read.c

   1 /*      $Id: read.c,v 1.81 2014/08/16 19:00:01 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
   5  * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
   6  *
   7  * Permission to use, copy, modify, and distribute this software for any
   8  * purpose with or without fee is hereby granted, provided that the above
   9  * copyright notice and this permission notice appear in all copies.
  10  *
  11  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  12  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  13  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  14  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  15  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  16  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  17  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  18  */
  19 #include "config.h"
  20
  21 #include <sys/types.h>
  22 #if HAVE_MMAP
  23 #include <sys/stat.h>
  24 #include <sys/mman.h>
  25 #endif
  26
  27 #include <assert.h>
  28 #include <ctype.h>
  29 #include <errno.h>
  30 #include <fcntl.h>
  31 #include <stdarg.h>
  32 #include <stdint.h>
  33 #include <stdio.h>
  34 #include <stdlib.h>
  35 #include <string.h>
  36 #include <unistd.h>
  37
  38 #include "mandoc.h"
  39 #include "mandoc_aux.h"
  40 #include "libmandoc.h"
  41 #include "mdoc.h"
  42 #include "man.h"
  43 #include "main.h"
  44
  45 #define REPARSE_LIMIT   1000
  46
  47 struct  buf {
  48         char             *buf; /* binary input buffer */
  49         size_t            sz; /* size of binary buffer */
  50 };
  51
  52 struct  mparse {
  53         enum mandoclevel  file_status; /* status of current parse */
  54         enum mandoclevel  wlevel; /* ignore messages below this */
  55         int               line; /* line number in the file */
  56         int               options; /* parser options */
  57         struct man       *pman; /* persistent man parser */
  58         struct mdoc      *pmdoc; /* persistent mdoc parser */
  59         struct man       *man; /* man parser */
  60         struct mdoc      *mdoc; /* mdoc parser */
  61         struct roff      *roff; /* roff parser (!NULL) */
  62         char             *sodest; /* filename pointed to by .so */
  63         int               reparse_count; /* finite interp. stack */
  64         mandocmsg         mmsg; /* warning/error message handler */
  65         const char       *file;
  66         struct buf       *secondary;
  67         const char       *defos; /* default operating system */
  68 };
  69
  70 static  void      resize_buf(struct buf *, size_t);
  71 static  void      mparse_buf_r(struct mparse *, struct buf, int);
  72 static  void      pset(const char *, int, struct mparse *);
  73 static  int       read_whole_file(struct mparse *, const char *, int,
  74                                 struct buf *, int *);
  75 static  void      mparse_end(struct mparse *);
  76 static  void      mparse_parse_buffer(struct mparse *, struct buf,
  77                         const char *);
  78
  79 static  const enum mandocerr    mandoclimits[MANDOCLEVEL_MAX] = {
  80         MANDOCERR_OK,
  81         MANDOCERR_WARNING,
  82         MANDOCERR_WARNING,
  83         MANDOCERR_ERROR,
  84         MANDOCERR_FATAL,
  85         MANDOCERR_MAX,
  86         MANDOCERR_MAX
  87 };
  88
  89 static  const char * const      mandocerrs[MANDOCERR_MAX] = {
  90         "ok",
  91
  92         "generic warning",
  93
  94         /* related to the prologue */
  95         "missing manual title, using UNTITLED",
  96         "missing manual title, using \"\"",
  97         "lower case character in document title",
  98         "missing manual section, using \"\"",
  99         "unknown manual section",
 100         "unknown manual volume or arch",
 101         "missing date, using today's date",
 102         "cannot parse date, using it verbatim",
 103         "missing Os macro, using \"\"",
 104         "duplicate prologue macro",
 105         "late prologue macro",
 106         "skipping late title macro",
 107         "prologue macros out of order",
 108
 109         /* related to document structure */
 110         ".so is fragile, better use ln(1)",
 111         "no document body",
 112         "content before first section header",
 113         "first section is not \"NAME\"",
 114         "bad NAME section contents",
 115         "sections out of conventional order",
 116         "duplicate section title",
 117         "unexpected section",
 118
 119         /* related to macros and nesting */
 120         "obsolete macro",
 121         "skipping paragraph macro",
 122         "moving paragraph macro out of list",
