gcc/cpptrad.c

   1 /* CPP Library - traditional lexical analysis and macro expansion.
   2    Copyright (C) 2002 Free Software Foundation, Inc.
   3    Contributed by Neil Booth, May 2002
   4
   5 This program is free software; you can redistribute it and/or modify it
   6 under the terms of the GNU General Public License as published by the
   7 Free Software Foundation; either version 2, or (at your option) any
   8 later version.
   9
  10 This program is distributed in the hope that it will be useful,
  11 but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 GNU General Public License for more details.
  14
  15 You should have received a copy of the GNU General Public License
  16 along with this program; if not, write to the Free Software
  17 Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
  18
  19 #include "config.h"
  20 #include "system.h"
  21 #include "cpplib.h"
  22 #include "cpphash.h"
  23
  24 /* The replacement text of a function-like macro is stored as a
  25    contiguous sequence of aligned blocks, each representing the text
  26    between subsequent parameters.
  27
  28    Each block comprises the text between its surrounding parameters,
  29    the length of that text, and the one-based index of the following
  30    parameter.  The final block in the replacement text is easily
  31    recognizable as it has an argument index of zero.  */
  32
  33 struct block
  34 {
  35   unsigned int text_len;
  36   unsigned short arg_index;
  37   uchar text[1];
  38 };
  39
  40 #define BLOCK_HEADER_LEN offsetof (struct block, text)
  41 #define BLOCK_LEN(TEXT_LEN) CPP_ALIGN (BLOCK_HEADER_LEN + (TEXT_LEN))
  42
  43 /* Structure holding information about a function-like macro
  44    invocation.  */
  45 struct fun_macro
  46 {
  47   /* Memory buffer holding the trad_arg array.  */
  48   _cpp_buff *buff;
  49
  50   /* An array of size the number of macro parameters + 1, containing
  51      the offsets of the start of each macro argument in the output
  52      buffer.  The argument continues until the character before the
  53      start of the next one.  */
  54   size_t *args;
  55
  56   /* The hashnode of the macro.  */
  57   cpp_hashnode *node;
  58
  59   /* The offset of the macro name in the output buffer.  */
  60   size_t offset;
  61
  62   /* The line the macro name appeared on.  */
  63   unsigned int line;
  64
  65   /* Zero-based index of argument being currently lexed.  */
  66   unsigned int argc;
  67 };
  68
  69 /* Lexing state.  It is mostly used to prevent macro expansion.  */
  70 enum ls {ls_none = 0,           /* Normal state.  */
  71          ls_fun_open,           /* When looking for '('.  */
  72          ls_fun_close,          /* When looking for ')'.  */
  73          ls_defined,            /* After defined.  */
  74          ls_defined_close,      /* Looking for ')' of defined().  */
  75          ls_hash,               /* After # in preprocessor conditional.  */
  76          ls_predicate,          /* After the predicate, maybe paren?  */
  77          ls_answer};            /* In answer to predicate.  */
  78
  79 /* Lexing TODO: Maybe handle space in escaped newlines.  Stop cpplex.c
  80    from recognizing comments and directives during its lexing pass.  */
  81
  82 static const uchar *handle_newline PARAMS ((cpp_reader *, const uchar *));
  83 static const uchar *skip_escaped_newlines PARAMS ((cpp_reader *,
  84                                                    const uchar *));
  85 static const uchar *skip_whitespace PARAMS ((cpp_reader *, const uchar *,
  86                                              int));
  87 static cpp_hashnode *lex_identifier PARAMS ((cpp_reader *, const uchar *));
  88 static const uchar *copy_comment PARAMS ((cpp_reader *, const uchar *, int));
  89 static void scan_out_logical_line PARAMS ((cpp_reader *pfile, cpp_macro *));
  90 static void check_output_buffer PARAMS ((cpp_reader *, size_t));
  91 static void push_replacement_text PARAMS ((cpp_reader *, cpp_hashnode *));
  92 static bool scan_parameters PARAMS ((cpp_reader *, cpp_macro *));
  93 static bool recursive_macro PARAMS ((cpp_reader *, cpp_hashnode *));
  94 static void save_replacement_text PARAMS ((cpp_reader *, cpp_macro *,
  95                                            unsigned int));
  96 static void maybe_start_funlike PARAMS ((cpp_reader *, cpp_hashnode *,
  97                                          const uchar *, struct fun_macro *));
  98 static void save_argument PARAMS ((struct fun_macro *, size_t));
  99 static void replace_args_and_push PARAMS ((cpp_reader *, struct fun_macro *));
 100 static size_t canonicalize_text PARAMS ((uchar *, const uchar *, size_t,
 101                                          uchar *));
 102
 103 /* Ensures we have N bytes' space in the output buffer, and
 104    reallocates it if not.  */
 105 static void
 106 check_output_buffer (pfile, n)
 107      cpp_reader *pfile;
 108      size_t n;
 109 {
 110   /* We might need two bytes to terminate an unterminated comment, and
 111      one more to terminate the line with a NUL.  */
 112   n += 2 + 1;
 113
 114   if (n > (size_t) (pfile->out.limit - pfile->out.cur))
 115     {
 116       size_t size = pfile->out.cur - pfile->out.base;
 117       size_t new_size = (size + n) * 3 / 2;
 118
 119       pfile->out.base
 120         = (uchar *) xrealloc (pfile->out.base, new_size);
 121       pfile->out.limit = pfile->out.base + new_size;
 122       pfile->out.cur = pfile->out.base + size;
 123     }
 124 }
 125
 126 /* To be called whenever a newline character is encountered in the
 127    input file, at CUR.  Handles DOS, Mac and Unix ends of line, and
 128    increments pfile->line.
