src/scan-gram.l

   1 /* Bison Grammar Scanner                             -*- C -*-
   2
   3    Copyright (C) 2002-2012 Free Software Foundation, Inc.
   4
   5    This file is part of Bison, the GNU Compiler Compiler.
   6
   7    This program is free software: you can redistribute it and/or modify
   8    it under the terms of the GNU General Public License as published by
   9    the Free Software Foundation, either version 3 of the License, or
  10    (at your option) any later version.
  11
  12    This program is distributed in the hope that it will be useful,
  13    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15    GNU General Public License for more details.
  16
  17    You should have received a copy of the GNU General Public License
  18    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
  19
  20 %option debug nodefault noinput nounput noyywrap never-interactive
  21 %option prefix="gram_" outfile="lex.yy.c"
  22
  23 %{
  24 /* Work around a bug in flex 2.5.31.  See Debian bug 333231
  25    <http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=333231>.  */
  26 #undef gram_wrap
  27 #define gram_wrap() 1
  28
  29 #define FLEX_PREFIX(Id) gram_ ## Id
  30 #include <src/flex-scanner.h>
  31
  32 #include <src/complain.h>
  33 #include <src/files.h>
  34 #include <src/gram.h>
  35 #include <quotearg.h>
  36 #include <src/reader.h>
  37 #include <src/uniqstr.h>
  38
  39 #include <ctype.h>
  40 #include <mbswidth.h>
  41 #include <quote.h>
  42 #include <streq.h>
  43
  44 #include <src/scan-gram.h>
  45
  46 #define YY_DECL GRAM_LEX_DECL
  47
  48 #define YY_USER_INIT                                    \
  49    code_start = scanner_cursor = loc->start;            \
  50
  51 /* Location of scanner cursor.  */
  52 static boundary scanner_cursor;
  53
  54 #define YY_USER_ACTION  location_compute (loc, &scanner_cursor, yytext, yyleng);
  55
  56 static size_t no_cr_read (FILE *, char *, size_t);
  57 #define YY_INPUT(buf, result, size) ((result) = no_cr_read (yyin, buf, size))
  58
  59 #define RETURN_PERCENT_PARAM(Value)                     \
  60   RETURN_VALUE(PERCENT_PARAM, param, param_ ## Value)
  61
  62 #define RETURN_PERCENT_FLAG(Value)                              \
  63   RETURN_VALUE(PERCENT_FLAG, uniqstr, uniqstr_new (Value))
  64
  65 #define RETURN_VALUE(Token, Field, Value)       \
  66   do {                                          \
  67     val->Field = Value;                         \
  68     return Token;                               \
  69   } while (0)
  70
  71 #define ROLLBACK_CURRENT_TOKEN                                  \
  72   do {                                                          \
  73     scanner_cursor.column -= mbsnwidth (yytext, yyleng, 0);     \
  74     yyless (0);                                                 \
  75   } while (0)
  76
  77 /* A string representing the most recently saved token.  */
  78 static char *last_string;
  79
  80 /* Bracketed identifier. */
  81 static uniqstr bracketed_id_str = 0;
  82 static location bracketed_id_loc;
  83 static boundary bracketed_id_start;
  84 static int bracketed_id_context_state = 0;
  85
  86 void
  87 gram_scanner_last_string_free (void)
  88 {
  89   STRING_FREE;
  90 }
  91
  92 static void handle_syncline (char *, location);
  93 static unsigned long int scan_integer (char const *p, int base, location loc);
  94 static int convert_ucn_to_byte (char const *hex_text);
  95 static void unexpected_eof (boundary, char const *);
  96 static void unexpected_newline (boundary, char const *);
  97
  98 %}
  99  /* A C-like comment in directives/rules. */
 100 %x SC_YACC_COMMENT
 101  /* Strings and characters in directives/rules. */
 102 %x SC_ESCAPED_STRING SC_ESCAPED_CHARACTER
 103  /* A identifier was just read in directives/rules.  Special state
 104     to capture the sequence 'identifier :'. */
 105 %x SC_AFTER_IDENTIFIER
 106  /* A complex tag, with nested angles brackets. */
 107 %x SC_TAG
 108
 109  /* Four types of user code:
 110     - prologue (code between '%{' '%}' in the first section, before %%);
 111     - actions, printers, union, etc, (between braced in the middle section);
 112     - epilogue (everything after the second %%).
