| | 1 | #ifdef RCSID |
| | 2 | static char RCSid[] = |
| | 3 | "$Header: d:/cvsroot/tads/tads3/VMSTR.CPP,v 1.3 1999/05/17 02:52:28 MJRoberts Exp $"; |
| | 4 | #endif |
| | 5 | |
| | 6 | /* |
| | 7 | * Copyright (c) 1998, 2002 Michael J. Roberts. All Rights Reserved. |
| | 8 | * |
| | 9 | * Please see the accompanying license file, LICENSE.TXT, for information |
| | 10 | * on using and copying this software. |
| | 11 | */ |
| | 12 | /* |
| | 13 | Name |
| | 14 | vmstr.cpp - VM string metaclass implementation |
| | 15 | Function |
| | 16 | |
| | 17 | Notes |
| | 18 | |
| | 19 | Modified |
| | 20 | 10/28/98 MJRoberts - Creation |
| | 21 | */ |
| | 22 | |
| | 23 | #include <stdlib.h> |
| | 24 | #include <stdio.h> |
| | 25 | #include <string.h> |
| | 26 | |
| | 27 | #include "t3std.h" |
| | 28 | #include "vmmcreg.h" |
| | 29 | #include "vmobj.h" |
| | 30 | #include "vmstr.h" |
| | 31 | #include "utf8.h" |
| | 32 | #include "vmerr.h" |
| | 33 | #include "vmerrnum.h" |
| | 34 | #include "vmfile.h" |
| | 35 | #include "vmstack.h" |
| | 36 | #include "vmpool.h" |
| | 37 | #include "vmmeta.h" |
| | 38 | #include "vmrun.h" |
| | 39 | #include "vmbif.h" |
| | 40 | #include "vmpredef.h" |
| | 41 | #include "vmlst.h" |
| | 42 | #include "vmuni.h" |
| | 43 | #include "vmcset.h" |
| | 44 | #include "vmbytarr.h" |
| | 45 | #include "charmap.h" |
| | 46 | |
| | 47 | |
| | 48 | /* ------------------------------------------------------------------------ */ |
| | 49 | /* |
| | 50 | * statics |
| | 51 | */ |
| | 52 | |
| | 53 | /* metaclass registration object */ |
| | 54 | static CVmMetaclassString metaclass_reg_obj; |
| | 55 | CVmMetaclass *CVmObjString::metaclass_reg_ = &metaclass_reg_obj; |
| | 56 | |
| | 57 | /* function table */ |
| | 58 | int (*CVmObjString::func_table_[])(VMG_ vm_val_t *retval, |
| | 59 | const vm_val_t *self_val, |
| | 60 | const char *str, uint *argc) = |
| | 61 | { |
| | 62 | &CVmObjString::getp_undef, |
| | 63 | &CVmObjString::getp_len, |
| | 64 | &CVmObjString::getp_substr, |
| | 65 | &CVmObjString::getp_upper, |
| | 66 | &CVmObjString::getp_lower, |
| | 67 | &CVmObjString::getp_find, |
| | 68 | &CVmObjString::getp_to_uni, |
| | 69 | &CVmObjString::getp_htmlify, |
| | 70 | &CVmObjString::getp_starts_with, |
| | 71 | &CVmObjString::getp_ends_with, |
| | 72 | &CVmObjString::getp_to_byte_array, |
| | 73 | &CVmObjString::getp_replace |
| | 74 | }; |
| | 75 | |
| | 76 | /* ------------------------------------------------------------------------ */ |
| | 77 | /* |
| | 78 | * Static creation methods |
| | 79 | */ |
| | 80 | |
| | 81 | |
| | 82 | /* create dynamically using stack arguments */ |
| | 83 | vm_obj_id_t CVmObjString::create_from_stack(VMG_ const uchar **, uint) |
| | 84 | { |
| | 85 | /* dynamic string construction is not currently supported */ |
| | 86 | err_throw(VMERR_BAD_DYNAMIC_NEW); |
| | 87 | |
| | 88 | /* the compiler doesn't know we won't make it here */ |
| | 89 | AFTER_ERR_THROW(return VM_INVALID_OBJ;) |
| | 90 | } |
| | 91 | |
| | 92 | /* create a string with no initial contents */ |
| | 93 | vm_obj_id_t CVmObjString::create(VMG_ int in_root_set) |
| | 94 | { |
| | 95 | vm_obj_id_t id = vm_new_id(vmg_ in_root_set, FALSE, FALSE); |
| | 96 | new (vmg_ id) CVmObjString(); |
| | 97 | return id; |
| | 98 | } |
| | 99 | |
| | 100 | /* create with a given buffer size */ |
| | 101 | vm_obj_id_t CVmObjString::create(VMG_ int in_root_set, size_t byte_size) |
| | 102 | { |
| | 103 | vm_obj_id_t id = vm_new_id(vmg_ in_root_set, FALSE, FALSE); |
| | 104 | new (vmg_ id) CVmObjString(vmg_ byte_size); |
| | 105 | return id; |
| | 106 | } |
| | 107 | |
| | 108 | /* create from a constant UTF-8 string */ |
| | 109 | vm_obj_id_t CVmObjString::create(VMG_ int in_root_set, |
| | 110 | const char *str, size_t bytelen) |
| | 111 | { |
| | 112 | vm_obj_id_t id = vm_new_id(vmg_ in_root_set, FALSE, FALSE); |
| | 113 | new (vmg_ id) CVmObjString(vmg_ str, bytelen); |
| | 114 | return id; |
| | 115 | } |
| | 116 | |
| | 117 | /* ------------------------------------------------------------------------ */ |
| | 118 | /* |
| | 119 | * Constructors |
| | 120 | */ |
| | 121 | |
| | 122 | /* |
| | 123 | * create a string object with a given buffer size |
| | 124 | */ |
| | 125 | CVmObjString::CVmObjString(VMG_ size_t len) |
| | 126 | { |
| | 127 | /* |
| | 128 | * the length is limited to an unsigned 16-bit value (NB: it really is |
| | 129 | * 65535 on ALL PLATFORMS - this is a portable limit imposed by the |
| | 130 | * portable storage format, not a local platform limit) |
| | 131 | */ |
| | 132 | if (len > 65535) |
| | 133 | { |
| | 134 | ext_ = 0; |
| | 135 | err_throw(VMERR_STR_TOO_LONG); |
| | 136 | } |
| | 137 | |
| | 138 | /* |
| | 139 | * allocate space for the buffer plus the length prefix in the |
| | 140 | * variable heap |
| | 141 | */ |
| | 142 | ext_ = (char *)G_mem->get_var_heap()->alloc_mem(len + VMB_LEN, this); |
| | 143 | |
| | 144 | /* set the length */ |
| | 145 | vmb_put_len(ext_, len); |
| | 146 | } |
| | 147 | |
| | 148 | /* |
| | 149 | * create a string object from a given UTF8 string constant |
| | 150 | */ |
| | 151 | CVmObjString::CVmObjString(VMG_ const char *str, size_t len) |
| | 152 | { |
| | 153 | /* check for the length limit */ |
| | 154 | if (len > 65535) |
| | 155 | { |
| | 156 | ext_ = 0; |
| | 157 | err_throw(VMERR_STR_TOO_LONG); |
| | 158 | } |
| | 159 | |
| | 160 | /* |
| | 161 | * allocate space for the string plus the length prefix in the |
| | 162 | * variable heap |
| | 163 | */ |
| | 164 | ext_ = (char *)G_mem->get_var_heap()->alloc_mem(len + VMB_LEN, this); |
| | 165 | |
| | 166 | /* |
| | 167 | * store the length prefix in portable format (so that we can easily |
| | 168 | * write our contents to a saved state file) |
| | 169 | */ |
| | 170 | vmb_put_len(ext_, len); |
| | 171 | |
| | 172 | /* copy the string's bytes */ |
| | 173 | memcpy(ext_ + VMB_LEN, str, len); |
| | 174 | } |
| | 175 | |
| | 176 | /* ------------------------------------------------------------------------ */ |
| | 177 | /* |
| | 178 | * receive notification of deletion |
| | 179 | */ |
| | 180 | void CVmObjString::notify_delete(VMG_ int in_root_set) |
| | 181 | { |
| | 182 | /* free our extension */ |
| | 183 | if (ext_ != 0 && !in_root_set) |
| | 184 | G_mem->get_var_heap()->free_mem(ext_); |
| | 185 | } |
| | 186 | |
| | 187 | /* ------------------------------------------------------------------------ */ |
| | 188 | /* |
| | 189 | * Set a property. Strings have no settable properties, so simply |
| | 190 | * signal an error indicating that the set-prop call is invalid. |
| | 191 | */ |
| | 192 | void CVmObjString::set_prop(VMG_ CVmUndo *, vm_obj_id_t, |
| | 193 | vm_prop_id_t, const vm_val_t *) |
| | 194 | { |
| | 195 | err_throw(VMERR_INVALID_SETPROP); |
| | 196 | } |
| | 197 | |
| | 198 | /* ------------------------------------------------------------------------ */ |
| | 199 | /* |
| | 200 | * Save the object to a file |
| | 201 | */ |
| | 202 | void CVmObjString::save_to_file(VMG_ CVmFile *fp) |
| | 203 | { |
| | 204 | size_t len; |
| | 205 | |
| | 206 | /* get our length */ |
| | 207 | len = vmb_get_len(ext_); |
| | 208 | |
| | 209 | /* write the length prefix and the string */ |
| | 210 | fp->write_bytes(ext_, len + VMB_LEN); |
| | 211 | } |
| | 212 | |
| | 213 | /* |
| | 214 | * Restore the object from a file |
| | 215 | */ |
| | 216 | void CVmObjString::restore_from_file(VMG_ vm_obj_id_t, |
| | 217 | CVmFile *fp, CVmObjFixup *) |
| | 218 | { |
| | 219 | size_t len; |
| | 220 | |
