libcpp/traditional.c

   1 /* CPP Library - traditional lexical analysis and macro expansion.
   2    Copyright (C) 2002-2018 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 3, 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; see the file COPYING3.  If not see
  17 <http://www.gnu.org/licenses/>.  */
  18
  19 #include "config.h"
  20 #include "system.h"
  21 #include "cpplib.h"
  22 #include "internal.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   source_location line;
  64
  65   /* Number of parameters.  */
  66   unsigned int paramc;
  67
  68   /* Zero-based index of argument being currently lexed.  */
  69   unsigned int argc;
  70 };
  71
  72 /* Lexing state.  It is mostly used to prevent macro expansion.  */
  73 enum ls {ls_none = 0,           /* Normal state.  */
  74          ls_fun_open,           /* When looking for '('.  */
  75          ls_fun_close,          /* When looking for ')'.  */
  76          ls_defined,            /* After defined.  */
  77          ls_defined_close,      /* Looking for ')' of defined().  */
  78          ls_hash,               /* After # in preprocessor conditional.  */
  79          ls_predicate,          /* After the predicate, maybe paren?  */
  80          ls_answer,             /* In answer to predicate.  */
  81          ls_has_include,        /* After __has_include__.  */
  82          ls_has_include_close}; /* Looking for ')' of __has_include__.  */
  83
  84 /* Lexing TODO: Maybe handle space in escaped newlines.  Stop lex.c
  85    from recognizing comments and directives during its lexing pass.  */
  86
  87 static const uchar *skip_whitespace (cpp_reader *, const uchar *, int);
  88 static cpp_hashnode *lex_identifier (cpp_reader *, const uchar *);
  89 static const uchar *copy_comment (cpp_reader *, const uchar *, int);
  90 static void check_output_buffer (cpp_reader *, size_t);
  91 static void push_replacement_text (cpp_reader *, cpp_hashnode *);
  92 static bool scan_parameters (cpp_reader *, cpp_macro *);
  93 static bool recursive_macro (cpp_reader *, cpp_hashnode *);
  94 static void save_replacement_text (cpp_reader *, cpp_macro *, unsigned int);
  95 static void maybe_start_funlike (cpp_reader *, cpp_hashnode *, const uchar *,
  96                                  struct fun_macro *);
  97 static void save_argument (struct fun_macro *, size_t);
  98 static void replace_args_and_push (cpp_reader *, struct fun_macro *);
  99 static size_t canonicalize_text (uchar *, const uchar *, size_t, uchar *);
 100
 101 /* Ensures we have N bytes' space in the output buffer, and
 102    reallocates it if not.  */
 103 static void
 104 check_output_buffer (cpp_reader *pfile, size_t n)
 105 {
 106   /* We might need two bytes to terminate an unterminated comment, and
 107      one more to terminate the line with a NUL.  */
 108   n += 2 + 1;
 109
 110   if (n > (size_t) (pfile->out.limit - pfile->out.cur))
 111     {
 112       size_t size = pfile->out.cur - pfile->out.base;
 113       size_t new_size = (size + n) * 3 / 2;
 114
 115       pfile->out.base = XRESIZEVEC (unsigned char, pfile->out.base, new_size);
 116       pfile->out.limit = pfile->out.base + new_size;
 117       pfile->out.cur = pfile->out.base + size;
 118     }
 119 }
 120
 121 /* Skip a C-style block comment in a macro as a result of -CC.
 122    Buffer->cur points to the initial asterisk of the comment.  */
 123 static void
 124 skip_macro_block_comment (cpp_reader *pfile)
 125 {
 126   const uchar *cur = pfile->buffer->cur;
 127
 128   cur++;
 129   if (*cur == '/')
 130     cur++;
 131
 132   /* People like decorating comments with '*', so check for '/'
 133      instead for efficiency.  */
 134   while(! (*cur++ == '/' && cur[-2] == '*') )
 135     ;
 136
 137   pfile->buffer->cur = cur;
 138 }
 139
 140 /* CUR points to the asterisk introducing a comment in the current
 141    context.  IN_DEFINE is true if we are in the replacement text of a
 142    macro.
 143
 144    The asterisk and following comment is copied to the buffer pointed
 145    to by pfile->out.cur, which must be of sufficient size.
 146    Unterminated comments are diagnosed, and correctly terminated in
 147    the output.  pfile->out.cur is updated depending upon IN_DEFINE,
 148    -C, -CC and pfile->state.in_directive.
 149
 150    Returns a pointer to the first character after the comment in the
 151    input buffer.  */
 152 static const uchar *
 153 copy_comment (cpp_reader *pfile, const uchar *cur, int in_define)
 154 {
 155   bool unterminated, copy = false;
 156   source_location src_loc = pfile->line_table->highest_line;
 157   cpp_buffer *buffer = pfile->buffer;
 158
 159   buffer->cur = cur;
 160   if (pfile->context->prev)
 161     unterminated = false, skip_macro_block_comment (pfile);
 162   else
 163     unterminated = _cpp_skip_block_comment (pfile);
 164
 165   if (unterminated)
 166     cpp_error_with_line (pfile, CPP_DL_ERROR, src_loc, 0,
 167                          "unterminated comment");
 168
 169   /* Comments in directives become spaces so that tokens are properly
 170      separated when the ISO preprocessor re-lexes the line.  The
 171      exception is #define.  */
 172   if (pfile->state.in_directive)
 173     {
 174       if (in_define)
 175         {
 176           if (CPP_OPTION (pfile, discard_comments_in_macro_exp))
 177             pfile->out.cur--;
 178           else
 179             copy = true;
 180         }
 181       else
 182         pfile->out.cur[-1] = ' ';
 183     }
 184   else if (CPP_OPTION (pfile, discard_comments))
 185     pfile->out.cur--;
 186   else
 187     copy = true;
 188
 189   if (copy)
 190     {
 191       size_t len = (size_t) (buffer->cur - cur);
 192       memcpy (pfile->out.cur, cur, len);
 193       pfile->out.cur += len;
 194       if (unterminated)
 195         {
 196           *pfile->out.cur++ = '*';
 197           *pfile->out.cur++ = '/';
 198         }
 199     }
 200
 201   return buffer->cur;
 202 }
 203
 204 /* CUR points to any character in the input buffer.  Skips over all
 205    contiguous horizontal white space and NULs, including comments if
 206    SKIP_COMMENTS, until reaching the first non-horizontal-whitespace
 207    character or the end of the current context.  Escaped newlines are
 208    removed.
