src/scan-gram.l

   1 /* Bison Grammar Scanner                             -*- C -*-
   2
   3    Copyright (C) 2002-2011 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
  43 #include <src/scan-gram.h>
  44
  45 #define YY_DECL GRAM_LEX_DECL
  46
  47 #define YY_USER_INIT                                    \
  48    code_start = scanner_cursor = loc->start;            \
  49
  50 /* Location of scanner cursor.  */
  51 static boundary scanner_cursor;
  52
  53 #define YY_USER_ACTION  location_compute (loc, &scanner_cursor, yytext, yyleng);
  54
  55 static size_t no_cr_read (FILE *, char *, size_t);
  56 #define YY_INPUT(buf, result, size) ((result) = no_cr_read (yyin, buf, size))
  57
  58 #define RETURN_PERCENT_PARAM(Value)                     \
  59   RETURN_VALUE(PERCENT_PARAM, param, param_ ## Value)
  60
  61 #define RETURN_PERCENT_FLAG(Value)                              \
  62   RETURN_VALUE(PERCENT_FLAG, uniqstr, uniqstr_new (Value))
  63
  64 #define RETURN_VALUE(Token, Field, Value)       \
  65   do {                                          \
  66     val->Field = Value;                         \
  67     return Token;                               \
  68   } while (0)
  69
  70 #define ROLLBACK_CURRENT_TOKEN                                  \
  71   do {                                                          \
  72     scanner_cursor.column -= mbsnwidth (yytext, yyleng, 0);     \
  73     yyless (0);                                                 \
  74   } while (0)
  75
  76 /* A string representing the most recently saved token.  */
  77 static char *last_string;
  78
  79 /* Bracketed identifier. */
  80 static uniqstr bracketed_id_str = 0;
  81 static location bracketed_id_loc;
  82 static boundary bracketed_id_start;
  83 static int bracketed_id_context_state = 0;
  84
  85 void
  86 gram_scanner_last_string_free (void)
  87 {
  88   STRING_FREE;
  89 }
  90
  91 static void handle_syncline (char *, location);
  92 static unsigned long int scan_integer (char const *p, int base, location loc);
  93 static int convert_ucn_to_byte (char const *hex_text);
  94 static void unexpected_eof (boundary, char const *);
  95 static void unexpected_newline (boundary, char const *);
  96
  97 %}
  98  /* A C-like comment in directives/rules. */
  99 %x SC_YACC_COMMENT
 100  /* Strings and characters in directives/rules. */
 101 %x SC_ESCAPED_STRING SC_ESCAPED_CHARACTER
 102  /* A identifier was just read in directives/rules.  Special state
 103     to capture the sequence `identifier :'. */
 104 %x SC_AFTER_IDENTIFIER
 105  /* A complex tag, with nested angles brackets. */
 106 %x SC_TAG
 107
 108  /* Four types of user code:
 109     - prologue (code between `%{' `%}' in the first section, before %%);
 110     - actions, printers, union, etc, (between braced in the middle section);
 111     - epilogue (everything after the second %%).
 112     - predicate (code between `%?{' and `{' in middle section); */
 113 %x SC_PROLOGUE SC_BRACED_CODE SC_EPILOGUE SC_PREDICATE
 114  /* C and C++ comments in code. */
 115 %x SC_COMMENT SC_LINE_COMMENT
 116  /* Strings and characters in code. */
 117 %x SC_STRING SC_CHARACTER
 118  /* Bracketed identifiers support. */
 119 %x SC_BRACKETED_ID SC_RETURN_BRACKETED_ID
 120
 121 letter    [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
 122 id        {letter}({letter}|[-0-9])*
 123 directive %{id}
 124 int       [0-9]+
 125
 126 /* POSIX says that a tag must be both an id and a C union member, but
 127    historically almost any character is allowed in a tag.  We disallow
 128    NUL, as this simplifies our implementation.  We disallow angle
 129    bracket to match them in nested pairs: several languages use them
 130    for generics/template types.  */
 131 tag      [^\0<>]+
 132
 133 /* Zero or more instances of backslash-newline.  Following GCC, allow
 134    white space between the backslash and the newline.  */
 135 splice   (\\[ \f\t\v]*\n)*
 136
 137 %%
 138 %{
 139   /* Nesting level.  Either for nested braces, or nested angle brackets
 140      (but not mixed).  */
 141   int nesting IF_LINT (= 0);
 142
 143   /* Parent context state, when applicable.  */
 144   int context_state IF_LINT (= 0);
 145
 146   /* Location of most recent identifier, when applicable.  */
 147   location id_loc IF_LINT (= empty_location);
 148
 149   /* Where containing code started, when applicable.  Its initial
 150      value is relevant only when yylex is invoked in the SC_EPILOGUE
 151      start condition.  */
 152   boundary code_start = scanner_cursor;
 153
 154   /* Where containing comment or string or character literal started,
 155      when applicable.  */
 156   boundary token_start IF_LINT (= scanner_cursor);
 157 %}
 158
 159
 160   /*-----------------------.
