| | 1 | /* |
| | 2 | * Copyright (c) 2002 by Michael J. Roberts. All Rights Reserved. |
| | 3 | * |
| | 4 | * Please see the accompanying license file, LICENSE.TXT, for information |
| | 5 | * on using and copying this software. |
| | 6 | */ |
| | 7 | /* |
| | 8 | Name |
| | 9 | vmstrcmp.cpp - T3 String Comparator intrinsic class |
| | 10 | Function |
| | 11 | |
| | 12 | Notes |
| | 13 | |
| | 14 | Modified |
| | 15 | 09/05/02 MJRoberts - Creation |
| | 16 | */ |
| | 17 | |
| | 18 | |
| | 19 | #include <stdlib.h> |
| | 20 | #include <os.h> |
| | 21 | #include "utf8.h" |
| | 22 | #include "vmuni.h" |
| | 23 | #include "vmtype.h" |
| | 24 | #include "vmobj.h" |
| | 25 | #include "vmmeta.h" |
| | 26 | #include "vmglob.h" |
| | 27 | #include "vmstrcmp.h" |
| | 28 | #include "vmstack.h" |
| | 29 | #include "vmbif.h" |
| | 30 | #include "vmfile.h" |
| | 31 | #include "vmlst.h" |
| | 32 | |
| | 33 | |
| | 34 | /* ------------------------------------------------------------------------ */ |
| | 35 | /* |
| | 36 | * Statics |
| | 37 | */ |
| | 38 | |
| | 39 | /* metaclass registration */ |
| | 40 | static CVmMetaclassStrComp metaclass_reg_obj; |
| | 41 | CVmMetaclass *CVmObjStrComp::metaclass_reg_ = &metaclass_reg_obj; |
| | 42 | |
| | 43 | /* function table */ |
| | 44 | int (CVmObjStrComp:: |
| | 45 | *CVmObjStrComp::func_table_[])(VMG_ vm_obj_id_t self, |
| | 46 | vm_val_t *retval, uint *argc) = |
| | 47 | { |
| | 48 | &CVmObjStrComp::getp_undef, |
| | 49 | &CVmObjStrComp::getp_calc_hash, |
| | 50 | &CVmObjStrComp::getp_match_values |
| | 51 | }; |
| | 52 | |
| | 53 | /* ------------------------------------------------------------------------ */ |
| | 54 | /* |
| | 55 | * notify of deletion |
| | 56 | */ |
| | 57 | void CVmObjStrComp::notify_delete(VMG_ int in_root_set) |
| | 58 | { |
| | 59 | /* delete my extension data */ |
| | 60 | delete_ext(vmg0_); |
| | 61 | } |
| | 62 | |
| | 63 | /* |
| | 64 | * Delete my extension data |
| | 65 | */ |
| | 66 | void CVmObjStrComp::delete_ext(VMG0_) |
| | 67 | { |
| | 68 | vmobj_strcmp_ext *ext = get_ext(); |
| | 69 | |
| | 70 | /* if I have an extension, delete it */ |
| | 71 | if (ext != 0) |
| | 72 | { |
| | 73 | size_t i; |
| | 74 | |
| | 75 | /* delete each first-level mapping table */ |
| | 76 | for (i = 0 ; i < sizeof(ext->equiv)/sizeof(ext->equiv[0]) ; ++i) |
| | 77 | { |
| | 78 | /* if this table is present, delete it */ |
| | 79 | if (ext->equiv[i] != 0) |
| | 80 | t3free(ext->equiv[i]); |
| | 81 | } |
| | 82 | |
| | 83 | /* delete and forget our extension */ |
| | 84 | G_mem->get_var_heap()->free_mem(ext_); |
| | 85 | ext_ = 0; |
| | 86 | } |
| | 87 | } |
| | 88 | |
| | 89 | /* ------------------------------------------------------------------------ */ |
| | 90 | /* |
| | 91 | * set a property |
| | 92 | */ |
| | 93 | void CVmObjStrComp::set_prop(VMG_ class CVmUndo *, |
| | 94 | vm_obj_id_t, vm_prop_id_t, |
| | 95 | const vm_val_t *) |
| | 96 | { |
| | 97 | /* we have no properties to set */ |
| | 98 | err_throw(VMERR_INVALID_SETPROP); |
| | 99 | } |
| | 100 | |
| | 101 | /* ------------------------------------------------------------------------ */ |
| | 102 | /* |
| | 103 | * Get a property |
| | 104 | */ |
| | 105 | int CVmObjStrComp::get_prop(VMG_ vm_prop_id_t prop, vm_val_t *val, |
| | 106 | vm_obj_id_t self, vm_obj_id_t *source_obj, |
| | 107 | uint *argc) |
| | 108 | { |
| | 109 | uint func_idx; |
| | 110 | |
| | 111 | /* translate the property into a function vector index */ |
| | 112 | func_idx = G_meta_table |
| | 113 | ->prop_to_vector_idx(metaclass_reg_->get_reg_idx(), prop); |
| | 114 | |
| | 115 | /* call the function, if we found it */ |
| | 116 | if ((this->*func_table_[func_idx])(vmg_ self, val, argc)) |
| | 117 | { |
| | 118 | *source_obj = metaclass_reg_->get_class_obj(vmg0_); |
| | 119 | return TRUE; |
| | 120 | } |
| | 121 | |
| | 122 | /* not found - inherit default handling */ |
| | 123 | return CVmObject::get_prop(vmg_ prop, val, self, source_obj, argc); |
| | 124 | } |
| | 125 | |
| | 126 | |
| | 127 | /* ------------------------------------------------------------------------ */ |
| | 128 | /* |
| | 129 | * Abstract equivalence mapping reader. This can be implemented on |
| | 130 | * different underlying data sources. |
| | 131 | */ |
| | 132 | class CVmObjStrCompMapReader |
| | 133 | { |
| | 134 | public: |
| | 135 | /* |
| | 136 | * Read the next equivalence mapping. We fill in *ref_ch with the |
| | 137 | * reference character, *uc_result_flags with the upper-case result |
| | 138 | * flags, and *lc_result_flags with the lower-case result flags. On |
