Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 1 | /* String object implementation */ |
| 2 | |
Guido van Rossum | 3f5da24 | 1990-12-20 15:06:42 +0000 | [diff] [blame] | 3 | #include "allobjects.h" |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 4 | |
| 5 | object * |
| 6 | newsizedstringobject(str, size) |
| 7 | char *str; |
| 8 | int size; |
| 9 | { |
| 10 | register stringobject *op = (stringobject *) |
| 11 | malloc(sizeof(stringobject) + size * sizeof(char)); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 12 | if (op == NULL) |
| 13 | return err_nomem(); |
| 14 | NEWREF(op); |
| 15 | op->ob_type = &Stringtype; |
| 16 | op->ob_size = size; |
| 17 | if (str != NULL) |
| 18 | memcpy(op->ob_sval, str, size); |
| 19 | op->ob_sval[size] = '\0'; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 20 | return (object *) op; |
| 21 | } |
| 22 | |
| 23 | object * |
| 24 | newstringobject(str) |
| 25 | char *str; |
| 26 | { |
| 27 | register unsigned int size = strlen(str); |
| 28 | register stringobject *op = (stringobject *) |
| 29 | malloc(sizeof(stringobject) + size * sizeof(char)); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 30 | if (op == NULL) |
| 31 | return err_nomem(); |
| 32 | NEWREF(op); |
| 33 | op->ob_type = &Stringtype; |
| 34 | op->ob_size = size; |
| 35 | strcpy(op->ob_sval, str); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 36 | return (object *) op; |
| 37 | } |
| 38 | |
| 39 | unsigned int |
| 40 | getstringsize(op) |
| 41 | register object *op; |
| 42 | { |
| 43 | if (!is_stringobject(op)) { |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 44 | err_badcall(); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 45 | return -1; |
| 46 | } |
| 47 | return ((stringobject *)op) -> ob_size; |
| 48 | } |
| 49 | |
| 50 | /*const*/ char * |
| 51 | getstringvalue(op) |
| 52 | register object *op; |
| 53 | { |
| 54 | if (!is_stringobject(op)) { |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 55 | err_badcall(); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 56 | return NULL; |
| 57 | } |
| 58 | return ((stringobject *)op) -> ob_sval; |
| 59 | } |
| 60 | |
| 61 | /* Methods */ |
| 62 | |
| 63 | static void |
| 64 | stringprint(op, fp, flags) |
| 65 | stringobject *op; |
| 66 | FILE *fp; |
| 67 | int flags; |
| 68 | { |
| 69 | int i; |
| 70 | char c; |
| 71 | if (flags & PRINT_RAW) { |
| 72 | fwrite(op->ob_sval, 1, (int) op->ob_size, fp); |
| 73 | return; |
| 74 | } |
| 75 | fprintf(fp, "'"); |
| 76 | for (i = 0; i < op->ob_size; i++) { |
| 77 | c = op->ob_sval[i]; |
| 78 | if (c == '\'' || c == '\\') |
| 79 | fprintf(fp, "\\%c", c); |
| 80 | else if (c < ' ' || c >= 0177) |
| 81 | fprintf(fp, "\\%03o", c&0377); |
| 82 | else |
| 83 | putc(c, fp); |
| 84 | } |
| 85 | fprintf(fp, "'"); |
| 86 | } |
| 87 | |
| 88 | static object * |
| 89 | stringrepr(op) |
| 90 | register stringobject *op; |
| 91 | { |
| 92 | /* XXX overflow? */ |
| 93 | int newsize = 2 + 4 * op->ob_size * sizeof(char); |
| 94 | object *v = newsizedstringobject((char *)NULL, newsize); |
| 95 | if (v == NULL) { |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 96 | return err_nomem(); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 97 | } |
| 98 | else { |
| 99 | register int i; |
| 100 | register char c; |
| 101 | register char *p; |
| 102 | NEWREF(v); |
| 103 | v->ob_type = &Stringtype; |
| 104 | ((stringobject *)v)->ob_size = newsize; |
| 105 | p = ((stringobject *)v)->ob_sval; |
| 106 | *p++ = '\''; |
| 107 | for (i = 0; i < op->ob_size; i++) { |
| 108 | c = op->ob_sval[i]; |
| 109 | if (c == '\'' || c == '\\') |
| 110 | *p++ = '\\', *p++ = c; |
| 111 | else if (c < ' ' || c >= 0177) { |
| 112 | sprintf(p, "\\%03o", c&0377); |
| 113 | while (*p != '\0') |
| 114 | p++; |
| 115 | |
| 116 | } |
| 117 | else |
