ob_size; |
| 327 | if (i == 0 && j == a->ob_size) { /* It's the same as a */ |
| 328 | INCREF(a); |
| 329 | return (object *)a; |
| 330 | } |
| 331 | if (j < i) |
| 332 | j = i; |
| 333 | return newsizedstringobject(a->ob_sval + i, (int) (j-i)); |
| 334 | } |
| 335 | |
| 336 | static object * |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 337 | string_item(a, i) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 338 | stringobject *a; |
| 339 | register int i; |
| 340 | { |
Guido van Rossum | daa8bb3 | 1991-04-04 10:48:33 +0000 | [diff] [blame] | 341 | int c; |
| 342 | object *v; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 343 | if (i < 0 || i >= a->ob_size) { |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 344 | err_setstr(IndexError, "string index out of range"); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 345 | return NULL; |
| 346 | } |
Guido van Rossum | daa8bb3 | 1991-04-04 10:48:33 +0000 | [diff] [blame] | 347 | c = a->ob_sval[i] & UCHAR_MAX; |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 348 | v = (object *) characters[c]; |
| 349 | #ifdef COUNT_ALLOCS |
| 350 | if (v != NULL) |
| 351 | one_strings++; |
| 352 | #endif |
Guido van Rossum | daa8bb3 | 1991-04-04 10:48:33 +0000 | [diff] [blame] | 353 | if (v == NULL) { |
| 354 | v = newsizedstringobject((char *)NULL, 1); |
| 355 | if (v == NULL) |
| 356 | return NULL; |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 357 | characters[c] = (stringobject *) v; |
Guido van Rossum | daa8bb3 | 1991-04-04 10:48:33 +0000 | [diff] [blame] | 358 | ((stringobject *)v)->ob_sval[0] = c; |
| 359 | } |
| 360 | INCREF(v); |
| 361 | return v; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 362 | } |
| 363 | |
| 364 | static int |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 365 | string_compare(a, b) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 366 | stringobject *a, *b; |
| 367 | { |
Guido van Rossum | 253919f | 1991-02-13 23:18:39 +0000 | [diff] [blame] | 368 | int len_a = a->ob_size, len_b = b->ob_size; |
| 369 | int min_len = (len_a < len_b) ? len_a : len_b; |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 370 | int cmp; |
| 371 | if (min_len > 0) { |
| 372 | cmp = *a->ob_sval - *b->ob_sval; |
| 373 | if (cmp == 0) |
| 374 | cmp = memcmp(a->ob_sval, b->ob_sval, min_len); |
| 375 | if (cmp != 0) |
| 376 | return cmp; |
| 377 | } |
Guido van Rossum | 253919f | 1991-02-13 23:18:39 +0000 | [diff] [blame] | 378 | return (len_a < len_b) ? -1 : (len_a > len_b) ? 1 : 0; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 379 | } |
| 380 | |
Guido van Rossum | 9bfef44 | 1993-03-29 10:43:31 +0000 | [diff] [blame] | 381 | static long |
| 382 | string_hash(a) |
| 383 | stringobject *a; |
| 384 | { |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 385 | register int len; |
| 386 | register unsigned char *p; |
| 387 | register long x; |
| 388 | |
| 389 | #ifdef CACHE_HASH |
| 390 | if (a->ob_shash != -1) |
| 391 | return a->ob_shash; |
| 392 | #endif |
| 393 | len = a->ob_size; |
| 394 | p = (unsigned char *) a->ob_sval; |
| 395 | x = *p << 7; |
Guido van Rossum | 9bfef44 | 1993-03-29 10:43:31 +0000 | [diff] [blame] | 396 | while (--len >= 0) |
| 397 | x = (x + x + x) ^ *p++; |
| 398 | x ^= a->ob_size; |
| 399 | if (x == -1) |
| 400 | x = -2; |
Sjoerd Mullender | 3bb8a05 | 1993-10-22 12:04:32 +0000 | [diff] [blame^] | 401 | #ifdef CACHE_HASH |
| 402 | a->ob_shash = x; |
| 403 | #endif |
Guido van Rossum | 9bfef44 | 1993-03-29 10:43:31 +0000 | [diff] [blame] | 404 | return x; |