 113     - predicate (code between '%?{' and '{' in middle section); */
 114 %x SC_PROLOGUE SC_BRACED_CODE SC_EPILOGUE SC_PREDICATE
 115  /* C and C++ comments in code. */
 116 %x SC_COMMENT SC_LINE_COMMENT
 117  /* Strings and characters in code. */
 118 %x SC_STRING SC_CHARACTER
 119  /* Bracketed identifiers support. */
 120 %x SC_BRACKETED_ID SC_RETURN_BRACKETED_ID
 121
 122 letter    [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
 123 id        {letter}({letter}|[-0-9])*
 124 directive %{id}
 125 int       [0-9]+
 126
 127 /* POSIX says that a tag must be both an id and a C union member, but
 128    historically almost any character is allowed in a tag.  We disallow
 129    NUL, as this simplifies our implementation.  We disallow angle
 130    bracket to match them in nested pairs: several languages use them
 131    for generics/template types.  */
 132 tag      [^\0<>]+
 133
 134 /* Zero or more instances of backslash-newline.  Following GCC, allow
 135    white space between the backslash and the newline.  */
 136 splice   (\\[ \f\t\v]*\n)*
 137
 138 %%
 139 %{
 140   /* Nesting level.  Either for nested braces, or nested angle brackets
 141      (but not mixed).  */
 142   int nesting PACIFY_CC (= 0);
 143
 144   /* Parent context state, when applicable.  */
 145   int context_state PACIFY_CC (= 0);
 146
 147   /* Location of most recent identifier, when applicable.  */
 148   location id_loc PACIFY_CC (= empty_location);
 149
 150   /* Where containing code started, when applicable.  Its initial
 151      value is relevant only when yylex is invoked in the SC_EPILOGUE
 152      start condition.  */
 153   boundary code_start = scanner_cursor;
 154
 155   /* Where containing comment or string or character literal started,
 156      when applicable.  */
 157   boundary token_start PACIFY_CC (= scanner_cursor);
 158 %}
 159
 160
 161   /*-----------------------.
 162   | Scanning white space.  |
 163   `-----------------------*/
 164
 165 <INITIAL,SC_AFTER_IDENTIFIER,SC_BRACKETED_ID,SC_RETURN_BRACKETED_ID>
 166 {
 167   /* Comments and white space.  */
 168   ","          warn_at (*loc, _("stray ',' treated as white space"));
 169   [ \f\n\t\v]  |
 170   "//".*       ;
 171   "/*" {
 172     token_start = loc->start;
 173     context_state = YY_START;
 174     BEGIN SC_YACC_COMMENT;
 175   }
 176
 177   /* #line directives are not documented, and may be withdrawn or
 178      modified in future versions of Bison.  */
 179   ^"#line "{int}" \"".*"\"\n" {
 180     handle_syncline (yytext + sizeof "#line " - 1, *loc);
 181   }
 182 }
 183
 184
 185   /*----------------------------.
 186   | Scanning Bison directives.  |
 187   `----------------------------*/
 188
 189   /* For directives that are also command line options, the regex must be
 190         "%..."
 191      after "[-_]"s are removed, and the directive must match the --long
 192      option name, with a single string argument.  Otherwise, add exceptions
 193      to ../build-aux/cross-options.pl.  */
 194
 195 <INITIAL>
 196 {
 197   "%binary"                         return PERCENT_NONASSOC;
 198   "%code"                           return PERCENT_CODE;
 199   "%debug"                          RETURN_PERCENT_FLAG("parse.trace");
 200   "%default"[-_]"prec"              return PERCENT_DEFAULT_PREC;
 201   "%define"                         return PERCENT_DEFINE;
 202   "%defines"                        return PERCENT_DEFINES;
 203   "%destructor"                     return PERCENT_DESTRUCTOR;
 204   "%dprec"                          return PERCENT_DPREC;
 205   "%error"[-_]"verbose"             return PERCENT_ERROR_VERBOSE;
 206   "%expect"                         return PERCENT_EXPECT;
 207   "%expect"[-_]"rr"                 return PERCENT_EXPECT_RR;
 208   "%file-prefix"                    return PERCENT_FILE_PREFIX;
 209   "%fixed"[-_]"output"[-_]"files"   return PERCENT_YACC;
 210   "%initial-action"                 return PERCENT_INITIAL_ACTION;
 211   "%glr-parser"                     return PERCENT_GLR_PARSER;
 212   "%language"                       return PERCENT_LANGUAGE;
 213   "%left"                           return PERCENT_LEFT;
 214   "%lex-param"                      RETURN_PERCENT_PARAM(lex);
 215   "%locations"                      RETURN_PERCENT_FLAG("locations");
 216   "%merge"                          return PERCENT_MERGE;
 217   "%name"[-_]"prefix"               return PERCENT_NAME_PREFIX;
