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 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/getargs.h>
  35 #include <src/gram.h>
  36 #include <quotearg.h>
  37 #include <src/reader.h>
  38 #include <src/uniqstr.h>
  39
  40 #include <c-ctype.h>
  41 #include <mbswidth.h>
  42 #include <quote.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 #define DEPRECATED(Msg)                                         \
  78   do {                                                          \
  79     size_t i;                                                   \
  80     complain (loc, Wdeprecated,                                 \
  81               _("deprecated directive: %s, use %s"),            \
  82               quote (yytext), quote_n (1, Msg));                \
  83     scanner_cursor.column -= mbsnwidth (Msg, strlen (Msg), 0);  \
  84     for (i = strlen (Msg); i != 0; --i)                         \
  85       unput (Msg[i - 1]);                                       \
  86   } while (0)
  87
  88 /* A string representing the most recently saved token.  */
  89 static char *last_string;
  90
  91 /* Bracketed identifier. */
  92 static uniqstr bracketed_id_str = 0;
  93 static location bracketed_id_loc;
  94 static boundary bracketed_id_start;
  95 static int bracketed_id_context_state = 0;
  96
  97 void
  98 gram_scanner_last_string_free (void)
  99 {
 100   STRING_FREE;
 101 }
 102
 103 static void handle_syncline (char *, location);
 104 static unsigned long int scan_integer (char const *p, int base, location loc);
 105 static int convert_ucn_to_byte (char const *hex_text);
 106 static void unexpected_eof (boundary, char const *);
 107 static void unexpected_newline (boundary, char const *);
 108
 109 %}
 110  /* A C-like comment in directives/rules. */
 111 %x SC_YACC_COMMENT
 112  /* Strings and characters in directives/rules. */
 113 %x SC_ESCAPED_STRING SC_ESCAPED_CHARACTER
 114  /* A identifier was just read in directives/rules.  Special state
 115     to capture the sequence 'identifier :'. */
 116 %x SC_AFTER_IDENTIFIER
 117  /* A complex tag, with nested angles brackets. */
 118 %x SC_TAG
 119
 120  /* Four types of user code:
 121     - prologue (code between '%{' '%}' in the first section, before %%);
 122     - actions, printers, union, etc, (between braced in the middle section);
 123     - epilogue (everything after the second %%).
 124     - predicate (code between '%?{' and '{' in middle section); */
 125 %x SC_PROLOGUE SC_BRACED_CODE SC_EPILOGUE SC_PREDICATE
 126  /* C and C++ comments in code. */
 127 %x SC_COMMENT SC_LINE_COMMENT
 128  /* Strings and characters in code. */
 129 %x SC_STRING SC_CHARACTER
 130  /* Bracketed identifiers support. */
 131 %x SC_BRACKETED_ID SC_RETURN_BRACKETED_ID
 132
 133 letter    [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
 134 notletter [^.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]{-}[%\{]
 135 id        {letter}({letter}|[-0-9])*
 136 int       [0-9]+
 137
 138 /* POSIX says that a tag must be both an id and a C union member, but
 139    historically almost any character is allowed in a tag.  We disallow
 140    NUL, as this simplifies our implementation.  We disallow angle
 141    bracket to match them in nested pairs: several languages use them
 142    for generics/template types.  */
 143 tag      [^\0<>]+
 144
 145 /* Zero or more instances of backslash-newline.  Following GCC, allow
 146    white space between the backslash and the newline.  */
 147 splice   (\\[ \f\t\v]*\n)*
 148
 149 /* An equal sign, with optional leading whitespaces. This is used in some
 150    deprecated constructs. */
 151 eqopt    ([[:space:]]*=)?
 152
 153 %%
 154 %{
 155   /* Nesting level.  Either for nested braces, or nested angle brackets
 156      (but not mixed).  */
 157   int nesting PACIFY_CC (= 0);
 158
 159   /* Parent context state, when applicable.  */
 160   int context_state PACIFY_CC (= 0);
 161
 162   /* Location of most recent identifier, when applicable.  */
 163   location id_loc PACIFY_CC (= empty_location);
 164
 165   /* Where containing code started, when applicable.  Its initial
 166      value is relevant only when yylex is invoked in the SC_EPILOGUE
 167      start condition.  */
 168   boundary code_start = scanner_cursor;
 169
 170   /* Where containing comment or string or character literal started,
 171      when applicable.  */
 172   boundary token_start PACIFY_CC (= scanner_cursor);
 173 %}
 174
 175
 176   /*-----------------------.