 161   | Scanning white space.  |
 162   `-----------------------*/
 163
 164 <INITIAL,SC_AFTER_IDENTIFIER,SC_BRACKETED_ID,SC_RETURN_BRACKETED_ID>
 165 {
 166   /* Comments and white space.  */
 167   ","          warn_at (*loc, _("stray `,' treated as white space"));
 168   [ \f\n\t\v]  |
 169   "//".*       ;
 170   "/*" {
 171     token_start = loc->start;
 172     context_state = YY_START;
 173     BEGIN SC_YACC_COMMENT;
 174   }
 175
 176   /* #line directives are not documented, and may be withdrawn or
 177      modified in future versions of Bison.  */
 178   ^"#line "{int}" \"".*"\"\n" {
 179     handle_syncline (yytext + sizeof "#line " - 1, *loc);
 180   }
 181 }
 182
 183
 184   /*----------------------------.
 185   | Scanning Bison directives.  |
 186   `----------------------------*/
 187
 188   /* For directives that are also command line options, the regex must be
 189         "%..."
 190      after "[-_]"s are removed, and the directive must match the --long
 191      option name, with a single string argument.  Otherwise, add exceptions
 192      to ../build-aux/cross-options.pl.  */
 193
 194 <INITIAL>
 195 {
 196   "%binary"                         return PERCENT_NONASSOC;
 197   "%code"                           return PERCENT_CODE;
 198   "%debug"                          RETURN_PERCENT_FLAG("parse.trace");
 199   "%default"[-_]"prec"              return PERCENT_DEFAULT_PREC;
 200   "%define"                         return PERCENT_DEFINE;
 201   "%defines"                        return PERCENT_DEFINES;
 202   "%destructor"                     return PERCENT_DESTRUCTOR;
 203   "%dprec"                          return PERCENT_DPREC;
 204   "%error"[-_]"verbose"             return PERCENT_ERROR_VERBOSE;
 205   "%expect"                         return PERCENT_EXPECT;
 206   "%expect"[-_]"rr"                 return PERCENT_EXPECT_RR;
 207   "%file-prefix"                    return PERCENT_FILE_PREFIX;
 208   "%fixed"[-_]"output"[-_]"files"   return PERCENT_YACC;
 209   "%initial-action"                 return PERCENT_INITIAL_ACTION;
 210   "%glr-parser"                     return PERCENT_GLR_PARSER;
 211   "%language"                       return PERCENT_LANGUAGE;
 212   "%left"                           return PERCENT_LEFT;
 213   "%lex-param"                      RETURN_PERCENT_PARAM(lex);
 214   "%locations"                      RETURN_PERCENT_FLAG("locations");
 215   "%merge"                          return PERCENT_MERGE;
 216   "%name"[-_]"prefix"               return PERCENT_NAME_PREFIX;
 217   "%no"[-_]"default"[-_]"prec"      return PERCENT_NO_DEFAULT_PREC;
 218   "%no"[-_]"lines"                  return PERCENT_NO_LINES;
 219   "%nonassoc"                       return PERCENT_NONASSOC;
 220   "%nondeterministic-parser"        return PERCENT_NONDETERMINISTIC_PARSER;
 221   "%nterm"                          return PERCENT_NTERM;
 222   "%output"                         return PERCENT_OUTPUT;
 223   "%param"                          RETURN_PERCENT_PARAM(both);
 224   "%parse-param"                    RETURN_PERCENT_PARAM(parse);
 225   "%prec"                           return PERCENT_PREC;
 226   "%precedence"                     return PERCENT_PRECEDENCE;
 227   "%printer"                        return PERCENT_PRINTER;
 228   "%pure"[-_]"parser"               RETURN_PERCENT_FLAG("api.pure");
 229   "%require"                        return PERCENT_REQUIRE;
 230   "%right"                          return PERCENT_RIGHT;
 231   "%skeleton"                       return PERCENT_SKELETON;
 232   "%start"                          return PERCENT_START;
 233   "%term"                           return PERCENT_TOKEN;
 234   "%token"                          return PERCENT_TOKEN;
 235   "%token"[-_]"table"               return PERCENT_TOKEN_TABLE;
 236   "%type"                           return PERCENT_TYPE;
 237   "%union"                          return PERCENT_UNION;
 238   "%verbose"                        return PERCENT_VERBOSE;
 239   "%yacc"                           return PERCENT_YACC;
