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" |
| 10 | |
| 11 | object * |
| 12 | newsizedstringobject(str, size) |
| 13 | char *str; |
| 14 | int size; |
| 15 | { |
| 16 | register stringobject *op = (stringobject *) |
| 17 | malloc(sizeof(stringobject) + size * sizeof(char)); |
| 18 | if (op == NULL) { |
| 19 | errno = ENOMEM; |
| 20 | } |
| 21 | else { |
| 22 | NEWREF(op); |
| 23 | op->ob_type = &Stringtype; |
| 24 | op->ob_size = size; |
| 25 | if (str != NULL) |
| 26 | memcpy(op->ob_sval, str, size); |
| 27 | op->ob_sval[size] = '\0'; |
| 28 | } |
| 29 | return (object *) op; |
| 30 | } |
| 31 | |
| 32 | object * |
| 33 | newstringobject(str) |
| 34 | char *str; |
| 35 | { |
| 36 | register unsigned int size = strlen(str); |
| 37 | register stringobject *op = (stringobject *) |
| 38 | malloc(sizeof(stringobject) + size * sizeof(char)); |
| 39 | if (op == NULL) { |
| 40 | errno = ENOMEM; |
| 41 | } |
| 42 | else { |
| 43 | NEWREF(op); |
| 44 | op->ob_type = &Stringtype; |
| 45 | op->ob_size = size; |
| 46 | strcpy(op->ob_sval, str); |
| 47 | } |
| 48 | return (object *) op; |
| 49 | } |
| 50 | |
| 51 | unsigned int |
| 52 | getstringsize(op) |
| 53 | register object *op; |
| 54 | { |
| 55 | if (!is_stringobject(op)) { |
| 56 | errno = EBADF; |
| 57 | return -1; |
| 58 | } |
| 59 | return ((stringobject *)op) -> ob_size; |
| 60 | } |
| 61 | |
| 62 | /*const*/ char * |
| 63 | getstringvalue(op) |
| 64 | register object *op; |
| 65 | { |
| 66 | if (!is_stringobject(op)) { |
| 67 | errno = EBADF; |
| 68 | return NULL; |
| 69 | } |
| 70 | return ((stringobject *)op) -> ob_sval; |
| 71 | } |
| 72 | |
| 73 | /* Methods */ |
| 74 | |
| 75 | static void |
| 76 | stringprint(op, fp, flags) |
| 77 | stringobject *op; |
| 78 | FILE *fp; |
| 79 | int flags; |
| 80 | { |
| 81 | int i; |
| 82 | char c; |
| 83 | if (flags & PRINT_RAW) { |
| 84 | fwrite(op->ob_sval, 1, (int) op->ob_size, fp); |
| 85 | return; |
| 86 | } |
| 87 | fprintf(fp, "'"); |
| 88 | for (i = 0; i < op->ob_size; i++) { |
| 89 | c = op->ob_sval[i]; |
| 90 | if (c == '\'' || c == '\\') |
| 91 | fprintf(fp, "\\%c", c); |
| 92 | else if (c < ' ' || c >= 0177) |
| 93 | fprintf(fp, "\\%03o", c&0377); |
| 94 | else |
| 95 | putc(c, fp); |
| 96 | } |
| 97 | fprintf(fp, "'"); |
| 98 | } |
| 99 | |
| 100 | static object * |
| 101 | stringrepr(op) |
| 102 | register stringobject *op; |
| 103 | { |
| 104 | /* XXX overflow? */ |
| 105 | int newsize = 2 + 4 * op->ob_size * sizeof(char); |
| 106 | object *v = newsizedstringobject((char *)NULL, newsize); |
| 107 | if (v == NULL) { |
| 108 | errno = ENOMEM; |
| 109 | } |
| 110 | else { |
| 111 | register int i; |
| 112 | register char c; |
| 113 | register char *p; |
| 114 | NEWREF(v); |
| 115 | v->ob_type = &Stringtype; |
| 116 | ((stringobject *)v)->ob_size = newsize; |
| 117 | p = ((stringobject *)v)->ob_sval; |
| 118 | *p++ = '\''; |
| 119 | for (i = 0; i < op->ob_size; i++) { |
| 120 | c = op->ob_sval[i]; |
| 121 | if (c == '\'' || c == '\\') |
| 122 | *p++ = '\\', *p++ = c; |