| | 139 | * input, *value_ch_cnt is the maximum number of value characters we |
| | 140 | * can store in the buffer at val_buf; on return, this is the number |
| | 141 | * of characters in the mapping, which might be higher than the number |
| | 142 | * originally in the buffer. In any case, we will write no more than |
| | 143 | * the allowed buffer size as given by *value_ch_cnt on input (so we |
| | 144 | * might indicate a higher number than we actually wrote: we always |
| | 145 | * return the actual value mapping size, even if we couldn't store the |
| | 146 | * whole thing because of a lack of buffer space). |
| | 147 | * |
| | 148 | * This routine should retrieve one mapping each time it's called, and |
| | 149 | * then move on to the next mapping. The caller is responsible for |
| | 150 | * making sure that this routine is called the correct number of |
| | 151 | * times, so we don't have to worry about running out of mappings. |
| | 152 | */ |
| | 153 | virtual void read_mapping(VMG_ wchar_t *ref_ch, |
| | 154 | unsigned long *uc_result_flags, |
| | 155 | unsigned long *lc_result_flags, |
| | 156 | wchar_t *val_buf, size_t *value_ch_cnt) = 0; |
| | 157 | }; |
| | 158 | |
| | 159 | /* |
| | 160 | * Stream-based mapping reader. This reads mappings from a stream that |
| | 161 | * uses our serialization format for image and saved-state files. |
| | 162 | */ |
| | 163 | class CVmObjStrCompMapReaderStream: public CVmObjStrCompMapReader |
| | 164 | { |
| | 165 | public: |
| | 166 | CVmObjStrCompMapReaderStream(CVmStream *str) { str_ = str; } |
| | 167 | |
| | 168 | /* read a mapping */ |
| | 169 | virtual void read_mapping(VMG_ wchar_t *ref_ch, |
| | 170 | unsigned long *uc_result_flags, |
| | 171 | unsigned long *lc_result_flags, |
| | 172 | wchar_t *val_buf, size_t *value_ch_cnt) |
| | 173 | { |
| | 174 | size_t copy_limit; |
| | 175 | size_t copy_size; |
| | 176 | size_t i; |
| | 177 | |
| | 178 | /* limit our value character copying to the actual buffer size */ |
| | 179 | copy_limit = *value_ch_cnt; |
| | 180 | |
| | 181 | /* read the header values */ |
| | 182 | *ref_ch = (wchar_t)str_->read_uint2(); |
| | 183 | copy_size = *value_ch_cnt = str_->read_byte(); |
| | 184 | *uc_result_flags = str_->read_uint4(); |
| | 185 | *lc_result_flags = str_->read_uint4(); |
| | 186 | |
| | 187 | /* limit copying to the actual buffer size */ |
| | 188 | if (copy_size > copy_limit) |
| | 189 | copy_size = copy_limit; |
| | 190 | |
| | 191 | /* read the values */ |
| | 192 | for (i = 0 ; i < copy_size ; ++i) |
| | 193 | *val_buf++ = (wchar_t)str_->read_uint2(); |
| | 194 | |
| | 195 | /* skip any values from the input that we weren't able to store */ |
| | 196 | for ( ; i < *value_ch_cnt ; ++i) |
| | 197 | str_->read_uint2(); |
| | 198 | } |
| | 199 | |
| | 200 | protected: |
| | 201 | /* my stream */ |
| | 202 | CVmStream *str_; |
| | 203 | }; |
| | 204 | |
| | 205 | /* |
| | 206 | * Constructor-list mapping reader. This reads mappings from list data in |
| | 207 | * the format that we use in our constructor. |
| | 208 | */ |
| | 209 | class CVmObjStrCompMapReaderList: public CVmObjStrCompMapReader |
| | 210 | { |
| | 211 | public: |
| | 212 | CVmObjStrCompMapReaderList(const char *lst) |
| | 213 | { |
| | 214 | /* remember the list */ |
| | 215 | lst_ = lst; |
| | 216 | |
| | 217 | /* start at the first element */ |
| | 218 | idx_ = 1; |
| | 219 | } |
| | 220 | |
| | 221 | /* read a mapping */ |
| | 222 | virtual void read_mapping(VMG_ wchar_t *ref_ch, |
| | 223 | unsigned long *uc_result_flags, |
| | 224 | unsigned long *lc_result_flags, |
| | 225 | wchar_t *val_buf, size_t *value_ch_cnt) |
| | 226 | { |
| | 227 | vm_val_t val; |
| | 228 | const char *sublst; |
| | 229 | const char *refstr; |
| | 230 | const char *valstr; |
| | 231 | size_t copy_rem; |
| | 232 | size_t rem; |
| | 233 | utf8_ptr p; |
| | 234 | |
| | 235 | /* note the size limit of the caller's value string buffer */ |
| | 236 | copy_rem = *value_ch_cnt; |
| | 237 | |
| | 238 | /* retrieve the next element of the list */ |
| | 239 | CVmObjList::index_list(vmg_ &val, lst_, idx_); |
| | 240 | |
| | 241 | /* get the value as a sublist */ |
| | 242 | if ((sublst = val.get_as_list(vmg0_)) == 0) |
| | 243 | err_throw(VMERR_BAD_TYPE_BIF); |
| | 244 | |
| | 245 | /* retrieve the reference character string (sublst's 1st element) */ |
| | 246 | CVmObjList::index_list(vmg_ &val, sublst, 1); |
| | 247 | refstr = val.get_as_string(vmg0_); |
| | 248 | |
| | 249 | /* retrieve the value string (sublst's 2nd element) */ |
| | 250 | CVmObjList::index_list(vmg_ &val, sublst, 2); |
| | 251 | valstr = val.get_as_string(vmg0_); |