 209
 210    The whitespace is copied verbatim to the output buffer, except that
 211    comments are handled as described in copy_comment().
 212    pfile->out.cur is updated.
 213
 214    Returns a pointer to the first character after the whitespace in
 215    the input buffer.  */
 216 static const uchar *
 217 skip_whitespace (cpp_reader *pfile, const uchar *cur, int skip_comments)
 218 {
 219   uchar *out = pfile->out.cur;
 220
 221   for (;;)
 222     {
 223       unsigned int c = *cur++;
 224       *out++ = c;
 225
 226       if (is_nvspace (c))
 227         continue;
 228
 229       if (c == '/' && *cur == '*' && skip_comments)
 230         {
 231           pfile->out.cur = out;
 232           cur = copy_comment (pfile, cur, false /* in_define */);
 233           out = pfile->out.cur;
 234           continue;
 235         }
 236
 237       out--;
 238       break;
 239     }
 240
 241   pfile->out.cur = out;
 242   return cur - 1;
 243 }
 244
 245 /* Lexes and outputs an identifier starting at CUR, which is assumed
 246    to point to a valid first character of an identifier.  Returns
 247    the hashnode, and updates out.cur.  */
 248 static cpp_hashnode *
 249 lex_identifier (cpp_reader *pfile, const uchar *cur)
 250 {
 251   size_t len;
 252   uchar *out = pfile->out.cur;
 253   cpp_hashnode *result;
 254
 255   do
 256     *out++ = *cur++;
 257   while (is_numchar (*cur));
 258
 259   CUR (pfile->context) = cur;
 260   len = out - pfile->out.cur;
 261   result = CPP_HASHNODE (ht_lookup (pfile->hash_table, pfile->out.cur,
 262                                     len, HT_ALLOC));
 263   pfile->out.cur = out;
 264   return result;
 265 }
 266
 267 /* Overlays the true file buffer temporarily with text of length LEN
 268    starting at START.  The true buffer is restored upon calling
 269    restore_buff().  */
 270 void
 271 _cpp_overlay_buffer (cpp_reader *pfile, const uchar *start, size_t len)
 272 {
 273   cpp_buffer *buffer = pfile->buffer;
 274
 275   pfile->overlaid_buffer = buffer;
 276   pfile->saved_cur = buffer->cur;
 277   pfile->saved_rlimit = buffer->rlimit;
 278   pfile->saved_line_base = buffer->next_line;
 279   buffer->need_line = false;
 280
 281   buffer->cur = start;
 282   buffer->line_base = start;
 283   buffer->rlimit = start + len;
 284 }
 285
 286 /* Restores a buffer overlaid by _cpp_overlay_buffer().  */
 287 void
 288 _cpp_remove_overlay (cpp_reader *pfile)
 289 {
 290   cpp_buffer *buffer = pfile->overlaid_buffer;
 291
 292   buffer->cur = pfile->saved_cur;
 293   buffer->rlimit = pfile->saved_rlimit;
 294   buffer->line_base = pfile->saved_line_base;
 295   buffer->need_line = true;
 296
 297   pfile->overlaid_buffer = NULL;
 298 }
 299
 300 /* Reads a logical line into the output buffer.  Returns TRUE if there
 301    is more text left in the buffer.  */
 302 bool
 303 _cpp_read_logical_line_trad (cpp_reader *pfile)
 304 {
 305   do
 306     {
 307       if (pfile->buffer->need_line && !_cpp_get_fresh_line (pfile))
 308         return false;
 309     }
 310   while (!_cpp_scan_out_logical_line (pfile, NULL, false)
 311          || pfile->state.skipping);
 312
 313   return pfile->buffer != NULL;
 314 }
 315
 316 /* Return true if NODE is a fun_like macro.  */
 317 static inline bool
 318 fun_like_macro (cpp_hashnode *node)
 319 {
 320   if (node->flags & NODE_BUILTIN)
 321     return node->value.builtin == BT_HAS_ATTRIBUTE;
 322   else
 323     return node->value.macro->fun_like;
 324 }
 325
 326 /* Set up state for finding the opening '(' of a function-like
 327    macro.  */
 328 static void
 329 maybe_start_funlike (cpp_reader *pfile, cpp_hashnode *node, const uchar *start,
 330                      struct fun_macro *macro)
 331 {
 332   unsigned int n;
 333   if (node->flags & NODE_BUILTIN)
 334     n = 1;
 335   else
 336     n = node->value.macro->paramc;
 337
 338   if (macro->buff)
 339     _cpp_release_buff (pfile, macro->buff);
 340   macro->buff = _cpp_get_buff (pfile, (n + 1) * sizeof (size_t));
 341   macro->args = (size_t *) BUFF_FRONT (macro->buff);
 342   macro->node = node;
 343   macro->offset = start - pfile->out.base;
 344   macro->paramc = n;
 345   macro->argc = 0;
 346 }
 347
 348 /* Save the OFFSET of the start of the next argument to MACRO.  */
 349 static void
 350 save_argument (struct fun_macro *macro, size_t offset)
 351 {
 352   macro->argc++;
 353   if (macro->argc <= macro->paramc)
 354     macro->args[macro->argc] = offset;
 355 }
 356
 357 /* Copies the next logical line in the current buffer (starting at
 358    buffer->cur) to the output buffer.  The output is guaranteed to
 359    terminate with a NUL character.  buffer->cur is updated.