 240
 241   {directive} {
 242     complain_at (*loc, _("invalid directive: %s"), quote (yytext));
 243   }
 244
 245   "="                     return EQUAL;
 246   "|"                     return PIPE;
 247   ";"                     return SEMICOLON;
 248
 249   {id} {
 250     val->uniqstr = uniqstr_new (yytext);
 251     id_loc = *loc;
 252     bracketed_id_str = NULL;
 253     BEGIN SC_AFTER_IDENTIFIER;
 254   }
 255
 256   {int} {
 257     val->integer = scan_integer (yytext, 10, *loc);
 258     return INT;
 259   }
 260   0[xX][0-9abcdefABCDEF]+ {
 261     val->integer = scan_integer (yytext, 16, *loc);
 262     return INT;
 263   }
 264
 265   /* Identifiers may not start with a digit.  Yet, don't silently
 266      accept "1FOO" as "1 FOO".  */
 267   {int}{id} {
 268     complain_at (*loc, _("invalid identifier: %s"), quote (yytext));
 269   }
 270
 271   /* Characters.  */
 272   "'"         token_start = loc->start; BEGIN SC_ESCAPED_CHARACTER;
 273
 274   /* Strings. */
 275   "\""        token_start = loc->start; BEGIN SC_ESCAPED_STRING;
 276
 277   /* Prologue. */
 278   "%{"        code_start = loc->start; BEGIN SC_PROLOGUE;
 279
 280   /* Code in between braces.  */
 281   "{" {
 282     STRING_GROW;
 283     nesting = 0;
 284     code_start = loc->start;
 285     BEGIN SC_BRACED_CODE;
 286   }
 287
 288   /* Semantic predicate. */
 289   "%?"[ \f\n\t\v]*"{" {
 290     nesting = 0;
 291     code_start = loc->start;
 292     BEGIN SC_PREDICATE;
 293   }
 294
 295   /* A type. */
 296   "<*>"       return TAG_ANY;
 297   "<>"        return TAG_NONE;
 298   "<"{tag}">" {
 299     obstack_grow (&obstack_for_string, yytext + 1, yyleng - 2);
 300     STRING_FINISH;
 301     val->uniqstr = uniqstr_new (last_string);
 302     STRING_FREE;
 303     return TAG;
 304   }
 305   "<"         {
 306     nesting = 0;
 307     token_start = loc->start;
 308     BEGIN SC_TAG;
 309   }
 310
 311   "%%" {
 312     static int percent_percent_count;
 313     if (++percent_percent_count == 2)
 314       BEGIN SC_EPILOGUE;
 315     return PERCENT_PERCENT;
 316   }
 317
 318   "[" {
 319     bracketed_id_str = NULL;
 320     bracketed_id_start = loc->start;
 321     bracketed_id_context_state = YY_START;
 322     BEGIN SC_BRACKETED_ID;
 323   }
 324
 325   . {
 326     complain_at (*loc, _("invalid character: %s"), quote (yytext));
 327   }
 328
 329   <<EOF>> {
 330     loc->start = loc->end = scanner_cursor;
 331     yyterminate ();
 332   }
 333 }
 334
 335
 336   /*--------------------------------------------------------------.
 337   | Supporting \0 complexifies our implementation for no expected |
 338   | added value.                                                  |
 339   `--------------------------------------------------------------*/
 340
 341 <SC_ESCAPED_CHARACTER,SC_ESCAPED_STRING,SC_TAG>
 342 {
 343   \0        complain_at (*loc, _("invalid null character"));
 344 }
 345
 346
 347   /*-----------------------------------------------------------------.
 348   | Scanning after an identifier, checking whether a colon is next.  |
 349   `-----------------------------------------------------------------*/
 350
 351 <SC_AFTER_IDENTIFIER>
 352 {
 353   "[" {
 354     if (bracketed_id_str)
 355       {
 356         ROLLBACK_CURRENT_TOKEN;
 357         BEGIN SC_RETURN_BRACKETED_ID;
 358         *loc = id_loc;
 359         return ID;
 360       }
 361     else
 362       {
 363         bracketed_id_start = loc->start;
 364         bracketed_id_context_state = YY_START;
 365         BEGIN SC_BRACKETED_ID;
 366       }
 367   }
 368   ":" {
 369     BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL);
 370     *loc = id_loc;
 371     return ID_COLON;
 372   }
 373   . {
 374     ROLLBACK_CURRENT_TOKEN;
 375     BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL);
 376     *loc = id_loc;
 377     return ID;
 378   }
 379   <<EOF>> {
 380     BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL);
 381     *loc = id_loc;
 382     return ID;
 383   }
 384 }
 385
 386   /*--------------------------------.