| | 252 | |
| | 253 | /* make sure the reference and value strings are indeed strings */ |
| | 254 | if (refstr == 0 || valstr == 0) |
| | 255 | err_throw(VMERR_BAD_TYPE_BIF); |
| | 256 | |
| | 257 | /* we need at least one character in each string */ |
| | 258 | if (vmb_get_len(refstr) == 0 || vmb_get_len(valstr) == 0) |
| | 259 | err_throw(VMERR_BAD_VAL_BIF); |
| | 260 | |
| | 261 | /* fill in the caller's reference character from the ref string */ |
| | 262 | *ref_ch = utf8_ptr::s_getch(refstr + VMB_LEN); |
| | 263 | |
| | 264 | /* store the characters of the value string, up to the buffer limit */ |
| | 265 | p.set((char *)valstr + VMB_LEN); |
| | 266 | rem = vmb_get_len(valstr); |
| | 267 | for (*value_ch_cnt = 0 ; rem != 0 ; p.inc(&rem)) |
| | 268 | { |
| | 269 | /* if we have room, copy this character */ |
| | 270 | if (copy_rem != 0) |
| | 271 | { |
| | 272 | /* copy the character */ |
| | 273 | *val_buf++ = p.getch(); |
| | 274 | |
| | 275 | /* count it */ |
| | 276 | --copy_rem; |
| | 277 | } |
| | 278 | |
| | 279 | /* count it in the actual length */ |
| | 280 | ++(*value_ch_cnt); |
| | 281 | } |
| | 282 | |
| | 283 | /* get the upper-case flags (sublst's 3rd element) */ |
| | 284 | CVmObjList::index_list(vmg_ &val, sublst, 3); |
| | 285 | if (val.typ != VM_INT) |
| | 286 | err_throw(VMERR_BAD_TYPE_BIF); |
| | 287 | *uc_result_flags = val.val.intval; |
| | 288 | |
| | 289 | /* get the lower-case flags (sublst's 4th element) */ |
| | 290 | CVmObjList::index_list(vmg_ &val, sublst, 4); |
| | 291 | if (val.typ != VM_INT) |
| | 292 | err_throw(VMERR_BAD_TYPE_BIF); |
| | 293 | *lc_result_flags = val.val.intval; |
| | 294 | |
| | 295 | /* we're done with this mapping, so advance to the next one */ |
| | 296 | ++idx_; |
| | 297 | } |
| | 298 | |
| | 299 | protected: |
| | 300 | /* my list data */ |
| | 301 | const char *lst_; |
| | 302 | |
| | 303 | /* the list index of the next mapping to retrieve */ |
| | 304 | size_t idx_; |
| | 305 | }; |
| | 306 | |
| | 307 | /* ------------------------------------------------------------------------ */ |
| | 308 | /* |
| | 309 | * Create from stack. We take the following constructor arguments: |
| | 310 | * |
| | 311 | * trunc_len (int: truncation length) |
| | 312 | *. case_sensitive (bool: case sensitivity flag) |
| | 313 | *. mappings (list: equivalence mappings) |
| | 314 | * |
| | 315 | * Our equivalence mappings are given as a list of lists; the main list |
| | 316 | * consists of one sublist per mapping. Each mapping sublist looks like |
| | 317 | * this: |
| | 318 | * |
| | 319 | * ['ref_char', 'val_string', uc_flags, lc_flag] |
| | 320 | * |
| | 321 | * The 'ref_char' is a one-character string giving the reference character |
| | 322 | * of the mapping, and 'val_string' is a string of one or more characters |
| | 323 | * that can match the reference character in a value (input) string. |
| | 324 | * uc_flags and lc_flags are integers giving the upper-case and lower-case |
| | 325 | * flags (respectively) that are to be added to the match result code when |
| | 326 | * the mapping is used to match a pair of strings. |
| | 327 | */ |
| | 328 | vm_obj_id_t CVmObjStrComp::create_from_stack( |
| | 329 | VMG_ const uchar **pc_ptr, uint argc) |
| | 330 | { |
| | 331 | size_t trunc_len; |
| | 332 | int case_sensitive; |
| | 333 | const char *lst; |
| | 334 | vm_obj_id_t id; |
| | 335 | CVmObjStrComp *obj; |
| | 336 | size_t equiv_cnt; |
| | 337 | size_t total_chars; |
| | 338 | |
| | 339 | /* check arguments */ |
| | 340 | if (argc != 3) |
| | 341 | err_throw(VMERR_WRONG_NUM_OF_ARGS); |
| | 342 | |
| | 343 | /* pop the truncation length parameter */ |
| | 344 | if (G_stk->get(0)->typ == VM_NIL) |
| | 345 | { |
| | 346 | /* it's nil, so truncation is not allowed */ |
| | 347 | trunc_len = 0; |
| | 348 | G_stk->discard(); |
| | 349 | } |
| | 350 | else |
| | 351 | { |
| | 352 | /* retrieve the truncation length as an integer */ |
| | 353 | trunc_len = CVmBif::pop_int_val(vmg0_); |
| | 354 | } |
| | 355 | |
| | 356 | /* get the case sensitivity flag */ |
| | 357 | case_sensitive = CVmBif::pop_bool_val(vmg0_); |
| | 358 | |
| | 359 | /* |
| | 360 | * retrieve the mapping list, but leave it on the stack (for gc |
| | 361 | * protection) |
| | 362 | */ |
| | 363 | if (G_stk->get(0)->typ == VM_NIL) |
| | 364 | { |
| | 365 | /* there are no mappings */ |
| | 366 | lst = 0; |
| | 367 | equiv_cnt = 0; |
| | 368 | total_chars = 0; |
| | 369 | } |
| | 370 | else |
| | 371 | { |
| | 372 | size_t i; |
| | 373 | |
| | 374 | /* get the list value from the argument */ |
| | 375 | lst = G_stk->get(0)->get_as_list(vmg0_); |