| 405 | } |
| 406 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 407 | static sequence_methods string_as_sequence = { |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 408 | string_length, /*sq_length*/ |
| 409 | string_concat, /*sq_concat*/ |
| 410 | string_repeat, /*sq_repeat*/ |
| 411 | string_item, /*sq_item*/ |
| 412 | string_slice, /*sq_slice*/ |
Guido van Rossum | f380e66 | 1991-06-04 19:36:32 +0000 | [diff] [blame] | 413 | 0, /*sq_ass_item*/ |
| 414 | 0, /*sq_ass_slice*/ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 415 | }; |
| 416 | |
| 417 | typeobject Stringtype = { |
| 418 | OB_HEAD_INIT(&Typetype) |
| 419 | 0, |
| 420 | "string", |
| 421 | sizeof(stringobject), |
| 422 | sizeof(char), |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 423 | string_dealloc, /*tp_dealloc*/ |
| 424 | string_print, /*tp_print*/ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 425 | 0, /*tp_getattr*/ |
| 426 | 0, /*tp_setattr*/ |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 427 | string_compare, /*tp_compare*/ |
| 428 | string_repr, /*tp_repr*/ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 429 | 0, /*tp_as_number*/ |
| 430 | &string_as_sequence, /*tp_as_sequence*/ |
| 431 | 0, /*tp_as_mapping*/ |
Guido van Rossum | 9bfef44 | 1993-03-29 10:43:31 +0000 | [diff] [blame] | 432 | string_hash, /*tp_hash*/ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 433 | }; |
| 434 | |
| 435 | void |
| 436 | joinstring(pv, w) |
| 437 | register object **pv; |
| 438 | register object *w; |
| 439 | { |
| 440 | register object *v; |
| 441 | if (*pv == NULL || w == NULL || !is_stringobject(*pv)) |
| 442 | return; |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 443 | v = string_concat((stringobject *) *pv, w); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 444 | DECREF(*pv); |
| 445 | *pv = v; |
| 446 | } |
| 447 | |
| 448 | /* The following function breaks the notion that strings are immutable: |
| 449 | it changes the size of a string. We get away with this only if there |
| 450 | is only one module referencing the object. You can also think of it |
| 451 | as creating a new string object and destroying the old one, only |
| 452 | more efficiently. In any case, don't use this if the string may |
| 453 | already be known to some other part of the code... */ |
| 454 | |
| 455 | int |
| 456 | resizestring(pv, newsize) |
| 457 | object **pv; |
| 458 | int newsize; |
| 459 | { |
Guido van Rossum | 921842f | 1990-11-18 17:30:23 +0000 | [diff] [blame] | 460 | register object *v; |
| 461 | register stringobject *sv; |
| 462 | v = *pv; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 463 | if (!is_stringobject(v) || v->ob_refcnt != 1) { |
| 464 | *pv = 0; |
| 465 | DECREF(v); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 466 | err_badcall(); |
| 467 | return -1; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 468 | } |
Guido van Rossum | 921842f | 1990-11-18 17:30:23 +0000 | [diff] [blame] | 469 | /* XXX UNREF/NEWREF interface should be more symmetrical */ |
Guido van Rossum | 392ab32 | 1990-11-18 17:41:19 +0000 | [diff] [blame] | 470 | #ifdef REF_DEBUG |
Guido van Rossum | 921842f | 1990-11-18 17:30:23 +0000 | [diff] [blame] | 471 | --ref_total; |
| 472 | #endif |
| 473 | UNREF(v); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 474 | *pv = (object *) |
| 475 | realloc((char *)v, |
| 476 | sizeof(stringobject) + newsize * sizeof(char)); |
| 477 | if (*pv == NULL) { |