| 118 | *p++ = c; |
| 119 | } |
| 120 | *p++ = '\''; |
| 121 | *p = '\0'; |
| 122 | resizestring(&v, (int) (p - ((stringobject *)v)->ob_sval)); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 123 | return v; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 124 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 125 | } |
| 126 | |
| 127 | static int |
| 128 | stringlength(a) |
| 129 | stringobject *a; |
| 130 | { |
| 131 | return a->ob_size; |
| 132 | } |
| 133 | |
| 134 | static object * |
| 135 | stringconcat(a, bb) |
| 136 | register stringobject *a; |
| 137 | register object *bb; |
| 138 | { |
| 139 | register unsigned int size; |
| 140 | register stringobject *op; |
| 141 | if (!is_stringobject(bb)) { |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 142 | err_badarg(); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 143 | return NULL; |
| 144 | } |
| 145 | #define b ((stringobject *)bb) |
| 146 | /* Optimize cases with empty left or right operand */ |
| 147 | if (a->ob_size == 0) { |
| 148 | INCREF(bb); |
| 149 | return bb; |
| 150 | } |
| 151 | if (b->ob_size == 0) { |
| 152 | INCREF(a); |
| 153 | return (object *)a; |
| 154 | } |
| 155 | size = a->ob_size + b->ob_size; |
| 156 | op = (stringobject *) |
| 157 | malloc(sizeof(stringobject) + size * sizeof(char)); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 158 | if (op == NULL) |
| 159 | return err_nomem(); |
| 160 | NEWREF(op); |
| 161 | op->ob_type = &Stringtype; |
| 162 | op->ob_size = size; |
| 163 | memcpy(op->ob_sval, a->ob_sval, (int) a->ob_size); |
| 164 | memcpy(op->ob_sval + a->ob_size, b->ob_sval, (int) b->ob_size); |
| 165 | op->ob_sval[size] = '\0'; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 166 | return (object *) op; |
| 167 | #undef b |
| 168 | } |
| 169 | |
| 170 | static object * |
| 171 | stringrepeat(a, n) |
| 172 | register stringobject *a; |
| 173 | register int n; |
| 174 | { |
| 175 | register int i; |
| 176 | register unsigned int size; |
| 177 | register stringobject *op; |
| 178 | if (n < 0) |
| 179 | n = 0; |
| 180 | size = a->ob_size * n; |
| 181 | if (size == a->ob_size) { |
| 182 | INCREF(a); |
| 183 | return (object *)a; |
| 184 | } |
| 185 | op = (stringobject *) |
| 186 | malloc(sizeof(stringobject) + size * sizeof(char)); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 187 | if (op == NULL) |
| 188 | return err_nomem(); |
| 189 | NEWREF(op); |
| 190 | op->ob_type = &Stringtype; |
| 191 | op->ob_size = size; |
| 192 | for (i = 0; i < size; i += a->ob_size) |
| 193 | memcpy(op->ob_sval+i, a->ob_sval, (int) a->ob_size); |
| 194 | op->ob_sval[size] = '\0'; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 195 | return (object *) op; |
| 196 | } |
| 197 | |
| 198 | /* String slice a[i:j] consists of characters a[i] ... a[j-1] */ |
| 199 | |
| 200 | static object * |
| 201 | stringslice(a, i, j) |
| 202 | register stringobject *a; |
| 203 | register int i, j; /* May be negative! */ |
| 204 | { |
| 205 | if (i < 0) |
| 206 | i = 0; |
| 207 | if (j < 0) |
| 208 | j = 0; /* Avoid signed/unsigned bug in next line */ |
| 209 | if (j > a->ob_size) |
| 210 | j = a->ob_size; |
| 211 | if (i == 0 && j == a->ob_size) { /* It's the same as a */ |
| 212 | INCREF(a); |
| 213 | return (object *)a; |
| 214 | } |
| 215 | if (j < i) |
| 216 | j = i; |
| 217 | return newsizedstringobject(a->ob_sval + i, (int) (j-i)); |
| 218 | } |
| 219 | |
| 220 | static object * |
| 221 | stringitem(a, i) |
| 222 | stringobject *a; |
| 223 | register int i; |
| 224 | { |
| 225 | if (i < 0 || i >= a->ob_size) { |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 226 | err_setstr(IndexError, "string index out of range"); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 227 | return NULL; |
| 228 | } |
| 229 | return stringslice(a, i, i+1); |
