usr/src/cmd/localedef/parser.y

   1 %{
   2 /*
   3  * This file and its contents are supplied under the terms of the
   4  * Common Development and Distribution License ("CDDL"), version 1.0.
   5  * You may only use this file in accordance with the terms of version
   6  * 1.0 of the CDDL.
   7  *
   8  * A full copy of the text of the CDDL should have accompanied this
   9  * source.  A copy of the CDDL is also available via the Internet at
  10  * http://www.illumos.org/license/CDDL.
  11  */
  12
  13 /*
  14  * Copyright 2010 Nexenta Systems, Inc.  All rights reserved.
  15  * Copyright 2013 DEY Storage Systems, Inc.
  16  */
  17
  18 /*
  19  * POSIX localedef grammar.
  20  */
  21
  22 #include <wchar.h>
  23 #include <stdio.h>
  24 #include <limits.h>
  25 #include "localedef.h"
  26
  27 %}
  28 %union {
  29         int             num;
  30         wchar_t         wc;
  31         char            *token;
  32         collsym_t       *collsym;
  33         collelem_t      *collelem;
  34 }
  35
  36 %token          T_CODE_SET
  37 %token          T_MB_CUR_MAX
  38 %token          T_MB_CUR_MIN
  39 %token          T_COM_CHAR
  40 %token          T_ESC_CHAR
  41 %token          T_LT
  42 %token          T_GT
  43 %token          T_NL
  44 %token          T_SEMI
  45 %token          T_COMMA
  46 %token          T_ELLIPSIS
  47 %token          T_RPAREN
  48 %token          T_LPAREN
  49 %token          T_QUOTE
  50 %token          T_NULL
  51 %token          T_WS
  52 %token          T_END
  53 %token          T_COPY
  54 %token          T_CHARMAP
  55 %token          T_WIDTH
  56 %token          T_CTYPE
  57 %token          T_ISUPPER
  58 %token          T_ISLOWER
  59 %token          T_ISALPHA
  60 %token          T_ISDIGIT
  61 %token          T_ISPUNCT
  62 %token          T_ISXDIGIT
  63 %token          T_ISSPACE
  64 %token          T_ISPRINT
  65 %token          T_ISGRAPH
  66 %token          T_ISBLANK
  67 %token          T_ISCNTRL
  68 %token          T_ISALNUM
  69 %token          T_ISSPECIAL
  70 %token          T_ISPHONOGRAM
  71 %token          T_ISIDEOGRAM
  72 %token          T_ISENGLISH
  73 %token          T_ISNUMBER
  74 %token          T_TOUPPER
  75 %token          T_TOLOWER
  76 %token          T_COLLATE
  77 %token          T_COLLATING_SYMBOL
  78 %token          T_COLLATING_ELEMENT
  79 %token          T_ORDER_START
  80 %token          T_ORDER_END
  81 %token          T_FORWARD
  82 %token          T_BACKWARD
  83 %token          T_POSITION
  84 %token          T_FROM
  85 %token          T_UNDEFINED
  86 %token          T_IGNORE
  87 %token          T_MESSAGES
  88 %token          T_YESSTR
  89 %token          T_NOSTR
  90 %token          T_YESEXPR
  91 %token          T_NOEXPR
  92 %token          T_MONETARY
  93 %token          T_INT_CURR_SYMBOL
  94 %token          T_CURRENCY_SYMBOL
  95 %token          T_MON_DECIMAL_POINT
  96 %token          T_MON_THOUSANDS_SEP
  97 %token          T_POSITIVE_SIGN
  98 %token          T_NEGATIVE_SIGN
  99 %token          T_MON_GROUPING
 100 %token          T_INT_FRAC_DIGITS
 101 %token          T_FRAC_DIGITS
 102 %token          T_P_CS_PRECEDES
 103 %token          T_P_SEP_BY_SPACE
 104 %token          T_N_CS_PRECEDES
 105 %token          T_N_SEP_BY_SPACE
 106 %token          T_P_SIGN_POSN
 107 %token          T_N_SIGN_POSN
 108 %token          T_INT_P_CS_PRECEDES
 109 %token          T_INT_N_CS_PRECEDES
 110 %token          T_INT_P_SEP_BY_SPACE
 111 %token          T_INT_N_SEP_BY_SPACE
 112 %token          T_INT_P_SIGN_POSN
 113 %token          T_INT_N_SIGN_POSN
 114 %token          T_NUMERIC
 115 %token          T_DECIMAL_POINT
 116 %token          T_THOUSANDS_SEP
 117 %token          T_GROUPING
 118 %token          T_TIME
 119 %token          T_ABDAY
 120 %token          T_DAY
 121 %token          T_ABMON
 122 %token          T_MON
 123 %token          T_ERA
 124 %token          T_ERA_D_FMT
 125 %token          T_ERA_T_FMT
 126 %token          T_ERA_D_T_FMT
 127 %token          T_ALT_DIGITS
 128 %token          T_D_T_FMT
 129 %token          T_D_FMT
 130 %token          T_T_FMT
 131 %token          T_AM_PM
 132 %token          T_T_FMT_AMPM
 133 %token          T_DATE_FMT
 134 %token  <wc>            T_CHAR
 135 %token  <token>         T_NAME
 136 %token  <num>           T_NUMBER
 137 %token  <token>         T_SYMBOL
 138 %token  <collsym>       T_COLLSYM
 139 %token  <collelem>      T_COLLELEM