| | 376 | if (lst == 0) |
| | 377 | err_throw(VMERR_BAD_TYPE_BIF); |
| | 378 | |
| | 379 | /* the list contains one entry per equivalence mapping */ |
| | 380 | equiv_cnt = vmb_get_len(lst); |
| | 381 | |
| | 382 | /* run through the list and count the value string characters */ |
| | 383 | for (i = 1, total_chars = 0 ; i <= equiv_cnt ; ++i) |
| | 384 | { |
| | 385 | vm_val_t val; |
| | 386 | const char *sublst; |
| | 387 | const char *strp; |
| | 388 | utf8_ptr ustrp; |
| | 389 | |
| | 390 | /* get this mapping from the list */ |
| | 391 | CVmObjList::index_list(vmg_ &val, lst, i); |
| | 392 | |
| | 393 | /* make sure it's a sublist */ |
| | 394 | if ((sublst = val.get_as_list(vmg0_)) == 0) |
| | 395 | err_throw(VMERR_BAD_TYPE_BIF); |
| | 396 | |
| | 397 | /* |
| | 398 | * get the second element of the mapping sublist - this is the |
| | 399 | * value string |
| | 400 | */ |
| | 401 | CVmObjList::index_list(vmg_ &val, sublst, 2); |
| | 402 | if ((strp = val.get_as_string(vmg0_)) == 0) |
| | 403 | err_throw(VMERR_BAD_TYPE_BIF); |
| | 404 | |
| | 405 | /* add the character length of the string to the total */ |
| | 406 | ustrp.set((char *)strp + VMB_LEN); |
| | 407 | total_chars += ustrp.len(vmb_get_len(strp)); |
| | 408 | } |
| | 409 | } |
| | 410 | |
| | 411 | /* create the new object */ |
| | 412 | id = vm_new_id(vmg_ FALSE, FALSE, FALSE); |
| | 413 | obj = new (vmg_ id) CVmObjStrComp(); |
| | 414 | |
| | 415 | /* set up a list-based mapping reader */ |
| | 416 | CVmObjStrCompMapReaderList reader(lst); |
| | 417 | |
| | 418 | /* allocate and initialize the new object's extension */ |
| | 419 | obj->alloc_ext(vmg_ trunc_len, case_sensitive, equiv_cnt, total_chars, |
| | 420 | &reader); |
| | 421 | |
| | 422 | /* discard the gc protection */ |
| | 423 | G_stk->discard(); |
| | 424 | |
| | 425 | /* return the new object */ |
| | 426 | return id; |
| | 427 | } |
| | 428 | |
| | 429 | /* ------------------------------------------------------------------------ */ |
| | 430 | /* |
| | 431 | * Load from an image file |
| | 432 | */ |
| | 433 | void CVmObjStrComp::load_from_image(VMG_ vm_obj_id_t /*self*/, |
| | 434 | const char *ptr, size_t len) |
| | 435 | { |
| | 436 | /* load my image data */ |
| | 437 | CVmReadOnlyMemoryStream str(ptr, len); |
| | 438 | load_from_stream(vmg_ &str); |
| | 439 | } |
| | 440 | |
| | 441 | /* |
| | 442 | * Load from an abstract stream |
| | 443 | */ |
| | 444 | void CVmObjStrComp::load_from_stream(VMG_ CVmStream *str) |
| | 445 | { |
| | 446 | unsigned int trunc_len; |
| | 447 | unsigned int flags; |
| | 448 | unsigned int equiv_cnt; |
| | 449 | unsigned int total_chars; |
| | 450 | |
| | 451 | /* load the fixed header */ |
| | 452 | trunc_len = str->read_uint2(); |
| | 453 | flags = str->read_uint2(); |
| | 454 | equiv_cnt = str->read_uint2(); |
| | 455 | total_chars = str->read_uint2(); |
| | 456 | |
| | 457 | /* set up a stream-based mapping reader */ |
| | 458 | CVmObjStrCompMapReaderStream reader(str); |
| | 459 | |
| | 460 | /* allocate and initialize our extension */ |
| | 461 | alloc_ext(vmg_ trunc_len, (flags & 0x0001) != 0, equiv_cnt, total_chars, |
| | 462 | &reader); |
| | 463 | } |
| | 464 | |
| | 465 | /* |
| | 466 | * Allocate and initialize our extension |
| | 467 | */ |
| | 468 | void CVmObjStrComp::alloc_ext(VMG_ size_t trunc_len, int case_sensitive, |
| | 469 | size_t equiv_cnt, size_t total_chars, |
| | 470 | CVmObjStrCompMapReader *reader) |
| | 471 | { |
| | 472 | size_t siz; |
| | 473 | vmobj_strcmp_ext *ext; |
| | 474 | vmobj_strcmp_equiv *nxt_equiv; |
| | 475 | wchar_t *nxt_ch; |
| | 476 | size_t ch_rem; |
| | 477 | size_t i; |
| | 478 | size_t idx1, idx2; |
| | 479 | |
| | 480 | /* delete my extension, if I have one already */ |
| | 481 | delete_ext(vmg0_); |
| | 482 | |
| | 483 | /* |
| | 484 | * Calculate how much space we need for our extension. In addition to |
| | 485 | * the fixed part, allocate space for one vmobj_strcmp_equiv structure |
| | 486 | * per mapping, plus the wchar_t's for the value mappings. |
| | 487 | */ |
| | 488 | siz = sizeof(vmobj_strcmp_ext) |
| | 489 | + (equiv_cnt * sizeof(vmobj_strcmp_equiv)) |
| | 490 | + (total_chars * sizeof(wchar_t)); |
| | 491 | |
| | 492 | /* allocate our new extension */ |
| | 493 | ext_ = (char *)G_mem->get_var_heap()->alloc_mem(siz, this); |
| | 494 | ext = get_ext(); |
| | 495 | |
| | 496 | /* |
| | 497 | * set up our suballocation pool pointers: put the equivalence mapping |
| | 498 | * structures after the fixed part of the extension, and put the |
| | 499 | * wchar_t's after the equivalence mappings |