| 230 | } |
| 231 | |
| 232 | static int |
| 233 | stringcompare(a, b) |
| 234 | stringobject *a, *b; |
| 235 | { |
Guido van Rossum | 253919f | 1991-02-13 23:18:39 +0000 | [diff] [blame^] | 236 | int len_a = a->ob_size, len_b = b->ob_size; |
| 237 | int min_len = (len_a < len_b) ? len_a : len_b; |
| 238 | int cmp = memcmp(a->ob_sval, b->ob_sval, min_len); |
| 239 | if (cmp != 0) |
| 240 | return cmp; |
| 241 | 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] | 242 | } |
| 243 | |
| 244 | static sequence_methods string_as_sequence = { |
| 245 | stringlength, /*tp_length*/ |
| 246 | stringconcat, /*tp_concat*/ |
| 247 | stringrepeat, /*tp_repeat*/ |
| 248 | stringitem, /*tp_item*/ |
| 249 | stringslice, /*tp_slice*/ |
| 250 | 0, /*tp_ass_item*/ |
| 251 | 0, /*tp_ass_slice*/ |
| 252 | }; |
| 253 | |
| 254 | typeobject Stringtype = { |
| 255 | OB_HEAD_INIT(&Typetype) |
| 256 | 0, |
| 257 | "string", |
| 258 | sizeof(stringobject), |
| 259 | sizeof(char), |
| 260 | free, /*tp_dealloc*/ |
| 261 | stringprint, /*tp_print*/ |
| 262 | 0, /*tp_getattr*/ |
| 263 | 0, /*tp_setattr*/ |
| 264 | stringcompare, /*tp_compare*/ |
| 265 | stringrepr, /*tp_repr*/ |
| 266 | 0, /*tp_as_number*/ |
| 267 | &string_as_sequence, /*tp_as_sequence*/ |
| 268 | 0, /*tp_as_mapping*/ |
| 269 | }; |
| 270 | |
| 271 | void |
| 272 | joinstring(pv, w) |
| 273 | register object **pv; |
| 274 | register object *w; |
| 275 | { |
| 276 | register object *v; |
| 277 | if (*pv == NULL || w == NULL || !is_stringobject(*pv)) |
| 278 | return; |
| 279 | v = stringconcat((stringobject *) *pv, w); |
| 280 | DECREF(*pv); |
| 281 | *pv = v; |
| 282 | } |
| 283 | |
| 284 | /* The following function breaks the notion that strings are immutable: |
| 285 | it changes the size of a string. We get away with this only if there |
| 286 | is only one module referencing the object. You can also think of it |
| 287 | as creating a new string object and destroying the old one, only |
| 288 | more efficiently. In any case, don't use this if the string may |
| 289 | already be known to some other part of the code... */ |
| 290 | |
| 291 | int |
| 292 | resizestring(pv, newsize) |
| 293 | object **pv; |
| 294 | int newsize; |
| 295 | { |
Guido van Rossum | 921842f | 1990-11-18 17:30:23 +0000 | [diff] [blame] | 296 | register object *v; |
| 297 | register stringobject *sv; |
| 298 | v = *pv; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 299 | if (!is_stringobject(v) || v->ob_refcnt != 1) { |
| 300 | *pv = 0; |
| 301 | DECREF(v); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 302 | err_badcall(); |
| 303 | return -1; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 304 | } |
Guido van Rossum | 921842f | 1990-11-18 17:30:23 +0000 | [diff] [blame] | 305 | /* XXX UNREF/NEWREF interface should be more symmetrical */ |
Guido van Rossum | 392ab32 | 1990-11-18 17:41:19 +0000 | [diff] [blame] | 306 | #ifdef REF_DEBUG |
Guido van Rossum | 921842f | 1990-11-18 17:30:23 +0000 | [diff] [blame] | 307 | --ref_total; |
| 308 | #endif |
| 309 | UNREF(v); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 310 | *pv = (object *) |
| 311 | realloc((char *)v, |
| 312 | sizeof(stringobject) + newsize * sizeof(char)); |
| 313 | if (*pv == NULL) { |
Guido van Rossum | 921842f | 1990-11-18 17:30:23 +0000 | [diff] [blame] | 314 | DEL(v); |
Guido van Rossum | 2a9096b | 1990-10-21 22:15:08 +0000 | [diff] [blame] | 315 | err_nomem(); |
| 316 | return -1; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 317 | } |
Guido van Rossum | 921842f | 1990-11-18 17:30:23 +0000 | [diff] [blame] | 318 | NEWREF(*pv); |
| 319 | sv = (stringobject *) *pv; |
| 320 | sv->ob_size = newsize; |
| 321 | sv->ob_sval[newsize] = '\0'; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 322 | return 0; |
| 323 | } |