 140
 141 %%
 142
 143 localedef       : setting_list categories
 144                 | categories
 145                 ;
 146
 147 string          : T_QUOTE charlist T_QUOTE
 148                 | T_QUOTE T_QUOTE
 149                 ;
 150
 151 charlist        : charlist T_CHAR
 152                 {
 153                         add_wcs($2);
 154                 }
 155                 | T_CHAR
 156                 {
 157                         add_wcs($1);
 158                 }
 159                 ;
 160
 161 setting_list    : setting_list setting
 162                 | setting
 163                 ;
 164
 165
 166 setting         : T_COM_CHAR T_CHAR T_NL
 167                 {
 168                         com_char = $2;
 169                 }
 170                 | T_ESC_CHAR T_CHAR T_NL
 171                 {
 172                         esc_char = $2;
 173                 }
 174                 | T_MB_CUR_MAX T_NUMBER T_NL
 175                 {
 176                         mb_cur_max = $2;
 177                 }
 178                 | T_MB_CUR_MIN T_NUMBER T_NL
 179                 {
 180                         mb_cur_min = $2;
 181                 }
 182                 | T_CODE_SET string T_NL
 183                 {
 184                         wchar_t *w = get_wcs();
 185                         set_wide_encoding(to_mb_string(w));
 186                         free(w);
 187                 }
 188                 | T_CODE_SET T_NAME T_NL
 189                 {
 190                         set_wide_encoding($2);
 191                 }
 192                 ;
 193
 194 copycat         : T_COPY T_NAME T_NL
 195                 {
 196                         copy_category($2);
 197                 }
 198                 | T_COPY string T_NL
 199                 {
 200                         wchar_t *w = get_wcs();
 201                         copy_category(to_mb_string(w));
 202                         free(w);
 203                 }
 204                 ;
 205
 206 categories      : categories category
 207                 | category
 208                 ;
 209
 210
 211 category        : charmap
 212                 | messages
 213                 | monetary
 214                 | ctype
 215                 | collate
 216                 | numeric
 217                 | time
 218                 ;
 219
 220
 221 charmap         : T_CHARMAP T_NL charmap_list T_END T_CHARMAP T_NL
 222                 | T_WIDTH T_NL width_list T_END T_WIDTH T_NL
 223                 ;
 224
 225
 226 charmap_list    : charmap_list charmap_entry
 227                 | charmap_entry
 228                 ;
 229
 230
 231 charmap_entry   : T_SYMBOL T_CHAR
 232                 {
 233                         add_charmap($1, $2);
 234                         scan_to_eol();
 235                 }
 236                 | T_SYMBOL T_ELLIPSIS T_SYMBOL T_CHAR
 237                 {
 238                         add_charmap_range($1, $3, $4);
 239                         scan_to_eol();
 240                 }
 241                 | T_NL
 242                 ;
 243
 244 width_list      : width_list width_entry
 245                 | width_entry
 246                 ;
 247
 248 width_entry     : T_CHAR T_NUMBER T_NL
 249                 {
 250                         add_width($1, $2);
 251                 }
 252                 | T_SYMBOL T_NUMBER T_NL
 253                 {
 254                         add_charmap_undefined($1);
 255                 }
 256                 | T_CHAR T_ELLIPSIS T_CHAR T_NUMBER T_NL
 257                 {
 258                         add_width_range($1, $3, $4);
 259                 }
 260                 | T_SYMBOL T_ELLIPSIS T_SYMBOL T_NUMBER T_NL
 261                 {
 262                         add_charmap_undefined($1);
 263                         add_charmap_undefined($3);
 264                 }
 265                 | T_CHAR T_ELLIPSIS T_SYMBOL T_NUMBER T_NL
 266                 {
 267                         add_width($1, $4);
 268                         add_charmap_undefined($3);
 269                 }
 270                 | T_SYMBOL T_ELLIPSIS T_CHAR T_NUMBER T_NL
 271                 {
 272                         add_width($3, $4);
 273                         add_charmap_undefined($1);
 274                 }
 275                 | T_NL
 276                 ;
 277
 278 ctype           : T_CTYPE T_NL ctype_list T_END T_CTYPE T_NL