| 123 | else if (c < ' ' || c >= 0177) { |
| 124 | sprintf(p, "\\%03o", c&0377); |
| 125 | while (*p != '\0') |
| 126 | p++; |
| 127 | |
| 128 | } |
| 129 | else |
| 130 | *p++ = c; |
| 131 | } |
| 132 | *p++ = '\''; |
| 133 | *p = '\0'; |
| 134 | resizestring(&v, (int) (p - ((stringobject *)v)->ob_sval)); |
| 135 | } |
| 136 | return v; |
| 137 | } |
| 138 | |
| 139 | static int |
| 140 | stringlength(a) |
| 141 | stringobject *a; |
| 142 | { |
| 143 | return a->ob_size; |
| 144 | } |
| 145 | |
| 146 | static object * |
| 147 | stringconcat(a, bb) |
| 148 | register stringobject *a; |
| 149 | register object *bb; |
| 150 | { |
| 151 | register unsigned int size; |
| 152 | register stringobject *op; |
| 153 | if (!is_stringobject(bb)) { |
| 154 | errno = EINVAL; |
| 155 | return NULL; |
| 156 | } |
| 157 | #define b ((stringobject *)bb) |
| 158 | /* Optimize cases with empty left or right operand */ |
| 159 | if (a->ob_size == 0) { |
| 160 | INCREF(bb); |
| 161 | return bb; |
| 162 | } |
| 163 | if (b->ob_size == 0) { |
| 164 | INCREF(a); |
| 165 | return (object *)a; |
| 166 | } |
| 167 | size = a->ob_size + b->ob_size; |
| 168 | op = (stringobject *) |
| 169 | malloc(sizeof(stringobject) + size * sizeof(char)); |
| 170 | if (op == NULL) { |
| 171 | errno = ENOMEM; |
| 172 | } |
| 173 | else { |
| 174 | NEWREF(op); |
| 175 | op->ob_type = &Stringtype; |
| 176 | op->ob_size = size; |
| 177 | memcpy(op->ob_sval, a->ob_sval, (int) a->ob_size); |
| 178 | memcpy(op->ob_sval + a->ob_size, b->ob_sval, (int) b->ob_size); |
| 179 | op->ob_sval[size] = '\0'; |
| 180 | } |
| 181 | return (object *) op; |
| 182 | #undef b |
| 183 | } |
| 184 | |
| 185 | static object * |
| 186 | stringrepeat(a, n) |
| 187 | register stringobject *a; |
| 188 | register int n; |
| 189 | { |
| 190 | register int i; |
| 191 | register unsigned int size; |
| 192 | register stringobject *op; |
| 193 | if (n < 0) |
| 194 | n = 0; |
| 195 | size = a->ob_size * n; |
| 196 | if (size == a->ob_size) { |
| 197 | INCREF(a); |
| 198 | return (object *)a; |
| 199 | } |
| 200 | op = (stringobject *) |
| 201 | malloc(sizeof(stringobject) + size * sizeof(char)); |
| 202 | if (op == NULL) { |
| 203 | errno = ENOMEM; |
| 204 | } |
| 205 | else { |
| 206 | NEWREF(op); |
| 207 | op->ob_type = &Stringtype; |
| 208 | op->ob_size = size; |
| 209 | for (i = 0; i < size; i += a->ob_size) |
| 210 | memcpy(op->ob_sval+i, a->ob_sval, (int) a->ob_size); |
| 211 | op->ob_sval[size] = '\0'; |
| 212 | } |
| 213 | return (object *) op; |
| 214 | } |
| 215 | |
| 216 | /* String slice a[i:j] consists of characters a[i] ... a[j-1] */ |
| 217 | |
| 218 | static object * |
| 219 | stringslice(a, i, j) |
| 220 | register stringobject *a; |
| 221 | register int i, j; /* May be negative! */ |
| 222 | { |
| 223 | if (i < 0) |
| 224 | i = 0; |
| 225 | if (j < 0) |
| 226 | j = 0; /* Avoid signed/unsigned bug in next line */ |
| 227 | if (j > a->ob_size) |
| 228 | j = a->ob_size; |
| 229 | if (i == 0 && j == a->ob_size) { /* It's the same as a */ |