 360
 361    If MACRO is non-NULL, then we are scanning the replacement list of
 362    MACRO, and we call save_replacement_text() every time we meet an
 363    argument.
 364
 365    If BUILTIN_MACRO_ARG is true, this is called to macro expand
 366    arguments of builtin function-like macros.  */
 367 bool
 368 _cpp_scan_out_logical_line (cpp_reader *pfile, cpp_macro *macro,
 369                             bool builtin_macro_arg)
 370 {
 371   bool result = true;
 372   cpp_context *context;
 373   const uchar *cur;
 374   uchar *out;
 375   struct fun_macro fmacro;
 376   unsigned int c, paren_depth = 0, quote;
 377   enum ls lex_state = ls_none;
 378   bool header_ok;
 379   const uchar *start_of_input_line;
 380
 381   fmacro.buff = NULL;
 382   fmacro.args = NULL;
 383   fmacro.node = NULL;
 384   fmacro.offset = 0;
 385   fmacro.line = 0;
 386   fmacro.paramc = 0;
 387   fmacro.argc = 0;
 388
 389   quote = 0;
 390   header_ok = pfile->state.angled_headers;
 391   CUR (pfile->context) = pfile->buffer->cur;
 392   RLIMIT (pfile->context) = pfile->buffer->rlimit;
 393   if (!builtin_macro_arg)
 394     {
 395       pfile->out.cur = pfile->out.base;
 396       pfile->out.first_line = pfile->line_table->highest_line;
 397     }
 398   /* start_of_input_line is needed to make sure that directives really,
 399      really start at the first character of the line.  */
 400   start_of_input_line = pfile->buffer->cur;
 401  new_context:
 402   context = pfile->context;
 403   cur = CUR (context);
 404   check_output_buffer (pfile, RLIMIT (context) - cur);
 405   out = pfile->out.cur;
 406
 407   for (;;)
 408     {
 409       if (!context->prev
 410           && !builtin_macro_arg
 411           && cur >= pfile->buffer->notes[pfile->buffer->cur_note].pos)
 412         {
 413           pfile->buffer->cur = cur;
 414           _cpp_process_line_notes (pfile, false);
 415         }
 416       c = *cur++;
 417       *out++ = c;
 418
 419       /* Whitespace should "continue" out of the switch,
 420          non-whitespace should "break" out of it.  */
 421       switch (c)
 422         {
 423         case ' ':
 424         case '\t':
 425         case '\f':
 426         case '\v':
 427         case '\0':
 428           continue;
 429
 430         case '\n':
 431           /* If this is a macro's expansion, pop it.  */
 432           if (context->prev)
 433             {
 434               pfile->out.cur = out - 1;
 435               _cpp_pop_context (pfile);
 436               goto new_context;
 437             }
 438
 439           /* Omit the newline from the output buffer.  */
 440           pfile->out.cur = out - 1;
 441           pfile->buffer->cur = cur;
 442           if (builtin_macro_arg)
 443             goto done;
 444           pfile->buffer->need_line = true;
 445           CPP_INCREMENT_LINE (pfile, 0);
 446
 447           if ((lex_state == ls_fun_open || lex_state == ls_fun_close)
 448               && !pfile->state.in_directive
 449               && _cpp_get_fresh_line (pfile))
 450             {
 451               /* Newlines in arguments become a space, but we don't
 452                  clear any in-progress quote.  */
 453               if (lex_state == ls_fun_close)
 454                 out[-1] = ' ';
 455               cur = pfile->buffer->cur;
 456               continue;
 457             }
 458           goto done;
 459
 460         case '<':
 461           if (header_ok)
 462             quote = '>';
 463           break;
 464         case '>':
 465           if (c == quote)
 466             quote = 0;
 467           break;
 468
 469         case '"':
 470         case '\'':
 471           if (c == quote)
 472             quote = 0;
 473           else if (!quote)
 474             quote = c;
 475           break;
 476
 477         case '\\':
 478           /* Skip escaped quotes here, it's easier than above.  */
 479           if (*cur == '\\' || *cur == '"' || *cur == '\'')
 480             *out++ = *cur++;
 481           break;
 482
 483         case '/':
 484           /* Traditional CPP does not recognize comments within
 485              literals.  */
 486           if (!quote && *cur == '*')
 487             {
 488               pfile->out.cur = out;
 489               cur = copy_comment (pfile, cur, macro != 0);
 490               out = pfile->out.cur;
 491               continue;
 492             }
 493           break;
 494
 495         case '_':
 496         case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
 497         case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
 498         case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
 499         case 's': case 't': case 'u': case 'v': case 'w': case 'x':
 500         case 'y': case 'z':
 501         case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
 502         case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
 503         case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
 504         case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
 505         case 'Y': case 'Z':
 506           if (!pfile->state.skipping && (quote == 0 || macro))
 507             {
 508               cpp_hashnode *node;
 509               uchar *out_start = out - 1;
 510
 511               pfile->out.cur = out_start;
 512               node = lex_identifier (pfile, cur - 1);
 513               out = pfile->out.cur;
 514               cur = CUR (context);
 515
 516               if (node->type == NT_MACRO
 517                   /* Should we expand for ls_answer?  */
 518                   && (lex_state == ls_none || lex_state == ls_fun_open)
 519                   && !pfile->state.prevent_expansion)
 520                 {
 521                   /* Macros invalidate MI optimization.  */
 522                   pfile->mi_valid = false;
 523                   if (fun_like_macro (node))
 524                     {
 525                       maybe_start_funlike (pfile, node, out_start, &fmacro);
 526                       lex_state = ls_fun_open;
 527                       fmacro.line = pfile->line_table->highest_line;
 528                       continue;
 529                     }
 530                   else if (!recursive_macro (pfile, node))
 531                     {
 532                       /* Remove the object-like macro's name from the
 533                          output, and push its replacement text.  */
 534                       pfile->out.cur = out_start;
 535                       push_replacement_text (pfile, node);
 536                       lex_state = ls_none;
 537                       goto new_context;
 538                     }
 539                 }
 540               else if (macro && (node->flags & NODE_MACRO_ARG) != 0)
 541                 {
 542                   /* Found a parameter in the replacement text of a
 543                      #define.  Remove its name from the output.  */
 544                   pfile->out.cur = out_start;
 545                   save_replacement_text (pfile, macro, node->value.arg_index);
 546                   out = pfile->out.base;
 547                 }
 548               else if (lex_state == ls_hash)