 177   | Scanning white space.  |
 178   `-----------------------*/
 179
 180 <INITIAL,SC_AFTER_IDENTIFIER,SC_BRACKETED_ID,SC_RETURN_BRACKETED_ID>
 181 {
 182   /* Comments and white space.  */
 183   "," {
 184      complain (loc, Wother, _("stray ',' treated as white space"));
 185   }
 186   [ \f\n\t\v]  |
 187   "//".*       continue;
 188   "/*" {
 189     token_start = loc->start;
 190     context_state = YY_START;
 191     BEGIN SC_YACC_COMMENT;
 192   }
 193
 194   /* #line directives are not documented, and may be withdrawn or
 195      modified in future versions of Bison.  */
 196   ^"#line "{int}(" \"".*"\"")?"\n" {
 197     handle_syncline (yytext + sizeof "#line " - 1, *loc);
 198   }
 199 }
 200
 201
 202   /*----------------------------.
 203   | Scanning Bison directives.  |
 204   `----------------------------*/
 205
 206   /* For directives that are also command line options, the regex must be
 207         "%..."
 208      after "[-_]"s are removed, and the directive must match the --long
 209      option name, with a single string argument.  Otherwise, add exceptions
 210      to ../build-aux/cross-options.pl.  */
 211
 212 <INITIAL>
 213 {
 214   "%binary"                         return PERCENT_NONASSOC;
 215   "%code"                           return PERCENT_CODE;
 216   "%debug"                          RETURN_PERCENT_FLAG("parse.trace");
 217   "%default-prec"                   return PERCENT_DEFAULT_PREC;
 218   "%define"                         return PERCENT_DEFINE;
 219   "%defines"                        return PERCENT_DEFINES;
 220   "%destructor"                     return PERCENT_DESTRUCTOR;
 221   "%dprec"                          return PERCENT_DPREC;
 222   "%error-verbose"                  return PERCENT_ERROR_VERBOSE;
 223   "%expect"                         return PERCENT_EXPECT;
 224   "%expect-rr"                      return PERCENT_EXPECT_RR;
 225   "%file-prefix"                    return PERCENT_FILE_PREFIX;
 226   "%fixed-output-files"             return PERCENT_YACC;
 227   "%initial-action"                 return PERCENT_INITIAL_ACTION;
 228   "%glr-parser"                     return PERCENT_GLR_PARSER;
 229   "%language"                       return PERCENT_LANGUAGE;
 230   "%left"                           return PERCENT_LEFT;
 231   "%lex-param"                      RETURN_PERCENT_PARAM(lex);
 232   "%locations"                      RETURN_PERCENT_FLAG("locations");
 233   "%merge"                          return PERCENT_MERGE;
 234   "%name-prefix"                    return PERCENT_NAME_PREFIX;
 235   "%no-default-prec"                return PERCENT_NO_DEFAULT_PREC;
 236   "%no-lines"                       return PERCENT_NO_LINES;
 237   "%nonassoc"                       return PERCENT_NONASSOC;
 238   "%nondeterministic-parser"        return PERCENT_NONDETERMINISTIC_PARSER;
 239   "%nterm"                          return PERCENT_NTERM;
 240   "%output"                         return PERCENT_OUTPUT;
 241   "%param"                          RETURN_PERCENT_PARAM(both);
 242   "%parse-param"                    RETURN_PERCENT_PARAM(parse);
 243   "%prec"                           return PERCENT_PREC;
 244   "%precedence"                     return PERCENT_PRECEDENCE;
 245   "%printer"                        return PERCENT_PRINTER;
 246   "%pure-parser"                    RETURN_PERCENT_FLAG("api.pure");
 247   "%require"                        return PERCENT_REQUIRE;
 248   "%right"                          return PERCENT_RIGHT;
 249   "%skeleton"                       return PERCENT_SKELETON;
 250   "%start"                          return PERCENT_START;
 251   "%term"                           return PERCENT_TOKEN;
 252   "%token"                          return PERCENT_TOKEN;
 253   "%token-table"                    return PERCENT_TOKEN_TABLE;
 254   "%type"                           return PERCENT_TYPE;
 255   "%union"                          return PERCENT_UNION;
 256   "%verbose"                        return PERCENT_VERBOSE;