| | 500 | */ |
| | 501 | nxt_equiv = (vmobj_strcmp_equiv *)(ext + 1); |
| | 502 | nxt_ch = (wchar_t *)(&nxt_equiv[equiv_cnt]); |
| | 503 | ch_rem = total_chars; |
| | 504 | |
| | 505 | /* initialize the extension structure */ |
| | 506 | ext->trunc_len = trunc_len; |
| | 507 | ext->case_sensitive = case_sensitive; |
| | 508 | |
| | 509 | /* |
| | 510 | * we have no equivalence mappings installed yet, so clear out the |
| | 511 | * first tier of the mapping array |
| | 512 | */ |
| | 513 | for (i = 0 ; i < sizeof(ext->equiv)/sizeof(ext->equiv[0]) ; ++i) |
| | 514 | ext->equiv[i] = 0; |
| | 515 | |
| | 516 | /* load the mappings */ |
| | 517 | for (i = 0 ; i < equiv_cnt ; ++i, ++nxt_equiv) |
| | 518 | { |
| | 519 | wchar_t ref_ch; |
| | 520 | |
| | 521 | /* |
| | 522 | * set up our equivalent's value buffer with the remainder of our |
| | 523 | * main buffer |
| | 524 | */ |
| | 525 | nxt_equiv->val_ch = nxt_ch; |
| | 526 | nxt_equiv->val_ch_cnt = ch_rem; |
| | 527 | |
| | 528 | /* read the mapping */ |
| | 529 | reader->read_mapping(vmg_ &ref_ch, |
| | 530 | &nxt_equiv->uc_result_flags, |
| | 531 | &nxt_equiv->lc_result_flags, |
| | 532 | nxt_equiv->val_ch, &nxt_equiv->val_ch_cnt); |
| | 533 | |
| | 534 | /* deduct this mapping from our main value character buffer */ |
| | 535 | nxt_ch += nxt_equiv->val_ch_cnt; |
| | 536 | ch_rem -= nxt_equiv->val_ch_cnt; |
| | 537 | |
| | 538 | /* if we don't have a first-tier table for this character, add one */ |
| | 539 | idx1 = (ref_ch >> 8) & 0xFF; |
| | 540 | idx2 = (ref_ch & 0xFF); |
| | 541 | if (ext->equiv[idx1] == 0) |
| | 542 | { |
| | 543 | vmobj_strcmp_equiv **p; |
| | 544 | size_t j; |
| | 545 | |
| | 546 | /* allocate a first-tier table for this index */ |
| | 547 | ext->equiv[idx1] = p = (vmobj_strcmp_equiv **)t3malloc( |
| | 548 | 256 * sizeof(vmobj_strcmp_equiv *)); |
| | 549 | |
| | 550 | /* clear out the first-tier table */ |
| | 551 | for (j = 0 ; j < 256 ; ++j, ++p) |
| | 552 | *p = 0; |
| | 553 | } |
| | 554 | |
| | 555 | /* set the mapping for this character */ |
| | 556 | ext->equiv[idx1][idx2] = nxt_equiv; |
| | 557 | } |
| | 558 | } |
| | 559 | |
| | 560 | /* ------------------------------------------------------------------------ */ |
| | 561 | /* |
| | 562 | * save to a file |
| | 563 | */ |
| | 564 | void CVmObjStrComp::save_to_file(VMG_ class CVmFile *fp) |
| | 565 | { |
| | 566 | /* write our data to the file */ |
| | 567 | CVmFileStream str(fp); |
| | 568 | write_to_stream(vmg_ &str, 0); |
| | 569 | } |
| | 570 | |
| | 571 | /* |
| | 572 | * Serialize to a stream |
| | 573 | */ |
| | 574 | ulong CVmObjStrComp::write_to_stream(VMG_ CVmStream *str, ulong *bytes_avail) |
| | 575 | { |
| | 576 | wchar_t ref_ch; |
| | 577 | vmobj_strcmp_ext *ext = get_ext(); |
| | 578 | size_t i; |
| | 579 | vmobj_strcmp_equiv ***p; |
| | 580 | size_t total_value_ch; |
| | 581 | size_t equiv_cnt; |
| | 582 | size_t need_size; |
| | 583 | |
| | 584 | /* get the mapping totals */ |
| | 585 | count_equiv_mappings(&equiv_cnt, &total_value_ch); |
| | 586 | |
| | 587 | /* |
| | 588 | * Calculate our space needs. We need 8 bytes for the fixed header, |
| | 589 | * 11 bytes per equivalent mapping, and 2 bytes per value string |
| | 590 | * character. |
| | 591 | */ |
| | 592 | need_size = 8 + (11 * equiv_cnt) + (2 * total_value_ch); |
| | 593 | |
| | 594 | /* if we have a size limit, check to make sure we can abide by it */ |
| | 595 | if (bytes_avail != 0 && need_size > *bytes_avail) |
| | 596 | { |
| | 597 | /* |
| | 598 | * there's not enough space in the output stream for us, so don't |
| | 599 | * write anything at all; simply return the amount of space we |
| | 600 | * need |
| | 601 | */ |
| | 602 | return need_size; |
| | 603 | } |
| | 604 | |
| | 605 | /* write out the serialization structure header */ |
| | 606 | str->write_int2(ext->trunc_len); |
| | 607 | str->write_int2(ext->case_sensitive ? 0x0001 : 0x0000); |
| | 608 | str->write_int2(equiv_cnt); |
| | 609 | str->write_int2(total_value_ch); |
| | 610 | |
| | 611 | /* run through our equivalence table again and write the mappings */ |
| | 612 | for (ref_ch = 0, i = 0, p = ext->equiv ; |
| | 613 | i < sizeof(ext->equiv)/sizeof(ext->equiv[0]) ; ++i, ++p, ++ref_ch) |
| | 614 | { |
| | 615 | vmobj_strcmp_equiv **ep; |
| | 616 | size_t j; |
| | 617 | |
| | 618 | /* if this first-tier mapping is unused, skip it */ |
| | 619 | if (*p == 0) |
| | 620 | continue; |
| | 621 | |
| | 622 | /* run through our second-level table */ |
| | 623 | for (j = 0, ep = *p ; j < 256 ; ++j, ++ep) |
| | 624 | { |
| | 625 | /* if this mapping is used, write it out */ |