Guido van Rossum | 921842f | 1990-11-18 17:30:23 +0000 | [diff] [blame] | 478 | DEL(v); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 479 | err_nomem(); |
| 480 | return -1; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 481 | } |
Guido van Rossum | 921842f | 1990-11-18 17:30:23 +0000 | [diff] [blame] | 482 | NEWREF(*pv); |
| 483 | sv = (stringobject *) *pv; |
| 484 | sv->ob_size = newsize; |
| 485 | sv->ob_sval[newsize] = '\0'; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 486 | return 0; |
| 487 | } |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 488 | |
| 489 | /* Helpers for formatstring */ |
| 490 | |
| 491 | static object * |
| 492 | getnextarg(args, arglen, p_argidx) |
| 493 | object *args; |
| 494 | int arglen; |
| 495 | int *p_argidx; |
| 496 | { |
| 497 | int argidx = *p_argidx; |
| 498 | if (argidx < arglen) { |
| 499 | (*p_argidx)++; |
| 500 | if (arglen < 0) |
| 501 | return args; |
| 502 | else |
| 503 | return gettupleitem(args, argidx); |
| 504 | } |
| 505 | err_setstr(TypeError, "not enough arguments for format string"); |
| 506 | return NULL; |
| 507 | } |
| 508 | |
| 509 | #define F_LJUST (1<<0) |
| 510 | #define F_SIGN (1<<1) |
| 511 | #define F_BLANK (1<<2) |
| 512 | #define F_ALT (1<<3) |
| 513 | #define F_ZERO (1<<4) |
| 514 | |
| 515 | extern double fabs PROTO((double)); |
| 516 | |
| 517 | static char * |
| 518 | formatfloat(flags, prec, type, v) |
| 519 | int flags; |
| 520 | int prec; |
| 521 | int type; |
| 522 | object *v; |
| 523 | { |
| 524 | char fmt[20]; |
| 525 | static char buf[120]; |
| 526 | double x; |
| 527 | if (!getargs(v, "d;float argument required", &x)) |
| 528 | return NULL; |
| 529 | if (prec < 0) |
| 530 | prec = 6; |
| 531 | if (prec > 50) |
| 532 | prec = 50; /* Arbitrary limitation */ |
| 533 | if (type == 'f' && fabs(x)/1e25 >= 1e25) |
| 534 | type = 'g'; |
| 535 | sprintf(fmt, "%%%s.%d%c", (flags&F_ALT) ? "#" : "", prec, type); |
| 536 | sprintf(buf, fmt, x); |
| 537 | return buf; |
| 538 | } |
| 539 | |
| 540 | static char * |
| 541 | formatint(flags, prec, type, v) |
| 542 | int flags; |
| 543 | int prec; |
| 544 | int type; |
| 545 | object *v; |
| 546 | { |
| 547 | char fmt[20]; |
| 548 | static char buf[50]; |
| 549 | long x; |
| 550 | if (!getargs(v, "l;int argument required", &x)) |
| 551 | return NULL; |
| 552 | if (prec < 0) |
| 553 | prec = 1; |
| 554 | sprintf(fmt, "%%%s.%dl%c", (flags&F_ALT) ? "#" : "", prec, type); |
| 555 | sprintf(buf, fmt, x); |
| 556 | return buf; |
| 557 | } |
| 558 | |
| 559 | static char * |
| 560 | formatchar(v) |
| 561 | object *v; |
| 562 | { |
| 563 | static char buf[2]; |
| 564 | if (is_stringobject(v)) { |
| 565 | if (!getargs(v, "c;%c requires int or char", &buf[0])) |
| 566 | return NULL; |
| 567 | } |
| 568 | else { |
| 569 | if (!getargs(v, "b;%c requires int or char", &buf[0])) |
| 570 | return NULL; |
| 571 | } |
| 572 | buf[1] = '\0'; |
| 573 | return buf; |
| 574 | } |
| 575 | |
| 576 | /* fmt%(v1,v2,...) is roughly equivalent to sprintf(fmt, v1, v2, ...) */ |
| 577 | |
| 578 | object * |
| 579 | formatstring(format, args) |
| 580 | object *format; |
| 581 | object *args; |
| 582 | { |
| 583 | char *fmt, *res; |
| 584 | int fmtcnt, rescnt, reslen, arglen, argidx; |
| 585 | object *result; |
| 586 | if (format == NULL || !is_stringobject(format) || args == NULL) { |
| 587 | err_badcall(); |
| 588 | return NULL; |
| 589 | } |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 590 | fmt = getstringvalue(format); |