| | 221 | /* read the length prefix */ |
| | 222 | len = fp->read_uint2(); |
| | 223 | |
| | 224 | /* free any existing extension */ |
| | 225 | if (ext_ != 0) |
| | 226 | { |
| | 227 | G_mem->get_var_heap()->free_mem(ext_); |
| | 228 | ext_ = 0; |
| | 229 | } |
| | 230 | |
| | 231 | /* |
| | 232 | * allocate our extension - make room for the length prefix plus the |
| | 233 | * bytes of the string |
| | 234 | */ |
| | 235 | ext_ = (char *)G_mem->get_var_heap()->alloc_mem(len + VMB_LEN, this); |
| | 236 | |
| | 237 | /* store our length prefix */ |
| | 238 | vmb_put_len(ext_, len); |
| | 239 | |
| | 240 | /* read the string */ |
| | 241 | fp->read_bytes(ext_ + VMB_LEN, len); |
| | 242 | } |
| | 243 | |
| | 244 | /* ------------------------------------------------------------------------ */ |
| | 245 | /* |
| | 246 | * Add a value to this string |
| | 247 | */ |
| | 248 | void CVmObjString::add_val(VMG_ vm_val_t *result, |
| | 249 | vm_obj_id_t self, const vm_val_t *val) |
| | 250 | { |
| | 251 | /* |
| | 252 | * Use the generic string adder, using my extension as the constant |
| | 253 | * string. We store our extension in the general string format |
| | 254 | * required by the static adder. |
| | 255 | */ |
| | 256 | add_to_str(vmg_ result, self, ext_, val); |
| | 257 | } |
| | 258 | |
| | 259 | /* |
| | 260 | * Static string adder. This creates a new string object that results |
| | 261 | * from appending the given value to the given string constant. This is |
| | 262 | * defined statically so that this same code can be shared for adding to |
| | 263 | * constant pool strings and adding to CVmObjString objects. |
| | 264 | * |
| | 265 | * 'strval' must point to a constant string. The first two bytes of the |
| | 266 | * string are stored in portable UINT2 format and give the length in |
| | 267 | * bytes of the string, not including the length prefix; immediately |
| | 268 | * following the length prefix are the bytes of the string. |
| | 269 | * |
| | 270 | * Note that we *always* create a new object to hold the result, even if |
| | 271 | * the new string is identical to the first, so that we consistently |
| | 272 | * return a distinct reference from the original. |
| | 273 | */ |
| | 274 | void CVmObjString::add_to_str(VMG_ vm_val_t *result, |
| | 275 | vm_obj_id_t self, const char *strval1, |
| | 276 | const vm_val_t *val) |
| | 277 | { |
| | 278 | const char *strval2; |
| | 279 | char buf[128]; |
| | 280 | vm_obj_id_t obj; |
| | 281 | size_t len1, len2; |
| | 282 | CVmObjString *objptr; |
| | 283 | vm_val_t new_obj2; |
| | 284 | |
| | 285 | /* convert the value to be appended to a string */ |
| | 286 | strval2 = cvt_to_str(vmg_ &new_obj2, buf, sizeof(buf), val, 10); |
| | 287 | |
| | 288 | /* |
| | 289 | * push the new string (if any) and self, to protect the two strings |
| | 290 | * from garbage collection |
| | 291 | */ |
| | 292 | G_stk->push()->set_obj(self); |
| | 293 | G_stk->push(&new_obj2); |
| | 294 | |
| | 295 | /* get the lengths of the two strings */ |
| | 296 | len1 = vmb_get_len(strval1); |
| | 297 | len2 = vmb_get_len(strval2); |
| | 298 | |
| | 299 | /* create a new string object to hold the result */ |
| | 300 | obj = create(vmg_ FALSE, len1 + len2); |
| | 301 | objptr = (CVmObjString *)vm_objp(vmg_ obj); |
| | 302 | |
| | 303 | /* copy the two strings into the new object's string buffer */ |
| | 304 | objptr->copy_into_str(0, strval1 + VMB_LEN, len1); |
| | 305 | objptr->copy_into_str(len1, strval2 + VMB_LEN, len2); |
| | 306 | |
| | 307 | /* we're done with the garbage collection protection */ |
| | 308 | G_stk->discard(2); |
| | 309 | |
| | 310 | /* return the new object in the result */ |
| | 311 | result->set_obj(obj); |
| | 312 | } |
| | 313 | |
| | 314 | |
| | 315 | /* ------------------------------------------------------------------------ */ |
| | 316 | /* |
| | 317 | * Allocate a string buffer large enough to hold a given value. We'll |
| | 318 | * use the provided buffer if possible. |
| | 319 | * |
| | 320 | * If the provided buffer is null or is not large enough, we'll allocate |
| | 321 | * a new string object with a large enough buffer to hold the value, and |
| | 322 | * return the object's extension as the buffer. This object will never |
| | 323 | * be referenced by anyone, so it will be deleted at the next garbage |
| | 324 | * collection. |
| | 325 | * |
| | 326 | * The buffer size and requested size are in bytes. |
| | 327 | */ |
| | 328 | char *CVmObjString::alloc_str_buf(VMG_ vm_val_t *new_obj, |
| | 329 | char *buf, size_t buf_size, |
| | 330 | size_t required_size) |
| | 331 | { |
| | 332 | vm_obj_id_t obj; |
| | 333 | |
| | 334 | /* if the provided buffer is large enough, use it */ |
| | 335 | if (buf != 0 && buf_size >= required_size) |
| | 336 | { |
| | 337 | /* there's no new object */ |
| | 338 | new_obj->set_nil(); |
| | 339 | |
| | 340 | /* return the buffer */ |
| | 341 | return buf; |
| | 342 | } |
| | 343 | |
| | 344 | /* allocate a new string object */ |
| | 345 | obj = create(vmg_ FALSE, required_size); |
| | 346 | |
| | 347 | /* return the new object's string buffer */ |
| | 348 | return (char *)vm_objp(vmg_ obj)->cast_to_string(vmg_ obj, new_obj); |
| | 349 | } |
| | 350 | |
| | 351 | /* ------------------------------------------------------------------------ */ |
| | 352 | /* |
| | 353 | * Convert a value to a string |
| | 354 | */ |
| | 355 | const char *CVmObjString::cvt_to_str(VMG_ vm_val_t *new_str, |
| | 356 | char *result_buf, |
| | 357 | size_t result_buf_size, |
| | 358 | const vm_val_t *val, int radix) |
| | 359 | { |
| | 360 | /* presume we won't need to create a new string object */ |
| | 361 | new_str->set_nil(); |
| | 362 | |
| | 363 | /* check the type of the value */ |
| | 364 | switch(val->typ) |
| | 365 | { |
| | 366 | case VM_SSTRING: |
| | 367 | /* it's a string constant - no conversion is necessary */ |
| | 368 | return G_const_pool->get_ptr(val->val.ofs); |
| | 369 | |
| | 370 | case VM_OBJ: |
| | 371 | /* it's an object - ask it for its string representation */ |
| | 372 | return vm_objp(vmg_ val->val.obj) |
| | 373 | ->cast_to_string(vmg_ val->val.obj, new_str); |
| | 374 | break; |
| | 375 | |
| | 376 | case VM_INT: |
| | 377 | /* |
| | 378 | * It's a number - convert it to a string. Use the provided result |
| | 379 | * buffer if possible, but make sure we have room for the number. |
| | 380 | * The unicode values we're storing are in the ascii range, so we |
| | 381 | * only need one byte per character. The longest buffer we'd need, |
| | 382 | * then, is 32 bytes, for a conversion to a binary digit string. |
| | 383 | * The conversion also needs two bytes for the length prefix; give |
| | 384 | * it a few extra bytes as insurance against future algorithm |
| | 385 | * changes that might need more padding. |
| | 386 | */ |
| | 387 | result_buf = alloc_str_buf(vmg_ new_str, |
| | 388 | result_buf, result_buf_size, 40); |
| | 389 | |
| | 390 | /* generate the string */ |
| | 391 | return cvt_int_to_str(result_buf, 40, val->val.intval, radix); |
| | 392 | |
| | 393 | case VM_NIL: |
| | 394 | /* nil - use the literal string "nil" */ |
| | 395 | return "\003\000nil"; |
| | 396 | break; |
| | 397 | |
| | 398 | case VM_TRUE: |
| | 399 | /* true - use the literal string "true" */ |
| | 400 | return "\004\000true"; |
| | 401 | break; |
| | 402 | |
| | 403 | default: |
| | 404 | /* other types cannot be added to a string */ |
| | 405 | err_throw(VMERR_NO_STR_CONV); |
| | 406 | |
| | 407 | /* we never really get here, but the compiler doesn't know that */ |