 218   "%no"[-_]"default"[-_]"prec"      return PERCENT_NO_DEFAULT_PREC;
 219   "%no"[-_]"lines"                  return PERCENT_NO_LINES;
 220   "%nonassoc"                       return PERCENT_NONASSOC;
 221   "%nondeterministic-parser"        return PERCENT_NONDETERMINISTIC_PARSER;
 222   "%nterm"                          return PERCENT_NTERM;
 223   "%output"                         return PERCENT_OUTPUT;
 224   "%param"                          RETURN_PERCENT_PARAM(both);
 225   "%parse-param"                    RETURN_PERCENT_PARAM(parse);
 226   "%prec"                           return PERCENT_PREC;
 227   "%precedence"                     return PERCENT_PRECEDENCE;
 228   "%printer"                        return PERCENT_PRINTER;
 229   "%pure"[-_]"parser"               RETURN_PERCENT_FLAG("api.pure");
 230   "%require"                        return PERCENT_REQUIRE;
 231   "%right"                          return PERCENT_RIGHT;
 232   "%skeleton"                       return PERCENT_SKELETON;
 233   "%start"                          return PERCENT_START;
 234   "%term"                           return PERCENT_TOKEN;
 235   "%token"                          return PERCENT_TOKEN;
 236   "%token"[-_]"table"               return PERCENT_TOKEN_TABLE;
 237   "%type"                           return PERCENT_TYPE;
 238   "%union"                          return PERCENT_UNION;
 239   "%verbose"                        return PERCENT_VERBOSE;
 240   "%yacc"                           return PERCENT_YACC;
 241
 242   {directive} {
 243     complain_at (*loc, _("invalid directive: %s"), quote (yytext));
 244   }
 245
 246   "="                     return EQUAL;
 247   "|"                     return PIPE;
 248   ";"                     return SEMICOLON;
 249
 250   {id} {
 251     val->uniqstr = uniqstr_new (yytext);
 252     id_loc = *loc;
 253     bracketed_id_str = NULL;
 254     BEGIN SC_AFTER_IDENTIFIER;
 255   }
 256
 257   {int} {
 258     val->integer = scan_integer (yytext, 10, *loc);
 259     return INT;
 260   }
 261   0[xX][0-9abcdefABCDEF]+ {
 262     val->integer = scan_integer (yytext, 16, *loc);
 263     return INT;
 264   }
 265
 266   /* Identifiers may not start with a digit.  Yet, don't silently
 267      accept "1FOO" as "1 FOO".  */
 268   {int}{id} {
 269     complain_at (*loc, _("invalid identifier: %s"), quote (yytext));
 270   }
 271
 272   /* Characters.  */
 273   "'"         token_start = loc->start; BEGIN SC_ESCAPED_CHARACTER;
 274
 275   /* Strings. */
 276   "\""        token_start = loc->start; BEGIN SC_ESCAPED_STRING;
 277
 278   /* Prologue. */
 279   "%{"        code_start = loc->start; BEGIN SC_PROLOGUE;
 280
 281   /* Code in between braces.  */
 282   "{" {
 283     STRING_GROW;
 284     nesting = 0;
 285     code_start = loc->start;
 286     BEGIN SC_BRACED_CODE;
 287   }
 288
 289   /* Semantic predicate. */
 290   "%?"[ \f\n\t\v]*"{" {
 291     nesting = 0;
 292     code_start = loc->start;
 293     BEGIN SC_PREDICATE;
 294   }
 295
 296   /* A type. */
 297   "<*>"       return TAG_ANY;
 298   "<>"        return TAG_NONE;
 299   "<"{tag}">" {
 300     obstack_grow (&obstack_for_string, yytext + 1, yyleng - 2);
 301     STRING_FINISH;
 302     val->uniqstr = uniqstr_new (last_string);
 303     STRING_FREE;
 304     return TAG;
 305   }
 306   "<"         {
 307     nesting = 0;
 308     token_start = loc->start;
 309     BEGIN SC_TAG;
 310   }
 311
 312   "%%" {
 313     static int percent_percent_count;
 314     if (++percent_percent_count == 2)
 315       BEGIN SC_EPILOGUE;
 316     return PERCENT_PERCENT;
 317   }
 318
 319   "[" {
 320     bracketed_id_str = NULL;
 321     bracketed_id_start = loc->start;
 322     bracketed_id_context_state = YY_START;
 323     BEGIN SC_BRACKETED_ID;
 324   }
 325
 326   . {
 327     complain_at (*loc, _("invalid character: %s"), quote (yytext));
 328   }
 329
 330   <<EOF>> {
 331     loc->start = loc->end = scanner_cursor;
 332     yyterminate ();
 333   }
 334 }
 335
 336
 337   /*--------------------------------------------------------------.
 338   | Supporting \0 complexifies our implementation for no expected |
 339   | added value.                                                  |
 340   `--------------------------------------------------------------*/
 341
 342 <SC_ESCAPED_CHARACTER,SC_ESCAPED_STRING,SC_TAG>
 343 {
 344   \0        complain_at (*loc, _("invalid null character"));
 345 }
 346
 347
 348   /*-----------------------------------------------------------------.