 549                 {
 550                   lex_state = ls_predicate;
 551                   continue;
 552                 }
 553               else if (pfile->state.in_expression
 554                        && node == pfile->spec_nodes.n_defined)
 555                 {
 556                   lex_state = ls_defined;
 557                   continue;
 558                 }
 559               else if (pfile->state.in_expression
 560                        && (node == pfile->spec_nodes.n__has_include__
 561                         || node == pfile->spec_nodes.n__has_include_next__))
 562                 {
 563                   lex_state = ls_has_include;
 564                   continue;
 565                 }
 566             }
 567           break;
 568
 569         case '(':
 570           if (quote == 0)
 571             {
 572               paren_depth++;
 573               if (lex_state == ls_fun_open)
 574                 {
 575                   if (recursive_macro (pfile, fmacro.node))
 576                     lex_state = ls_none;
 577                   else
 578                     {
 579                       lex_state = ls_fun_close;
 580                       paren_depth = 1;
 581                       out = pfile->out.base + fmacro.offset;
 582                       fmacro.args[0] = fmacro.offset;
 583                     }
 584                 }
 585               else if (lex_state == ls_predicate)
 586                 lex_state = ls_answer;
 587               else if (lex_state == ls_defined)
 588                 lex_state = ls_defined_close;
 589               else if (lex_state == ls_has_include)
 590                 lex_state = ls_has_include_close;
 591             }
 592           break;
 593
 594         case ',':
 595           if (quote == 0 && lex_state == ls_fun_close && paren_depth == 1)
 596             save_argument (&fmacro, out - pfile->out.base);
 597           break;
 598
 599         case ')':
 600           if (quote == 0)
 601             {
 602               paren_depth--;
 603               if (lex_state == ls_fun_close && paren_depth == 0)
 604                 {
 605                   if (fmacro.node->flags & NODE_BUILTIN)
 606                     {
 607                       /* Handle builtin function-like macros like
 608                          __has_attribute.  The already parsed arguments
 609                          are put into a buffer, which is then preprocessed
 610                          and the result is fed to _cpp_push_text_context
 611                          with disabled expansion, where the ISO preprocessor
 612                          parses it.  While in traditional preprocessing
 613                          macro arguments aren't immediately expanded, they in
 614                          the end are because the macro with replaced arguments
 615                          is preprocessed again.  For the builtin function-like
 616                          macros we need the argument immediately though,
 617                          if we don't preprocess them, they would behave
 618                          very differently from ISO preprocessor handling
 619                          of those builtin macros.  So, this handling is
 620                          more similar to traditional preprocessing of
 621                          #if directives, where we also keep preprocessing
 622                          until everything is expanded, and then feed the
 623                          result with disabled expansion to ISO preprocessor
 624                          for handling the directives.  */
 625                       lex_state = ls_none;
 626                       save_argument (&fmacro, out - pfile->out.base);
 627                       cpp_macro m;
 628                       memset (&m, '\0', sizeof (m));
 629                       m.paramc = fmacro.paramc;
 630                       if (_cpp_arguments_ok (pfile, &m, fmacro.node,
 631                                              fmacro.argc))
 632                         {
 633                           size_t len = fmacro.args[1] - fmacro.args[0];
 634                           uchar *buf;
 635
 636                           /* Remove the macro's invocation from the
 637                              output, and push its replacement text.  */
 638                           pfile->out.cur = pfile->out.base + fmacro.offset;
 639                           CUR (context) = cur;
 640                           buf = _cpp_unaligned_alloc (pfile, len + 2);
 641                           buf[0] = '(';
 642                           memcpy (buf + 1, pfile->out.base + fmacro.args[0],
 643                                   len);
 644                           buf[len + 1] = '\n';
 645
 646                           const unsigned char *ctx_rlimit = RLIMIT (context);
 647                           const unsigned char *saved_cur = pfile->buffer->cur;
 648                           const unsigned char *saved_rlimit
 649                             = pfile->buffer->rlimit;
 650                           const unsigned char *saved_line_base
 651                             = pfile->buffer->line_base;
 652                           bool saved_need_line = pfile->buffer->need_line;
 653                           cpp_buffer *saved_overlaid_buffer
 654                             = pfile->overlaid_buffer;
 655                           pfile->buffer->cur = buf;
 656                           pfile->buffer->line_base = buf;
 657                           pfile->buffer->rlimit = buf + len + 1;
 658                           pfile->buffer->need_line = false;
 659                           pfile->overlaid_buffer = pfile->buffer;
 660                           bool saved_in_directive = pfile->state.in_directive;
 661                           pfile->state.in_directive = true;
 662                           cpp_context *saved_prev_context = context->prev;
 663                           context->prev = NULL;
 664
 665                           _cpp_scan_out_logical_line (pfile, NULL, true);
 666
 667                           pfile->state.in_directive = saved_in_directive;
 668                           check_output_buffer (pfile, 1);
 669                           *pfile->out.cur = '\n';
 670                           pfile->buffer->cur = pfile->out.base + fmacro.offset;
 671                           pfile->buffer->line_base = pfile->buffer->cur;
 672                           pfile->buffer->rlimit = pfile->out.cur;
 673                           CUR (context) = pfile->buffer->cur;
 674                           RLIMIT (context) = pfile->buffer->rlimit;
 675
 676                           pfile->state.prevent_expansion++;
 677                           const uchar *text
 678                             = _cpp_builtin_macro_text (pfile, fmacro.node);
 679                           pfile->state.prevent_expansion--;
 680
 681                           context->prev = saved_prev_context;
 682                           pfile->buffer->cur = saved_cur;
 683                           pfile->buffer->rlimit = saved_rlimit;
 684                           pfile->buffer->line_base = saved_line_base;