 257   "%yacc"                           return PERCENT_YACC;
 258
 259   /* deprecated */
 260   "%default"[-_]"prec"              DEPRECATED("%default-prec");
 261   "%error"[-_]"verbose"             DEPRECATED("%define parse.error verbose");
 262   "%expect"[-_]"rr"                 DEPRECATED("%expect-rr");
 263   "%file-prefix"{eqopt}             DEPRECATED("%file-prefix");
 264   "%fixed"[-_]"output"[-_]"files"   DEPRECATED("%fixed-output-files");
 265   "%name"[-_]"prefix"{eqopt}        DEPRECATED("%name-prefix");
 266   "%no"[-_]"default"[-_]"prec"      DEPRECATED("%no-default-prec");
 267   "%no"[-_]"lines"                  DEPRECATED("%no-lines");
 268   "%output"{eqopt}                  DEPRECATED("%output");
 269   "%pure"[-_]"parser"               DEPRECATED("%pure-parser");
 270   "%token"[-_]"table"               DEPRECATED("%token-table");
 271
 272   "%"{id}|"%"{notletter}([[:graph:]])+ {
 273     complain (loc, complaint, _("invalid directive: %s"), quote (yytext));
 274   }
 275
 276   "="                     return EQUAL;
 277   "|"                     return PIPE;
 278   ";"                     return SEMICOLON;
 279
 280   {id} {
 281     val->uniqstr = uniqstr_new (yytext);
 282     id_loc = *loc;
 283     bracketed_id_str = NULL;
 284     BEGIN SC_AFTER_IDENTIFIER;
 285   }
 286
 287   {int} {
 288     val->integer = scan_integer (yytext, 10, *loc);
 289     return INT;
 290   }
 291   0[xX][0-9abcdefABCDEF]+ {
 292     val->integer = scan_integer (yytext, 16, *loc);
 293     return INT;
 294   }
 295
 296   /* Identifiers may not start with a digit.  Yet, don't silently
 297      accept "1FOO" as "1 FOO".  */
 298   {int}{id} {
 299     complain (loc, complaint, _("invalid identifier: %s"), quote (yytext));
 300   }
 301
 302   /* Characters.  */
 303   "'"         token_start = loc->start; BEGIN SC_ESCAPED_CHARACTER;
 304
 305   /* Strings. */
 306   "\""        token_start = loc->start; BEGIN SC_ESCAPED_STRING;
 307
 308   /* Prologue. */
 309   "%{"        code_start = loc->start; BEGIN SC_PROLOGUE;
 310
 311   /* Code in between braces.  */
 312   "{" {
 313     STRING_GROW;
 314     nesting = 0;
 315     code_start = loc->start;
 316     BEGIN SC_BRACED_CODE;
 317   }
 318
 319   /* Semantic predicate. */
 320   "%?"[ \f\n\t\v]*"{" {
 321     nesting = 0;
 322     code_start = loc->start;
 323     BEGIN SC_PREDICATE;
 324   }
 325
 326   /* A type. */
 327   "<*>"       return TAG_ANY;
 328   "<>"        return TAG_NONE;
 329   "<"{tag}">" {
 330     obstack_grow (&obstack_for_string, yytext + 1, yyleng - 2);
 331     STRING_FINISH;
 332     val->uniqstr = uniqstr_new (last_string);
 333     STRING_FREE;
 334     return TAG;
 335   }
 336   "<"         {
 337     nesting = 0;
 338     token_start = loc->start;
 339     BEGIN SC_TAG;
 340   }
 341
 342   "%%" {
 343     static int percent_percent_count;
 344     if (++percent_percent_count == 2)
 345       BEGIN SC_EPILOGUE;
 346     return PERCENT_PERCENT;
 347   }
 348
 349   "[" {
 350     bracketed_id_str = NULL;
 351     bracketed_id_start = loc->start;
 352     bracketed_id_context_state = YY_START;
 353     BEGIN SC_BRACKETED_ID;
 354   }
 355
 356   [^\[%A-Za-z0-9_<>{}\"\'*;|=/, \f\n\t\v]+|. {
 357     complain (loc, complaint, "%s: %s",
 358               ngettext ("invalid character", "invalid characters", yyleng),
 359               quote_mem (yytext, yyleng));
 360   }
 361
 362   <<EOF>> {
 363     loc->start = loc->end = scanner_cursor;
 364     yyterminate ();
 365   }
 366 }
 367
 368
 369   /*--------------------------------------------------------------.
 370   | Supporting \0 complexifies our implementation for no expected |
 371   | added value.                                                  |
 372   `--------------------------------------------------------------*/
 373
 374 <SC_ESCAPED_CHARACTER,SC_ESCAPED_STRING,SC_TAG>
 375 {
 376   \0        complain (loc, complaint, _("invalid null character"));
 377 }
 378
 379
 380   /*-----------------------------------------------------------------.