 387   | Scanning bracketed identifiers. |
 388   `--------------------------------*/
 389
 390 <SC_BRACKETED_ID>
 391 {
 392   {id} {
 393     if (bracketed_id_str)
 394       {
 395         complain_at (*loc, _("unexpected identifier in bracketed name: %s"),
 396                      quote (yytext));
 397       }
 398     else
 399       {
 400         bracketed_id_str = uniqstr_new (yytext);
 401         bracketed_id_loc = *loc;
 402       }
 403   }
 404   "]" {
 405     BEGIN bracketed_id_context_state;
 406     if (bracketed_id_str)
 407       {
 408         if (INITIAL == bracketed_id_context_state)
 409           {
 410             val->uniqstr = bracketed_id_str;
 411             bracketed_id_str = 0;
 412             *loc = bracketed_id_loc;
 413             return BRACKETED_ID;
 414           }
 415       }
 416     else
 417       complain_at (*loc, _("an identifier expected"));
 418   }
 419   . {
 420     complain_at (*loc, _("invalid character in bracketed name: %s"),
 421                  quote (yytext));
 422   }
 423   <<EOF>> {
 424     BEGIN bracketed_id_context_state;
 425     unexpected_eof (bracketed_id_start, "]");
 426   }
 427 }
 428
 429 <SC_RETURN_BRACKETED_ID>
 430 {
 431   . {
 432     ROLLBACK_CURRENT_TOKEN;
 433     val->uniqstr = bracketed_id_str;
 434     bracketed_id_str = 0;
 435     *loc = bracketed_id_loc;
 436     BEGIN INITIAL;
 437     return BRACKETED_ID;
 438   }
 439 }
 440
 441
 442   /*---------------------------------------------------------------.
 443   | Scanning a Yacc comment.  The initial `/ *' is already eaten.  |
 444   `---------------------------------------------------------------*/
 445
 446 <SC_YACC_COMMENT>
 447 {
 448   "*/"     BEGIN context_state;
 449   .|\n     ;
 450   <<EOF>>  unexpected_eof (token_start, "*/"); BEGIN context_state;
 451 }
 452
 453
 454   /*------------------------------------------------------------.
 455   | Scanning a C comment.  The initial `/ *' is already eaten.  |
 456   `------------------------------------------------------------*/
 457
 458 <SC_COMMENT>
 459 {
 460   "*"{splice}"/"  STRING_GROW; BEGIN context_state;
 461   <<EOF>>         unexpected_eof (token_start, "*/"); BEGIN context_state;
 462 }
 463
 464
 465   /*--------------------------------------------------------------.
 466   | Scanning a line comment.  The initial `//' is already eaten.  |
 467   `--------------------------------------------------------------*/
 468
 469 <SC_LINE_COMMENT>
 470 {
 471   "\n"           STRING_GROW; BEGIN context_state;
 472   {splice}       STRING_GROW;
 473   <<EOF>>        BEGIN context_state;
 474 }
 475
 476
 477   /*------------------------------------------------.
 478   | Scanning a Bison string, including its escapes. |
 479   | The initial quote is already eaten.             |
 480   `------------------------------------------------*/
 481
 482 <SC_ESCAPED_STRING>
 483 {
 484   "\""|"\n" {
 485     if (yytext[0] == '\n')
 486       unexpected_newline (token_start, "\"");
 487     STRING_FINISH;
 488     loc->start = token_start;
 489     val->chars = last_string;
 490     BEGIN INITIAL;
 491     return STRING;
 492   }
 493   <<EOF>> {
 494     unexpected_eof (token_start, "\"");
 495     STRING_FINISH;
 496     loc->start = token_start;
 497     val->chars = last_string;
 498     BEGIN INITIAL;
 499     return STRING;
 500   }
 501 }
 502
 503   /*----------------------------------------------------------.