 685                           pfile->buffer->need_line = saved_need_line;
 686                           pfile->overlaid_buffer = saved_overlaid_buffer;
 687                           pfile->out.cur = pfile->out.base + fmacro.offset;
 688                           CUR (context) = cur;
 689                           RLIMIT (context) = ctx_rlimit;
 690                           len = ustrlen (text);
 691                           buf = _cpp_unaligned_alloc (pfile, len + 1);
 692                           memcpy (buf, text, len);
 693                           buf[len] = '\n';
 694                           text = buf;
 695                           _cpp_push_text_context (pfile, fmacro.node,
 696                                                   text, len);
 697                           goto new_context;
 698                         }
 699                       break;
 700                     }
 701
 702                   cpp_macro *m = fmacro.node->value.macro;
 703
 704                   m->used = 1;
 705                   lex_state = ls_none;
 706                   save_argument (&fmacro, out - pfile->out.base);
 707
 708                   /* A single zero-length argument is no argument.  */
 709                   if (fmacro.argc == 1
 710                       && m->paramc == 0
 711                       && out == pfile->out.base + fmacro.offset + 1)
 712                     fmacro.argc = 0;
 713
 714                   if (_cpp_arguments_ok (pfile, m, fmacro.node, fmacro.argc))
 715                     {
 716                       /* Remove the macro's invocation from the
 717                          output, and push its replacement text.  */
 718                       pfile->out.cur = pfile->out.base + fmacro.offset;
 719                       CUR (context) = cur;
 720                       replace_args_and_push (pfile, &fmacro);
 721                       goto new_context;
 722                     }
 723                 }
 724               else if (lex_state == ls_answer || lex_state == ls_defined_close
 725                         || lex_state == ls_has_include_close)
 726                 lex_state = ls_none;
 727             }
 728           break;
 729
 730         case '#':
 731           if (cur - 1 == start_of_input_line
 732               /* A '#' from a macro doesn't start a directive.  */
 733               && !pfile->context->prev
 734               && !pfile->state.in_directive)
 735             {
 736               /* A directive.  With the way _cpp_handle_directive
 737                  currently works, we only want to call it if either we
 738                  know the directive is OK, or we want it to fail and
 739                  be removed from the output.  If we want it to be
 740                  passed through (the assembler case) then we must not
 741                  call _cpp_handle_directive.  */
 742               pfile->out.cur = out;
 743               cur = skip_whitespace (pfile, cur, true /* skip_comments */);
 744               out = pfile->out.cur;
 745
 746               if (*cur == '\n')
 747                 {
 748                   /* Null directive.  Ignore it and don't invalidate
 749                      the MI optimization.  */
 750                   pfile->buffer->need_line = true;
 751                   CPP_INCREMENT_LINE (pfile, 0);
 752                   result = false;
 753                   goto done;
 754                 }
 755               else
 756                 {
 757                   bool do_it = false;
 758
 759                   if (is_numstart (*cur)
 760                       && CPP_OPTION (pfile, lang) != CLK_ASM)
 761                     do_it = true;
 762                   else if (is_idstart (*cur))
 763                     /* Check whether we know this directive, but don't
 764                        advance.  */
 765                     do_it = lex_identifier (pfile, cur)->is_directive;
 766
 767                   if (do_it || CPP_OPTION (pfile, lang) != CLK_ASM)
 768                     {
 769                       /* This is a kludge.  We want to have the ISO
 770                          preprocessor lex the next token.  */
 771                       pfile->buffer->cur = cur;
 772                       _cpp_handle_directive (pfile, false /* indented */);
 773                       result = false;
 774                       goto done;
 775                     }
 776                 }
 777             }
 778
 779           if (pfile->state.in_expression)
 780             {
 781               lex_state = ls_hash;
 782               continue;
 783             }
 784           break;
 785
 786         default:
 787           break;
 788         }
 789
 790       /* Non-whitespace disables MI optimization and stops treating
 791          '<' as a quote in #include.  */
 792       header_ok = false;
 793       if (!pfile->state.in_directive)
 794         pfile->mi_valid = false;
 795
 796       if (lex_state == ls_none)
 797         continue;
 798
 799       /* Some of these transitions of state are syntax errors.  The
 800          ISO preprocessor will issue errors later.  */
 801       if (lex_state == ls_fun_open)
 802         /* Missing '('.  */
 803         lex_state = ls_none;
 804       else if (lex_state == ls_hash
 805                || lex_state == ls_predicate
 806                || lex_state == ls_defined
 807                || lex_state == ls_has_include)
 808         lex_state = ls_none;
 809
 810       /* ls_answer and ls_defined_close keep going until ')'.  */
 811     }
 812
 813  done:
 814   if (fmacro.buff)
 815     _cpp_release_buff (pfile, fmacro.buff);
 816
 817   if (lex_state == ls_fun_close)
 818     cpp_error_with_line (pfile, CPP_DL_ERROR, fmacro.line, 0,
 819                          "unterminated argument list invoking macro \"%s\"",
 820                          NODE_NAME (fmacro.node));
 821   return result;
 822 }
 823
 824 /* Push a context holding the replacement text of the macro NODE on
 825    the context stack.  NODE is either object-like, or a function-like
 826    macro with no arguments.  */
 827 static void
 828 push_replacement_text (cpp_reader *pfile, cpp_hashnode *node)
 829 {
 830   size_t len;
 831   const uchar *text;
 832   uchar *buf;
 833
 834   if (node->flags & NODE_BUILTIN)
 835     {
 836       text = _cpp_builtin_macro_text (pfile, node);
 837       len = ustrlen (text);
 838       buf = _cpp_unaligned_alloc (pfile, len + 1);
 839       memcpy (buf, text, len);
 840       buf[len] = '\n';
 841       text = buf;
 842     }
 843   else
 844     {
 845       cpp_macro *macro = node->value.macro;
 846       macro->used = 1;
 847       text = macro->exp.text;
 848       macro->traditional = 1;
 849       len = macro->count;
 850     }
 851
 852   _cpp_push_text_context (pfile, node, text, len);
 853 }
 854
 855 /* Returns TRUE if traditional macro recursion is detected.  */
 856 static bool
 857 recursive_macro (cpp_reader *pfile, cpp_hashnode *node)
 858 {
 859   bool recursing = !!(node->flags & NODE_DISABLED);
 860
 861   /* Object-like macros that are already expanding are necessarily
 862      recursive.