 123         "skipping no-space macro",
 124         "blocks badly nested",
 125         "nested displays are not portable",
 126         "moving content out of list",
 127         ".Vt block has child macro",
 128         "fill mode already enabled, skipping",
 129         "fill mode already disabled, skipping",
 130         "line scope broken",
 131
 132         /* related to missing macro arguments */
 133         "skipping empty request",
 134         "conditional request controls empty scope",
 135         "skipping empty macro",
 136         "empty argument, using 0n",
 137         "argument count wrong",
 138         "missing display type, using -ragged",
 139         "list type is not the first argument",
 140         "missing -width in -tag list, using 8n",
 141         "missing utility name, using \"\"",
 142         "empty head in list item",
 143         "empty list item",
 144         "missing font type, using \\fR",
 145         "unknown font type, using \\fR",
 146         "missing -std argument, adding it",
 147
 148         /* related to bad macro arguments */
 149         "unterminated quoted argument",
 150         "duplicate argument",
 151         "skipping duplicate argument",
 152         "skipping duplicate display type",
 153         "skipping duplicate list type",
 154         "skipping -width argument",
 155         "unknown AT&T UNIX version",
 156         "invalid content in Rs block",
 157         "invalid Boolean argument",
 158         "unknown font, skipping request",
 159
 160         /* related to plain text */
 161         "blank line in fill mode, using .sp",
 162         "tab in filled text",
 163         "whitespace at end of input line",
 164         "bad comment style",
 165         "invalid escape sequence",
 166         "undefined string, using \"\"",
 167
 168         "generic error",
 169
 170         /* related to equations */
 171         "unexpected equation scope closure",
 172         "equation scope open on exit",
 173         "overlapping equation scopes",
 174         "unexpected end of equation",
 175         "equation syntax error",
 176
 177         /* related to tables */
 178         "bad table syntax",
 179         "bad table option",
 180         "bad table layout",
 181         "no table layout cells specified",
 182         "no table data cells specified",
 183         "ignore data in cell",
 184         "data block still open",
 185         "ignoring extra data cells",
 186
 187         /* related to document structure and macros */
 188         "input stack limit exceeded, infinite loop?",
 189         "skipping bad character",
 190         "skipping unknown macro",
 191         "skipping item outside list",
 192         "skipping column outside column list",
 193         "skipping end of block that is not open",
 194         "inserting missing end of block",
 195         "appending missing end of block",
 196
 197         /* related to request and macro arguments */
 198         "escaped character not allowed in a name",
 199         "argument count wrong",
 200         "missing list type, using -item",
 201         "missing manual name, using \"\"",
 202         "uname(3) system call failed, using UNKNOWN",
 203         "unknown standard specifier",
 204         "skipping request without numeric argument",
 205         "skipping all arguments",
 206         "skipping excess arguments",
 207
 208         "generic fatal error",
 209
 210         "input too large",
 211         "NOT IMPLEMENTED: Bd -file",
 212         "NOT IMPLEMENTED: .so with absolute path or \"..\"",
 213         ".so request failed",
 214
 215         /* system errors */
 216         NULL,
 217         "cannot stat file",
 218         "cannot read file",
 219 };
 220
 221 static  const char * const      mandoclevels[MANDOCLEVEL_MAX] = {
 222         "SUCCESS",
 223         "RESERVED",
 224         "WARNING",
 225         "ERROR",
 226         "FATAL",
 227         "BADARG",
 228         "SYSERR"
 229 };
 230
 231
 232 static void
 233 resize_buf(struct buf *buf, size_t initial)
 234 {
 235
 236         buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
 237         buf->buf = mandoc_realloc(buf->buf, buf->sz);
 238 }
 239
 240 static void
 241 pset(const char *buf, int pos, struct mparse *curp)
 242 {
 243         int              i;
 244
 245         /*
 246          * Try to intuit which kind of manual parser should be used.  If
 247          * passed in by command-line (-man, -mdoc), then use that
 248          * explicitly.  If passed as -mandoc, then try to guess from the
 249          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 250          * default to -man, which is more lenient.