| | 626 | if (*ep != 0) |
| | 627 | { |
| | 628 | size_t k; |
| | 629 | wchar_t *vp; |
| | 630 | |
| | 631 | /* write the fixed part of the mapping */ |
| | 632 | str->write_int2(ref_ch); |
| | 633 | str->write_byte((uchar)(*ep)->val_ch_cnt); |
| | 634 | str->write_int4((*ep)->uc_result_flags); |
| | 635 | str->write_int4((*ep)->lc_result_flags); |
| | 636 | |
| | 637 | /* write the value mapping characters */ |
| | 638 | for (k = (*ep)->val_ch_cnt, vp = (*ep)->val_ch ; k != 0 ; |
| | 639 | --k, ++vp) |
| | 640 | { |
| | 641 | /* write this character */ |
| | 642 | str->write_int2(*vp); |
| | 643 | } |
| | 644 | } |
| | 645 | } |
| | 646 | } |
| | 647 | |
| | 648 | /* return our space needs */ |
| | 649 | return need_size; |
| | 650 | } |
| | 651 | |
| | 652 | /* |
| | 653 | * Count the equivalence mappings. |
| | 654 | */ |
| | 655 | void CVmObjStrComp::count_equiv_mappings(size_t *equiv_cnt, |
| | 656 | size_t *total_value_ch) |
| | 657 | { |
| | 658 | vmobj_strcmp_ext *ext = get_ext(); |
| | 659 | size_t i; |
| | 660 | vmobj_strcmp_equiv ***p; |
| | 661 | |
| | 662 | /* run through our table and count up the mappings */ |
| | 663 | for (*total_value_ch = 0, *equiv_cnt = 0, i = 0, p = ext->equiv ; |
| | 664 | i < sizeof(ext->equiv)/sizeof(ext->equiv[0]) ; ++i, ++p) |
| | 665 | { |
| | 666 | vmobj_strcmp_equiv **ep; |
| | 667 | size_t j; |
| | 668 | |
| | 669 | /* if this first-tier mapping is unused, skip it */ |
| | 670 | if (*p == 0) |
| | 671 | continue; |
| | 672 | |
| | 673 | /* run through our second-level table */ |
| | 674 | for (j = 0, ep = *p ; j < 256 ; ++j, ++ep) |
| | 675 | { |
| | 676 | /* if this mapping is used, count it */ |
| | 677 | if (*ep != 0) |
| | 678 | { |
| | 679 | /* count this equivalent mapping */ |
| | 680 | ++(*equiv_cnt); |
| | 681 | |
| | 682 | /* count its value mapping characters in the total */ |
| | 683 | *total_value_ch += (*ep)->val_ch_cnt; |
| | 684 | } |
| | 685 | } |
| | 686 | } |
| | 687 | } |
| | 688 | |
| | 689 | /* ------------------------------------------------------------------------ */ |
| | 690 | /* |
| | 691 | * restore from a file |
| | 692 | */ |
| | 693 | void CVmObjStrComp::restore_from_file(VMG_ vm_obj_id_t /*self*/, |
| | 694 | class CVmFile *fp, CVmObjFixup *) |
| | 695 | { |
| | 696 | /* load from the file */ |
| | 697 | CVmFileStream str(fp); |
| | 698 | load_from_stream(vmg_ &str); |
| | 699 | } |
| | 700 | |
| | 701 | /* ------------------------------------------------------------------------ */ |
| | 702 | /* |
| | 703 | * property evaluator - calculate a hash value |
| | 704 | */ |
| | 705 | int CVmObjStrComp::getp_calc_hash(VMG_ vm_obj_id_t /*self*/, |
| | 706 | vm_val_t *retval, uint *argc) |
| | 707 | { |
| | 708 | static CVmNativeCodeDesc desc(1); |
| | 709 | const char *strp; |
| | 710 | |
| | 711 | /* check arguments */ |
| | 712 | if (get_prop_check_argc(retval, argc, &desc)) |
| | 713 | return TRUE; |
| | 714 | |
| | 715 | /* |
| | 716 | * retrieve the string argument (it must be a string), but leave it on |
| | 717 | * the stack for now, for gc protection |
| | 718 | */ |
| | 719 | strp = G_stk->get(0)->get_as_string(vmg0_); |
| | 720 | if (strp == 0) |
| | 721 | err_throw(VMERR_BAD_TYPE_BIF); |
| | 722 | |
| | 723 | /* calculate the hash value and return it */ |
| | 724 | retval->set_int(calc_str_hash(strp + VMB_LEN, vmb_get_len(strp))); |
| | 725 | |
| | 726 | /* discard gc protection */ |
| | 727 | G_stk->discard(); |
| | 728 | |
| | 729 | /* handled */ |
| | 730 | return TRUE; |
| | 731 | } |
| | 732 | |
| | 733 | /* |
| | 734 | * Calculate a hash value |
| | 735 | */ |
| | 736 | unsigned int CVmObjStrComp::calc_str_hash(const char *strp, size_t len) |
| | 737 | { |
| | 738 | vmobj_strcmp_ext *ext = get_ext(); |
| | 739 | utf8_ptr p; |
| | 740 | unsigned int hash; |
| | 741 | size_t char_limit; |
| | 742 | |
| | 743 | /* set up to scan the string */ |
| | 744 | p.set((char *)strp); |
| | 745 | |
| | 746 | /* |
| | 747 | * Limit the scan to our truncation length, because we can't |
| | 748 | * distinguish hash buckets beyond the truncation point (if we did, a |
| | 749 | * truncated string wouldn't hash into the same bucket as a longer |
| | 750 | * string it matches; but all matching strings are required to go into |
| | 751 | * the same bucket, so such a hash mismatch is not allowed). |
| | 752 | */ |
| | 753 | char_limit = ext->trunc_len; |
| | 754 | |
| | 755 | /* scan the string */ |
| | 756 | for (hash = 0 ; len != 0 ; p.inc(&len)) |
| | 757 | { |
| | 758 | wchar_t ch; |
| | 759 | vmobj_strcmp_equiv **t1; |
| | 760 | vmobj_strcmp_equiv *eq; |
| | 761 | |