 504   | Scanning a Bison character literal, decoding its escapes. |
 505   | The initial quote is already eaten.                       |
 506   `----------------------------------------------------------*/
 507
 508 <SC_ESCAPED_CHARACTER>
 509 {
 510   "'"|"\n" {
 511     STRING_FINISH;
 512     loc->start = token_start;
 513     val->character = last_string[0];
 514     {
 515       /* FIXME: Eventually, make these errors.  */
 516       if (last_string[0] == '\0')
 517         {
 518           warn_at (*loc, _("empty character literal"));
 519           /* '\0' seems dangerous even if we are about to complain.  */
 520           val->character = '\'';
 521         }
 522       else if (last_string[1] != '\0')
 523         warn_at (*loc, _("extra characters in character literal"));
 524     }
 525     if (yytext[0] == '\n')
 526       unexpected_newline (token_start, "'");
 527     STRING_FREE;
 528     BEGIN INITIAL;
 529     return CHAR;
 530   }
 531   <<EOF>> {
 532     STRING_FINISH;
 533     loc->start = token_start;
 534     val->character = last_string[0];
 535     {
 536       /* FIXME: Eventually, make these errors.  */
 537       if (last_string[0] == '\0')
 538         {
 539           warn_at (*loc, _("empty character literal"));
 540           /* '\0' seems dangerous even if we are about to complain.  */
 541           val->character = '\'';
 542         }
 543       else if (last_string[1] != '\0')
 544         warn_at (*loc, _("extra characters in character literal"));
 545     }
 546     unexpected_eof (token_start, "'");
 547     STRING_FREE;
 548     BEGIN INITIAL;
 549     return CHAR;
 550   }
 551 }
 552
 553   /*-----------------------------------------------------------.
 554   | Scanning a Bison nested tag.  The initial angle bracket is |
 555   | already eaten.                                             |
 556   `-----------------------------------------------------------*/
 557
 558 <SC_TAG>
 559 {
 560   ">" {
 561     --nesting;
 562     if (nesting < 0)
 563       {
 564         STRING_FINISH;
 565         loc->start = token_start;
 566         val->uniqstr = uniqstr_new (last_string);
 567         STRING_FREE;
 568         BEGIN INITIAL;
 569         return TAG;
 570       }
 571     STRING_GROW;
 572   }
 573
 574   [^<>]+ STRING_GROW;
 575   "<"+   STRING_GROW; nesting += yyleng;
 576
 577   <<EOF>> {
 578     unexpected_eof (token_start, ">");
 579     STRING_FINISH;
 580     loc->start = token_start;
 581     val->uniqstr = uniqstr_new (last_string);
 582     STRING_FREE;
 583     BEGIN INITIAL;
 584     return TAG;
 585   }
 586 }
 587
 588   /*----------------------------.
 589   | Decode escaped characters.  |
 590   `----------------------------*/
 591
 592 <SC_ESCAPED_STRING,SC_ESCAPED_CHARACTER>
 593 {
 594   \\[0-7]{1,3} {
 595     unsigned long int c = strtoul (yytext + 1, NULL, 8);
 596     if (!c || UCHAR_MAX < c)
 597       complain_at (*loc, _("invalid number after \\-escape: %s"),
 598                    yytext+1);
 599     else
 600       obstack_1grow (&obstack_for_string, c);
 601   }
 602
 603   \\x[0-9abcdefABCDEF]+ {
 604     verify (UCHAR_MAX < ULONG_MAX);
 605     unsigned long int c = strtoul (yytext + 2, NULL, 16);
 606     if (!c || UCHAR_MAX < c)
 607       complain_at (*loc, _("invalid number after \\-escape: %s"),
 608                    yytext+1);
 609     else
 610       obstack_1grow (&obstack_for_string, c);
 611   }
 612
 613   \\a   obstack_1grow (&obstack_for_string, '\a');
 614   \\b   obstack_1grow (&obstack_for_string, '\b');
 615   \\f   obstack_1grow (&obstack_for_string, '\f');
 616   \\n   obstack_1grow (&obstack_for_string, '\n');
 617   \\r   obstack_1grow (&obstack_for_string, '\r');
 618   \\t   obstack_1grow (&obstack_for_string, '\t');
 619   \\v   obstack_1grow (&obstack_for_string, '\v');
 620
 621   /* \\[\"\'?\\] would be shorter, but it confuses xgettext.  */
 622   \\("\""|"'"|"?"|"\\")  obstack_1grow (&obstack_for_string, yytext[1]);
 623
 624   \\(u|U[0-9abcdefABCDEF]{4})[0-9abcdefABCDEF]{4} {
 625     int c = convert_ucn_to_byte (yytext);
 626     if (c <= 0)
 627       complain_at (*loc, _("invalid number after \\-escape: %s"),
 628                    yytext+1);
 629     else
 630       obstack_1grow (&obstack_for_string, c);
 631   }
 632   \\(.|\n)      {
 633     char const *p = yytext + 1;
 634     /* Quote only if escaping won't make the character visible.  */
 635     if (isspace ((unsigned char) *p) && isprint ((unsigned char) *p))
 636       p = quote (p);
 637     else
 638       p = quotearg_style_mem (escape_quoting_style, p, 1);
 639     complain_at (*loc, _("invalid character after \\-escape: %s"), p);
 640   }
 641 }
 642
 643   /*--------------------------------------------.