| 230 | INCREF(a); |
| 231 | return (object *)a; |
| 232 | } |
| 233 | if (j < i) |
| 234 | j = i; |
| 235 | return newsizedstringobject(a->ob_sval + i, (int) (j-i)); |
| 236 | } |
| 237 | |
| 238 | static object * |
| 239 | stringitem(a, i) |
| 240 | stringobject *a; |
| 241 | register int i; |
| 242 | { |
| 243 | if (i < 0 || i >= a->ob_size) { |
| 244 | errno = EDOM; |
| 245 | return NULL; |
| 246 | } |
| 247 | return stringslice(a, i, i+1); |
| 248 | } |
| 249 | |
| 250 | static int |
| 251 | stringcompare(a, b) |
| 252 | stringobject *a, *b; |
| 253 | { |
| 254 | /* XXX should use memcmp on shortest size, then compare lengths */ |
| 255 | return strcmp(a->ob_sval, b->ob_sval); |
| 256 | } |
| 257 | |
| 258 | static sequence_methods string_as_sequence = { |
| 259 | stringlength, /*tp_length*/ |
| 260 | stringconcat, /*tp_concat*/ |
| 261 | stringrepeat, /*tp_repeat*/ |
| 262 | stringitem, /*tp_item*/ |
| 263 | stringslice, /*tp_slice*/ |
| 264 | 0, /*tp_ass_item*/ |
| 265 | 0, /*tp_ass_slice*/ |
| 266 | }; |
| 267 | |
| 268 | typeobject Stringtype = { |
| 269 | OB_HEAD_INIT(&Typetype) |
| 270 | 0, |
| 271 | "string", |
| 272 | sizeof(stringobject), |
| 273 | sizeof(char), |
| 274 | free, /*tp_dealloc*/ |
| 275 | stringprint, /*tp_print*/ |
| 276 | 0, /*tp_getattr*/ |
| 277 | 0, /*tp_setattr*/ |
| 278 | stringcompare, /*tp_compare*/ |
| 279 | stringrepr, /*tp_repr*/ |
| 280 | 0, /*tp_as_number*/ |
| 281 | &string_as_sequence, /*tp_as_sequence*/ |
| 282 | 0, /*tp_as_mapping*/ |
| 283 | }; |
| 284 | |
| 285 | void |
| 286 | joinstring(pv, w) |
| 287 | register object **pv; |
| 288 | register object *w; |
| 289 | { |
| 290 | register object *v; |
| 291 | if (*pv == NULL || w == NULL || !is_stringobject(*pv)) |
| 292 | return; |
| 293 | v = stringconcat((stringobject *) *pv, w); |
| 294 | DECREF(*pv); |
| 295 | *pv = v; |
| 296 | } |
| 297 | |
| 298 | /* The following function breaks the notion that strings are immutable: |
| 299 | it changes the size of a string. We get away with this only if there |
| 300 | is only one module referencing the object. You can also think of it |
| 301 | as creating a new string object and destroying the old one, only |
| 302 | more efficiently. In any case, don't use this if the string may |
| 303 | already be known to some other part of the code... */ |
| 304 | |
| 305 | int |
| 306 | resizestring(pv, newsize) |
| 307 | object **pv; |
| 308 | int newsize; |
| 309 | { |
| 310 | register stringobject *v; |
| 311 | v = (stringobject *) *pv; |
| 312 | if (!is_stringobject(v) || v->ob_refcnt != 1) { |
| 313 | *pv = 0; |
| 314 | DECREF(v); |
| 315 | return errno = EBADF; |
| 316 | } |
| 317 | *pv = (object *) |
| 318 | realloc((char *)v, |
| 319 | sizeof(stringobject) + newsize * sizeof(char)); |
| 320 | if (*pv == NULL) { |
| 321 | DECREF(v); |
| 322 | return errno = ENOMEM; |
| 323 | } |
| 324 | v = (stringobject *) *pv; |
| 325 | v->ob_size = newsize; |
| 326 | v->ob_sval[newsize] = '\0'; |
| 327 | return 0; |
| 328 | } |