| | 762 | /* get the current character */ |
| | 763 | ch = p.getch(); |
| | 764 | |
| | 765 | /* check for a substitution mapping for this character */ |
| | 766 | if ((t1 = ext->equiv[(ch >> 8) & 0xFF]) != 0 |
| | 767 | && (eq = t1[ch & 0xFF]) != 0) |
| | 768 | { |
| | 769 | wchar_t *vp; |
| | 770 | size_t vlen; |
| | 771 | |
| | 772 | /* |
| | 773 | * This character has a mapping, so add the contribution from |
| | 774 | * the canonical form of the character, which is the value |
| | 775 | * side of the mapping. |
| | 776 | */ |
| | 777 | for (vp = eq->val_ch, vlen = eq->val_ch_cnt ; vlen != 0 ; |
| | 778 | ++vp, --vlen) |
| | 779 | { |
| | 780 | /* get this character */ |
| | 781 | ch = *vp; |
| | 782 | |
| | 783 | /* convert to lower case if we're insensitive to case */ |
| | 784 | if (!ext->case_sensitive) |
| | 785 | ch = t3_to_lower(ch); |
| | 786 | |
| | 787 | /* add it to the hash code */ |
| | 788 | hash += ch; |
| | 789 | hash &= 0xFFFF; |
| | 790 | |
| | 791 | /* if we've reached the truncation limit, we're done */ |
| | 792 | if (char_limit == 1) |
| | 793 | return hash; |
| | 794 | else if (char_limit != 0) |
| | 795 | --char_limit; |
| | 796 | } |
| | 797 | } |
| | 798 | else |
| | 799 | { |
| | 800 | /* |
| | 801 | * if we are not sensitive to case, always use the lower-case |
| | 802 | * representation of a character for its hash value |
| | 803 | */ |
| | 804 | if (!ext->case_sensitive) |
| | 805 | ch = t3_to_lower(ch); |
| | 806 | |
| | 807 | /* add the contribution from this character */ |
| | 808 | hash += ch; |
| | 809 | hash &= 0xFFFF; |
| | 810 | |
| | 811 | /* if we've reached the truncation limit, we're done */ |
| | 812 | if (char_limit == 1) |
| | 813 | return hash; |
| | 814 | else if (char_limit != 0) |
| | 815 | --char_limit; |
| | 816 | } |
| | 817 | } |
| | 818 | |
| | 819 | /* return the hash code */ |
| | 820 | return hash; |
| | 821 | } |
| | 822 | |
| | 823 | /* ------------------------------------------------------------------------ */ |
| | 824 | /* |
| | 825 | * Pre-defined return flag values. |
| | 826 | */ |
| | 827 | |
| | 828 | #define RF_MATCH 0x0001 /* the string matched */ |
| | 829 | #define RF_CASEFOLD 0x0002 /* matched with case folding */ |
| | 830 | #define RF_TRUNC 0x0004 /* matched with truncation */ |
| | 831 | |
| | 832 | |
| | 833 | /* |
| | 834 | * property evaluator - calculate a hash value |
| | 835 | */ |
| | 836 | int CVmObjStrComp::getp_match_values(VMG_ vm_obj_id_t /*self*/, |
| | 837 | vm_val_t *retval, uint *argc) |
| | 838 | { |
| | 839 | static CVmNativeCodeDesc desc(2); |
| | 840 | const char *valstr; |
| | 841 | const char *refstr; |
| | 842 | |
| | 843 | /* check arguments */ |
| | 844 | if (get_prop_check_argc(retval, argc, &desc)) |
| | 845 | return TRUE; |
| | 846 | |
| | 847 | /* retrieve the strings, but leave them on the stack for gc protection */ |
| | 848 | valstr = G_stk->get(0)->get_as_string(vmg0_); |
| | 849 | refstr = G_stk->get(1)->get_as_string(vmg0_); |
| | 850 | |
| | 851 | /* make sure they're valid strings */ |
| | 852 | if (valstr == 0 || refstr == 0) |
| | 853 | err_throw(VMERR_BAD_TYPE_BIF); |
| | 854 | |
| | 855 | /* compare the strings and return the result */ |
| | 856 | retval->set_int(match_strings(valstr + VMB_LEN, vmb_get_len(valstr), |
| | 857 | refstr + VMB_LEN, vmb_get_len(refstr))); |
| | 858 | |
| | 859 | /* discard the gc protection */ |
| | 860 | G_stk->discard(2); |
| | 861 | |
| | 862 | /* handled */ |
| | 863 | return TRUE; |
| | 864 | } |
| | 865 | |
| | 866 | /* |
| | 867 | * Match two strings |
| | 868 | */ |
| | 869 | unsigned long CVmObjStrComp::match_strings(const char *valstr, size_t vallen, |
| | 870 | const char *refstr, size_t reflen) |
| | 871 | { |
| | 872 | vmobj_strcmp_ext *ext = get_ext(); |
| | 873 | utf8_ptr valp; |
| | 874 | utf8_ptr refp; |
| | 875 | unsigned long ret; |
| | 876 | int fold_case = !(ext->case_sensitive); |
| | 877 | size_t valcharlen; |
| | 878 | |
| | 879 | /* set up to scan the strings */ |
| | 880 | valp.set((char *)valstr); |
| | 881 | refp.set((char *)refstr); |
| | 882 | |
| | 883 | /* start with no return flags */ |
| | 884 | ret = 0; |
| | 885 | |
| | 886 | /* scan the strings */ |
| | 887 | for (valcharlen = 0 ; vallen != 0 && reflen != 0 ; refp.inc(&reflen)) |
| | 888 | { |
| | 889 | wchar_t valch; |
| | 890 | wchar_t refch; |
| | 891 | vmobj_strcmp_equiv **t1; |
| | 892 | vmobj_strcmp_equiv *eq; |
| | 893 | |
| | 894 | /* get each character */ |
| | 895 | valch = valp.getch(); |
| | 896 | refch = refp.getch(); |
| | 897 | |
| | 898 | /* check for an exact match first */ |
| | 899 | if (refch == valch) |