 129
 130    Returns a pointer the character after the newline sequence.  */
 131 static const uchar *
 132 handle_newline (pfile, cur)
 133      cpp_reader *pfile;
 134      const uchar *cur;
 135 {
 136   pfile->line++;
 137   if (cur[0] + cur[1] == '\r' + '\n')
 138     cur++;
 139   return cur + 1;
 140 }
 141
 142 /* CUR points to any character in the current context, not necessarily
 143    a backslash.  Advances CUR until all escaped newlines are skipped,
 144    and returns the new position without updating the context.
 145
 146    Warns if a file buffer ends in an escaped newline.  */
 147 static const uchar *
 148 skip_escaped_newlines (pfile, cur)
 149      cpp_reader *pfile;
 150      const uchar *cur;
 151 {
 152   const uchar *orig_cur = cur;
 153
 154   while (*cur == '\\' && is_vspace (cur[1]))
 155     cur = handle_newline (pfile, cur + 1);
 156
 157   if (cur != orig_cur && cur == RLIMIT (pfile->context) && pfile->buffer->inc)
 158     cpp_error (pfile, DL_PEDWARN, "backslash-newline at end of file");
 159
 160   return cur;
 161 }
 162
 163 /* CUR points to the asterisk introducing a comment in the current
 164    context.  IN_DEFINE is true if we are in the replacement text of a
 165    macro.
 166
 167    The asterisk and following comment is copied to the buffer pointed
 168    to by pfile->out.cur, which must be of sufficient size.
 169    Unterminated comments are diagnosed, and correctly terminated in
 170    the output.  pfile->out.cur is updated depending upon IN_DEFINE,
 171    -C, -CC and pfile->state.in_directive.
 172
 173    Returns a pointer to the first character after the comment in the
 174    input buffer.  */
 175 static const uchar *
 176 copy_comment (pfile, cur, in_define)
 177      cpp_reader *pfile;
 178      const uchar *cur;
 179      int in_define;
 180 {
 181   unsigned int from_line = pfile->line;
 182   const uchar *limit = RLIMIT (pfile->context);
 183   uchar *out = pfile->out.cur;
 184
 185   do
 186     {
 187       unsigned int c = *cur++;
 188       *out++ = c;
 189
 190       if (c == '/')
 191         {
 192           /* An immediate slash does not terminate the comment.  */
 193           if (out[-2] == '*' && out - 2 > pfile->out.cur)
 194             goto done;
 195
 196           if (*cur == '*' && cur[1] != '/'
 197               && CPP_OPTION (pfile, warn_comments))
 198             cpp_error_with_line (pfile, DL_WARNING, pfile->line, 0,
 199                                  "\"/*\" within comment");
 200         }
 201       else if (is_vspace (c))
 202         {
 203           cur = handle_newline (pfile, cur - 1);
 204           /* Canonicalize newline sequences and skip escaped ones.  */
 205           if (out[-2] == '\\')
 206             out -= 2;
 207           else
 208             out[-1] = '\n';
 209         }
 210     }
 211   while (cur < limit);
 212
 213   cpp_error_with_line (pfile, DL_ERROR, from_line, 0, "unterminated comment");
 214   *out++ = '*';
 215   *out++ = '/';
 216
 217  done:
 218   /* Comments in directives become spaces so that tokens are properly
 219      separated when the ISO preprocessor re-lexes the line.  The
 220      exception is #define.  */
 221   if (pfile->state.in_directive)
 222     {
 223       if (in_define)
 224         {
 225           if (CPP_OPTION (pfile, discard_comments_in_macro_exp))
 226             pfile->out.cur--;
 227           else
 228             pfile->out.cur = out;
 229         }
 230       else
 231         pfile->out.cur[-1] = ' ';
 232     }
 233   else if (CPP_OPTION (pfile, discard_comments))
 234     pfile->out.cur--;
 235   else
 236     pfile->out.cur = out;
 237
 238   return cur;
 239 }
 240
 241 /* CUR points to any character in the input buffer.  Skips over all
 242    contiguous horizontal white space and NULs, including comments if
 243    SKIP_COMMENTS, until reaching the first non-horizontal-whitespace
 244    character or the end of the current context.  Escaped newlines are
 245    removed.
 246
 247    The whitespace is copied verbatim to the output buffer, except that
 248    comments are handled as described in copy_comment().
 249    pfile->out.cur is updated.