| 591 | fmtcnt = getstringsize(format); |
| 592 | reslen = rescnt = fmtcnt + 100; |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 593 | result = newsizedstringobject((char *)NULL, reslen); |
| 594 | if (result == NULL) |
| 595 | return NULL; |
| 596 | res = getstringvalue(result); |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 597 | if (is_tupleobject(args)) { |
| 598 | arglen = gettuplesize(args); |
| 599 | argidx = 0; |
| 600 | } |
| 601 | else { |
| 602 | arglen = -1; |
| 603 | argidx = -2; |
| 604 | } |
| 605 | while (--fmtcnt >= 0) { |
| 606 | if (*fmt != '%') { |
| 607 | if (--rescnt < 0) { |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 608 | rescnt = fmtcnt + 100; |
| 609 | reslen += rescnt; |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 610 | if (resizestring(&result, reslen) < 0) |
| 611 | return NULL; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 612 | res = getstringvalue(result) + reslen - rescnt; |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 613 | } |
| 614 | *res++ = *fmt++; |
| 615 | } |
| 616 | else { |
| 617 | /* Got a format specifier */ |
| 618 | int flags = 0; |
| 619 | char *fmtstart = fmt++; |
| 620 | int width = -1; |
| 621 | int prec = -1; |
| 622 | int size = 0; |
| 623 | int c; |
| 624 | int fill; |
| 625 | object *v; |
| 626 | char *buf; |
| 627 | int sign; |
| 628 | int len; |
| 629 | while (--fmtcnt >= 0) { |
| 630 | switch (c = *fmt++) { |
| 631 | case '-': flags |= F_LJUST; continue; |
| 632 | case '+': flags |= F_SIGN; continue; |
| 633 | case ' ': flags |= F_BLANK; continue; |
| 634 | case '#': flags |= F_ALT; continue; |
| 635 | case '0': flags |= F_ZERO; continue; |
| 636 | } |
| 637 | break; |
| 638 | } |
| 639 | if (c == '*') { |
| 640 | v = getnextarg(args, arglen, &argidx); |
| 641 | if (v == NULL) |
| 642 | goto error; |
| 643 | if (!is_intobject(v)) { |
| 644 | err_setstr(TypeError, "* wants int"); |
| 645 | goto error; |
| 646 | } |
| 647 | width = getintvalue(v); |
| 648 | if (width < 0) |
| 649 | width = 0; |
| 650 | if (--fmtcnt >= 0) |
| 651 | c = *fmt++; |
| 652 | } |
| 653 | else if (isdigit(c)) { |
| 654 | width = c - '0'; |
| 655 | while (--fmtcnt >= 0) { |
| 656 | c = *fmt++; |
| 657 | if (!isdigit(c)) |
| 658 | break; |
| 659 | if ((width*10) / 10 != width) { |
| 660 | err_setstr(ValueError, |
| 661 | "width too big"); |
| 662 | goto error; |
| 663 | } |
| 664 | width = width*10 + (c - '0'); |
| 665 | } |
| 666 | } |
| 667 | if (c == '.') { |
| 668 | prec = 0; |
| 669 | if (--fmtcnt >= 0) |
| 670 | c = *fmt++; |
| 671 | if (c == '*') { |
| 672 | v = getnextarg(args, arglen, &argidx); |
| 673 | if (v == NULL) |
| 674 | goto error; |
| 675 | if (!is_intobject(v)) { |
| 676 | err_setstr(TypeError, |
| 677 | "* wants int"); |
| 678 | goto error; |
| 679 | } |
| 680 | prec = getintvalue(v); |
| 681 | if (prec < 0) |
| 682 | prec = 0; |
| 683 | if (--fmtcnt >= 0) |
| 684 | c = *fmt++; |
| 685 | } |
| 686 | else if (isdigit(c)) { |
| 687 | prec = c - '0'; |
| 688 | while (--fmtcnt >= 0) { |
| 689 | c = *fmt++; |
| 690 | if (!isdigit(c)) |
| 691 | break; |
| 692 | if ((prec*10) / 10 != prec) { |
| 693 | err_setstr(ValueError, |
| 694 | "prec too big"); |
| 695 | goto error; |
| 696 | } |
| 697 | prec = prec*10 + (c - '0'); |
| 698 | } |
| 699 | } |
| 700 | } /* prec */ |
| 701 | if (fmtcnt >= 0) { |
| 702 | if (c == 'h' || c == 'l' || c == 'L') { |
| 703 | size = c; |
| 704 | if (--fmtcnt >= 0) |
| 705 | c = *fmt++; |
| 706 | } |
| 707 | } |