| | 900 | { |
| | 901 | /* it's an exact match - skip this input character */ |
| | 902 | valp.inc(&vallen); |
| | 903 | ++valcharlen; |
| | 904 | continue; |
| | 905 | } |
| | 906 | |
| | 907 | /* check for a case-folded match if we're insensitive to case */ |
| | 908 | if (fold_case && t3_to_lower(valch) == t3_to_lower(refch)) |
| | 909 | { |
| | 910 | /* it's a case-folded match - skip this input character */ |
| | 911 | valp.inc(&vallen); |
| | 912 | ++valcharlen; |
| | 913 | |
| | 914 | /* note in the flags that we have differing cases in the match */ |
| | 915 | ret |= RF_CASEFOLD; |
| | 916 | |
| | 917 | /* keep going */ |
| | 918 | continue; |
| | 919 | } |
| | 920 | |
| | 921 | /* check for a reference equivalence mapping */ |
| | 922 | if ((t1 = ext->equiv[(refch >> 8) & 0xFF]) != 0 |
| | 923 | && (eq = t1[refch & 0xFF]) != 0) |
| | 924 | { |
| | 925 | wchar_t *vp; |
| | 926 | size_t vlen; |
| | 927 | |
| | 928 | /* |
| | 929 | * In case we match, apply the appropriate flags added for the |
| | 930 | * equivalence mapping, based on the case of the first value |
| | 931 | * character we're testing. (If we don't match, we'll simply |
| | 932 | * return failure, so it won't matter that we messed with the |
| | 933 | * flags.) |
| | 934 | */ |
| | 935 | ret |= (t3_is_upper(valch) ? eq->uc_result_flags |
| | 936 | : eq->lc_result_flags); |
| | 937 | |
| | 938 | /* match each character from the mapping string */ |
| | 939 | for (vp = eq->val_ch, vlen = eq->val_ch_cnt ; |
| | 940 | vallen != 0 && vlen != 0 ; ++vp, --vlen) |
| | 941 | { |
| | 942 | /* get this character */ |
| | 943 | refch = *vp; |
| | 944 | |
| | 945 | /* if we have an exact match, keep going */ |
| | 946 | if (refch == valch) |
| | 947 | { |
| | 948 | /* matched - skip this character and keep going */ |
| | 949 | valp.inc(&vallen); |
| | 950 | ++valcharlen; |
| | 951 | continue; |
| | 952 | } |
| | 953 | |
| | 954 | /* check for a case-folded match if appropriate */ |
| | 955 | if (fold_case && t3_to_lower(valch) == t3_to_lower(refch)) |
| | 956 | { |
| | 957 | /* matched - skip this input character */ |
| | 958 | valp.inc(&vallen); |
| | 959 | ++valcharlen; |
| | 960 | |
| | 961 | /* note the case-folded match and keep going */ |
| | 962 | ret |= RF_CASEFOLD; |
| | 963 | continue; |
| | 964 | } |
| | 965 | |
| | 966 | /* no match */ |
| | 967 | return 0; |
| | 968 | } |
| | 969 | |
| | 970 | /* |
| | 971 | * if we make it here, we matched the equivalence mapping - |
| | 972 | * we've already skipped the input we matched, so simply keep |
| | 973 | * going |
| | 974 | */ |
| | 975 | continue; |
| | 976 | } |
| | 977 | |
| | 978 | /* we don't have anything else to try, so we don't have a match */ |
| | 979 | return 0; |
| | 980 | } |
| | 981 | |
| | 982 | /* |
| | 983 | * If we ran out of reference string before we ran out of value |
| | 984 | * string, we definitely do not have a match. If we ran out of value |
| | 985 | * string before we ran out reference string, we have a match as long |
| | 986 | * as we matched at least the truncation length. |
| | 987 | */ |
| | 988 | if (reflen == 0 && vallen == 0) |
| | 989 | { |
| | 990 | /* |
| | 991 | * We ran out of both at the same time - it's a match. Return the |
| | 992 | * result code up to this point OR'd with RF_MATCH, which is our |
| | 993 | * pre-defined bit that we set for every match. |
| | 994 | */ |
| | 995 | return (ret | RF_MATCH); |
| | 996 | } |
| | 997 | else if (vallen != 0) |
| | 998 | { |
| | 999 | /* we ran out of reference string first - it's not a match */ |
| | 1000 | return 0; |
| | 1001 | } |
| | 1002 | else |
| | 1003 | { |
| | 1004 | /* |
| | 1005 | * We ran out of value string first, so it's a truncated match if |
| | 1006 | * we matched at least up to the truncation length (assuming we |
| | 1007 | * allow truncation at all). If we didn't make it to the |
| | 1008 | * truncation length, or we don't allow truncation, it's not a |
| | 1009 | * match. |
| | 1010 | */ |
| | 1011 | if (ext->trunc_len != 0 && valcharlen >= ext->trunc_len) |
| | 1012 | { |
| | 1013 | /* |
| | 1014 | * it's a truncated match - return the result code up to this |
| | 1015 | * point, OR'd with RF_MATCH (our pre-defined bit we set for |
| | 1016 | * every match) and RF_TRUNC (our pre-defined bit we set for |
| | 1017 | * truncated matches) |
| | 1018 | */ |
| | 1019 | return (ret | RF_MATCH | RF_TRUNC); |
| | 1020 | } |
| | 1021 | else |
| | 1022 | { |
| | 1023 | /* didn't make it to the truncation length, so it's not a match */ |
| | 1024 | return 0; |
| | 1025 | } |
| | 1026 | } |
| | 1027 | } |
| | 1028 | |