 349   | Scanning after an identifier, checking whether a colon is next.  |
 350   `-----------------------------------------------------------------*/
 351
 352 <SC_AFTER_IDENTIFIER>
 353 {
 354   "[" {
 355     if (bracketed_id_str)
 356       {
 357         ROLLBACK_CURRENT_TOKEN;
 358         BEGIN SC_RETURN_BRACKETED_ID;
 359         *loc = id_loc;
 360         return ID;
 361       }
 362     else
 363       {
 364         bracketed_id_start = loc->start;
 365         bracketed_id_context_state = YY_START;
 366         BEGIN SC_BRACKETED_ID;
 367       }
 368   }
 369   ":" {
 370     BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL);
 371     *loc = id_loc;
 372     return ID_COLON;
 373   }
 374   . {
 375     ROLLBACK_CURRENT_TOKEN;
 376     BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL);
 377     *loc = id_loc;
 378     return ID;
 379   }
 380   <<EOF>> {
 381     BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL);
 382     *loc = id_loc;
 383     return ID;
 384   }
 385 }
 386
 387   /*--------------------------------.
 388   | Scanning bracketed identifiers. |
 389   `--------------------------------*/
 390
 391 <SC_BRACKETED_ID>
 392 {
 393   {id} {
 394     if (bracketed_id_str)
 395       {
 396         complain_at (*loc, _("unexpected identifier in bracketed name: %s"),
 397                      quote (yytext));
 398       }
 399     else
 400       {
 401         bracketed_id_str = uniqstr_new (yytext);
 402         bracketed_id_loc = *loc;
 403       }
 404   }
 405   "]" {
 406     BEGIN bracketed_id_context_state;
 407     if (bracketed_id_str)
 408       {
 409         if (INITIAL == bracketed_id_context_state)
 410           {
 411             val->uniqstr = bracketed_id_str;
 412             bracketed_id_str = 0;
 413             *loc = bracketed_id_loc;
 414             return BRACKETED_ID;
 415           }
 416       }
 417     else
 418       complain_at (*loc, _("an identifier expected"));
 419   }
 420   . {
 421     complain_at (*loc, _("invalid character in bracketed name: %s"),
 422                  quote (yytext));
 423   }
 424   <<EOF>> {
 425     BEGIN bracketed_id_context_state;
 426     unexpected_eof (bracketed_id_start, "]");
 427   }
 428 }
 429
 430 <SC_RETURN_BRACKETED_ID>
 431 {
 432   . {
 433     ROLLBACK_CURRENT_TOKEN;
 434     val->uniqstr = bracketed_id_str;
 435     bracketed_id_str = 0;
 436     *loc = bracketed_id_loc;
 437     BEGIN INITIAL;
 438     return BRACKETED_ID;
 439   }
 440 }
 441
 442
 443   /*---------------------------------------------------------------.
 444   | Scanning a Yacc comment.  The initial '/ *' is already eaten.  |
 445   `---------------------------------------------------------------*/
 446
 447 <SC_YACC_COMMENT>
 448 {
 449   "*/"     BEGIN context_state;
 450   .|\n     ;
 451   <<EOF>>  unexpected_eof (token_start, "*/"); BEGIN context_state;
 452 }
 453
 454
 455   /*------------------------------------------------------------.
 456   | Scanning a C comment.  The initial '/ *' is already eaten.  |
 457   `------------------------------------------------------------*/
 458
 459 <SC_COMMENT>
 460 {
 461   "*"{splice}"/"  STRING_GROW; BEGIN context_state;
 462   <<EOF>>         unexpected_eof (token_start, "*/"); BEGIN context_state;
 463 }
 464
 465
 466   /*--------------------------------------------------------------.
 467   | Scanning a line comment.  The initial '//' is already eaten.  |
 468   `--------------------------------------------------------------*/
 469
 470 <SC_LINE_COMMENT>
 471 {
 472   "\n"           STRING_GROW; BEGIN context_state;
 473   {splice}       STRING_GROW;
 474   <<EOF>>        BEGIN context_state;
 475 }
 476
 477
 478   /*------------------------------------------------.
 479   | Scanning a Bison string, including its escapes. |
 480   | The initial quote is already eaten.             |
 481   `------------------------------------------------*/
 482
 483 <SC_ESCAPED_STRING>
 484 {
 485   "\""|"\n" {
 486     if (yytext[0] == '\n')
 487       unexpected_newline (token_start, "\"");
 488     STRING_FINISH;
 489     loc->start = token_start;
 490     val->chars = last_string;
 491     BEGIN INITIAL;
 492     return STRING;
 493   }
 494   <<EOF>> {
 495     unexpected_eof (token_start, "\"");
 496     STRING_FINISH;
 497     loc->start = token_start;
 498     val->chars = last_string;
 499     BEGIN INITIAL;
 500     return STRING;
 501   }
 502 }
 503
 504   /*----------------------------------------------------------.