 381   | Scanning after an identifier, checking whether a colon is next.  |
 382   `-----------------------------------------------------------------*/
 383
 384 <SC_AFTER_IDENTIFIER>
 385 {
 386   "[" {
 387     if (bracketed_id_str)
 388       {
 389         ROLLBACK_CURRENT_TOKEN;
 390         BEGIN SC_RETURN_BRACKETED_ID;
 391         *loc = id_loc;
 392         return ID;
 393       }
 394     else
 395       {
 396         bracketed_id_start = loc->start;
 397         bracketed_id_context_state = YY_START;
 398         BEGIN SC_BRACKETED_ID;
 399       }
 400   }
 401   ":" {
 402     BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL);
 403     *loc = id_loc;
 404     return ID_COLON;
 405   }
 406   . {
 407     ROLLBACK_CURRENT_TOKEN;
 408     BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL);
 409     *loc = id_loc;
 410     return ID;
 411   }
 412   <<EOF>> {
 413     BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL);
 414     *loc = id_loc;
 415     return ID;
 416   }
 417 }
 418
 419   /*--------------------------------.
 420   | Scanning bracketed identifiers. |
 421   `--------------------------------*/
 422
 423 <SC_BRACKETED_ID>
 424 {
 425   {id} {
 426     if (bracketed_id_str)
 427       {
 428         complain (loc, complaint,
 429                   _("unexpected identifier in bracketed name: %s"),
 430                   quote (yytext));
 431       }
 432     else
 433       {
 434         bracketed_id_str = uniqstr_new (yytext);
 435         bracketed_id_loc = *loc;
 436       }
 437   }
 438   "]" {
 439     BEGIN bracketed_id_context_state;
 440     if (bracketed_id_str)
 441       {
 442         if (INITIAL == bracketed_id_context_state)
 443           {
 444             val->uniqstr = bracketed_id_str;
 445             bracketed_id_str = 0;
 446             *loc = bracketed_id_loc;
 447             return BRACKETED_ID;
 448           }
 449       }
 450     else
 451       complain (loc, complaint, _("an identifier expected"));
 452   }
 453
 454   [^\].A-Za-z0-9_/ \f\n\t\v]+|. {
 455     complain (loc, complaint, "%s: %s",
 456               ngettext ("invalid character in bracketed name",
 457                         "invalid characters in bracketed name", yyleng),
 458               quote_mem (yytext, yyleng));
 459   }
 460
 461   <<EOF>> {
 462     BEGIN bracketed_id_context_state;
 463     unexpected_eof (bracketed_id_start, "]");
 464   }
 465 }
 466
 467 <SC_RETURN_BRACKETED_ID>
 468 {
 469   . {
 470     ROLLBACK_CURRENT_TOKEN;
 471     val->uniqstr = bracketed_id_str;
 472     bracketed_id_str = 0;
 473     *loc = bracketed_id_loc;
 474     BEGIN INITIAL;
 475     return BRACKETED_ID;
 476   }
 477 }
 478
 479
 480   /*---------------------------------------------------------------.
 481   | Scanning a Yacc comment.  The initial '/ *' is already eaten.  |
 482   `---------------------------------------------------------------*/
 483
 484 <SC_YACC_COMMENT>
 485 {
 486   "*/"     BEGIN context_state;
 487   .|\n     continue;
 488   <<EOF>>  unexpected_eof (token_start, "*/"); BEGIN context_state;
 489 }
 490
 491
 492   /*------------------------------------------------------------.
 493   | Scanning a C comment.  The initial '/ *' is already eaten.  |
 494   `------------------------------------------------------------*/
 495
 496 <SC_COMMENT>
 497 {
 498   "*"{splice}"/"  STRING_GROW; BEGIN context_state;
 499   <<EOF>>         unexpected_eof (token_start, "*/"); BEGIN context_state;
 500 }
 501
 502
 503   /*--------------------------------------------------------------.
 504   | Scanning a line comment.  The initial '//' is already eaten.  |
 505   `--------------------------------------------------------------*/
 506
 507 <SC_LINE_COMMENT>
 508 {
 509   "\n"           STRING_GROW; BEGIN context_state;
 510   {splice}       STRING_GROW;
 511   <<EOF>>        BEGIN context_state;
 512 }
 513
 514
 515   /*------------------------------------------------.
 516   | Scanning a Bison string, including its escapes. |
 517   | The initial quote is already eaten.             |
 518   `------------------------------------------------*/
 519
 520 <SC_ESCAPED_STRING>
 521 {
 522   "\""|"\n" {
 523     if (yytext[0] == '\n')
 524       unexpected_newline (token_start, "\"");
 525     STRING_FINISH;
 526     loc->start = token_start;
 527     val->chars = last_string;
 528     BEGIN INITIAL;
 529     return STRING;
 530   }
 531   <<EOF>> {
 532     unexpected_eof (token_start, "\"");
 533     STRING_FINISH;
 534     loc->start = token_start;
 535     val->chars = last_string;
 536     BEGIN INITIAL;
 537     return STRING;
 538   }
 539 }
 540
 541   /*----------------------------------------------------------.