 863
 864      However, it is possible to have traditional function-like macros
 865      that are not infinitely recursive but recurse to any given depth.
 866      Further, it is easy to construct examples that get ever longer
 867      until the point they stop recursing.  So there is no easy way to
 868      detect true recursion; instead we assume any expansion more than
 869      20 deep since the first invocation of this macro must be
 870      recursing.  */
 871   if (recursing && fun_like_macro (node))
 872     {
 873       size_t depth = 0;
 874       cpp_context *context = pfile->context;
 875
 876       do
 877         {
 878           depth++;
 879           if (context->c.macro == node && depth > 20)
 880             break;
 881           context = context->prev;
 882         }
 883       while (context);
 884       recursing = context != NULL;
 885     }
 886
 887   if (recursing)
 888     cpp_error (pfile, CPP_DL_ERROR,
 889                "detected recursion whilst expanding macro \"%s\"",
 890                NODE_NAME (node));
 891
 892   return recursing;
 893 }
 894
 895 /* Return the length of the replacement text of a function-like or
 896    object-like non-builtin macro.  */
 897 size_t
 898 _cpp_replacement_text_len (const cpp_macro *macro)
 899 {
 900   size_t len;
 901
 902   if (macro->fun_like && (macro->paramc != 0))
 903     {
 904       const uchar *exp;
 905
 906       len = 0;
 907       for (exp = macro->exp.text;;)
 908         {
 909           struct block *b = (struct block *) exp;
 910
 911           len += b->text_len;
 912           if (b->arg_index == 0)
 913             break;
 914           len += NODE_LEN (macro->params[b->arg_index - 1]);
 915           exp += BLOCK_LEN (b->text_len);
 916         }
 917     }
 918   else
 919     len = macro->count;
 920
 921   return len;
 922 }
 923
 924 /* Copy the replacement text of MACRO to DEST, which must be of
 925    sufficient size.  It is not NUL-terminated.  The next character is
 926    returned.  */
 927 uchar *
 928 _cpp_copy_replacement_text (const cpp_macro *macro, uchar *dest)
 929 {
 930   if (macro->fun_like && (macro->paramc != 0))
 931     {
 932       const uchar *exp;
 933
 934       for (exp = macro->exp.text;;)
 935         {
 936           struct block *b = (struct block *) exp;
 937           cpp_hashnode *param;
 938
 939           memcpy (dest, b->text, b->text_len);
 940           dest += b->text_len;
 941           if (b->arg_index == 0)
 942             break;
 943           param = macro->params[b->arg_index - 1];
 944           memcpy (dest, NODE_NAME (param), NODE_LEN (param));
 945           dest += NODE_LEN (param);
 946           exp += BLOCK_LEN (b->text_len);
 947         }
 948     }
 949   else
 950     {
 951       memcpy (dest, macro->exp.text, macro->count);
 952       dest += macro->count;
 953     }
 954
 955   return dest;
 956 }
 957
 958 /* Push a context holding the replacement text of the macro NODE on
 959    the context stack.  NODE is either object-like, or a function-like
 960    macro with no arguments.  */
 961 static void
 962 replace_args_and_push (cpp_reader *pfile, struct fun_macro *fmacro)
 963 {
 964   cpp_macro *macro = fmacro->node->value.macro;
 965
 966   if (macro->paramc == 0)
 967     push_replacement_text (pfile, fmacro->node);
 968   else
 969     {
 970       const uchar *exp;
 971       uchar *p;
 972       _cpp_buff *buff;
 973       size_t len = 0;
 974       int cxtquote = 0;
 975
 976       /* Get an estimate of the length of the argument-replaced text.