 644   | Scanning user-code characters and strings.  |
 645   `--------------------------------------------*/
 646
 647 <SC_CHARACTER,SC_STRING>
 648 {
 649   {splice}|\\{splice}[^\n\[\]]  STRING_GROW;
 650 }
 651
 652 <SC_CHARACTER>
 653 {
 654   "'"           STRING_GROW; BEGIN context_state;
 655   \n            unexpected_newline (token_start, "'"); BEGIN context_state;
 656   <<EOF>>       unexpected_eof (token_start, "'"); BEGIN context_state;
 657 }
 658
 659 <SC_STRING>
 660 {
 661   "\""          STRING_GROW; BEGIN context_state;
 662   \n            unexpected_newline (token_start, "\""); BEGIN context_state;
 663   <<EOF>>       unexpected_eof (token_start, "\""); BEGIN context_state;
 664 }
 665
 666
 667   /*---------------------------------------------------.
 668   | Strings, comments etc. can be found in user code.  |
 669   `---------------------------------------------------*/
 670
 671 <SC_BRACED_CODE,SC_PROLOGUE,SC_EPILOGUE,SC_PREDICATE>
 672 {
 673   "'" {
 674     STRING_GROW;
 675     context_state = YY_START;
 676     token_start = loc->start;
 677     BEGIN SC_CHARACTER;
 678   }
 679   "\"" {
 680     STRING_GROW;
 681     context_state = YY_START;
 682     token_start = loc->start;
 683     BEGIN SC_STRING;
 684   }
 685   "/"{splice}"*" {
 686     STRING_GROW;
 687     context_state = YY_START;
 688     token_start = loc->start;
 689     BEGIN SC_COMMENT;
 690   }
 691   "/"{splice}"/" {
 692     STRING_GROW;
 693     context_state = YY_START;
 694     BEGIN SC_LINE_COMMENT;
 695   }
 696 }
 697
 698
 699
 700   /*-----------------------------------------------------------.
 701   | Scanning some code in braces (actions, predicates). The    |
 702   | initial "{" is already eaten.                              |
 703   `-----------------------------------------------------------*/
 704
 705 <SC_BRACED_CODE,SC_PREDICATE>
 706 {
 707   "{"|"<"{splice}"%"  STRING_GROW; nesting++;
 708   "%"{splice}">"      STRING_GROW; nesting--;
 709
 710   /* Tokenize `<<%' correctly (as `<<' `%') rather than incorrrectly
 711      (as `<' `<%').  */
 712   "<"{splice}"<"  STRING_GROW;
 713
 714   <<EOF>> {
 715     int token = (YY_START == SC_BRACED_CODE) ? BRACED_CODE : BRACED_PREDICATE;
 716     unexpected_eof (code_start, "}");
 717     STRING_FINISH;
 718     loc->start = code_start;
 719     val->code = last_string;
 720     BEGIN INITIAL;
 721     return token;
 722   }
 723 }
 724
 725 <SC_BRACED_CODE>
 726 {
 727   "}" {
 728     obstack_1grow (&obstack_for_string, '}');
 729
 730     --nesting;
 731     if (nesting < 0)
 732       {
 733         STRING_FINISH;
 734         loc->start = code_start;
 735         val->code = last_string;
 736         BEGIN INITIAL;
 737         return BRACED_CODE;
 738       }
 739   }
 740 }
 741
 742 <SC_PREDICATE>
 743 {
 744   "}" {
 745     --nesting;
 746     if (nesting < 0)
 747       {
 748         STRING_FINISH;
 749         loc->start = code_start;
 750         val->code = last_string;
 751         BEGIN INITIAL;
 752         return BRACED_PREDICATE;
 753       }
 754     else
 755       obstack_1grow (&obstack_for_string, '}');
 756   }
 757 }
 758
 759   /*--------------------------------------------------------------.