 505   | Scanning a Bison character literal, decoding its escapes. |
 506   | The initial quote is already eaten.                       |
 507   `----------------------------------------------------------*/
 508
 509 <SC_ESCAPED_CHARACTER>
 510 {
 511   "'"|"\n" {
 512     STRING_FINISH;
 513     loc->start = token_start;
 514     val->character = last_string[0];
 515     {
 516       /* FIXME: Eventually, make these errors.  */
 517       if (last_string[0] == '\0')
 518         {
 519           warn_at (*loc, _("empty character literal"));
 520           /* '\0' seems dangerous even if we are about to complain.  */
 521           val->character = '\'';
 522         }
 523       else if (last_string[1] != '\0')
 524         warn_at (*loc, _("extra characters in character literal"));
 525     }
 526     if (yytext[0] == '\n')
 527       unexpected_newline (token_start, "'");
 528     STRING_FREE;
 529     BEGIN INITIAL;
 530     return CHAR;
 531   }
 532   <<EOF>> {
 533     STRING_FINISH;
 534     loc->start = token_start;
 535     val->character = last_string[0];
 536     {
 537       /* FIXME: Eventually, make these errors.  */
 538       if (last_string[0] == '\0')
 539         {
 540           warn_at (*loc, _("empty character literal"));
 541           /* '\0' seems dangerous even if we are about to complain.  */
 542           val->character = '\'';
 543         }
 544       else if (last_string[1] != '\0')
 545         warn_at (*loc, _("extra characters in character literal"));
 546     }
 547     unexpected_eof (token_start, "'");
 548     STRING_FREE;
 549     BEGIN INITIAL;
 550     return CHAR;
 551   }
 552 }
 553
 554   /*-----------------------------------------------------------.
 555   | Scanning a Bison nested tag.  The initial angle bracket is |
 556   | already eaten.                                             |
 557   `-----------------------------------------------------------*/
 558
 559 <SC_TAG>
 560 {
 561   ">" {
 562     --nesting;
 563     if (nesting < 0)
 564       {
 565         STRING_FINISH;
 566         loc->start = token_start;
 567         val->uniqstr = uniqstr_new (last_string);
 568         STRING_FREE;
 569         BEGIN INITIAL;
 570         return TAG;
 571       }
 572     STRING_GROW;
 573   }
 574
 575   [^<>]+ STRING_GROW;
 576   "<"+   STRING_GROW; nesting += yyleng;
 577
 578   <<EOF>> {
 579     unexpected_eof (token_start, ">");
 580     STRING_FINISH;
 581     loc->start = token_start;
 582     val->uniqstr = uniqstr_new (last_string);
 583     STRING_FREE;
 584     BEGIN INITIAL;
 585     return TAG;
 586   }
 587 }
 588
 589   /*----------------------------.
 590   | Decode escaped characters.  |
 591   `----------------------------*/
 592
 593 <SC_ESCAPED_STRING,SC_ESCAPED_CHARACTER>
 594 {
 595   \\[0-7]{1,3} {
 596     unsigned long int c = strtoul (yytext + 1, NULL, 8);
 597     if (!c || UCHAR_MAX < c)
 598       complain_at (*loc, _("invalid number after \\-escape: %s"),
 599                    yytext+1);
 600     else
 601       obstack_1grow (&obstack_for_string, c);
 602   }
 603
 604   \\x[0-9abcdefABCDEF]+ {
 605     verify (UCHAR_MAX < ULONG_MAX);
 606     unsigned long int c = strtoul (yytext + 2, NULL, 16);
 607     if (!c || UCHAR_MAX < c)
 608       complain_at (*loc, _("invalid number after \\-escape: %s"),
 609                    yytext+1);
 610     else
 611       obstack_1grow (&obstack_for_string, c);
 612   }
 613
 614   \\a   obstack_1grow (&obstack_for_string, '\a');
 615   \\b   obstack_1grow (&obstack_for_string, '\b');
 616   \\f   obstack_1grow (&obstack_for_string, '\f');
 617   \\n   obstack_1grow (&obstack_for_string, '\n');
 618   \\r   obstack_1grow (&obstack_for_string, '\r');
 619   \\t   obstack_1grow (&obstack_for_string, '\t');
 620   \\v   obstack_1grow (&obstack_for_string, '\v');
 621
 622   /* \\[\"\'?\\] would be shorter, but it confuses xgettext.  */
 623   \\("\""|"'"|"?"|"\\")  obstack_1grow (&obstack_for_string, yytext[1]);
 624
 625   \\(u|U[0-9abcdefABCDEF]{4})[0-9abcdefABCDEF]{4} {
 626     int c = convert_ucn_to_byte (yytext);
 627     if (c <= 0)
 628       complain_at (*loc, _("invalid number after \\-escape: %s"),
 629                    yytext+1);
 630     else
 631       obstack_1grow (&obstack_for_string, c);
 632   }
 633   \\(.|\n)      {
 634     char const *p = yytext + 1;
 635     /* Quote only if escaping won't make the character visible.  */
 636     if (isspace ((unsigned char) *p) && isprint ((unsigned char) *p))
 637       p = quote (p);
 638     else
 639       p = quotearg_style_mem (escape_quoting_style, p, 1);
 640     complain_at (*loc, _("invalid character after \\-escape: %s"), p);
 641   }
 642 }
 643
 644   /*--------------------------------------------.