 542   | Scanning a Bison character literal, decoding its escapes. |
 543   | The initial quote is already eaten.                       |
 544   `----------------------------------------------------------*/
 545
 546 <SC_ESCAPED_CHARACTER>
 547 {
 548   "'"|"\n" {
 549     STRING_FINISH;
 550     loc->start = token_start;
 551     val->character = last_string[0];
 552     {
 553       /* FIXME: Eventually, make these errors.  */
 554       if (last_string[0] == '\0')
 555         {
 556           complain (loc, Wother, _("empty character literal"));
 557           /* '\0' seems dangerous even if we are about to complain.  */
 558           val->character = '\'';
 559         }
 560       else if (last_string[1] != '\0')
 561         complain (loc, Wother,
 562                   _("extra characters in character literal"));
 563     }
 564     if (yytext[0] == '\n')
 565       unexpected_newline (token_start, "'");
 566     STRING_FREE;
 567     BEGIN INITIAL;
 568     return CHAR;
 569   }
 570   <<EOF>> {
 571     STRING_FINISH;
 572     loc->start = token_start;
 573     val->character = last_string[0];
 574     {
 575       /* FIXME: Eventually, make these errors.  */
 576       if (last_string[0] == '\0')
 577         {
 578           complain (loc, Wother, _("empty character literal"));
 579           /* '\0' seems dangerous even if we are about to complain.  */
 580           val->character = '\'';
 581         }
 582       else if (last_string[1] != '\0')
 583         complain (loc, Wother,
 584                   _("extra characters in character literal"));
 585     }
 586     unexpected_eof (token_start, "'");
 587     STRING_FREE;
 588     BEGIN INITIAL;
 589     return CHAR;
 590   }
 591 }
 592
 593   /*-----------------------------------------------------------.
 594   | Scanning a Bison nested tag.  The initial angle bracket is |
 595   | already eaten.                                             |
 596   `-----------------------------------------------------------*/
 597
 598 <SC_TAG>
 599 {
 600   ">" {
 601     --nesting;
 602     if (nesting < 0)
 603       {
 604         STRING_FINISH;
 605         loc->start = token_start;
 606         val->uniqstr = uniqstr_new (last_string);
 607         STRING_FREE;
 608         BEGIN INITIAL;
 609         return TAG;
 610       }
 611     STRING_GROW;
 612   }
 613
 614   [^<>]+ STRING_GROW;
 615   "<"+   STRING_GROW; nesting += yyleng;
 616
 617   <<EOF>> {
 618     unexpected_eof (token_start, ">");
 619     STRING_FINISH;
 620     loc->start = token_start;
 621     val->uniqstr = uniqstr_new (last_string);
 622     STRING_FREE;
 623     BEGIN INITIAL;
 624     return TAG;
 625   }
 626 }
 627
 628   /*----------------------------.
 629   | Decode escaped characters.  |
 630   `----------------------------*/
 631
 632 <SC_ESCAPED_STRING,SC_ESCAPED_CHARACTER>
 633 {
 634   \\[0-7]{1,3} {
 635     unsigned long int c = strtoul (yytext + 1, NULL, 8);
 636     if (!c || UCHAR_MAX < c)
 637       complain (loc, complaint, _("invalid number after \\-escape: %s"),
 638                    yytext+1);
 639     else
 640       obstack_1grow (&obstack_for_string, c);
 641   }
 642
 643   \\x[0-9abcdefABCDEF]+ {
 644     verify (UCHAR_MAX < ULONG_MAX);
 645     unsigned long int c = strtoul (yytext + 2, NULL, 16);
 646     if (!c || UCHAR_MAX < c)
 647       complain (loc, complaint, _("invalid number after \\-escape: %s"),
 648                    yytext+1);
 649     else
 650       obstack_1grow (&obstack_for_string, c);
 651   }
 652
 653   \\a   obstack_1grow (&obstack_for_string, '\a');
 654   \\b   obstack_1grow (&obstack_for_string, '\b');
 655   \\f   obstack_1grow (&obstack_for_string, '\f');
 656   \\n   obstack_1grow (&obstack_for_string, '\n');
 657   \\r   obstack_1grow (&obstack_for_string, '\r');
 658   \\t   obstack_1grow (&obstack_for_string, '\t');
 659   \\v   obstack_1grow (&obstack_for_string, '\v');
 660
 661   /* \\[\"\'?\\] would be shorter, but it confuses xgettext.  */
 662   \\("\""|"'"|"?"|"\\")  obstack_1grow (&obstack_for_string, yytext[1]);
 663
 664   \\(u|U[0-9abcdefABCDEF]{4})[0-9abcdefABCDEF]{4} {
 665     int c = convert_ucn_to_byte (yytext);
 666     if (c <= 0)
 667       complain (loc, complaint, _("invalid number after \\-escape: %s"),
 668                    yytext+1);
 669     else
 670       obstack_1grow (&obstack_for_string, c);
 671   }
 672   \\(.|\n)      {
 673     char const *p = yytext + 1;
 674     /* Quote only if escaping won't make the character visible.  */
 675     if (c_isspace ((unsigned char) *p) && c_isprint ((unsigned char) *p))
 676       p = quote (p);
 677     else
 678       p = quotearg_style_mem (escape_quoting_style, p, 1);
 679     complain (loc, complaint, _("invalid character after \\-escape: %s"),
 680                  p);
 681   }
 682 }
 683
 684   /*--------------------------------------------.
