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