 250
 251    Returns a pointer to the first character after the whitespace in
 252    the input buffer.  */
 253 static const uchar *
 254 skip_whitespace (pfile, cur, skip_comments)
 255      cpp_reader *pfile;
 256      const uchar *cur;
 257      int skip_comments;
 258 {
 259   uchar *out = pfile->out.cur;
 260
 261   for (;;)
 262     {
 263       unsigned int c = *cur++;
 264       *out++ = c;
 265
 266       if (is_nvspace (c) && c)
 267         continue;
 268
 269       if (!c && cur - 1 != RLIMIT (pfile->context))
 270         continue;
 271
 272       if (c == '/' && skip_comments)
 273         {
 274           const uchar *tmp = skip_escaped_newlines (pfile, cur);
 275           if (*tmp == '*')
 276             {
 277               pfile->out.cur = out;
 278               cur = copy_comment (pfile, tmp, false /* in_define */);
 279               out = pfile->out.cur;
 280               continue;
 281             }
 282         }
 283
 284       out--;
 285       if (c == '\\' && is_vspace (*cur))
 286         {
 287           cur = skip_escaped_newlines (pfile, cur - 1);
 288           continue;
 289         }
 290
 291       break;
 292     }
 293
 294   pfile->out.cur = out;
 295   return cur - 1;
 296 }
 297
 298 /* Lexes and outputs an identifier starting at CUR, which is assumed
 299    to point to a valid first character of an identifier.  Returns
 300    the hashnode, and updates out.cur.  */
 301 static cpp_hashnode *
 302 lex_identifier (pfile, cur)
 303      cpp_reader *pfile;
 304      const uchar *cur;
 305 {
 306   size_t len;
 307   uchar *out = pfile->out.cur;
 308   cpp_hashnode *result;
 309
 310   do
 311     {
 312       do
 313         *out++ = *cur++;
 314       while (is_numchar (*cur));
 315       cur = skip_escaped_newlines (pfile, cur);
 316     }
 317   while (is_numchar (*cur));
 318
 319   CUR (pfile->context) = cur;
 320   len = out - pfile->out.cur;
 321   result = (cpp_hashnode *) ht_lookup (pfile->hash_table, pfile->out.cur,
 322                                        len, HT_ALLOC);
 323   pfile->out.cur = out;
 324   return result;
 325 }
 326
 327 /* Overlays the true file buffer temporarily with text of length LEN
 328    starting at START.  The true buffer is restored upon calling
 329    restore_buff().  */
 330 void
 331 _cpp_overlay_buffer (pfile, start, len)
 332      cpp_reader *pfile;
 333      const uchar *start;
 334      size_t len;
 335 {
 336   cpp_buffer *buffer = pfile->buffer;
 337
 338   pfile->overlaid_buffer = buffer;
 339   buffer->saved_cur = buffer->cur;
 340   buffer->saved_rlimit = buffer->rlimit;
 341
 342   buffer->cur = start;
 343   buffer->rlimit = start + len;
 344
 345   pfile->saved_line = pfile->line;
 346 }
 347
 348 /* Restores a buffer overlaid by _cpp_overlay_buffer().  */
 349 void
 350 _cpp_remove_overlay (pfile)
 351      cpp_reader *pfile;
 352 {
 353   cpp_buffer *buffer = pfile->overlaid_buffer;
 354
 355   buffer->cur = buffer->saved_cur;
 356   buffer->rlimit = buffer->saved_rlimit;
 357
 358   pfile->line = pfile->saved_line;
 359 }
 360
 361 /* Reads a logical line into the output buffer.  Returns TRUE if there
 362    is more text left in the buffer.  */
 363 bool
 364 _cpp_read_logical_line_trad (pfile)
 365      cpp_reader *pfile;
 366 {
 367   do
 368     {
 369       if (pfile->buffer->cur == pfile->buffer->rlimit)
 370         {
 371           bool stop = true;
 372
 373           /* Don't pop the last buffer.  */
 374           if (pfile->buffer->prev)
 375             {
 376               stop = pfile->buffer->return_at_eof;
 377               _cpp_pop_buffer (pfile);
 378             }
 379
 380           if (stop)
 381             return false;
 382         }
 383
 384       scan_out_logical_line (pfile, NULL);
 385     }
 386   while (pfile->state.skipping);
 387
 388   return true;
 389 }
 390
 391 /* Set up state for finding the opening '(' of a function-like
 392    macro.  */
 393 static void
 394 maybe_start_funlike (pfile, node, start, macro)
 395      cpp_reader *pfile;
 396      cpp_hashnode *node;
 397      const uchar *start;
 398      struct fun_macro *macro;
 399 {
 400   unsigned int n = node->value.macro->paramc + 1;
 401
 402   if (macro->buff)
 403     _cpp_release_buff (pfile, macro->buff);
 404   macro->buff = _cpp_get_buff (pfile, n * sizeof (size_t));
 405   macro->args = (size_t *) BUFF_FRONT (macro->buff);
 406   macro->node = node;
 407   macro->offset = start - pfile->out.base;
 408   macro->argc = 0;
 409 }
 410
 411 /* Save the OFFSET of the start of the next argument to MACRO.  */
 412 static void
 413 save_argument (macro, offset)
 414      struct fun_macro *macro;
 415      size_t offset;
 416 {
 417   macro->argc++;
 418   if (macro->argc <= macro->node->value.macro->paramc)
 419     macro->args[macro->argc] = offset;
 420 }
 421
 422 /* Copies the next logical line in the current buffer (starting at
 423    buffer->cur) to the output buffer.  The output is guaranteed to
 424    terminate with a NUL character.  buffer->cur is updated.