 760   | Scanning some prologue: from "%{" (already scanned) to "%}".  |
 761   `--------------------------------------------------------------*/
 762
 763 <SC_PROLOGUE>
 764 {
 765   "%}" {
 766     STRING_FINISH;
 767     loc->start = code_start;
 768     val->chars = last_string;
 769     BEGIN INITIAL;
 770     return PROLOGUE;
 771   }
 772
 773   <<EOF>> {
 774     unexpected_eof (code_start, "%}");
 775     STRING_FINISH;
 776     loc->start = code_start;
 777     val->chars = last_string;
 778     BEGIN INITIAL;
 779     return PROLOGUE;
 780   }
 781 }
 782
 783
 784   /*---------------------------------------------------------------.
 785   | Scanning the epilogue (everything after the second "%%", which |
 786   | has already been eaten).                                       |
 787   `---------------------------------------------------------------*/
 788
 789 <SC_EPILOGUE>
 790 {
 791   <<EOF>> {
 792     STRING_FINISH;
 793     loc->start = code_start;
 794     val->chars = last_string;
 795     BEGIN INITIAL;
 796     return EPILOGUE;
 797   }
 798 }
 799
 800
 801   /*-----------------------------------------------------.
 802   | By default, grow the string obstack with the input.  |
 803   `-----------------------------------------------------*/
 804
 805 <SC_COMMENT,SC_LINE_COMMENT,SC_BRACED_CODE,SC_PREDICATE,SC_PROLOGUE,SC_EPILOGUE,SC_STRING,SC_CHARACTER,SC_ESCAPED_STRING,SC_ESCAPED_CHARACTER>. |
 806   <SC_COMMENT,SC_LINE_COMMENT,SC_BRACED_CODE,SC_PREDICATE,SC_PROLOGUE,SC_EPILOGUE>\n    STRING_GROW;
 807
 808 %%
 809
 810 /* Read bytes from FP into buffer BUF of size SIZE.  Return the
 811    number of bytes read.  Remove '\r' from input, treating \r\n
 812    and isolated \r as \n.  */
 813
 814 static size_t
 815 no_cr_read (FILE *fp, char *buf, size_t size)
 816 {
 817   size_t bytes_read = fread (buf, 1, size, fp);
 818   if (bytes_read)
 819     {
 820       char *w = memchr (buf, '\r', bytes_read);
 821       if (w)
 822         {
 823           char const *r = ++w;
 824           char const *lim = buf + bytes_read;
 825
 826           for (;;)
 827             {
 828               /* Found an '\r'.  Treat it like '\n', but ignore any
 829                  '\n' that immediately follows.  */
 830               w[-1] = '\n';
 831               if (r == lim)
 832                 {
 833                   int ch = getc (fp);
 834                   if (ch != '\n' && ungetc (ch, fp) != ch)
 835                     break;
 836                 }
 837               else if (*r == '\n')
 838                 r++;
 839
 840               /* Copy until the next '\r'.  */
 841               do
 842                 {
 843                   if (r == lim)
 844                     return w - buf;
 845                 }
 846               while ((*w++ = *r++) != '\r');
 847             }
 848
 849           return w - buf;
 850         }
 851     }
 852
 853   return bytes_read;
 854 }
 855
 856
 857
 858 /*------------------------------------------------------.
 859 | Scan NUMBER for a base-BASE integer at location LOC.  |
 860 `------------------------------------------------------*/
 861
 862 static unsigned long int
 863 scan_integer (char const *number, int base, location loc)
 864 {
 865   verify (INT_MAX < ULONG_MAX);
 866   unsigned long int num = strtoul (number, NULL, base);
 867
 868   if (INT_MAX < num)
 869     {
 870       complain_at (loc, _("integer out of range: %s"), quote (number));
 871       num = INT_MAX;
 872     }
 873
 874   return num;
 875 }
 876
 877
 878 /*------------------------------------------------------------------.
 879 | Convert universal character name UCN to a single-byte character,  |
 880 | and return that character.  Return -1 if UCN does not correspond  |
 881 | to a single-byte character.                                       |
 882 `------------------------------------------------------------------*/
 883
 884 static int
 885 convert_ucn_to_byte (char const *ucn)
 886 {
 887   verify (UCHAR_MAX <= INT_MAX);
 888   unsigned long int code = strtoul (ucn + 2, NULL, 16);
 889
 890   /* FIXME: Currently we assume Unicode-compatible unibyte characters
 891      on ASCII hosts (i.e., Latin-1 on hosts with 8-bit bytes).  On
 892      non-ASCII hosts we support only the portable C character set.