 645   | Scanning user-code characters and strings.  |
 646   `--------------------------------------------*/
 647
 648 <SC_CHARACTER,SC_STRING>
 649 {
 650   {splice}|\\{splice}[^\n\[\]]  STRING_GROW;
 651 }
 652
 653 <SC_CHARACTER>
 654 {
 655   "'"           STRING_GROW; BEGIN context_state;
 656   \n            unexpected_newline (token_start, "'"); BEGIN context_state;
 657   <<EOF>>       unexpected_eof (token_start, "'"); BEGIN context_state;
 658 }
 659
 660 <SC_STRING>
 661 {
 662   "\""          STRING_GROW; BEGIN context_state;
 663   \n            unexpected_newline (token_start, "\""); BEGIN context_state;
 664   <<EOF>>       unexpected_eof (token_start, "\""); BEGIN context_state;
 665 }
 666
 667
 668   /*---------------------------------------------------.
 669   | Strings, comments etc. can be found in user code.  |
 670   `---------------------------------------------------*/
 671
 672 <SC_BRACED_CODE,SC_PROLOGUE,SC_EPILOGUE,SC_PREDICATE>
 673 {
 674   "'" {
 675     STRING_GROW;
 676     context_state = YY_START;
 677     token_start = loc->start;
 678     BEGIN SC_CHARACTER;
 679   }
 680   "\"" {
 681     STRING_GROW;
 682     context_state = YY_START;
 683     token_start = loc->start;
 684     BEGIN SC_STRING;
 685   }
 686   "/"{splice}"*" {
 687     STRING_GROW;
 688     context_state = YY_START;
 689     token_start = loc->start;
 690     BEGIN SC_COMMENT;
 691   }
 692   "/"{splice}"/" {
 693     STRING_GROW;
 694     context_state = YY_START;
 695     BEGIN SC_LINE_COMMENT;
 696   }
 697 }
 698
 699
 700
 701   /*-----------------------------------------------------------.
 702   | Scanning some code in braces (actions, predicates). The    |
 703   | initial "{" is already eaten.                              |
 704   `-----------------------------------------------------------*/
 705
 706 <SC_BRACED_CODE,SC_PREDICATE>
 707 {
 708   "{"|"<"{splice}"%"  STRING_GROW; nesting++;
 709   "%"{splice}">"      STRING_GROW; nesting--;
 710
 711   /* Tokenize '<<%' correctly (as '<<' '%') rather than incorrrectly
 712      (as '<' '<%').  */
 713   "<"{splice}"<"  STRING_GROW;
 714
 715   <<EOF>> {
 716     int token = (YY_START == SC_BRACED_CODE) ? BRACED_CODE : BRACED_PREDICATE;
 717     unexpected_eof (code_start, "}");
 718     STRING_FINISH;
 719     loc->start = code_start;
 720     val->code = last_string;
 721     BEGIN INITIAL;
 722     return token;
 723   }
 724 }
 725
 726 <SC_BRACED_CODE>
 727 {
 728   "}" {
 729     obstack_1grow (&obstack_for_string, '}');
 730
 731     --nesting;
 732     if (nesting < 0)
 733       {
 734         STRING_FINISH;
 735         loc->start = code_start;
 736         val->code = last_string;
 737         BEGIN INITIAL;
 738         return BRACED_CODE;
 739       }
 740   }
 741 }
 742
 743 <SC_PREDICATE>
 744 {
 745   "}" {
 746     --nesting;
 747     if (nesting < 0)
 748       {
 749         STRING_FINISH;
 750         loc->start = code_start;
 751         val->code = last_string;
 752         BEGIN INITIAL;
 753         return BRACED_PREDICATE;
 754       }
 755     else
 756       obstack_1grow (&obstack_for_string, '}');
 757   }
 758 }
 759
 760   /*--------------------------------------------------------------.