| | 408 | AFTER_ERR_THROW(return 0;) |
| | 409 | } |
| | 410 | } |
| | 411 | |
| | 412 | /* ------------------------------------------------------------------------ */ |
| | 413 | /* |
| | 414 | * Convert an integer to a string, storing the result in the given |
| | 415 | * buffer in portable string format (with length prefix). The radix |
| | 416 | * must be 8, 10, or 16. |
| | 417 | * |
| | 418 | * Decimal numbers are treated as signed, and a leading dash is included |
| | 419 | * if the number is negative. Octal and hex numbers are treated as |
| | 420 | * unsigned. |
| | 421 | * |
| | 422 | * For efficiency, we store the number at the end of the buffer (this |
| | 423 | * makes it easy to generate the number, since we need to generate |
| | 424 | * numerals in reverse order). We return a pointer to the result, which |
| | 425 | * may not start at the beginning of the buffer. |
| | 426 | */ |
| | 427 | char *CVmObjString::cvt_int_to_str(char *buf, size_t buflen, |
| | 428 | int32 inval, int radix) |
| | 429 | { |
| | 430 | int neg; |
| | 431 | uint32 val; |
| | 432 | char *p; |
| | 433 | size_t len; |
| | 434 | |
| | 435 | /* start at the end of the buffer */ |
| | 436 | p = buf + buflen; |
| | 437 | |
| | 438 | /* |
| | 439 | * if it's negative, and we're converting to decimal representation, |
| | 440 | * treat the value as signed and use a leading minus sign; |
| | 441 | * otherwise, treat the value as unsigned |
| | 442 | */ |
| | 443 | if (radix == 10 && inval < 0) |
| | 444 | { |
| | 445 | /* note that we need a minus sign */ |
| | 446 | neg = TRUE; |
| | 447 | |
| | 448 | /* use the positive value for the conversion */ |
| | 449 | val = (uint32)(-inval); |
| | 450 | } |
| | 451 | else |
| | 452 | { |
| | 453 | /* the value is positive (or at least unsigned) */ |
| | 454 | neg = FALSE; |
| | 455 | |
| | 456 | /* use the value as-is */ |
| | 457 | val = (uint32)inval; |
| | 458 | } |
| | 459 | |
| | 460 | /* store numerals in reverse order */ |
| | 461 | do |
| | 462 | { |
| | 463 | char c; |
| | 464 | |
| | 465 | /* if we have no more room, throw an error */ |
| | 466 | if (p == buf) |
| | 467 | err_throw(VMERR_CONV_BUF_OVF); |
| | 468 | |
| | 469 | /* move on to the next available character in the buffer */ |
| | 470 | --p; |
| | 471 | |
| | 472 | /* figure the character representation of this numeral */ |
| | 473 | c = (char)(val % radix); |
| | 474 | if (c < 10) |
| | 475 | c += '0'; |
| | 476 | else |
| | 477 | c += 'A' - 10; |
| | 478 | |
| | 479 | /* store the numeral at the current location */ |
| | 480 | *p = c; |
| | 481 | |
| | 482 | /* divide the remaining number by the radix */ |
| | 483 | val /= radix; |
| | 484 | } while (val != 0); |
| | 485 | |
| | 486 | /* store the leading minus sign if necessary */ |
| | 487 | if (neg) |
| | 488 | { |
| | 489 | /* if we don't have room, throw an error */ |
| | 490 | if (p == buf) |
| | 491 | err_throw(VMERR_CONV_BUF_OVF); |
| | 492 | |
| | 493 | /* move to the next byte */ |
| | 494 | --p; |
| | 495 | |
| | 496 | /* store the minus sign */ |
| | 497 | *p = '-'; |
| | 498 | } |
| | 499 | |
| | 500 | /* calculate the length */ |
| | 501 | len = buflen - (p - buf); |
| | 502 | |
| | 503 | /* make sure we have room for the length prefix */ |
| | 504 | if (p < buf + 2) |
| | 505 | err_throw(VMERR_CONV_BUF_OVF); |
| | 506 | |
| | 507 | /* store the length prefix */ |
| | 508 | p -= 2; |
| | 509 | vmb_put_len(p, len); |
| | 510 | |
| | 511 | /* return the pointer to the start of the number */ |
| | 512 | return p; |
| | 513 | } |
| | 514 | |
| | 515 | /* ------------------------------------------------------------------------ */ |
| | 516 | /* |
| | 517 | * Check a value for equality |
| | 518 | */ |
| | 519 | int CVmObjString::equals(VMG_ vm_obj_id_t self, |
| | 520 | const vm_val_t *val, int /*depth*/) const |
| | 521 | { |
| | 522 | /* if the other value is a reference to myself, we certainly match */ |
| | 523 | if (val->typ == VM_OBJ && val->val.obj == self) |
| | 524 | return TRUE; |
| | 525 | |
| | 526 | /* |
| | 527 | * use the constant string comparison routine, using our underlying |
| | 528 | * string as the constant string data |
| | 529 | */ |
| | 530 | return const_equals(vmg_ ext_, val); |
| | 531 | } |
| | 532 | |
| | 533 | /* |
| | 534 | * Constant string equality test |
| | 535 | */ |
| | 536 | int CVmObjString::const_equals(VMG_ const char *str, const vm_val_t *val) |
| | 537 | { |
| | 538 | const char *str2; |
| | 539 | size_t len; |
| | 540 | |
| | 541 | /* get the other value as a string */ |
| | 542 | str2 = val->get_as_string(vmg0_); |
| | 543 | |
| | 544 | /* if the object doesn't have an underlying string, we don't match */ |
| | 545 | if (str2 == 0) |
| | 546 | return FALSE; |
| | 547 | |
| | 548 | /* |
| | 549 | * if their lengths match, and the bytes match exactly, we have a |
| | 550 | * match; otherwise, they're not equal |
| | 551 | */ |
| | 552 | len = vmb_get_len(str); |
| | 553 | return (len == vmb_get_len(str2) |
| | 554 | && memcmp(str + VMB_LEN, str2 + VMB_LEN, len) == 0); |
| | 555 | } |
| | 556 | |
| | 557 | /* ------------------------------------------------------------------------ */ |
| | 558 | /* |
| | 559 | * Hash value |
| | 560 | */ |
| | 561 | uint CVmObjString::calc_hash(VMG_ vm_obj_id_t self, int /*depth*/) const |
| | 562 | { |
| | 563 | return const_calc_hash(ext_); |
| | 564 | } |
| | 565 | |
| | 566 | /* |
| | 567 | * Hash value calculation |
| | 568 | */ |
| | 569 | uint CVmObjString::const_calc_hash(const char *str) |
| | 570 | { |
| | 571 | size_t len; |
| | 572 | uint hash; |
| | 573 | utf8_ptr p; |
| | 574 | |
| | 575 | /* get and skip the length prefix */ |
| | 576 | len = vmb_get_len(str); |
| | 577 | str += VMB_LEN; |
| | 578 | |
| | 579 | /* scan the string and calculate the hash */ |
| | 580 | for (p.set((char *)str), hash = 0 ; len != 0 ; p.inc(&len)) |
| | 581 | hash += p.getch(); |
| | 582 | |
| | 583 | /* return the result */ |
| | 584 | return hash; |
| | 585 | } |
| | 586 | |
| | 587 | |
| | 588 | /* ------------------------------------------------------------------------ */ |
| | 589 | /* |
| | 590 | * Compare this string to another value |
| | 591 | */ |
| | 592 | int CVmObjString::compare_to(VMG_ vm_obj_id_t /*self*/, |
| | 593 | const vm_val_t *val) const |
| | 594 | { |
| | 595 | /* use the static string magnitude comparison routine */ |
| | 596 | return const_compare(vmg_ ext_, val); |
| | 597 | } |
| | 598 | |
| | 599 | /* |
| | 600 | * Compare a constant string value to another value. Returns a positive |
| | 601 | * number if the constant string is lexically greater than the other |
| | 602 | * value, a negative number if the constant string is lexically less |
| | 603 | * than the other value, or zero if the constant string is lexically |
| | 604 | * identical to the other value. |
| | 605 | * |
| | 606 | * The other value must be a string constant or an object with an |
| | 607 | * underlying string value. We'll throw an error for any other type of |
| | 608 | * value. |
| | 609 | */ |
| | 610 | int CVmObjString::const_compare(VMG_ const char *str1, const vm_val_t *val) |
| | 611 | { |
| | 612 | const char *str2; |
| | 613 | size_t len1, len2; |
| | 614 | |
| | 615 | /* get the other value as a string */ |
| | 616 | str2 = val->get_as_string(vmg0_); |
| | 617 | |
| | 618 | /* if it's not a string, we can't compare it */ |
| | 619 | if (str2 == 0) |
| | 620 | err_throw(VMERR_INVALID_COMPARISON); |
| | 621 | |
| | 622 | /* get the lengths of the two strings */ |