 893      These limitations should be removed once we add support for
 894      multibyte characters.  */
 895
 896   if (UCHAR_MAX < code)
 897     return -1;
 898
 899 #if ! ('$' == 0x24 && '@' == 0x40 && '`' == 0x60 && '~' == 0x7e)
 900   {
 901     /* A non-ASCII host.  Use CODE to index into a table of the C
 902        basic execution character set, which is guaranteed to exist on
 903        all Standard C platforms.  This table also includes '$', '@',
 904        and '`', which are not in the basic execution character set but
 905        which are unibyte characters on all the platforms that we know
 906        about.  */
 907     static signed char const table[] =
 908       {
 909         '\0',   -1,   -1,   -1,   -1,   -1,   -1, '\a',
 910         '\b', '\t', '\n', '\v', '\f', '\r',   -1,   -1,
 911           -1,   -1,   -1,   -1,   -1,   -1,   -1,   -1,
 912           -1,   -1,   -1,   -1,   -1,   -1,   -1,   -1,
 913          ' ',  '!',  '"',  '#',  '$',  '%',  '&', '\'',
 914          '(',  ')',  '*',  '+',  ',',  '-',  '.',  '/',
 915          '0',  '1',  '2',  '3',  '4',  '5',  '6',  '7',
 916          '8',  '9',  ':',  ';',  '<',  '=',  '>',  '?',
 917          '@',  'A',  'B',  'C',  'D',  'E',  'F',  'G',
 918          'H',  'I',  'J',  'K',  'L',  'M',  'N',  'O',
 919          'P',  'Q',  'R',  'S',  'T',  'U',  'V',  'W',
 920          'X',  'Y',  'Z',  '[', '\\',  ']',  '^',  '_',
 921          '`',  'a',  'b',  'c',  'd',  'e',  'f',  'g',
 922          'h',  'i',  'j',  'k',  'l',  'm',  'n',  'o',
 923          'p',  'q',  'r',  's',  't',  'u',  'v',  'w',
 924          'x',  'y',  'z',  '{',  '|',  '}',  '~'
 925       };
 926
 927     code = code < sizeof table ? table[code] : -1;
 928   }
 929 #endif
 930
 931   return code;
 932 }
 933
 934
 935 /*----------------------------------------------------------------.
 936 | Handle `#line INT "FILE"'.  ARGS has already skipped `#line '.  |
 937 `----------------------------------------------------------------*/
 938
 939 static void
 940 handle_syncline (char *args, location loc)
 941 {
 942   char *after_num;
 943   unsigned long int lineno = strtoul (args, &after_num, 10);
 944   char *file = strchr (after_num, '"') + 1;
 945   *strchr (file, '"') = '\0';
 946   if (INT_MAX <= lineno)
 947     {
 948       warn_at (loc, _("line number overflow"));
 949       lineno = INT_MAX;
 950     }
 951   current_file = uniqstr_new (file);
 952   boundary_set (&scanner_cursor, current_file, lineno, 1);
 953 }
 954
 955
 956 /*----------------------------------------------------------------.
 957 | For a token or comment starting at START, report message MSGID, |
 958 | which should say that an end marker was found before            |
 959 | the expected TOKEN_END.                                         |
 960 `----------------------------------------------------------------*/
 961
 962 static void
 963 unexpected_end (boundary start, char const *msgid, char const *token_end)
 964 {
 965   location loc;
 966   loc.start = start;
 967   loc.end = scanner_cursor;
 968   complain_at (loc, _(msgid), token_end);
 969 }
 970
 971
 972 /*------------------------------------------------------------------------.
 973 | Report an unexpected EOF in a token or comment starting at START.       |
 974 | An end of file was encountered and the expected TOKEN_END was missing.  |
 975 `------------------------------------------------------------------------*/
 976
 977 static void
 978 unexpected_eof (boundary start, char const *token_end)
 979 {
 980   unexpected_end (start, N_("missing `%s' at end of file"), token_end);
 981 }
 982
 983
 984 /*----------------------------------------.
 985 | Likewise, but for unexpected newlines.  |
 986 `----------------------------------------*/
 987
 988 static void
 989 unexpected_newline (boundary start, char const *token_end)
 990 {
 991   unexpected_end (start, N_("missing `%s' at end of line"), token_end);
 992 }
 993
 994
 995 /*-------------------------.
 996 | Initialize the scanner.  |
 997 `-------------------------*/
 998
 999 void
1000 gram_scanner_initialize (void)
1001 {
1002   obstack_init (&obstack_for_string);
1003 }
1004
1005
1006 /*-----------------------------------------------.
1007 | Free all the memory allocated to the scanner.  |
1008 `-----------------------------------------------*/
1009
1010 void
1011 gram_scanner_free (void)
1012 {
1013   obstack_free (&obstack_for_string, 0);
1014   /* Reclaim Flex's buffers.  */
1015   yylex_destroy ();
1016 }