 761   | Scanning some prologue: from "%{" (already scanned) to "%}".  |
 762   `--------------------------------------------------------------*/
 763
 764 <SC_PROLOGUE>
 765 {
 766   "%}" {
 767     STRING_FINISH;
 768     loc->start = code_start;
 769     val->chars = last_string;
 770     BEGIN INITIAL;
 771     return PROLOGUE;
 772   }
 773
 774   <<EOF>> {
 775     unexpected_eof (code_start, "%}");
 776     STRING_FINISH;
 777     loc->start = code_start;
 778     val->chars = last_string;
 779     BEGIN INITIAL;
 780     return PROLOGUE;
 781   }
 782 }
 783
 784
 785   /*---------------------------------------------------------------.
 786   | Scanning the epilogue (everything after the second "%%", which |
 787   | has already been eaten).                                       |
 788   `---------------------------------------------------------------*/
 789
 790 <SC_EPILOGUE>
 791 {
 792   <<EOF>> {
 793     STRING_FINISH;
 794     loc->start = code_start;
 795     val->chars = last_string;
 796     BEGIN INITIAL;
 797     return EPILOGUE;
 798   }
 799 }
 800
 801
 802   /*-----------------------------------------------------.
 803   | By default, grow the string obstack with the input.  |
 804   `-----------------------------------------------------*/
 805
 806 <SC_COMMENT,SC_LINE_COMMENT,SC_BRACED_CODE,SC_PREDICATE,SC_PROLOGUE,SC_EPILOGUE,SC_STRING,SC_CHARACTER,SC_ESCAPED_STRING,SC_ESCAPED_CHARACTER>. |
 807   <SC_COMMENT,SC_LINE_COMMENT,SC_BRACED_CODE,SC_PREDICATE,SC_PROLOGUE,SC_EPILOGUE>\n    STRING_GROW;
 808
 809 %%
 810
 811 /* Read bytes from FP into buffer BUF of size SIZE.  Return the
 812    number of bytes read.  Remove '\r' from input, treating \r\n
 813    and isolated \r as \n.  */
 814
 815 static size_t
 816 no_cr_read (FILE *fp, char *buf, size_t size)
 817 {
 818   size_t bytes_read = fread (buf, 1, size, fp);
 819   if (bytes_read)
 820     {
 821       char *w = memchr (buf, '\r', bytes_read);
 822       if (w)
 823         {
 824           char const *r = ++w;
 825           char const *lim = buf + bytes_read;
 826
 827           for (;;)
 828             {
 829               /* Found an '\r'.  Treat it like '\n', but ignore any
 830                  '\n' that immediately follows.  */
 831               w[-1] = '\n';
 832               if (r == lim)
 833                 {
 834                   int ch = getc (fp);
 835                   if (ch != '\n' && ungetc (ch, fp) != ch)
 836                     break;
 837                 }
 838               else if (*r == '\n')
 839                 r++;
 840
 841               /* Copy until the next '\r'.  */
 842               do
 843                 {
 844                   if (r == lim)
 845                     return w - buf;
 846                 }
 847               while ((*w++ = *r++) != '\r');
 848             }
 849
 850           return w - buf;
 851         }
 852     }
 853
 854   return bytes_read;
 855 }
 856
 857
 858
 859 /*------------------------------------------------------.
 860 | Scan NUMBER for a base-BASE integer at location LOC.  |
 861 `------------------------------------------------------*/
 862
 863 static unsigned long int
 864 scan_integer (char const *number, int base, location loc)
 865 {
 866   verify (INT_MAX < ULONG_MAX);
 867   unsigned long int num = strtoul (number, NULL, base);
 868
 869   if (INT_MAX < num)
 870     {
 871       complain_at (loc, _("integer out of range: %s"), quote (number));
 872       num = INT_MAX;
 873     }
 874
 875   return num;
 876 }
 877
 878
 879 /*------------------------------------------------------------------.
 880 | Convert universal character name UCN to a single-byte character,  |
 881 | and return that character.  Return -1 if UCN does not correspond  |
 882 | to a single-byte character.                                       |
 883 `------------------------------------------------------------------*/
 884
 885 static int
 886 convert_ucn_to_byte (char const *ucn)
 887 {
 888   verify (UCHAR_MAX <= INT_MAX);
 889   unsigned long int code = strtoul (ucn + 2, NULL, 16);
 890
 891   /* FIXME: Currently we assume Unicode-compatible unibyte characters
 892      on ASCII hosts (i.e., Latin-1 on hosts with 8-bit bytes).  On
 893      non-ASCII hosts we support only the portable C character set.