 425
 426    If MACRO is non-NULL, then we are scanning the replacement list of
 427    MACRO, and we call save_replacement_text() every time we meet an
 428    argument.  */
 429 static void
 430 scan_out_logical_line (pfile, macro)
 431      cpp_reader *pfile;
 432      cpp_macro *macro;
 433 {
 434   cpp_context *context;
 435   const uchar *cur;
 436   uchar *out;
 437   struct fun_macro fmacro;
 438   unsigned int c, paren_depth = 0, quote;
 439   enum ls lex_state = ls_none;
 440   bool header_ok;
 441
 442   fmacro.buff = NULL;
 443
 444  start_logical_line:
 445   quote = 0;
 446   header_ok = pfile->state.angled_headers;
 447   CUR (pfile->context) = pfile->buffer->cur;
 448   RLIMIT (pfile->context) = pfile->buffer->rlimit;
 449   pfile->out.cur = pfile->out.base;
 450   pfile->out.first_line = pfile->line;
 451  new_context:
 452   context = pfile->context;
 453   cur = CUR (context);
 454   check_output_buffer (pfile, RLIMIT (context) - cur);
 455   out = pfile->out.cur;
 456
 457   for (;;)
 458     {
 459       c = *cur++;
 460       *out++ = c;
 461
 462       /* Whitespace should "continue" out of the switch,
 463          non-whitespace should "break" out of it.  */
 464       switch (c)
 465         {
 466         case ' ':
 467         case '\t':
 468         case '\f':
 469         case '\v':
 470           continue;
 471
 472         case '\0':
 473           if (cur - 1 != RLIMIT (context))
 474             continue;
 475
 476           /* If this is a macro's expansion, pop it.  */
 477           if (context->prev)
 478             {
 479               pfile->out.cur = out - 1;
 480               _cpp_pop_context (pfile);
 481               goto new_context;
 482             }
 483
 484           /* Premature end of file.  Fake a new line.  */
 485           cur--;
 486           if (!pfile->buffer->from_stage3)
 487             cpp_error (pfile, DL_PEDWARN, "no newline at end of file");
 488           pfile->line++;
 489           goto done;
 490
 491         case '\r': case '\n':
 492           cur = handle_newline (pfile, cur - 1);
 493           if ((lex_state == ls_fun_open || lex_state == ls_fun_close)
 494               && !pfile->state.in_directive)
 495             {
 496               /* Newlines in arguments become a space, but we don't
 497                  clear any in-progress quote.  */
 498               if (lex_state == ls_fun_close)
 499                 out[-1] = ' ';
 500               continue;
 501             }
 502           goto done;
 503
 504         case '<':
 505           if (header_ok)
 506             quote = '>';
 507           break;
 508         case '>':
 509           if (c == quote)
 510             quote = 0;
 511           break;
 512
 513         case '"':
 514         case '\'':
 515           if (c == quote)
 516             quote = 0;
 517           else if (!quote)
 518             quote = c;
 519           break;
 520
 521         case '\\':
 522           if (is_vspace (*cur))
 523             {
 524               out--;
 525               cur = skip_escaped_newlines (pfile, cur - 1);
 526               continue;
 527             }
 528           else
 529             {
 530               /* Skip escaped quotes here, it's easier than above, but
 531                  take care to first skip escaped newlines.  */
 532               cur = skip_escaped_newlines (pfile, cur);
 533               if (*cur == '\\' || *cur == '"' || *cur == '\'')
 534                 *out++ = *cur++;
 535             }
 536           break;
 537
 538         case '/':
 539           /* Traditional CPP does not recognize comments within
 540              literals.  */
 541           if (!quote)
 542             {
 543               cur = skip_escaped_newlines (pfile, cur);
 544               if (*cur == '*')
 545                 {
 546                   pfile->out.cur = out;
 547                   cur = copy_comment (pfile, cur, macro != 0);
 548                   out = pfile->out.cur;
 549                   continue;
 550                 }
 551             }
 552           break;
 553
 554         case '_':
 555         case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
 556         case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
 557         case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
 558         case 's': case 't': case 'u': case 'v': case 'w': case 'x':
 559         case 'y': case 'z':
 560         case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
 561         case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
 562         case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
 563         case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
 564         case 'Y': case 'Z':
 565           if (!pfile->state.skipping && (quote == 0 || macro))
 566             {
 567               cpp_hashnode *node;
 568               uchar *out_start = out - 1;
 569
 570               pfile->out.cur = out_start;
 571               node = lex_identifier (pfile, cur - 1);
 572               out = pfile->out.cur;
 573               cur = CUR (context);
 574
 575               if (node->type == NT_MACRO
 576                   /* Should we expand for ls_answer?  */
 577                   && (lex_state == ls_none || lex_state == ls_fun_open)
 578                   && !pfile->state.prevent_expansion)
 579                 {
 580                   /* Macros invalidate MI optimization.  */
 581                   pfile->mi_valid = false;
