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