| | 623 | len1 = vmb_get_len(str1); |
| | 624 | len2 = vmb_get_len(str2); |
| | 625 | |
| | 626 | /* perform a lexical comparison and return the result */ |
| | 627 | return utf8_ptr::s_compare_to(str1 + VMB_LEN, len1, str2 + VMB_LEN, len2); |
| | 628 | } |
| | 629 | |
| | 630 | /* ------------------------------------------------------------------------ */ |
| | 631 | /* |
| | 632 | * Find a substring within a string |
| | 633 | */ |
| | 634 | const char *CVmObjString::find_substr(VMG_ const char *str, int start_idx, |
| | 635 | const char *substr, size_t *idxp) |
| | 636 | { |
| | 637 | utf8_ptr p; |
| | 638 | size_t rem; |
| | 639 | size_t sublen; |
| | 640 | size_t char_ofs; |
| | 641 | int i; |
| | 642 | |
| | 643 | /* get the lengths */ |
| | 644 | rem = vmb_get_len(str); |
| | 645 | sublen = vmb_get_len(substr); |
| | 646 | |
| | 647 | /* set up utf8 pointer into the string */ |
| | 648 | p.set((char *)str + 2); |
| | 649 | |
| | 650 | /* skip to the starting index */ |
| | 651 | for (i = start_idx ; i > 0 && rem >= sublen ; --i, p.inc(&rem)) ; |
| | 652 | |
| | 653 | /* scan for the substring */ |
| | 654 | for (char_ofs = 0 ; rem != 0 && rem >= sublen ; ++char_ofs, p.inc(&rem)) |
| | 655 | { |
| | 656 | /* check for a match */ |
| | 657 | if (memcmp(p.getptr(), substr + VMB_LEN, sublen) == 0) |
| | 658 | { |
| | 659 | /* it's a match - set the return index if they are interested */ |
| | 660 | if (idxp != 0) |
| | 661 | *idxp = char_ofs + start_idx; |
| | 662 | |
| | 663 | /* return the current pointer */ |
| | 664 | return p.getptr(); |
| | 665 | } |
| | 666 | } |
| | 667 | |
| | 668 | /* we didn't find it - so indicate by returning null */ |
| | 669 | return 0; |
| | 670 | } |
| | 671 | |
| | 672 | /* ------------------------------------------------------------------------ */ |
| | 673 | /* |
| | 674 | * Evaluate a property |
| | 675 | */ |
| | 676 | int CVmObjString::get_prop(VMG_ vm_prop_id_t prop, vm_val_t *retval, |
| | 677 | vm_obj_id_t self, vm_obj_id_t *source_obj, |
| | 678 | uint *argc) |
| | 679 | { |
| | 680 | vm_val_t self_val; |
| | 681 | |
| | 682 | /* use the constant evaluator */ |
| | 683 | self_val.set_obj(self); |
| | 684 | if (const_get_prop(vmg_ retval, &self_val, ext_, prop, source_obj, argc)) |
| | 685 | { |
| | 686 | *source_obj = metaclass_reg_->get_class_obj(vmg0_); |
| | 687 | return TRUE; |
| | 688 | } |
| | 689 | |
| | 690 | /* inherit default handling from the base object class */ |
| | 691 | return CVmObject::get_prop(vmg_ prop, retval, self, source_obj, argc); |
| | 692 | } |
| | 693 | |
| | 694 | /* ------------------------------------------------------------------------ */ |
| | 695 | /* |
| | 696 | * Evaluate a property of a constant string value |
| | 697 | */ |
| | 698 | int CVmObjString::const_get_prop(VMG_ vm_val_t *retval, |
| | 699 | const vm_val_t *self_val, const char *str, |
| | 700 | vm_prop_id_t prop, vm_obj_id_t *src_obj, |
| | 701 | uint *argc) |
| | 702 | { |
| | 703 | uint func_idx; |
| | 704 | |
| | 705 | /* presume no source object */ |
| | 706 | *src_obj = VM_INVALID_OBJ; |
| | 707 | |
| | 708 | /* translate the property index to an index into our function table */ |
| | 709 | func_idx = G_meta_table |
| | 710 | ->prop_to_vector_idx(metaclass_reg_->get_reg_idx(), prop); |
| | 711 | |
| | 712 | /* call the appropriate function */ |
| | 713 | if ((*func_table_[func_idx])(vmg_ retval, self_val, str, argc)) |
| | 714 | return TRUE; |
| | 715 | |
| | 716 | /* |
| | 717 | * If this is a constant string (which is indicated by an invalid |
| | 718 | * 'self' object ID), try inheriting the default object |
| | 719 | * interpretation, passing the constant string placeholder object |
| | 720 | * for its type information. |
| | 721 | */ |
| | 722 | if (self_val->typ != VM_OBJ) |
| | 723 | { |
| | 724 | /* try going to CVmObject directly */ |
| | 725 | if (vm_objp(vmg_ G_predef->const_str_obj) |
| | 726 | ->CVmObject::get_prop(vmg_ prop, retval, G_predef->const_str_obj, |
| | 727 | src_obj, argc)) |
| | 728 | return TRUE; |
| | 729 | } |
| | 730 | |
| | 731 | /* not handled */ |
| | 732 | return FALSE; |
| | 733 | } |
| | 734 | |
| | 735 | /* ------------------------------------------------------------------------ */ |
| | 736 | /* |
| | 737 | * property evaluator - get the length |
| | 738 | */ |
| | 739 | int CVmObjString::getp_len(VMG_ vm_val_t *retval, const vm_val_t *self_val, |
| | 740 | const char *str, uint *argc) |
| | 741 | { |
| | 742 | utf8_ptr p; |
| | 743 | static CVmNativeCodeDesc desc(0); |
| | 744 | |
| | 745 | /* check arguments */ |
| | 746 | if (get_prop_check_argc(retval, argc, &desc)) |
| | 747 | return TRUE; |
| | 748 | |
| | 749 | /* set up a utf-8 pointer to the string's contents */ |
| | 750 | p.set((char *)str + VMB_LEN); |
| | 751 | |
| | 752 | /* return the character length of the string */ |
| | 753 | retval->set_int(p.len(vmb_get_len(str))); |
| | 754 | |
| | 755 | /* handled */ |
| | 756 | return TRUE; |
| | 757 | } |
| | 758 | |
| | 759 | /* ------------------------------------------------------------------------ */ |
| | 760 | /* |
| | 761 | * property evaluator - extract a substring |
| | 762 | */ |
| | 763 | int CVmObjString::getp_substr(VMG_ vm_val_t *retval, const vm_val_t *self_val, |
| | 764 | const char *str, uint *in_argc) |
| | 765 | { |
| | 766 | long start; |
| | 767 | ulong len = 0; |
| | 768 | size_t rem; |
| | 769 | utf8_ptr p; |
| | 770 | utf8_ptr start_p; |
| | 771 | size_t start_rem; |
| | 772 | size_t new_len; |
| | 773 | vm_obj_id_t obj; |
| | 774 | uint argc = (in_argc == 0 ? 0 : *in_argc); |
| | 775 | static CVmNativeCodeDesc desc(1, 1); |
| | 776 | |
| | 777 | /* check arguments */ |
| | 778 | if (get_prop_check_argc(retval, in_argc, &desc)) |
| | 779 | return TRUE; |
| | 780 | |
| | 781 | /* pop the starting index */ |
| | 782 | start = CVmBif::pop_long_val(vmg0_); |
| | 783 | |
| | 784 | /* pop the length, if present */ |
| | 785 | if (argc >= 2) |
| | 786 | len = CVmBif::pop_long_val(vmg0_); |
| | 787 | |
| | 788 | /* push a self-reference to protect against GC */ |
| | 789 | G_stk->push(self_val); |
| | 790 | |
| | 791 | /* set up a utf8 pointer to traverse the string */ |
| | 792 | p.set((char *)str + VMB_LEN); |
| | 793 | |
| | 794 | /* get the byte length of the string */ |
| | 795 | rem = vmb_get_len(str); |
| | 796 | |
| | 797 | /* |
| | 798 | * Skip ahead to the starting index. If the index is positive, it's |
| | 799 | * an index from the start of the string; if it's negative, it's an |
| | 800 | * offset from the end of the string. |
| | 801 | */ |
| | 802 | if (start > 0) |
| | 803 | { |
| | 804 | /* |
| | 805 | * it's an index from the start - skip ahead by start-1 characters |
| | 806 | * (since a start value of 1 tells us to start at the first |
| | 807 | * character) |
| | 808 | */ |
| | 809 | for ( ; start > 1 && rem != 0 ; --start) |
| | 810 | p.inc(&rem); |
| | 811 | } |
| | 812 | else if (start < 0) |
| | 813 | { |
| | 814 | /* |
| | 815 | * It's an index from the end of the string: -1 tells us to start |
| | 816 | * at the last character, -2 at the second to last, and so on. |
| | 817 | * Move to the first byte past the end of the string, and work |
| | 818 | * backwards by the given number of characters. |
| | 819 | */ |
| | 820 | for (p.set((char *)str + VMB_LEN + rem), rem = 0 ; |
| | 821 | start < 0 && p.getptr() != (char *)str + VMB_LEN ; ++start) |
| | 822 | { |
| | 823 | /* move back one character */ |
| | 824 | p.dec(&rem); |
| | 825 | } |
| | 826 | } |
| | 827 | |