 582                   if (! (node->flags & NODE_BUILTIN)
 583                       && node->value.macro->fun_like)
 584                     {
 585                       maybe_start_funlike (pfile, node, out_start, &fmacro);
 586                       lex_state = ls_fun_open;
 587                       fmacro.line = pfile->line;
 588                       continue;
 589                     }
 590                   else if (!recursive_macro (pfile, node))
 591                     {
 592                       /* Remove the object-like macro's name from the
 593                          output, and push its replacement text.  */
 594                       pfile->out.cur = out_start;
 595                       push_replacement_text (pfile, node);
 596                       lex_state = ls_none;
 597                       goto new_context;
 598                     }
 599                 }
 600               else if (macro && node->arg_index)
 601                 {
 602                   /* Found a parameter in the replacement text of a
 603                      #define.  Remove its name from the output.  */
 604                   pfile->out.cur = out_start;
 605                   save_replacement_text (pfile, macro, node->arg_index);
 606                   out = pfile->out.base;
 607                 }
 608               else if (lex_state == ls_hash)
 609                 {
 610                   lex_state = ls_predicate;
 611                   continue;
 612                 }
 613               else if (pfile->state.in_expression
 614                        && node == pfile->spec_nodes.n_defined)
 615                 {
 616                   lex_state = ls_defined;
 617                   continue;
 618                 }
 619             }
 620           break;
 621
 622         case '(':
 623           if (quote == 0)
 624             {
 625               paren_depth++;
 626               if (lex_state == ls_fun_open)
 627                 {
 628                   if (recursive_macro (pfile, fmacro.node))
 629                     lex_state = ls_none;
 630                   else
 631                     {
 632                       lex_state = ls_fun_close;
 633                       paren_depth = 1;
 634                       out = pfile->out.base + fmacro.offset;
 635                       fmacro.args[0] = fmacro.offset;
 636                     }
 637                 }
 638               else if (lex_state == ls_predicate)
 639                 lex_state = ls_answer;
 640               else if (lex_state == ls_defined)
 641                 lex_state = ls_defined_close;
 642             }
 643           break;
 644
 645         case ',':
 646           if (quote == 0 && lex_state == ls_fun_close && paren_depth == 1)
 647             save_argument (&fmacro, out - pfile->out.base);
 648           break;
 649
 650         case ')':
 651           if (quote == 0)
 652             {
 653               paren_depth--;
 654               if (lex_state == ls_fun_close && paren_depth == 0)
 655                 {
 656                   cpp_macro *m = fmacro.node->value.macro;
 657
 658                   lex_state = ls_none;
 659                   save_argument (&fmacro, out - pfile->out.base);
 660
 661                   /* A single zero-length argument is no argument.  */
 662                   if (fmacro.argc == 1
 663                       && m->paramc == 0
 664                       && out == pfile->out.base + fmacro.offset + 1)
 665                     fmacro.argc = 0;
 666
 667                   if (_cpp_arguments_ok (pfile, m, fmacro.node, fmacro.argc))
 668                     {
 669                       /* Remove the macro's invocation from the
 670                          output, and push its replacement text.  */
 671                       pfile->out.cur = (pfile->out.base
 672                                              + fmacro.offset);
 673                       CUR (context) = cur;
 674                       replace_args_and_push (pfile, &fmacro);
 675                       goto new_context;
 676                     }
 677                 }
 678               else if (lex_state == ls_answer || lex_state == ls_defined_close)
 679                 lex_state = ls_none;
 680             }
 681           break;
 682
 683         case '#':
 684           if (out - 1 == pfile->out.base && !pfile->state.in_directive)
 685             {
 686               /* A directive.  With the way _cpp_handle_directive
 687                  currently works, we only want to call it if either we
 688                  know the directive is OK, or we want it to fail and
 689                  be removed from the output.  If we want it to be
 690                  passed through (the assembler case) then we must not
 691                  call _cpp_handle_directive.  */
 692               pfile->out.cur = out;
 693               cur = skip_whitespace (pfile, cur, true /* skip_comments */);
 694               out = pfile->out.cur;
 695
 696               if (is_vspace (*cur))
 697                 {
 698                   /* Null directive.  Ignore it and don't invalidate
 699                      the MI optimization.  */
 700                   out = pfile->out.base;
 701                   continue;
 702                 }
 703               else
 704                 {
 705                   bool do_it = false;
 706
 707                   if (is_numstart (*cur))
 708                     do_it = true;