 894      These limitations should be removed once we add support for
 895      multibyte characters.  */
 896
 897   if (UCHAR_MAX < code)
 898     return -1;
 899
 900 #if ! ('$' == 0x24 && '@' == 0x40 && '`' == 0x60 && '~' == 0x7e)
 901   {
 902     /* A non-ASCII host.  Use CODE to index into a table of the C
 903        basic execution character set, which is guaranteed to exist on
 904        all Standard C platforms.  This table also includes '$', '@',
 905        and '`', which are not in the basic execution character set but
 906        which are unibyte characters on all the platforms that we know
 907        about.  */
 908     static signed char const table[] =
 909       {
 910         '\0',   -1,   -1,   -1,   -1,   -1,   -1, '\a',
 911         '\b', '\t', '\n', '\v', '\f', '\r',   -1,   -1,
 912           -1,   -1,   -1,   -1,   -1,   -1,   -1,   -1,
 913           -1,   -1,   -1,   -1,   -1,   -1,   -1,   -1,
 914          ' ',  '!',  '"',  '#',  '$',  '%',  '&', '\'',
 915          '(',  ')',  '*',  '+',  ',',  '-',  '.',  '/',
 916          '0',  '1',  '2',  '3',  '4',  '5',  '6',  '7',
 917          '8',  '9',  ':',  ';',  '<',  '=',  '>',  '?',
 918          '@',  'A',  'B',  'C',  'D',  'E',  'F',  'G',
 919          'H',  'I',  'J',  'K',  'L',  'M',  'N',  'O',
 920          'P',  'Q',  'R',  'S',  'T',  'U',  'V',  'W',
 921          'X',  'Y',  'Z',  '[', '\\',  ']',  '^',  '_',
 922          '`',  'a',  'b',  'c',  'd',  'e',  'f',  'g',
 923          'h',  'i',  'j',  'k',  'l',  'm',  'n',  'o',
 924          'p',  'q',  'r',  's',  't',  'u',  'v',  'w',
 925          'x',  'y',  'z',  '{',  '|',  '}',  '~'
 926       };
 927
 928     code = code < sizeof table ? table[code] : -1;
 929   }
 930 #endif
 931
 932   return code;
 933 }
 934
 935
 936 /*----------------------------------------------------------------.
 937 | Handle '#line INT "FILE"'.  ARGS has already skipped '#line '.  |
 938 `----------------------------------------------------------------*/
 939
 940 static void
 941 handle_syncline (char *args, location loc)
 942 {
 943   char *after_num;
 944   unsigned long int lineno = strtoul (args, &after_num, 10);
 945   char *file = mbschr (after_num, '"') + 1;
 946   *mbschr (file, '"') = '\0';
 947   if (INT_MAX <= lineno)
 948     {
 949       warn_at (loc, _("line number overflow"));
 950       lineno = INT_MAX;
 951     }
 952   current_file = uniqstr_new (file);
 953   boundary_set (&scanner_cursor, current_file, lineno, 1);
 954 }
 955
 956
 957 /*----------------------------------------------------------------.
 958 | For a token or comment starting at START, report message MSGID, |
 959 | which should say that an end marker was found before            |
 960 | the expected TOKEN_END.                                         |
 961 `----------------------------------------------------------------*/
 962
 963 static void
 964 unexpected_end (boundary start, char const *msgid, char const *token_end)
 965 {
 966   location loc;
 967   loc.start = start;
 968   loc.end = scanner_cursor;
 969   token_end = quote (token_end);
 970   // Instead of '\'', display "'".
 971   if (STREQ (token_end, "'\\''", '\'', '\\', '\'', '\'', 0,0,0,0,0))
 972     token_end = "\"'\"";
 973   complain_at (loc, _(msgid), token_end);
 974 }
 975
 976
 977 /*------------------------------------------------------------------------.
 978 | Report an unexpected EOF in a token or comment starting at START.       |
 979 | An end of file was encountered and the expected TOKEN_END was missing.  |
 980 `------------------------------------------------------------------------*/
 981
 982 static void
 983 unexpected_eof (boundary start, char const *token_end)
 984 {
 985   unexpected_end (start, N_("missing %s at end of file"), token_end);
 986 }
 987
 988
 989 /*----------------------------------------.
 990 | Likewise, but for unexpected newlines.  |
 991 `----------------------------------------*/
 992
 993 static void
 994 unexpected_newline (boundary start, char const *token_end)
 995 {
 996   unexpected_end (start, N_("missing %s at end of line"), token_end);
 997 }
 998
 999
1000 /*-------------------------.
1001 | Initialize the scanner.  |
1002 `-------------------------*/
1003
1004 void
1005 gram_scanner_initialize (void)
1006 {
1007   obstack_init (&obstack_for_string);
1008 }
1009
1010
1011 /*-----------------------------------------------.
1012 | Free all the memory allocated to the scanner.  |
1013 `-----------------------------------------------*/
1014
1015 void
1016 gram_scanner_free (void)
1017 {
1018   obstack_free (&obstack_for_string, 0);
1019   /* Reclaim Flex's buffers.  */
1020   yylex_destroy ();
1021 }