| | 828 | /* this is the starting position */ |
| | 829 | start_p = p; |
| | 830 | start_rem = rem; |
| | 831 | |
| | 832 | /* |
| | 833 | * if a length was specified, calculate the number of bytes in the |
| | 834 | * given length; otherwise, use the entire remainder of the string |
| | 835 | */ |
| | 836 | if (argc >= 2) |
| | 837 | { |
| | 838 | /* keep skipping ahead by the desired length */ |
| | 839 | for ( ; len > 0 && rem != 0 ; --len) |
| | 840 | p.inc(&rem); |
| | 841 | |
| | 842 | /* use the difference in lengths from the starting point to here */ |
| | 843 | new_len = start_rem - rem; |
| | 844 | } |
| | 845 | else |
| | 846 | { |
| | 847 | /* use the entire remainder of the string */ |
| | 848 | new_len = start_rem; |
| | 849 | } |
| | 850 | |
| | 851 | /* create the new string */ |
| | 852 | obj = CVmObjString::create(vmg_ FALSE, start_p.getptr(), new_len); |
| | 853 | |
| | 854 | /* return the new object */ |
| | 855 | retval->set_obj(obj); |
| | 856 | |
| | 857 | /* discard the GC protection references */ |
| | 858 | G_stk->discard(); |
| | 859 | |
| | 860 | /* handled */ |
| | 861 | return TRUE; |
| | 862 | } |
| | 863 | |
| | 864 | /* ------------------------------------------------------------------------ */ |
| | 865 | /* |
| | 866 | * property evaluator - toUpper |
| | 867 | */ |
| | 868 | int CVmObjString::getp_upper(VMG_ vm_val_t *retval, const vm_val_t *self_val, |
| | 869 | const char *str, uint *argc) |
| | 870 | { |
| | 871 | size_t srclen; |
| | 872 | size_t dstlen; |
| | 873 | size_t rem; |
| | 874 | utf8_ptr srcp; |
| | 875 | utf8_ptr dstp; |
| | 876 | vm_obj_id_t result_obj; |
| | 877 | static CVmNativeCodeDesc desc(0); |
| | 878 | |
| | 879 | /* check arguments */ |
| | 880 | if (get_prop_check_argc(retval, argc, &desc)) |
| | 881 | return TRUE; |
| | 882 | |
| | 883 | /* get my length */ |
| | 884 | srclen = vmb_get_len(str); |
| | 885 | |
| | 886 | /* leave the string on the stack as GC protection */ |
| | 887 | G_stk->push(self_val); |
| | 888 | |
| | 889 | /* |
| | 890 | * Scan the string to determine how long the result will be. The |
| | 891 | * result won't necessarily be the same length as the original, |
| | 892 | * because a two-byte character in the original could turn into a |
| | 893 | * three-byte character in the result, and vice versa. (We could |
| | 894 | * allocate a result buffer three times the length of the original, |
| | 895 | * but this seems more wasteful of space than scanning the string |
| | 896 | * twice is wasteful of time. It's a trade-off, though.) |
| | 897 | */ |
| | 898 | for (dstlen = 0, srcp.set((char *)str + VMB_LEN), rem = srclen ; |
| | 899 | rem != 0 ; srcp.inc(&rem)) |
| | 900 | { |
| | 901 | /* get the size of the mapping for this character */ |
| | 902 | dstlen += utf8_ptr::s_wchar_size(t3_to_upper(srcp.getch())); |
| | 903 | } |
| | 904 | |
| | 905 | /* allocate the result string */ |
| | 906 | result_obj = CVmObjString::create(vmg_ FALSE, dstlen); |
| | 907 | |
| | 908 | /* get a pointer to the result buffer */ |
| | 909 | dstp.set(((CVmObjString *)vm_objp(vmg_ result_obj))->cons_get_buf()); |
| | 910 | |
| | 911 | /* write the string */ |
| | 912 | for (srcp.set((char *)str + VMB_LEN), rem = srclen ; |
| | 913 | rem != 0 ; srcp.inc(&rem)) |
| | 914 | { |
| | 915 | /* write the next character */ |
| | 916 | dstp.setch(t3_to_upper(srcp.getch())); |
| | 917 | } |
| | 918 | |
| | 919 | /* return the value */ |
| | 920 | retval->set_obj(result_obj); |
| | 921 | |
| | 922 | /* discard GC protection */ |
| | 923 | G_stk->discard(); |
| | 924 | |
| | 925 | /* handled */ |
| | 926 | return TRUE; |
| | 927 | } |
| | 928 | |
| | 929 | /* ------------------------------------------------------------------------ */ |
| | 930 | /* |
| | 931 | * property evaluator - toLower |
| | 932 | */ |
| | 933 | int CVmObjString::getp_lower(VMG_ vm_val_t *retval, const vm_val_t *self_val, |
| | 934 | const char *str, uint *argc) |
| | 935 | { |
| | 936 | size_t srclen; |
| | 937 | size_t dstlen; |
| | 938 | size_t rem; |
| | 939 | utf8_ptr srcp; |
| | 940 | utf8_ptr dstp; |
| | 941 | vm_obj_id_t result_obj; |
| | 942 | static CVmNativeCodeDesc desc(0); |
| | 943 | |
| | 944 | /* check arguments */ |
| | 945 | if (get_prop_check_argc(retval, argc, &desc)) |
| | 946 | return TRUE; |
| | 947 | |
| | 948 | /* get my length */ |
| | 949 | srclen = vmb_get_len(str); |
| | 950 | |
| | 951 | /* leave the string on the stack as GC protection */ |
| | 952 | G_stk->push(self_val); |
| | 953 | |
| | 954 | /* |
| | 955 | * Scan the string to determine how long the result will be. The |
| | 956 | * result won't necessarily be the same length as the original, |
| | 957 | * because a two-byte character in the original could turn into a |
| | 958 | * three-byte character in the result, and vice versa. (We could |
| | 959 | * allocate a result buffer three times the length of the original, |
| | 960 | * but this seems more wasteful of space than scanning the string |
| | 961 | * twice is wasteful of time. It's a trade-off, though.) |
| | 962 | */ |
| | 963 | for (dstlen = 0, srcp.set((char *)str + VMB_LEN), rem = srclen ; |
| | 964 | rem != 0 ; srcp.inc(&rem)) |
| | 965 | { |
| | 966 | /* get the size of the mapping for this character */ |
| | 967 | dstlen += utf8_ptr::s_wchar_size(t3_to_lower(srcp.getch())); |
| | 968 | } |
| | 969 | |
| | 970 | /* allocate the result string */ |
| | 971 | result_obj = CVmObjString::create(vmg_ FALSE, dstlen); |
| | 972 | |
| | 973 | /* get a pointer to the result buffer */ |
| | 974 | dstp.set(((CVmObjString *)vm_objp(vmg_ result_obj))->cons_get_buf()); |
| | 975 | |
| | 976 | /* write the string */ |
| | 977 | for (srcp.set((char *)str + VMB_LEN), rem = srclen ; |
| | 978 | rem != 0 ; srcp.inc(&rem)) |
| | 979 | { |
| | 980 | /* write the next character */ |
| | 981 | dstp.setch(t3_to_lower(srcp.getch())); |
| | 982 | } |
| | 983 | |
| | 984 | /* return the value */ |
| | 985 | retval->set_obj(result_obj); |
| | 986 | |
| | 987 | /* discard GC protection */ |
| | 988 | G_stk->discard(); |
| | 989 | |
| | 990 | /* handled */ |
| | 991 | return TRUE; |
| | 992 | } |
| | 993 | |
| | 994 | /* ------------------------------------------------------------------------ */ |
| | 995 | /* |
| | 996 | * property evaluator - find |
| | 997 | */ |
| | 998 | int CVmObjString::getp_find(VMG_ vm_val_t *retval, const vm_val_t *self_val, |
| | 999 | const char *str, uint *argc) |
| | 1000 | { |
| | 1001 | const char *str2; |
| | 1002 | size_t idx; |
| | 1003 | uint orig_argc = (argc != 0 ? *argc : 0); |
| | 1004 | static CVmNativeCodeDesc desc(1, 1); |
| | 1005 | int start_idx; |
| | 1006 | |
| | 1007 | /* check arguments */ |
| | 1008 | if (get_prop_check_argc(retval, argc, &desc)) |
| | 1009 | return TRUE; |
| | 1010 | |
| | 1011 | /* retrieve the string to find */ |
| | 1012 | str2 = CVmBif::pop_str_val(vmg0_); |
| | 1013 | |
| | 1014 | /* if there's a starting index, retrieve it */ |
| | 1015 | start_idx = (orig_argc >= 2 ? CVmBif::pop_int_val(vmg0_) - 1 : 0); |
| | 1016 | |
| | 1017 | /* find the substring */ |
| | 1018 | if (find_substr(vmg_ str, start_idx, str2, &idx) != 0) |
| | 1019 | { |
| | 1020 | /* we found it - adjust to a 1-based value for return */ |
| | 1021 | retval->set_int(idx + 1); |
| | 1022 | } |
| | 1023 | else |
| | 1024 | { |
| | 1025 | /* didn't find it - return nil */ |
| | 1026 | retval->set_nil(); |
| | 1027 | } |
| | 1028 | |
| | 1029 | /* handled */ |
| | 1030 | return TRUE; |
| | 1031 | } |
| | 1032 | |