 709                   else if (is_idstart (*cur))
 710                     /* Check whether we know this directive, but don't
 711                        advance.  */
 712                     do_it = lex_identifier (pfile, cur)->directive_index != 0;
 713
 714                   if (do_it || CPP_OPTION (pfile, lang) != CLK_ASM)
 715                     {
 716                       /* This is a kludge.  We want to have the ISO
 717                          preprocessor lex the next token.  */
 718                       pfile->buffer->cur = cur;
 719                       _cpp_handle_directive (pfile, false /* indented */);
 720                       /* #include changes pfile->buffer so we need to
 721                          update the limits of the current context.  */
 722                       goto start_logical_line;
 723                     }
 724                 }
 725             }
 726
 727           if (pfile->state.in_expression)
 728             {
 729               lex_state = ls_hash;
 730               continue;
 731             }
 732           break;
 733
 734         default:
 735           break;
 736         }
 737
 738       /* Non-whitespace disables MI optimization and stops treating
 739          '<' as a quote in #include.  */
 740       header_ok = false;
 741       if (!pfile->state.in_directive)
 742         pfile->mi_valid = false;
 743
 744       if (lex_state == ls_none)
 745         continue;
 746
 747       /* Some of these transitions of state are syntax errors.  The
 748          ISO preprocessor will issue errors later.  */
 749       if (lex_state == ls_fun_open)
 750         /* Missing '('.  */
 751         lex_state = ls_none;
 752       else if (lex_state == ls_hash
 753                || lex_state == ls_predicate
 754                || lex_state == ls_defined)
 755         lex_state = ls_none;
 756
 757       /* ls_answer and ls_defined_close keep going until ')'.  */
 758     }
 759
 760  done:
 761   out[-1] = '\0';
 762   pfile->buffer->cur = cur;
 763   pfile->out.cur = out - 1;
 764   if (fmacro.buff)
 765     _cpp_release_buff (pfile, fmacro.buff);
 766
 767   if (lex_state == ls_fun_close)
 768     cpp_error_with_line (pfile, DL_ERROR, fmacro.line, 0,
 769                          "unterminated argument list invoking macro \"%s\"",
 770                          NODE_NAME (fmacro.node));
 771 }
 772
 773 /* Push a context holding the replacement text of the macro NODE on
 774    the context stack.  NODE is either object-like, or a function-like
 775    macro with no arguments.  */
 776 static void
 777 push_replacement_text (pfile, node)
 778      cpp_reader *pfile;
 779      cpp_hashnode *node;
 780 {
 781   size_t len;
 782   const uchar *text;
 783
 784   if (node->flags & NODE_BUILTIN)
 785     {
 786       text = _cpp_builtin_macro_text (pfile, node);
 787       len = ustrlen (text);
 788     }
 789   else
 790     {
 791       cpp_macro *macro = node->value.macro;
 792       text = macro->exp.text;
 793       len = macro->count;
 794     }
 795
 796   _cpp_push_text_context (pfile, node, text, len);
 797 }
 798
 799 /* Returns TRUE if traditional macro recursion is detected.  */
 800 static bool
 801 recursive_macro (pfile, node)
 802      cpp_reader *pfile;
 803      cpp_hashnode *node;
 804 {
 805   bool recursing = node->flags & NODE_DISABLED;
 806
 807   /* Object-like macros that are already expanding are necessarily
 808      recursive.
 809
 810      However, it is possible to have traditional function-like macros
 811      that are not infinitely recursive but recurse to any given depth.
 812      Further, it is easy to construct examples that get ever longer
 813      until the point they stop recursing.  So there is no easy way to
 814      detect true recursion; instead we assume any expansion more than
 815      20 deep since the first invocation of this macro must be
 816      recursing.  */
 817   if (recursing && node->value.macro->fun_like)
 818     {
 819       size_t depth = 0;
 820       cpp_context *context = pfile->context;
 821
 822       do
 823         {
 824           depth++;
 825           if (context->macro == node && depth > 20)
 826             break;
 827           context = context->prev;
 828         }
 829       while (context);
 830       recursing = context != NULL;
 831     }
 832
 833   if (recursing)
 834     cpp_error (pfile, DL_ERROR,
 835                "detected recursion whilst expanding macro \"%s\"",
 836                NODE_NAME (node));
 837
 838   return recursing;
 839 }
 840
 841 /* Return the length of the replacement text of a function-like or
 842    object-like non-builtin macro.  */
 843 size_t
 844 _cpp_replacement_text_len (macro)
 845      const cpp_macro *macro;
 846 {
 847   size_t len;
 848
 849   if (macro->fun_like)
 850     {
 851       const uchar *exp;
 852
 853       len = 0;
 854       for (exp = macro->exp.text;;)
 855         {
 856           struct block *b = (struct block *) exp;
 857
 858           len += b->text_len;
 859           if (b->arg_index == 0)
 860             break;
 861           len += NODE_LEN (macro->params[b->arg_index - 1]);
 862           exp += BLOCK_LEN (b->text_len);
 863         }
 864     }
 865   else
 866     len = macro->count;
 867
 868   return len;
 869 }
 870
 871 /* Copy the replacement text of MACRO to DEST, which must be of
 872    sufficient size.  It is not NUL-terminated.  The next character is
 873    returned.  */
 874 uchar *