 685   | Scanning user-code characters and strings.  |
 686   `--------------------------------------------*/
 687
 688 <SC_CHARACTER,SC_STRING>
 689 {
 690   {splice}|\\{splice}[^\n\[\]]  STRING_GROW;
 691 }
 692
 693 <SC_CHARACTER>
 694 {
 695   "'"           STRING_GROW; BEGIN context_state;
 696   \n            unexpected_newline (token_start, "'"); BEGIN context_state;
 697   <<EOF>>       unexpected_eof (token_start, "'"); BEGIN context_state;
 698 }
 699
 700 <SC_STRING>
 701 {
 702   "\""          STRING_GROW; BEGIN context_state;
 703   \n            unexpected_newline (token_start, "\""); BEGIN context_state;
 704   <<EOF>>       unexpected_eof (token_start, "\""); BEGIN context_state;
 705 }
 706
 707
 708   /*---------------------------------------------------.
 709   | Strings, comments etc. can be found in user code.  |
 710   `---------------------------------------------------*/
 711
 712 <SC_BRACED_CODE,SC_PROLOGUE,SC_EPILOGUE,SC_PREDICATE>
 713 {
 714   "'" {
 715     STRING_GROW;
 716     context_state = YY_START;
 717     token_start = loc->start;
 718     BEGIN SC_CHARACTER;
 719   }
 720   "\"" {
 721     STRING_GROW;
 722     context_state = YY_START;
 723     token_start = loc->start;
 724     BEGIN SC_STRING;
 725   }
 726   "/"{splice}"*" {
 727     STRING_GROW;
 728     context_state = YY_START;
 729     token_start = loc->start;
 730     BEGIN SC_COMMENT;
 731   }
 732   "/"{splice}"/" {
 733     STRING_GROW;
 734     context_state = YY_START;
 735     BEGIN SC_LINE_COMMENT;
 736   }
 737 }
 738
 739
 740
 741   /*-----------------------------------------------------------.
 742   | Scanning some code in braces (actions, predicates). The    |
 743   | initial "{" is already eaten.                              |
 744   `-----------------------------------------------------------*/
 745
 746 <SC_BRACED_CODE,SC_PREDICATE>
 747 {
 748   "{"|"<"{splice}"%"  STRING_GROW; nesting++;
 749   "%"{splice}">"      STRING_GROW; nesting--;
 750
 751   /* Tokenize '<<%' correctly (as '<<' '%') rather than incorrrectly
 752      (as '<' '<%').  */
 753   "<"{splice}"<"  STRING_GROW;
 754
 755   <<EOF>> {
 756     int token = (YY_START == SC_BRACED_CODE) ? BRACED_CODE : BRACED_PREDICATE;
 757     unexpected_eof (code_start, "}");
 758     STRING_FINISH;
 759     loc->start = code_start;
 760     val->code = last_string;
 761     BEGIN INITIAL;
 762     return token;
 763   }
 764 }
 765
 766 <SC_BRACED_CODE>
 767 {
 768   "}" {
 769     obstack_1grow (&obstack_for_string, '}');
 770
 771     --nesting;
 772     if (nesting < 0)
 773       {
 774         STRING_FINISH;
 775         loc->start = code_start;
 776         val->code = last_string;
 777         BEGIN INITIAL;
 778         return BRACED_CODE;
 779       }
 780   }
 781 }
 782
 783 <SC_PREDICATE>
 784 {
 785   "}" {
 786     --nesting;
 787     if (nesting < 0)
 788       {
 789         STRING_FINISH;
 790         loc->start = code_start;
 791         val->code = last_string;
 792         BEGIN INITIAL;
 793         return BRACED_PREDICATE;
 794       }
 795     else
 796       obstack_1grow (&obstack_for_string, '}');
 797   }
 798 }
 799
 800   /*--------------------------------------------------------------.