| 708 | if (fmtcnt < 0) { |
| 709 | err_setstr(ValueError, "incomplete format"); |
| 710 | goto error; |
| 711 | } |
| 712 | if (c != '%') { |
| 713 | v = getnextarg(args, arglen, &argidx); |
| 714 | if (v == NULL) |
| 715 | goto error; |
| 716 | } |
| 717 | sign = 0; |
| 718 | fill = ' '; |
| 719 | switch (c) { |
| 720 | case '%': |
| 721 | buf = "%"; |
| 722 | len = 1; |
| 723 | break; |
| 724 | case 's': |
| 725 | if (!is_stringobject(v)) { |
| 726 | err_setstr(TypeError, |
| 727 | "%s wants string"); |
| 728 | goto error; |
| 729 | } |
| 730 | buf = getstringvalue(v); |
| 731 | len = getstringsize(v); |
| 732 | if (prec >= 0 && len > prec) |
| 733 | len = prec; |
| 734 | break; |
| 735 | case 'i': |
| 736 | case 'd': |
| 737 | case 'u': |
| 738 | case 'o': |
| 739 | case 'x': |
| 740 | case 'X': |
| 741 | if (c == 'i') |
| 742 | c = 'd'; |
| 743 | buf = formatint(flags, prec, c, v); |
| 744 | if (buf == NULL) |
| 745 | goto error; |
| 746 | len = strlen(buf); |
| 747 | sign = (c == 'd'); |
| 748 | if (flags&F_ZERO) |
| 749 | fill = '0'; |
| 750 | break; |
| 751 | case 'e': |
| 752 | case 'E': |
| 753 | case 'f': |
| 754 | case 'g': |
| 755 | case 'G': |
| 756 | buf = formatfloat(flags, prec, c, v); |
| 757 | if (buf == NULL) |
| 758 | goto error; |
| 759 | len = strlen(buf); |
| 760 | sign = 1; |
| 761 | if (flags&F_ZERO) |
| 762 | fill = '0'; |
| 763 | break; |
| 764 | case 'c': |
| 765 | buf = formatchar(v); |
| 766 | if (buf == NULL) |
| 767 | goto error; |
| 768 | len = strlen(buf); |
| 769 | break; |
| 770 | default: |
| 771 | err_setstr(ValueError, |
| 772 | "unsupported format character"); |
| 773 | goto error; |
| 774 | } |
Guido van Rossum | 234f942 | 1993-06-17 12:35:49 +0000 | [diff] [blame] | 775 | /* XXX There's a bug somewhere here so that |
| 776 | XXX '%4d'%-1 yields '- 1' ... */ |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 777 | if (sign) { |
| 778 | if (*buf == '-' || *buf == '+') { |
| 779 | sign = *buf++; |
| 780 | len--; |
| 781 | } |
| 782 | else if (flags & F_SIGN) |
| 783 | sign = '+'; |
| 784 | else if (flags & F_BLANK) |
| 785 | sign = ' '; |
| 786 | else |
| 787 | sign = '\0'; |
| 788 | } |
| 789 | if (width < len) |
| 790 | width = len; |
| 791 | if (rescnt < width + (sign != '\0')) { |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 792 | reslen -= rescnt; |
| 793 | rescnt = width + fmtcnt + 100; |
| 794 | reslen += rescnt; |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 795 | if (resizestring(&result, reslen) < 0) |
| 796 | return NULL; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 797 | res = getstringvalue(result) + reslen - rescnt; |
Guido van Rossum | e537240 | 1993-03-16 12:15:04 +0000 | [diff] [blame] | 798 | } |
| 799 | if (sign) { |
| 800 | *res++ = sign; |
| 801 | rescnt--; |
| 802 | if (width > len) |
| 803 | width--; |
| 804 | } |
| 805 | if (width > len && !(flags&F_LJUST)) { |
| 806 | do { |
| 807 | --rescnt; |
| 808 | *res++ = fill; |
| 809 | } while (--width > len); |
| 810 | } |
| 811 | memcpy(res, buf, len); |
| 812 | res += len; |
| 813 | rescnt -= len; |
| 814 | while (--width >= len) { |
| 815 | --rescnt; |
| 816 | *res++ = ' '; |
| 817 | } |
| 818 | } /* '%' */ |
| 819 | } /* until end */ |
| 820 | if (argidx < arglen) { |
| 821 | err_setstr(TypeError, "not all arguments converted"); |
| 822 | goto error; |
| 823 | } |
| 824 | resizestring(&result, reslen - rescnt); |
| 825 | return result; |
| 826 | error: |
| 827 | DECREF(result); |
| 828 | return NULL; |
| 829 | } |