 977          This is a worst case estimate, assuming that every replacement
 978          text character needs quoting.  */
 979       for (exp = macro->exp.text;;)
 980         {
 981           struct block *b = (struct block *) exp;
 982
 983           len += b->text_len;
 984           if (b->arg_index == 0)
 985             break;
 986           len += 2 * (fmacro->args[b->arg_index]
 987                       - fmacro->args[b->arg_index - 1] - 1);
 988           exp += BLOCK_LEN (b->text_len);
 989         }
 990
 991       /* Allocate room for the expansion plus \n.  */
 992       buff = _cpp_get_buff (pfile, len + 1);
 993
 994       /* Copy the expansion and replace arguments.  */
 995       /* Accumulate actual length, including quoting as necessary */
 996       p = BUFF_FRONT (buff);
 997       len = 0;
 998       for (exp = macro->exp.text;;)
 999         {
1000           struct block *b = (struct block *) exp;
1001           size_t arglen;
1002           int argquote;
1003           uchar *base;
1004           uchar *in;
1005
1006           len += b->text_len;
1007           /* Copy the non-argument text literally, keeping
1008              track of whether matching quotes have been seen. */
1009           for (arglen = b->text_len, in = b->text; arglen > 0; arglen--)
1010             {
1011               if (*in == '"')
1012                 cxtquote = ! cxtquote;
1013               *p++ = *in++;
1014             }
1015           /* Done if no more arguments */
1016           if (b->arg_index == 0)
1017             break;
1018           arglen = (fmacro->args[b->arg_index]
1019                     - fmacro->args[b->arg_index - 1] - 1);
1020           base = pfile->out.base + fmacro->args[b->arg_index - 1];
1021           in = base;
1022 #if 0
1023           /* Skip leading whitespace in the text for the argument to
1024              be substituted. To be compatible with gcc 2.95, we would
1025              also need to trim trailing whitespace. Gcc 2.95 trims
1026              leading and trailing whitespace, which may be a bug.  The
1027              current gcc testsuite explicitly checks that this leading
1028              and trailing whitespace in actual arguments is
1029              preserved. */
1030           while (arglen > 0 && is_space (*in))
1031             {
1032               in++;
1033               arglen--;
1034             }
1035 #endif
1036           for (argquote = 0; arglen > 0; arglen--)
1037             {
1038               if (cxtquote && *in == '"')
1039                 {
1040                   if (in > base && *(in-1) != '\\')
1041                     argquote = ! argquote;
1042                   /* Always add backslash before double quote if argument
1043                      is expanded in a quoted context */
1044                   *p++ = '\\';
1045                   len++;
1046                 }
1047               else if (cxtquote && argquote && *in == '\\')
1048                 {
1049                   /* Always add backslash before a backslash in an argument
1050                      that is expanded in a quoted context and also in the
1051                      range of a quoted context in the argument itself. */
1052                   *p++ = '\\';
1053                   len++;
1054                 }
1055               *p++ = *in++;
1056               len++;
1057             }
1058           exp += BLOCK_LEN (b->text_len);
1059         }
1060
1061       /* \n-terminate.  */
1062       *p = '\n';
1063       _cpp_push_text_context (pfile, fmacro->node, BUFF_FRONT (buff), len);
1064
1065       /* So we free buffer allocation when macro is left.  */
1066       pfile->context->buff = buff;
1067     }
1068 }
1069
1070 /* Read and record the parameters, if any, of a function-like macro
1071    definition.  Destroys pfile->out.cur.
1072
1073    Returns true on success, false on failure (syntax error or a
1074    duplicate parameter).  On success, CUR (pfile->context) is just
1075    past the closing parenthesis.  */
1076 static bool
1077 scan_parameters (cpp_reader *pfile, cpp_macro *macro)
1078 {
1079   const uchar *cur = CUR (pfile->context) + 1;
1080   bool ok;
1081
1082   for (;;)
1083     {
1084       cur = skip_whitespace (pfile, cur, true /* skip_comments */);
1085
1086       if (is_idstart (*cur))
1087         {
1088           struct cpp_hashnode *id = lex_identifier (pfile, cur);
1089           ok = false;
1090           if (_cpp_save_parameter (pfile, macro, id, id))
1091             break;
1092           cur = skip_whitespace (pfile, CUR (pfile->context),
1093                                  true /* skip_comments */);
1094           if (*cur == ',')
1095             {
1096               cur++;
1097               continue;
1098             }
1099           ok = (*cur == ')');
1100           break;
1101         }
1102
1103       ok = (*cur == ')' && macro->paramc == 0);
1104       break;
1105     }
1106
1107   if (!ok)
1108     cpp_error (pfile, CPP_DL_ERROR, "syntax error in macro parameter list");
1109
1110   CUR (pfile->context) = cur + (*cur == ')');
1111
1112   return ok;
1113 }
1114
1115 /* Save the text from pfile->out.base to pfile->out.cur as
1116    the replacement text for the current macro, followed by argument
1117    ARG_INDEX, with zero indicating the end of the replacement
1118    text.  */
1119 static void
1120 save_replacement_text (cpp_reader *pfile, cpp_macro *macro,
1121                        unsigned int arg_index)
1122 {
1123   size_t len = pfile->out.cur - pfile->out.base;
1124   uchar *exp;
1125
1126   if (macro->paramc == 0)
1127     {
1128       /* Object-like and function-like macros without parameters
1129          simply store their \n-terminated replacement text.  */
1130       exp = _cpp_unaligned_alloc (pfile, len + 1);
1131       memcpy (exp, pfile->out.base, len);
1132       exp[len] = '\n';
1133       macro->exp.text = exp;