 279                 {
 280                         dump_ctype();
 281                 }
 282                 | T_CTYPE T_NL copycat  T_END T_CTYPE T_NL
 283                 ;
 284
 285 ctype_list      : ctype_list ctype_kw
 286                 | ctype_kw
 287                 ;
 288
 289 ctype_kw        : T_ISUPPER cc_list T_NL
 290                 | T_ISLOWER cc_list T_NL
 291                 | T_ISALPHA cc_list T_NL
 292                 | T_ISDIGIT cc_list T_NL
 293                 | T_ISPUNCT cc_list T_NL
 294                 | T_ISXDIGIT cc_list T_NL
 295                 | T_ISSPACE cc_list T_NL
 296                 | T_ISPRINT cc_list T_NL
 297                 | T_ISGRAPH cc_list T_NL
 298                 | T_ISBLANK cc_list T_NL
 299                 | T_ISCNTRL cc_list T_NL
 300                 | T_ISALNUM cc_list T_NL
 301                 | T_ISSPECIAL cc_list T_NL
 302                 | T_ISENGLISH cc_list T_NL
 303                 | T_ISNUMBER cc_list T_NL
 304                 | T_ISIDEOGRAM cc_list T_NL
 305                 | T_ISPHONOGRAM cc_list T_NL
 306                 | T_TOUPPER conv_list T_NL
 307                 | T_TOLOWER conv_list T_NL
 308                 ;
 309
 310 cc_list         : cc_list T_SEMI cc_range_end
 311                 | cc_list T_SEMI cc_char
 312                 | cc_char
 313                 ;
 314
 315 cc_range_end    : T_ELLIPSIS T_SEMI T_CHAR
 316                 {
 317                         add_ctype_range($3);
 318                 }
 319                 ;
 320
 321 cc_char         : T_CHAR
 322                 {
 323                         add_ctype($1);
 324                 }
 325                 | T_SYMBOL
 326                 {
 327                         add_charmap_undefined($1);
 328                 }
 329                 ;
 330
 331 conv_list       : conv_list T_SEMI conv_pair
 332                 | conv_pair
 333                 ;
 334
 335
 336 conv_pair       : T_LPAREN T_CHAR T_COMMA T_CHAR T_RPAREN
 337                 {
 338                         add_caseconv($2, $4);
 339                 }
 340                 | T_LPAREN T_SYMBOL T_COMMA T_CHAR T_RPAREN
 341                 {
 342                         add_charmap_undefined($2);
 343                 }
 344                 | T_LPAREN T_SYMBOL T_COMMA T_SYMBOL T_RPAREN
 345                 {
 346                         add_charmap_undefined($2);
 347                         add_charmap_undefined($4);
 348                 }
 349                 | T_LPAREN T_CHAR T_COMMA T_SYMBOL T_RPAREN
 350                 {
 351                         add_charmap_undefined($4);
 352                 }
 353                 ;
 354
 355 collate         : T_COLLATE T_NL coll_order T_END T_COLLATE T_NL
 356                 {
 357                         dump_collate();
 358                 }
 359                 | T_COLLATE T_NL coll_optional coll_order T_END T_COLLATE T_NL
 360                 {
 361                         dump_collate();
 362                 }
 363                 | T_COLLATE T_NL copycat T_END T_COLLATE T_NL
 364                 ;
 365
 366
 367 coll_optional   : coll_optional coll_symbols
 368                 | coll_optional coll_elements
 369                 | coll_symbols
 370                 | coll_elements
 371                 ;
 372
 373
 374 coll_symbols    : T_COLLATING_SYMBOL T_SYMBOL T_NL
 375                 {
 376                         define_collsym($2);
 377                 }
 378                 ;
 379
 380
 381 coll_elements   : T_COLLATING_ELEMENT T_SYMBOL T_FROM string T_NL
 382                 {
 383                         define_collelem($2, get_wcs());
 384                 }
 385                 ;
 386
 387 coll_order      : T_ORDER_START T_NL order_list T_ORDER_END T_NL
 388                 {
 389                         /* If no order list supplied default to one forward */
 390                         add_order_bit(T_FORWARD);
 391                         add_order_directive();
 392                 }
 393                 | T_ORDER_START order_args T_NL order_list T_ORDER_END T_NL
 394                 ;
 395
 396
 397 order_args      : order_args T_SEMI order_arg
 398                 {
 399                         add_order_directive();
 400                 }
 401                 | order_arg
 402                 {
 403                         add_order_directive();
 404                 }
 405                 ;
 406