 875 _cpp_copy_replacement_text (macro, dest)
 876      const cpp_macro *macro;
 877      uchar *dest;
 878 {
 879   if (macro->fun_like)
 880     {
 881       const uchar *exp;
 882
 883       for (exp = macro->exp.text;;)
 884         {
 885           struct block *b = (struct block *) exp;
 886           cpp_hashnode *param;
 887
 888           memcpy (dest, b->text, b->text_len);
 889           dest += b->text_len;
 890           if (b->arg_index == 0)
 891             break;
 892           param = macro->params[b->arg_index - 1];
 893           memcpy (dest, NODE_NAME (param), NODE_LEN (param));
 894           dest += NODE_LEN (param);
 895           exp += BLOCK_LEN (b->text_len);
 896         }
 897     }
 898   else
 899     {
 900       memcpy (dest, macro->exp.text, macro->count);
 901       dest += macro->count;
 902     }
 903
 904   return dest;
 905 }
 906
 907 /* Push a context holding the replacement text of the macro NODE on
 908    the context stack.  NODE is either object-like, or a function-like
 909    macro with no arguments.  */
 910 static void
 911 replace_args_and_push (pfile, fmacro)
 912      cpp_reader *pfile;
 913      struct fun_macro *fmacro;
 914 {
 915   cpp_macro *macro = fmacro->node->value.macro;
 916
 917   if (macro->paramc == 0)
 918     push_replacement_text (pfile, fmacro->node);
 919   else
 920     {
 921       const uchar *exp;
 922       uchar *p;
 923       _cpp_buff *buff;
 924       size_t len = 0;
 925
 926       /* Calculate the length of the argument-replaced text.  */
 927       for (exp = macro->exp.text;;)
 928         {
 929           struct block *b = (struct block *) exp;
 930
 931           len += b->text_len;
 932           if (b->arg_index == 0)
 933             break;
 934           len += (fmacro->args[b->arg_index]
 935                   - fmacro->args[b->arg_index - 1] - 1);
 936           exp += BLOCK_LEN (b->text_len);
 937         }
 938
 939       /* Allocate room for the expansion plus NUL.  */
 940       buff = _cpp_get_buff (pfile, len + 1);
 941
 942       /* Copy the expansion and replace arguments.  */
 943       p = BUFF_FRONT (buff);
 944       for (exp = macro->exp.text;;)
 945         {
 946           struct block *b = (struct block *) exp;
 947           size_t arglen;
 948
 949           memcpy (p, b->text, b->text_len);
 950           p += b->text_len;
 951           if (b->arg_index == 0)
 952             break;
 953           arglen = (fmacro->args[b->arg_index]
 954                     - fmacro->args[b->arg_index - 1] - 1);
 955           memcpy (p, pfile->out.base + fmacro->args[b->arg_index - 1],
 956                   arglen);
 957           p += arglen;
 958           exp += BLOCK_LEN (b->text_len);
 959         }
 960
 961       /* NUL-terminate.  */
 962       *p = '\0';
 963       _cpp_push_text_context (pfile, fmacro->node, BUFF_FRONT (buff), len);
 964
 965       /* So we free buffer allocation when macro is left.  */
 966       pfile->context->buff = buff;
 967     }
 968 }
 969
 970 /* Read and record the parameters, if any, of a function-like macro
 971    definition.  Destroys pfile->out.cur.
 972
 973    Returns true on success, false on failure (syntax error or a
 974    duplicate parameter).  On success, CUR (pfile->context) is just
 975    past the closing parenthesis.  */
 976 static bool
 977 scan_parameters (pfile, macro)
 978      cpp_reader *pfile;
 979      cpp_macro *macro;
 980 {
 981   const uchar *cur = CUR (pfile->context) + 1;
 982   bool ok;
 983
 984   for (;;)
 985     {
 986       cur = skip_whitespace (pfile, cur, true /* skip_comments */);
 987
 988       if (is_idstart (*cur))
 989         {
 990           ok = false;
 991           if (_cpp_save_parameter (pfile, macro, lex_identifier (pfile, cur)))
 992             break;
 993           cur = skip_whitespace (pfile, CUR (pfile->context),
 994                                  true /* skip_comments */);
 995           if (*cur == ',')
 996             {
 997               cur++;
 998               continue;
 999             }
1000           ok = (*cur == ')');
1001           break;
1002         }
1003
1004       ok = (*cur == ')' && macro->paramc == 0);
1005       break;
1006     }
1007
1008   CUR (pfile->context) = cur + (*cur == ')');
1009
1010   return ok;
1011 }
1012
1013 /* Save the text from pfile->out.base to pfile->out.cur as
1014    the replacement text for the current macro, followed by argument
1015    ARG_INDEX, with zero indicating the end of the replacement
1016    text.  */
1017 static void
1018 save_replacement_text (pfile, macro, arg_index)
1019      cpp_reader *pfile;
1020      cpp_macro *macro;
1021      unsigned int arg_index;
1022 {
1023   size_t len = pfile->out.cur - pfile->out.base;
1024   uchar *exp;
1025
1026   if (macro->paramc == 0)
1027     {
1028       /* Object-like and function-like macros without parameters
1029          simply store their NUL-terminated replacement text.  */
1030       exp = _cpp_unaligned_alloc (pfile, len + 1);
1031       memcpy (exp, pfile->out.base, len);
1032       exp[len] = '\0';
1033       macro->exp.text = exp;
1034       macro->count = len;
1035     }
1036   else
1037     {
1038       /* Store the text's length (unsigned int), the argument index
1039          (unsigned short, base 1) and then the text.  */
1040       size_t blen = BLOCK_LEN (len);
1041       struct block *block;
1042
1043       if (macro->count + blen > BUFF_ROOM (pfile->a_buff))