1134       macro->traditional = 1;
1135       macro->count = len;
1136     }
1137   else
1138     {
1139       /* Store the text's length (unsigned int), the argument index
1140          (unsigned short, base 1) and then the text.  */
1141       size_t blen = BLOCK_LEN (len);
1142       struct block *block;
1143
1144       if (macro->count + blen > BUFF_ROOM (pfile->a_buff))
1145         _cpp_extend_buff (pfile, &pfile->a_buff, macro->count + blen);
1146
1147       exp = BUFF_FRONT (pfile->a_buff);
1148       block = (struct block *) (exp + macro->count);
1149       macro->exp.text = exp;
1150       macro->traditional = 1;
1151
1152       /* Write out the block information.  */
1153       block->text_len = len;
1154       block->arg_index = arg_index;
1155       memcpy (block->text, pfile->out.base, len);
1156
1157       /* Lex the rest into the start of the output buffer.  */
1158       pfile->out.cur = pfile->out.base;
1159
1160       macro->count += blen;
1161
1162       /* If we've finished, commit the memory.  */
1163       if (arg_index == 0)
1164         BUFF_FRONT (pfile->a_buff) += macro->count;
1165     }
1166 }
1167
1168 /* Analyze and save the replacement text of a macro.  Returns true on
1169    success.  */
1170 bool
1171 _cpp_create_trad_definition (cpp_reader *pfile, cpp_macro *macro)
1172 {
1173   const uchar *cur;
1174   uchar *limit;
1175   cpp_context *context = pfile->context;
1176
1177   /* The context has not been set up for command line defines, and CUR
1178      has not been updated for the macro name for in-file defines.  */
1179   pfile->out.cur = pfile->out.base;
1180   CUR (context) = pfile->buffer->cur;
1181   RLIMIT (context) = pfile->buffer->rlimit;
1182   check_output_buffer (pfile, RLIMIT (context) - CUR (context));
1183
1184   /* Is this a function-like macro?  */
1185   if (* CUR (context) == '(')
1186     {
1187       bool ok = scan_parameters (pfile, macro);
1188
1189       /* Remember the params so we can clear NODE_MACRO_ARG flags.  */
1190       macro->params = (cpp_hashnode **) BUFF_FRONT (pfile->a_buff);
1191
1192       /* Setting macro to NULL indicates an error occurred, and
1193          prevents unnecessary work in _cpp_scan_out_logical_line.  */
1194       if (!ok)
1195         macro = NULL;
1196       else
1197         {
1198           BUFF_FRONT (pfile->a_buff) = (uchar *) &macro->params[macro->paramc];
1199           macro->fun_like = 1;
1200         }
1201     }
1202
1203   /* Skip leading whitespace in the replacement text.  */
1204   pfile->buffer->cur
1205     = skip_whitespace (pfile, CUR (context),
1206                        CPP_OPTION (pfile, discard_comments_in_macro_exp));
1207
1208   pfile->state.prevent_expansion++;
1209   _cpp_scan_out_logical_line (pfile, macro, false);
1210   pfile->state.prevent_expansion--;
1211
1212   if (!macro)
1213     return false;
1214
1215   /* Skip trailing white space.  */
1216   cur = pfile->out.base;
1217   limit = pfile->out.cur;
1218   while (limit > cur && is_space (limit[-1]))
1219     limit--;
1220   pfile->out.cur = limit;
1221   save_replacement_text (pfile, macro, 0);
1222
1223   return true;
1224 }
1225
1226 /* Copy SRC of length LEN to DEST, but convert all contiguous
1227    whitespace to a single space, provided it is not in quotes.  The
1228    quote currently in effect is pointed to by PQUOTE, and is updated
1229    by the function.  Returns the number of bytes copied.  */
1230 static size_t
1231 canonicalize_text (uchar *dest, const uchar *src, size_t len, uchar *pquote)
1232 {
1233   uchar *orig_dest = dest;
1234   uchar quote = *pquote;
1235
1236   while (len)
1237     {
1238       if (is_space (*src) && !quote)
1239         {
1240           do
1241             src++, len--;
1242           while (len && is_space (*src));
1243           *dest++ = ' ';
1244         }
1245       else
1246         {
1247           if (*src == '\'' || *src == '"')
1248             {
1249               if (!quote)
1250                 quote = *src;
1251               else if (quote == *src)
1252                 quote = 0;
1253             }
1254           *dest++ = *src++, len--;
1255         }
1256     }
1257
1258   *pquote = quote;
1259   return dest - orig_dest;
1260 }
1261
1262 /* Returns true if MACRO1 and MACRO2 have expansions different other
1263    than in the form of their whitespace.  */
1264 bool
1265 _cpp_expansions_different_trad (const cpp_macro *macro1,
1266                                 const cpp_macro *macro2)
1267 {
1268   uchar *p1 = XNEWVEC (uchar, macro1->count + macro2->count);
1269   uchar *p2 = p1 + macro1->count;
1270   uchar quote1 = 0, quote2 = 0;
1271   bool mismatch;
1272   size_t len1, len2;
1273
1274   if (macro1->paramc > 0)
1275     {
1276       const uchar *exp1 = macro1->exp.text, *exp2 = macro2->exp.text;
1277
1278       mismatch = true;
1279       for (;;)
1280         {
1281           struct block *b1 = (struct block *) exp1;
1282           struct block *b2 = (struct block *) exp2;
1283
1284           if (b1->arg_index != b2->arg_index)
1285             break;
1286
1287           len1 = canonicalize_text (p1, b1->text, b1->text_len, &quote1);
1288           len2 = canonicalize_text (p2, b2->text, b2->text_len, &quote2);
1289           if (len1 != len2 || memcmp (p1, p2, len1))
1290             break;
1291           if (b1->arg_index == 0)
1292             {
1293               mismatch = false;
1294               break;
1295             }
1296           exp1 += BLOCK_LEN (b1->text_len);
1297           exp2 += BLOCK_LEN (b2->text_len);
1298         }
1299     }
1300   else
1301     {
1302       len1 = canonicalize_text (p1, macro1->exp.text, macro1->count, &quote1);
1303       len2 = canonicalize_text (p2, macro2->exp.text, macro2->count, &quote2);
1304       mismatch = (len1 != len2 || memcmp (p1, p2, len1));
1305     }
1306
1307   free (p1);
1308   return mismatch;
1309 }