| | 1033 | /* ------------------------------------------------------------------------ */ |
| | 1034 | /* |
| | 1035 | * replace flags |
| | 1036 | */ |
| | 1037 | #define GETP_RPL_ALL 0x0001 |
| | 1038 | |
| | 1039 | /* |
| | 1040 | * property evaluator - replace |
| | 1041 | */ |
| | 1042 | int CVmObjString::getp_replace(VMG_ vm_val_t *retval, |
| | 1043 | const vm_val_t *self_val, |
| | 1044 | const char *str, uint *argc) |
| | 1045 | { |
| | 1046 | vm_val_t arg1; |
| | 1047 | vm_val_t arg2; |
| | 1048 | const char *substr; |
| | 1049 | const char *rplstr; |
| | 1050 | size_t sublen; |
| | 1051 | size_t rpllen; |
| | 1052 | uint orig_argc = (argc != 0 ? *argc : 0); |
| | 1053 | static CVmNativeCodeDesc desc(3, 1); |
| | 1054 | int flags; |
| | 1055 | utf8_ptr p; |
| | 1056 | size_t rem; |
| | 1057 | size_t new_len; |
| | 1058 | int found; |
| | 1059 | int start_idx; |
| | 1060 | const char *rpl_start; |
| | 1061 | |
| | 1062 | /* check arguments */ |
| | 1063 | if (get_prop_check_argc(retval, argc, &desc)) |
| | 1064 | return TRUE; |
| | 1065 | |
| | 1066 | /* |
| | 1067 | * make copies of the string references, so we can put them back on the |
| | 1068 | * stack as gc protection while we're working |
| | 1069 | */ |
| | 1070 | arg1 = *G_stk->get(0); |
| | 1071 | arg2 = *G_stk->get(1); |
| | 1072 | |
| | 1073 | /* retrieve the search and replacement substrings */ |
| | 1074 | substr = CVmBif::pop_str_val(vmg0_); |
| | 1075 | rplstr = CVmBif::pop_str_val(vmg0_); |
| | 1076 | |
| | 1077 | /* note the string lengths */ |
| | 1078 | sublen = vmb_get_len(substr); |
| | 1079 | rpllen = vmb_get_len(rplstr); |
| | 1080 | |
| | 1081 | /* get the flags */ |
| | 1082 | flags = CVmBif::pop_int_val(vmg0_); |
| | 1083 | |
| | 1084 | /* if there's a starting index, retrieve it */ |
| | 1085 | start_idx = (orig_argc >= 4 ? CVmBif::pop_int_val(vmg0_) - 1 : 0); |
| | 1086 | |
| | 1087 | /* put the string references back on the stack for gc protection */ |
| | 1088 | G_stk->push(&arg1); |
| | 1089 | G_stk->push(&arg2); |
| | 1090 | |
| | 1091 | /* start at the beginning of the string to search */ |
| | 1092 | rem = new_len = vmb_get_len(str); |
| | 1093 | p.set((char *)str + 2); |
| | 1094 | |
| | 1095 | /* skip ahead to the starting index */ |
| | 1096 | for ( ; start_idx > 0 && rem >= sublen ; --start_idx, p.inc(&rem)) ; |
| | 1097 | |
| | 1098 | /* |
| | 1099 | * note the starting index for replacements - we don't want to replace |
| | 1100 | * anything before this point |
| | 1101 | */ |
| | 1102 | rpl_start = p.getptr(); |
| | 1103 | |
| | 1104 | /* |
| | 1105 | * Scan for instances of the substring, so we can figure out how big |
| | 1106 | * the result string will be. Don't actually do any replacements yet; |
| | 1107 | * we'll scan again once we know how the result size. |
| | 1108 | */ |
| | 1109 | for (found = FALSE ; rem >= sublen ; ) |
| | 1110 | { |
| | 1111 | /* if this is a match for the substring, note it */ |
| | 1112 | if (memcmp(p.getptr(), substr + VMB_LEN, vmb_get_len(substr)) == 0) |
| | 1113 | { |
| | 1114 | /* note the find */ |
| | 1115 | found = TRUE; |
| | 1116 | |
| | 1117 | /* it's a match - adjust the result length for the replacement */ |
| | 1118 | new_len += rpllen - sublen; |
| | 1119 | |
| | 1120 | /* if we're replacing one instance only, look no further */ |
| | 1121 | if ((flags & GETP_RPL_ALL) == 0) |
| | 1122 | break; |
| | 1123 | |
| | 1124 | /* skip the entire substring in the source */ |
| | 1125 | p.set(p.getptr() + sublen); |
| | 1126 | rem -= sublen; |
| | 1127 | } |
| | 1128 | else |
| | 1129 | { |
| | 1130 | /* skip one character */ |
| | 1131 | p.inc(&rem); |
| | 1132 | } |
| | 1133 | } |
| | 1134 | |
| | 1135 | /* |
| | 1136 | * if we found no instances of the search substring, the result is |
| | 1137 | * simply the source string; otherwise, we must create a new string |
| | 1138 | * with the substitution(s) |
| | 1139 | */ |
| | 1140 | if (found) |
| | 1141 | { |
| | 1142 | utf8_ptr dst; |
| | 1143 | |
| | 1144 | /* allocate the new string */ |
| | 1145 | retval->set_obj(create(vmg_ FALSE, new_len)); |
| | 1146 | |
| | 1147 | /* get a pointer to the buffer */ |
| | 1148 | dst.set(((CVmObjString *)vm_objp(vmg_ retval->val.obj)) |
| | 1149 | ->cons_get_buf()); |
| | 1150 | |
| | 1151 | /* scan the string for replacements */ |
| | 1152 | for (p.set((char *)str + 2), rem = vmb_get_len(str) ; |
| | 1153 | rem >= sublen ; ) |
| | 1154 | { |
| | 1155 | /* |
| | 1156 | * If this is a match for the substring, and we've reached the |
| | 1157 | * starting point for replacements, replace the substring. |
| | 1158 | */ |
| | 1159 | if (p.getptr() >= rpl_start |
| | 1160 | && memcmp(p.getptr(), substr + VMB_LEN, sublen) == 0) |
| | 1161 | { |
| | 1162 | /* it's a match - copy the replacement into the result */ |
| | 1163 | memcpy(dst.getptr(), rplstr + VMB_LEN, rpllen); |
| | 1164 | |
| | 1165 | /* move past the replacement in the result */ |
| | 1166 | dst.set(dst.getptr() + rpllen); |
| | 1167 | |
| | 1168 | /* move past the search substring in the source */ |
| | 1169 | p.set(p.getptr() + sublen); |
| | 1170 | rem -= sublen; |
| | 1171 | |
| | 1172 | /* if we're replacing one instance only, look no further */ |
| | 1173 | if ((flags & GETP_RPL_ALL) == 0) |
| | 1174 | break; |
| | 1175 | } |
| | 1176 | else |
| | 1177 | { |
| | 1178 | /* copy the current character to the result */ |
| | 1179 | dst.setch(p.getch()); |
| | 1180 | |
| | 1181 | /* skip the current character of input */ |
| | 1182 | p.inc(&rem); |
| | 1183 | } |
| | 1184 | } |
| | 1185 | |
| | 1186 | /* copy the remaining source into the result */ |
| | 1187 | if (rem != 0) |
| | 1188 | memcpy(dst.getptr(), p.getptr(), rem); |
| | 1189 | } |
| | 1190 | else |
| | 1191 | { |
| | 1192 | /* we didn't find it - the result is simply the original string */ |
| | 1193 | *retval = *self_val; |
| | 1194 | } |
| | 1195 | |
| | 1196 | /* discard the gc protection */ |
| | 1197 | G_stk->discard(2); |
| | 1198 | |
| | 1199 | /* handled */ |
| | 1200 | return TRUE; |
| | 1201 | } |
| | 1202 | |
| | 1203 | /* ------------------------------------------------------------------------ */ |
| | 1204 | /* |
| | 1205 | * property evaluator - convert to unicode |
| | 1206 | */ |
| | 1207 | int CVmObjString::getp_to_uni(VMG_ vm_val_t *retval, |
| | 1208 | const vm_val_t *self_val, |
| | 1209 | const char *str, uint *in_argc) |
| | 1210 | { |
| | 1211 | uint argc = (in_argc != 0 ? *in_argc : 0); |
| | 1212 | size_t bytelen; |
| | 1213 | ulong idx = 0; |
| | 1214 | utf8_ptr p; |
| | 1215 | static CVmNativeCodeDesc desc(0, 1); |
| | 1216 | |
| | 1217 | /* check arguments */ |
| | 1218 | if (get_prop_check_argc(retval, in_argc, &desc)) |
| | 1219 | return TRUE; |
| | 1220 | |
| | 1221 | /* retrieve the index argument if present */ |
| | 1222 | if (argc >= 1) |
| | 1223 | idx = CVmBif::pop_long_val(vmg0_); |
| | 1224 | |
| | 1225 | /* push a self-reference as GC protection */ |
| | 1226 | G_stk->push(self_val); |
| | 1227 | |
| | 1228 | /* get and skip the string's length prefix */ |
| | 1229 | bytelen = vmb_get_len(str); |
| | 1230 | str += VMB_LEN; |
| | 1231 | |
| | 1232 | /* set up a utf8 pointer to the string */ |
| | 1233 | p.set((char *)str); |
| | 1234 | |
| | 1235 | /* check for an index argument */ |
| | 1236 | if (argc >= 1) |
| | 1237 | { |
| | 1238 | /* skip through the string until we get to the desired index */ |
| | 1239 | for ( ; idx > 1 && bytelen != 0 ; --idx, p.inc(&bytelen)) ; |
| | 1240 | |
| | 1241 | /* check to see if we have a character available */ |