1044         _cpp_extend_buff (pfile, &pfile->a_buff, macro->count + blen);
1045
1046       exp = BUFF_FRONT (pfile->a_buff);
1047       block = (struct block *) (exp + macro->count);
1048       macro->exp.text = exp;
1049
1050       /* Write out the block information.  */
1051       block->text_len = len;
1052       block->arg_index = arg_index;
1053       memcpy (block->text, pfile->out.base, len);
1054
1055       /* Lex the rest into the start of the output buffer.  */
1056       pfile->out.cur = pfile->out.base;
1057
1058       macro->count += blen;
1059
1060       /* If we've finished, commit the memory.  */
1061       if (arg_index == 0)
1062         BUFF_FRONT (pfile->a_buff) += macro->count;
1063     }
1064 }
1065
1066 /* Analyze and save the replacement text of a macro.  Returns true on
1067    success.  */
1068 bool
1069 _cpp_create_trad_definition (pfile, macro)
1070      cpp_reader *pfile;
1071      cpp_macro *macro;
1072 {
1073   const uchar *cur;
1074   uchar *limit;
1075   cpp_context *context = pfile->context;
1076
1077   /* The context has not been set up for command line defines, and CUR
1078      has not been updated for the macro name for in-file defines.  */
1079   pfile->out.cur = pfile->out.base;
1080   CUR (context) = pfile->buffer->cur;
1081   RLIMIT (context) = pfile->buffer->rlimit;
1082   check_output_buffer (pfile, RLIMIT (context) - CUR (context));
1083
1084   /* Is this a function-like macro?  */
1085   if (* CUR (context) == '(')
1086     {
1087       /* Setting macro to NULL indicates an error occurred, and
1088          prevents unnecessary work in scan_out_logical_line.  */
1089       if (!scan_parameters (pfile, macro))
1090         macro = NULL;
1091       else
1092         {
1093           /* Success.  Commit the parameter array.  */
1094           macro->params = (cpp_hashnode **) BUFF_FRONT (pfile->a_buff);
1095           BUFF_FRONT (pfile->a_buff) = (uchar *) &macro->params[macro->paramc];
1096           macro->fun_like = 1;
1097         }
1098     }
1099
1100   /* Skip leading whitespace in the replacement text.  */
1101   pfile->buffer->cur
1102     = skip_whitespace (pfile, CUR (context),
1103                        CPP_OPTION (pfile, discard_comments_in_macro_exp));
1104
1105   pfile->state.prevent_expansion++;
1106   scan_out_logical_line (pfile, macro);
1107   pfile->state.prevent_expansion--;
1108
1109   if (!macro)
1110     return false;
1111
1112   /* Skip trailing white space.  */
1113   cur = pfile->out.base;
1114   limit = pfile->out.cur;
1115   while (limit > cur && is_space (limit[-1]))
1116     limit--;
1117   pfile->out.cur = limit;
1118   save_replacement_text (pfile, macro, 0);
1119
1120   return true;
1121 }
1122
1123 /* Copy SRC of length LEN to DEST, but convert all contiguous
1124    whitespace to a single space, provided it is not in quotes.  The
1125    quote currently in effect is pointed to by PQUOTE, and is updated
1126    by the function.  Returns the number of bytes copied.  */
1127 static size_t
1128 canonicalize_text (dest, src, len, pquote)
1129      uchar *dest;
1130      const uchar *src;
1131      size_t len;
1132      uchar *pquote;
1133 {
1134   uchar *orig_dest = dest;
1135   uchar quote = *pquote;
1136
1137   while (len)
1138     {
1139       if (is_space (*src) && !quote)
1140         {
1141           do
1142             src++, len--;
1143           while (len && is_space (*src));
1144           *dest++ = ' ';
1145         }
1146       else
1147         {
1148           if (*src == '\'' || *src == '"')
1149             {
1150               if (!quote)
1151                 quote = *src;
1152               else if (quote == *src)
1153                 quote = 0;
1154             }
1155           *dest++ = *src++, len--;
1156         }
1157     }
1158
1159   *pquote = quote;
1160   return dest - orig_dest;
1161 }
1162
1163 /* Returns true if MACRO1 and MACRO2 have expansions different other
1164    than in the form of their whitespace.  */
1165 bool
1166 _cpp_expansions_different_trad (macro1, macro2)
1167      const cpp_macro *macro1, *macro2;
1168 {
1169   uchar *p1 = xmalloc (macro1->count + macro2->count);
1170   uchar *p2 = p1 + macro1->count;
1171   uchar quote1 = 0, quote2 = 0;
1172   bool mismatch;
1173   size_t len1, len2;
1174
1175   if (macro1->paramc > 0)
1176     {
1177       const uchar *exp1 = macro1->exp.text, *exp2 = macro2->exp.text;
1178
1179       mismatch = true;
1180       for (;;)
1181         {
1182           struct block *b1 = (struct block *) exp1;
1183           struct block *b2 = (struct block *) exp2;
1184
1185           if (b1->arg_index != b2->arg_index)
1186             break;
1187
1188           len1 = canonicalize_text (p1, b1->text, b1->text_len, &quote1);
1189           len2 = canonicalize_text (p2, b2->text, b2->text_len, &quote2);
1190           if (len1 != len2 || memcmp (p1, p2, len1))
1191             break;
1192           if (b1->arg_index == 0)
1193             {
1194               mismatch = false;
1195               break;
1196             }
1197           exp1 += BLOCK_LEN (b1->text_len);
1198           exp2 += BLOCK_LEN (b2->text_len);
1199         }
1200     }
1201   else
1202     {
1203       len1 = canonicalize_text (p1, macro1->exp.text, macro1->count, &quote1);
1204       len2 = canonicalize_text (p2, macro2->exp.text, macro2->count, &quote2);
1205       mismatch = (len1 != len2 || memcmp (p1, p2, len1));
1206     }
1207
1208   free (p1);
1209   return mismatch;
1210 }