 801   | Scanning some prologue: from "%{" (already scanned) to "%}".  |
 802   `--------------------------------------------------------------*/
 803
 804 <SC_PROLOGUE>
 805 {
 806   "%}" {
 807     STRING_FINISH;
 808     loc->start = code_start;
 809     val->chars = last_string;
 810     BEGIN INITIAL;
 811     return PROLOGUE;
 812   }
 813
 814   <<EOF>> {
 815     unexpected_eof (code_start, "%}");
 816     STRING_FINISH;
 817     loc->start = code_start;
 818     val->chars = last_string;
 819     BEGIN INITIAL;
 820     return PROLOGUE;
 821   }
 822 }
 823
 824
 825   /*---------------------------------------------------------------.
 826   | Scanning the epilogue (everything after the second "%%", which |
 827   | has already been eaten).                                       |
 828   `---------------------------------------------------------------*/
 829
 830 <SC_EPILOGUE>
 831 {
 832   <<EOF>> {
 833     STRING_FINISH;
 834     loc->start = code_start;
 835     val->chars = last_string;
 836     BEGIN INITIAL;
 837     return EPILOGUE;
 838   }
 839 }
 840
 841
 842   /*-----------------------------------------------------.
 843   | By default, grow the string obstack with the input.  |
 844   `-----------------------------------------------------*/
 845
 846 <SC_COMMENT,SC_LINE_COMMENT,SC_BRACED_CODE,SC_PREDICATE,SC_PROLOGUE,SC_EPILOGUE,SC_STRING,SC_CHARACTER,SC_ESCAPED_STRING,SC_ESCAPED_CHARACTER>. |
 847   <SC_COMMENT,SC_LINE_COMMENT,SC_BRACED_CODE,SC_PREDICATE,SC_PROLOGUE,SC_EPILOGUE>\n    STRING_GROW;
 848
 849 %%
 850
 851 /* Read bytes from FP into buffer BUF of size SIZE.  Return the
 852    number of bytes read.  Remove '\r' from input, treating \r\n
 853    and isolated \r as \n.  */
 854
 855 static size_t
 856 no_cr_read (FILE *fp, char *buf, size_t size)
 857 {
 858   size_t bytes_read = fread (buf, 1, size, fp);
 859   if (bytes_read)
 860     {
 861       char *w = memchr (buf, '\r', bytes_read);
 862       if (w)
 863         {
 864           char const *r = ++w;
 865           char const *lim = buf + bytes_read;
 866
 867           for (;;)
 868             {
 869               /* Found an '\r'.  Treat it like '\n', but ignore any
 870                  '\n' that immediately follows.  */
 871               w[-1] = '\n';
 872               if (r == lim)
 873                 {
 874                   int ch = getc (fp);
 875                   if (ch != '\n' && ungetc (ch, fp) != ch)
 876                     break;
 877                 }
 878               else if (*r == '\n')
 879                 r++;
 880
 881               /* Copy until the next '\r'.  */
 882               do
 883                 {
 884                   if (r == lim)
 885                     return w - buf;
 886                 }
 887               while ((*w++ = *r++) != '\r');
 888             }
 889
 890           return w - buf;
 891         }
 892     }
 893
 894   return bytes_read;
 895 }
 896
 897
 898
 899 /*------------------------------------------------------.
 900 | Scan NUMBER for a base-BASE integer at location LOC.  |
 901 `------------------------------------------------------*/
 902
 903 static unsigned long int
 904 scan_integer (char const *number, int base, location loc)
 905 {
 906   verify (INT_MAX < ULONG_MAX);
 907   unsigned long int num = strtoul (number, NULL, base);
 908
 909   if (INT_MAX < num)
 910     {
 911       complain (&loc, complaint, _("integer out of range: %s"),
 912                    quote (number));
 913       num = INT_MAX;
 914     }
 915
 916   return num;
 917 }
 918
 919
 920 /*------------------------------------------------------------------.
 921 | Convert universal character name UCN to a single-byte character,  |
 922 | and return that character.  Return -1 if UCN does not correspond  |
 923 | to a single-byte character.                                       |
 924 `------------------------------------------------------------------*/
 925
 926 static int
 927 convert_ucn_to_byte (char const *ucn)
 928 {
 929   verify (UCHAR_MAX <= INT_MAX);
 930   unsigned long int code = strtoul (ucn + 2, NULL, 16);
 931
 932   /* FIXME: Currently we assume Unicode-compatible unibyte characters
 933      on ASCII hosts (i.e., Latin-1 on hosts with 8-bit bytes).  On
 934      non-ASCII hosts we support only the portable C character set.