| | 1242 | if (idx == 1 && bytelen != 0) |
| | 1243 | { |
| | 1244 | /* the index is valid - return the character here */ |
| | 1245 | retval->set_int((long)p.getch()); |
| | 1246 | } |
| | 1247 | else |
| | 1248 | { |
| | 1249 | /* |
| | 1250 | * the index is past the end of the string or is less than 1 |
| | 1251 | * - return nil to indicate that there's no character here |
| | 1252 | */ |
| | 1253 | retval->set_nil(); |
| | 1254 | } |
| | 1255 | } |
| | 1256 | else |
| | 1257 | { |
| | 1258 | size_t charlen; |
| | 1259 | vm_obj_id_t lst_obj; |
| | 1260 | CVmObjList *lst; |
| | 1261 | size_t i; |
| | 1262 | |
| | 1263 | /* |
| | 1264 | * There's no index argument - they want a list of all of the |
| | 1265 | * code points in the string. First, get the number of |
| | 1266 | * characters in the string. |
| | 1267 | */ |
| | 1268 | charlen = p.len(bytelen); |
| | 1269 | |
| | 1270 | /* create a list to hold the results */ |
| | 1271 | lst_obj = CVmObjList::create(vmg_ FALSE, charlen); |
| | 1272 | lst = (CVmObjList *)vm_objp(vmg_ lst_obj); |
| | 1273 | |
| | 1274 | /* set the list's elements to the unicode characters values */ |
| | 1275 | for (i = 0 ; i < charlen ; ++i, p.inc()) |
| | 1276 | { |
| | 1277 | wchar_t ch; |
| | 1278 | vm_val_t ele_val; |
| | 1279 | |
| | 1280 | /* get this character */ |
| | 1281 | ch = p.getch(); |
| | 1282 | |
| | 1283 | /* set this list element */ |
| | 1284 | ele_val.set_int((long)ch); |
| | 1285 | lst->cons_set_element(i, &ele_val); |
| | 1286 | } |
| | 1287 | |
| | 1288 | /* return the list object */ |
| | 1289 | retval->set_obj(lst_obj); |
| | 1290 | } |
| | 1291 | |
| | 1292 | /* discard the GC protection */ |
| | 1293 | G_stk->discard(); |
| | 1294 | |
| | 1295 | /* handled */ |
| | 1296 | return TRUE; |
| | 1297 | } |
| | 1298 | |
| | 1299 | /* ------------------------------------------------------------------------ */ |
| | 1300 | /* |
| | 1301 | * property evaluator - htmlify |
| | 1302 | */ |
| | 1303 | |
| | 1304 | /* |
| | 1305 | * htmlify flags |
| | 1306 | */ |
| | 1307 | |
| | 1308 | /* preserve spaces */ |
| | 1309 | #define VMSTR_HTMLIFY_KEEP_SPACES 0x0001 |
| | 1310 | |
| | 1311 | /* preserve newlines */ |
| | 1312 | #define VMSTR_HTMLIFY_KEEP_NEWLINES 0x0002 |
| | 1313 | |
| | 1314 | /* preserve tabs */ |
| | 1315 | #define VMSTR_HTMLIFY_KEEP_TABS 0x0004 |
| | 1316 | |
| | 1317 | /* |
| | 1318 | * htmlify implementation |
| | 1319 | */ |
| | 1320 | int CVmObjString::getp_htmlify(VMG_ vm_val_t *retval, |
| | 1321 | const vm_val_t *self_val, |
| | 1322 | const char *str, uint *in_argc) |
| | 1323 | { |
| | 1324 | uint argc = (in_argc != 0 ? *in_argc : 0); |
| | 1325 | size_t bytelen; |
| | 1326 | utf8_ptr p; |
| | 1327 | utf8_ptr dstp; |
| | 1328 | size_t rem; |
| | 1329 | size_t extra; |
| | 1330 | long flags; |
| | 1331 | vm_obj_id_t result_obj; |
| | 1332 | int prv_was_sp; |
| | 1333 | static CVmNativeCodeDesc desc(0, 1); |
| | 1334 | |
| | 1335 | /* check arguments */ |
| | 1336 | if (get_prop_check_argc(retval, in_argc, &desc)) |
| | 1337 | return TRUE; |
| | 1338 | |
| | 1339 | /* if they specified flags, pop them */ |
| | 1340 | if (argc >= 1) |
| | 1341 | { |
| | 1342 | /* retrieve the flags */ |
| | 1343 | flags = CVmBif::pop_long_val(vmg0_); |
| | 1344 | } |
| | 1345 | else |
| | 1346 | { |
| | 1347 | /* no flags */ |
| | 1348 | flags = 0; |
| | 1349 | } |
| | 1350 | |
| | 1351 | /* push a self-reference as GC protection */ |
| | 1352 | G_stk->push(self_val); |
| | 1353 | |
| | 1354 | /* get and skip the string's length prefix */ |
| | 1355 | bytelen = vmb_get_len(str); |
| | 1356 | str += VMB_LEN; |
| | 1357 | |
| | 1358 | /* |
| | 1359 | * scan the string to determine how much space we'll have to add to |
| | 1360 | * generate the htmlified version |
| | 1361 | */ |
| | 1362 | for (prv_was_sp = FALSE, extra = 0, p.set((char *)str), rem = bytelen ; |
| | 1363 | rem != 0 ; p.inc(&rem)) |
| | 1364 | { |
| | 1365 | int this_is_sp; |
| | 1366 | |
| | 1367 | /* presume it's not a space */ |
| | 1368 | this_is_sp = FALSE; |
| | 1369 | |
| | 1370 | /* check what we have */ |
| | 1371 | switch(p.getch()) |
| | 1372 | { |
| | 1373 | case '&': |
| | 1374 | /* we must replace '&' with '&' - this adds four bytes */ |
| | 1375 | extra += 4; |
| | 1376 | break; |
| | 1377 | |
| | 1378 | case '<': |
| | 1379 | /* we must replace '<' with '<' - this adds three bytes */ |
| | 1380 | extra += 3; |
| | 1381 | break; |
| | 1382 | |
| | 1383 | case ' ': |
| | 1384 | /* |
| | 1385 | * If we're in preserve-spaces mode, and the previous space |
| | 1386 | * was some kind of whitespace character, change this to |
| | 1387 | * ' ' - this adds five bytes |
| | 1388 | */ |
| | 1389 | if (prv_was_sp && (flags & VMSTR_HTMLIFY_KEEP_SPACES) != 0) |
| | 1390 | extra += 5; |
| | 1391 | |
| | 1392 | /* note that this was a whitespace character */ |
| | 1393 | this_is_sp = TRUE; |
| | 1394 | break; |
| | 1395 | |
| | 1396 | case '\t': |
| | 1397 | /* if we're in preserve-tabs mode, change this to '<tab>' */ |
| | 1398 | if ((flags & VMSTR_HTMLIFY_KEEP_TABS) != 0) |
| | 1399 | extra += 4; |
| | 1400 | |
| | 1401 | /* note that this was a whitespace character */ |
| | 1402 | this_is_sp = TRUE; |
| | 1403 | break; |
| | 1404 | |
| | 1405 | case '\n': |
| | 1406 | case 0x2028: |
| | 1407 | /* if we're in preserve-newlines mode, change this to '<br>' */ |
| | 1408 | if ((flags & VMSTR_HTMLIFY_KEEP_NEWLINES) != 0) |
| | 1409 | extra += 3; |
| | 1410 | |
| | 1411 | /* note that this was a whitespace character */ |
| | 1412 | this_is_sp = TRUE; |
| | 1413 | break; |
| | 1414 | } |
| | 1415 | |
| | 1416 | /* for next time, remember whether this is a space */ |
| | 1417 | prv_was_sp = this_is_sp; |
| | 1418 | } |
| | 1419 | |
| | 1420 | /* allocate space for the new string */ |
| | 1421 | result_obj = create(vmg_ FALSE, bytelen + extra); |
| | 1422 | |
| | 1423 | /* get a pointer to the result buffer */ |
| | 1424 | dstp.set(((CVmObjString *)vm_objp(vmg_ result_obj))->cons_get_buf()); |
| | 1425 | |
| | 1426 | /* translate the string and write the result */ |
| | 1427 | for (prv_was_sp = FALSE, p.set((char *)str), rem = bytelen ; |
| | 1428 | rem != 0 ; p.inc(&rem)) |
| | 1429 | { |
| | 1430 | wchar_t ch; |
| | 1431 | int this_is_sp; |
| | 1432 | |
| | 1433 | /* get this character */ |
| | 1434 | ch = p.getch(); |
| | 1435 | |
| | 1436 | /* presume it's not a space */ |
| | 1437 | this_is_sp = FALSE; |
| | 1438 | |
| | 1439 | /* check what we have */ |
| | 1440 | switch(ch) |
| | 1441 | { |
| | 1442 | case '&': |
| | 1443 | /* replace '&' with '&' */ |
| | 1444 | dstp.setch_str("&"); |
| | 1445 | break; |
| | 1446 | |
| | 1447 | case '<': |
| | 1448 | /* we must replace '<' with '<' - this adds three bytes */ |
| | 1449 | dstp.setch_str("<"); |
| | 1450 | break; |
| | 1451 | |
| | 1452 | case ' ': |
| | 1453 | /* note that this was a whitespace character */ |
| | 1454 | this_is_sp = TRUE; |
| | 1455 | |
| | 1456 | /* |
| | 1457 | * ignore it if not in preserve-spaces mode, or if the |
| | 1458 | * previous character wasn't whitespace of some kind |
| | 1459 | */ |
| | 1460 | if (!prv_was_sp || (flags & VMSTR_HTMLIFY_KEEP_SPACES) == 0) |
| | 1461 | goto do_default; |
| | 1462 | |
| | 1463 | /* add the nbsp */ |
| | 1464 | dstp.setch_str(" "); |
| | 1465 | break; |
| | 1466 | |
| | 1467 | case '\t': |
| | 1468 | /* note that this was a whitespace character */ |
| | 1469 | this_is_sp = TRUE; |
| | 1470 | |
| | 1471 | /* ignore if not in preserve-tabs mode */ |
| | 1472 | if ((flags & VMSTR_HTMLIFY_KEEP_TABS) == 0) |