 407 order_arg       : order_arg T_COMMA order_dir
 408                 | order_dir
 409                 ;
 410
 411 order_dir       : T_FORWARD
 412                 {
 413                         add_order_bit(T_FORWARD);
 414                 }
 415                 | T_BACKWARD
 416                 {
 417                         add_order_bit(T_BACKWARD);
 418                 }
 419                 | T_POSITION
 420                 {
 421                         add_order_bit(T_POSITION);
 422                 }
 423                 ;
 424
 425 order_list      : order_list order_item
 426                 | order_item
 427                 ;
 428
 429 order_item      : T_COLLSYM T_NL
 430                 {
 431                         end_order_collsym($1);
 432                 }
 433                 | order_itemkw T_NL
 434                 {
 435                         end_order();
 436                 }
 437                 | order_itemkw order_weights T_NL
 438                 {
 439                         end_order();
 440                 }
 441                 ;
 442
 443 order_itemkw    : T_CHAR
 444                 {
 445                         start_order_char($1);
 446                 }
 447                 | T_ELLIPSIS
 448                 {
 449                         start_order_ellipsis();
 450                 }
 451                 | T_COLLELEM
 452                 {
 453                         start_order_collelem($1);
 454                 }
 455                 | T_UNDEFINED
 456                 {
 457                         start_order_undefined();
 458                 }
 459                 | T_SYMBOL
 460                 {
 461                         start_order_symbol($1);
 462                 }
 463                 ;
 464
 465 order_weights   : order_weights T_SEMI order_weight
 466                 | order_weights T_SEMI
 467                 | order_weight
 468                 ;
 469
 470 order_weight    : T_COLLELEM
 471                 {
 472                         add_order_collelem($1);
 473                 }
 474                 | T_COLLSYM
 475                 {
 476                         add_order_collsym($1);
 477                 }
 478                 | T_CHAR
 479                 {
 480                         add_order_char($1);
 481                 }
 482                 | T_ELLIPSIS
 483                 {
 484                         add_order_ellipsis();
 485                 }
 486                 | T_IGNORE
 487                 {
 488                         add_order_ignore();
 489                 }
 490                 | T_SYMBOL
 491                 {
 492                         add_order_symbol($1);
 493                 }
 494                 | T_QUOTE order_str T_QUOTE
 495                 {
 496                         add_order_subst();
 497                 }
 498                 ;
 499
 500 order_str       : order_str order_stritem
 501                 | order_stritem
 502                 ;
 503
 504 order_stritem   : T_CHAR
 505                 {
 506                         add_subst_char($1);
 507                 }
 508                 | T_COLLSYM
 509                 {
 510                         add_subst_collsym($1);
 511                 }
 512                 | T_COLLELEM
 513                 {
 514                         add_subst_collelem($1);
 515                 }
 516                 | T_SYMBOL
 517                 {
 518                         add_subst_symbol($1);
 519                 }
 520                 ;
 521
 522 messages        : T_MESSAGES T_NL messages_list T_END T_MESSAGES T_NL
 523                 {
 524                         dump_messages();
 525                 }
 526                 | T_MESSAGES T_NL copycat T_END T_MESSAGES T_NL
 527                 ;
 528
 529 messages_list   : messages_list messages_item
 530                 | messages_item
 531                 ;
 532
 533 messages_kw     : T_YESSTR
 534                 | T_NOSTR
 535                 | T_YESEXPR
 536                 | T_NOEXPR
 537                 ;
 538
 539 messages_item   : messages_kw string T_NL
 540                 {
 541                         add_message(get_wcs());
 542                 }
 543                 ;
 544
 545 monetary        : T_MONETARY T_NL monetary_list T_END T_MONETARY T_NL
 546                 {
 547                         dump_monetary();
 548                 }
 549                 | T_MONETARY T_NL copycat T_END T_MONETARY T_NL