 935      These limitations should be removed once we add support for
 936      multibyte characters.  */
 937
 938   if (UCHAR_MAX < code)
 939     return -1;
 940
 941 #if ! ('$' == 0x24 && '@' == 0x40 && '`' == 0x60 && '~' == 0x7e)
 942   {
 943     /* A non-ASCII host.  Use CODE to index into a table of the C
 944        basic execution character set, which is guaranteed to exist on
 945        all Standard C platforms.  This table also includes '$', '@',
 946        and '`', which are not in the basic execution character set but
 947        which are unibyte characters on all the platforms that we know
 948        about.  */
 949     static signed char const table[] =
 950       {
 951         '\0',   -1,   -1,   -1,   -1,   -1,   -1, '\a',
 952         '\b', '\t', '\n', '\v', '\f', '\r',   -1,   -1,
 953           -1,   -1,   -1,   -1,   -1,   -1,   -1,   -1,
 954           -1,   -1,   -1,   -1,   -1,   -1,   -1,   -1,
 955          ' ',  '!',  '"',  '#',  '$',  '%',  '&', '\'',
 956          '(',  ')',  '*',  '+',  ',',  '-',  '.',  '/',
 957          '0',  '1',  '2',  '3',  '4',  '5',  '6',  '7',
 958          '8',  '9',  ':',  ';',  '<',  '=',  '>',  '?',
 959          '@',  'A',  'B',  'C',  'D',  'E',  'F',  'G',
 960          'H',  'I',  'J',  'K',  'L',  'M',  'N',  'O',
 961          'P',  'Q',  'R',  'S',  'T',  'U',  'V',  'W',
 962          'X',  'Y',  'Z',  '[', '\\',  ']',  '^',  '_',
 963          '`',  'a',  'b',  'c',  'd',  'e',  'f',  'g',
 964          'h',  'i',  'j',  'k',  'l',  'm',  'n',  'o',
 965          'p',  'q',  'r',  's',  't',  'u',  'v',  'w',
 966          'x',  'y',  'z',  '{',  '|',  '}',  '~'
 967       };
 968
 969     code = code < sizeof table ? table[code] : -1;
 970   }
 971 #endif
 972
 973   return code;
 974 }
 975
 976
 977 /*---------------------------------------------------------------------.
 978 | Handle '#line INT( "FILE")?\n'.  ARGS has already skipped '#line '.  |
 979 `---------------------------------------------------------------------*/
 980
 981 static void
 982 handle_syncline (char *args, location loc)
 983 {
 984   char *file;
 985   unsigned long int lineno = strtoul (args, &file, 10);
 986   if (INT_MAX <= lineno)
 987     {
 988       complain (&loc, Wother, _("line number overflow"));
 989       lineno = INT_MAX;
 990     }
 991
 992   file = strchr (file, '"');
 993   if (file)
 994     {
 995       *strchr (file + 1, '"') = '\0';
 996       current_file = uniqstr_new (file + 1);
 997     }
 998   boundary_set (&scanner_cursor, current_file, lineno, 1);
 999 }
1000
1001
1002 /*----------------------------------------------------------------.
1003 | For a token or comment starting at START, report message MSGID, |
1004 | which should say that an end marker was found before            |
1005 | the expected TOKEN_END.                                         |
1006 `----------------------------------------------------------------*/
1007
1008 static void
1009 unexpected_end (boundary start, char const *msgid, char const *token_end)
1010 {
1011   location loc;
1012   loc.start = start;
1013   loc.end = scanner_cursor;
1014   token_end = quote (token_end);
1015   // Instead of '\'', display "'".
1016   if (STREQ (token_end, "'\\''"))
1017     token_end = "\"'\"";
1018   complain (&loc, complaint, _(msgid), token_end);
1019 }
1020
1021
1022 /*------------------------------------------------------------------------.
1023 | Report an unexpected EOF in a token or comment starting at START.       |
1024 | An end of file was encountered and the expected TOKEN_END was missing.  |
1025 `------------------------------------------------------------------------*/
1026
1027 static void
1028 unexpected_eof (boundary start, char const *token_end)
1029 {
1030   unexpected_end (start, N_("missing %s at end of file"), token_end);
1031 }
1032
1033
1034 /*----------------------------------------.
1035 | Likewise, but for unexpected newlines.  |
1036 `----------------------------------------*/
1037
1038 static void
1039 unexpected_newline (boundary start, char const *token_end)
1040 {
1041   unexpected_end (start, N_("missing %s at end of line"), token_end);
1042 }
1043
1044
1045 /*-------------------------.
1046 | Initialize the scanner.  |
1047 `-------------------------*/
1048
1049 void
1050 gram_scanner_initialize (void)
1051 {
1052   obstack_init (&obstack_for_string);
1053 }
1054
1055
1056 /*-----------------------------------------------.
1057 | Free all the memory allocated to the scanner.  |
1058 `-----------------------------------------------*/
1059
1060 void
1061 gram_scanner_free (void)
1062 {
1063   obstack_free (&obstack_for_string, 0);
1064   /* Reclaim Flex's buffers.  */
1065   yylex_destroy ();
1066 }