| | 1473 | goto do_default; |
| | 1474 | |
| | 1475 | /* add the <tab> */ |
| | 1476 | dstp.setch_str("<tab>"); |
| | 1477 | break; |
| | 1478 | |
| | 1479 | case '\n': |
| | 1480 | case 0x2028: |
| | 1481 | /* note that this was a whitespace character */ |
| | 1482 | this_is_sp = TRUE; |
| | 1483 | |
| | 1484 | /* if we're not in preserve-newlines mode, ignore it */ |
| | 1485 | if ((flags & VMSTR_HTMLIFY_KEEP_NEWLINES) == 0) |
| | 1486 | goto do_default; |
| | 1487 | |
| | 1488 | /* add the <br> */ |
| | 1489 | dstp.setch_str("<br>"); |
| | 1490 | break; |
| | 1491 | |
| | 1492 | default: |
| | 1493 | do_default: |
| | 1494 | /* copy this character unchanged */ |
| | 1495 | dstp.setch(ch); |
| | 1496 | break; |
| | 1497 | } |
| | 1498 | |
| | 1499 | /* for next time, remember whether this is a space */ |
| | 1500 | prv_was_sp = this_is_sp; |
| | 1501 | } |
| | 1502 | |
| | 1503 | /* return the new string */ |
| | 1504 | retval->set_obj(result_obj); |
| | 1505 | |
| | 1506 | /* discard the GC protection */ |
| | 1507 | G_stk->discard(); |
| | 1508 | |
| | 1509 | /* handled */ |
| | 1510 | return TRUE; |
| | 1511 | } |
| | 1512 | |
| | 1513 | /* ------------------------------------------------------------------------ */ |
| | 1514 | /* |
| | 1515 | * property evaluator - startsWith |
| | 1516 | */ |
| | 1517 | int CVmObjString::getp_starts_with(VMG_ vm_val_t *retval, |
| | 1518 | const vm_val_t *self_val, |
| | 1519 | const char *str, uint *argc) |
| | 1520 | { |
| | 1521 | static CVmNativeCodeDesc desc(1); |
| | 1522 | const char *str2; |
| | 1523 | size_t len; |
| | 1524 | size_t len2; |
| | 1525 | |
| | 1526 | /* check arguments */ |
| | 1527 | if (get_prop_check_argc(retval, argc, &desc)) |
| | 1528 | return TRUE; |
| | 1529 | |
| | 1530 | /* retrieve the other string */ |
| | 1531 | str2 = CVmBif::pop_str_val(vmg0_); |
| | 1532 | |
| | 1533 | /* get the lengths of the two strings */ |
| | 1534 | len = vmb_get_len(str); |
| | 1535 | len2 = vmb_get_len(str2); |
| | 1536 | |
| | 1537 | /* move to the contents of each string */ |
| | 1538 | str += VMB_LEN; |
| | 1539 | str2 += VMB_LEN; |
| | 1540 | |
| | 1541 | /* |
| | 1542 | * if the other string is no longer than our string, and the other |
| | 1543 | * string matches our string exactly for the other string's entire |
| | 1544 | * length, we start with the other string |
| | 1545 | */ |
| | 1546 | retval->set_logical(len2 <= len && memcmp(str, str2, len2) == 0); |
| | 1547 | |
| | 1548 | /* handled */ |
| | 1549 | return TRUE; |
| | 1550 | } |
| | 1551 | |
| | 1552 | /* |
| | 1553 | * property evaluator - endsWith |
| | 1554 | */ |
| | 1555 | int CVmObjString::getp_ends_with(VMG_ vm_val_t *retval, |
| | 1556 | const vm_val_t *self_val, |
| | 1557 | const char *str, uint *argc) |
| | 1558 | { |
| | 1559 | static CVmNativeCodeDesc desc(1); |
| | 1560 | const char *str2; |
| | 1561 | size_t len; |
| | 1562 | size_t len2; |
| | 1563 | |
| | 1564 | /* check arguments */ |
| | 1565 | if (get_prop_check_argc(retval, argc, &desc)) |
| | 1566 | return TRUE; |
| | 1567 | |
| | 1568 | /* retrieve the other string */ |
| | 1569 | str2 = CVmBif::pop_str_val(vmg0_); |
| | 1570 | |
| | 1571 | /* get the lengths of the two strings */ |
| | 1572 | len = vmb_get_len(str); |
| | 1573 | len2 = vmb_get_len(str2); |
| | 1574 | |
| | 1575 | /* move to the contents of each string */ |
| | 1576 | str += VMB_LEN; |
| | 1577 | str2 += VMB_LEN; |
| | 1578 | |
| | 1579 | /* |
| | 1580 | * If the other string is no longer than our string, and the other |
| | 1581 | * string matches our string at the end exactly for the other string's |
| | 1582 | * entire length, we start with the other string. Note we don't need |
| | 1583 | * to worry about finding a valid character index in our string for |
| | 1584 | * the ending offset, because all we care about is whether or not we |
| | 1585 | * have an exact byte match between our suffix and the other string. |
| | 1586 | */ |
| | 1587 | retval->set_logical(len2 <= len |
| | 1588 | && memcmp(str + len - len2, str2, len2) == 0); |
| | 1589 | |
| | 1590 | /* handled */ |
| | 1591 | return TRUE; |
| | 1592 | } |
| | 1593 | |
| | 1594 | /* ------------------------------------------------------------------------ */ |
| | 1595 | /* |
| | 1596 | * property evaluator - mapToByteArray |
| | 1597 | */ |
| | 1598 | int CVmObjString::getp_to_byte_array(VMG_ vm_val_t *retval, |
| | 1599 | const vm_val_t *self_val, |
| | 1600 | const char *str, uint *argc) |
| | 1601 | { |
| | 1602 | static CVmNativeCodeDesc desc(1); |
| | 1603 | size_t len; |
| | 1604 | CCharmapToLocal *mapper; |
| | 1605 | vm_val_t *arg; |
| | 1606 | size_t byte_len; |
| | 1607 | size_t src_bytes_used; |
| | 1608 | size_t out_idx; |
| | 1609 | CVmObjByteArray *arr; |
| | 1610 | |
| | 1611 | /* check arguments */ |
| | 1612 | if (get_prop_check_argc(retval, argc, &desc)) |
| | 1613 | return TRUE; |
| | 1614 | |
| | 1615 | /* retrieve the CharacterSet object and make sure it's valid */ |
| | 1616 | arg = G_stk->get(0); |
| | 1617 | if (arg->typ != VM_OBJ || !CVmObjCharSet::is_charset(vmg_ arg->val.obj)) |
| | 1618 | err_throw(VMERR_BAD_TYPE_BIF); |
| | 1619 | |
| | 1620 | /* get the to-local mapping from the character set */ |
| | 1621 | mapper = ((CVmObjCharSet *)vm_objp(vmg_ arg->val.obj)) |
| | 1622 | ->get_to_local(vmg0_); |
| | 1623 | |
| | 1624 | /* get my length and skip the length prefix */ |
| | 1625 | len = vmb_get_len(str); |
| | 1626 | str += VMB_LEN; |
| | 1627 | |
| | 1628 | /* |
| | 1629 | * first, do a mapping with a null output buffer to determine how many |
| | 1630 | * bytes we need for the mapping |
| | 1631 | */ |
| | 1632 | byte_len = mapper->map_utf8(0, 0, str, len, &src_bytes_used); |
| | 1633 | |
| | 1634 | /* allocate a new ByteArray with the required number of bytes */ |
| | 1635 | retval->set_obj(CVmObjByteArray::create(vmg_ FALSE, byte_len)); |
| | 1636 | arr = (CVmObjByteArray *)vm_objp(vmg_ retval->val.obj); |
| | 1637 | |
| | 1638 | /* convert it again, this time storing the bytes */ |
| | 1639 | for (out_idx = 1 ; len != 0 ; ) |
| | 1640 | { |
| | 1641 | char buf[128]; |
| | 1642 | |
| | 1643 | /* convert a buffer-full */ |
| | 1644 | byte_len = mapper->map_utf8(buf, sizeof(buf), str, len, |
| | 1645 | &src_bytes_used); |
| | 1646 | |
| | 1647 | /* store the bytes in the byte array */ |
| | 1648 | arr->cons_copy_from_buf((unsigned char *)buf, out_idx, byte_len); |
| | 1649 | |
| | 1650 | /* advance past the output bytes we used */ |
| | 1651 | out_idx += byte_len; |
| | 1652 | |
| | 1653 | /* advance past the source bytes we used */ |
| | 1654 | str += src_bytes_used; |
| | 1655 | len -= src_bytes_used; |
| | 1656 | } |
| | 1657 | |
| | 1658 | /* discard arguments */ |
| | 1659 | G_stk->discard(); |
| | 1660 | |
| | 1661 | /* handled */ |
| | 1662 | return TRUE; |
| | 1663 | } |
| | 1664 | |
| | 1665 | /* ------------------------------------------------------------------------ */ |
| | 1666 | /* |
| | 1667 | * Constant-pool string object |
| | 1668 | */ |
| | 1669 | |
| | 1670 | /* |
| | 1671 | * create |
| | 1672 | */ |
| | 1673 | vm_obj_id_t CVmObjStringConst::create(VMG_ const char *const_ptr) |
| | 1674 | { |
| | 1675 | /* create our new ID */ |
| | 1676 | vm_obj_id_t id = vm_new_id(vmg_ FALSE, FALSE, FALSE); |
| | 1677 | |
| | 1678 | /* create our string object, pointing directly to the constant pool */ |
| | 1679 | new (vmg_ id) CVmObjStringConst(vmg_ const_ptr); |
| | 1680 | |
| | 1681 | /* return the new ID */ |
| | 1682 | return id; |
| | 1683 | } |