 550                 ;
 551
 552 monetary_list   : monetary_list monetary_kw
 553                 | monetary_kw
 554                 ;
 555
 556 monetary_strkw  : T_INT_CURR_SYMBOL
 557                 | T_CURRENCY_SYMBOL
 558                 | T_MON_DECIMAL_POINT
 559                 | T_MON_THOUSANDS_SEP
 560                 | T_POSITIVE_SIGN
 561                 | T_NEGATIVE_SIGN
 562                 ;
 563
 564 monetary_numkw  : T_INT_FRAC_DIGITS
 565                 | T_FRAC_DIGITS
 566                 | T_P_CS_PRECEDES
 567                 | T_P_SEP_BY_SPACE
 568                 | T_N_CS_PRECEDES
 569                 | T_N_SEP_BY_SPACE
 570                 | T_P_SIGN_POSN
 571                 | T_N_SIGN_POSN
 572                 | T_INT_P_CS_PRECEDES
 573                 | T_INT_N_CS_PRECEDES
 574                 | T_INT_P_SEP_BY_SPACE
 575                 | T_INT_N_SEP_BY_SPACE
 576                 | T_INT_P_SIGN_POSN
 577                 | T_INT_N_SIGN_POSN
 578                 ;
 579
 580 monetary_kw     : monetary_strkw string T_NL
 581                 {
 582                         add_monetary_str(get_wcs());
 583                 }
 584                 | monetary_numkw T_NUMBER T_NL
 585                 {
 586                         add_monetary_num($2);
 587                 }
 588                 | T_MON_GROUPING mon_group_list T_NL
 589                 ;
 590
 591 mon_group_list  : T_NUMBER
 592                 {
 593                         reset_monetary_group();
 594                         add_monetary_group($1);
 595                 }
 596                 | mon_group_list T_SEMI T_NUMBER
 597                 {
 598                         add_monetary_group($3);
 599                 }
 600                 ;
 601
 602
 603 numeric         : T_NUMERIC T_NL numeric_list T_END T_NUMERIC T_NL
 604                 {
 605                         dump_numeric();
 606                 }
 607                 | T_NUMERIC T_NL copycat T_END T_NUMERIC T_NL
 608                 ;
 609
 610
 611 numeric_list    : numeric_list numeric_item
 612                 | numeric_item
 613                 ;
 614
 615
 616 numeric_item    : numeric_strkw string T_NL
 617                 {
 618                         add_numeric_str(get_wcs());
 619                 }
 620                 | T_GROUPING group_list T_NL
 621                 ;
 622
 623 numeric_strkw   : T_DECIMAL_POINT
 624                 | T_THOUSANDS_SEP
 625                 ;
 626
 627
 628 group_list      : T_NUMBER
 629                 {
 630                         reset_numeric_group();
 631                         add_numeric_group($1);
 632                 }
 633                 | group_list T_SEMI T_NUMBER
 634                 {
 635                         add_numeric_group($3);
 636                 }
 637                 ;
 638
 639
 640 time            : T_TIME T_NL time_kwlist T_END T_TIME T_NL
 641                 {
 642                         dump_time();
 643                 }
 644                 | T_TIME T_NL copycat T_END T_NUMERIC T_NL
 645                 ;
 646
 647 time_kwlist     : time_kwlist time_kw
 648                 | time_kw
 649                 ;
 650
 651 time_kw         : time_strkw string T_NL
 652                 {
 653                         add_time_str(get_wcs());
 654                 }
 655                 | time_listkw time_list T_NL
 656                 {
 657                         check_time_list();
 658                 }
 659                 ;
 660
 661 time_listkw     : T_ABDAY
 662                 | T_DAY
 663                 | T_ABMON
 664                 | T_MON
 665                 | T_ERA
 666                 | T_ALT_DIGITS
 667                 | T_AM_PM
 668                 ;
 669
 670 time_strkw      : T_ERA_D_T_FMT
 671                 | T_ERA_T_FMT
 672                 | T_ERA_D_FMT
 673                 | T_D_T_FMT
 674                 | T_D_FMT
 675                 | T_T_FMT
 676                 | T_T_FMT_AMPM
 677                 | T_DATE_FMT
 678                 ;
 679
 680 time_list       : time_list T_SEMI string
 681                 {
 682                         add_time_list(get_wcs());
 683                 }
 684                 | string
 685                 {
 686                         reset_time_list();
 687                         add_time_list(get_wcs());
 688                 }
 689                 ;