 251          *
 252          * Separate out pmdoc/pman from mdoc/man: the first persists
 253          * through all parsers, while the latter is used per-parse.
 254          */
 255
 256         if ('.' == buf[0] || '\'' == buf[0]) {
 257                 for (i = 1; buf[i]; i++)
 258                         if (' ' != buf[i] && '\t' != buf[i])
 259                                 break;
 260                 if ('\0' == buf[i])
 261                         return;
 262         }
 263
 264         if (MPARSE_MDOC & curp->options) {
 265                 curp->mdoc = curp->pmdoc;
 266                 return;
 267         } else if (MPARSE_MAN & curp->options) {
 268                 curp->man = curp->pman;
 269                 return;
 270         }
 271
 272         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 273                 if (NULL == curp->pmdoc)
 274                         curp->pmdoc = mdoc_alloc(
 275                             curp->roff, curp, curp->defos,
 276                             MPARSE_QUICK & curp->options ? 1 : 0);
 277                 assert(curp->pmdoc);
 278                 curp->mdoc = curp->pmdoc;
 279                 return;
 280         }
 281
 282         if (NULL == curp->pman)
 283                 curp->pman = man_alloc(curp->roff, curp,
 284                     MPARSE_QUICK & curp->options ? 1 : 0);
 285         assert(curp->pman);
 286         curp->man = curp->pman;
 287 }
 288
 289 /*
 290  * Main parse routine for an opened file.  This is called for each
 291  * opened file and simply loops around the full input file, possibly
 292  * nesting (i.e., with `so').
 293  */
 294 static void
 295 mparse_buf_r(struct mparse *curp, struct buf blk, int start)
 296 {
 297         const struct tbl_span   *span;
 298         struct buf       ln;
 299         enum rofferr     rr;
 300         int              i, of, rc;
 301         int              pos; /* byte number in the ln buffer */
 302         int              lnn; /* line number in the real file */
 303         unsigned char    c;
 304
 305         memset(&ln, 0, sizeof(struct buf));
 306
 307         lnn = curp->line;
 308         pos = 0;
 309
 310         for (i = 0; i < (int)blk.sz; ) {
 311                 if (0 == pos && '\0' == blk.buf[i])
 312                         break;
 313
 314                 if (start) {
 315                         curp->line = lnn;
 316                         curp->reparse_count = 0;
 317                 }
 318
 319                 while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
 320
 321                         /*
 322                          * When finding an unescaped newline character,
 323                          * leave the character loop to process the line.
 324                          * Skip a preceding carriage return, if any.
 325                          */
 326
 327                         if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
 328                             '\n' == blk.buf[i + 1])
 329                                 ++i;
 330                         if ('\n' == blk.buf[i]) {
 331                                 ++i;
 332                                 ++lnn;
 333                                 break;
 334                         }
 335
 336                         /*
 337                          * Make sure we have space for at least
 338                          * one backslash and one other character
 339                          * and the trailing NUL byte.
 340                          */
 341
 342                         if (pos + 2 >= (int)ln.sz)
 343                                 resize_buf(&ln, 256);
 344
 345                         /*
 346                          * Warn about bogus characters.  If you're using
 347                          * non-ASCII encoding, you're screwing your
 348                          * readers.  Since I'd rather this not happen,
 349                          * I'll be helpful and replace these characters
 350                          * with "?", so we don't display gibberish.
 351                          * Note to manual writers: use special characters.
 352                          */
 353
 354                         c = (unsigned char) blk.buf[i];
 355
 356                         if ( ! (isascii(c) &&
 357                             (isgraph(c) || isblank(c)))) {
 358                                 mandoc_vmsg(MANDOCERR_BADCHAR, curp,
 359                                     curp->line, pos, "0x%x", c);
 360                                 i++;
 361                                 ln.buf[pos++] = '?';
 362                                 continue;
 363                         }
 364
 365                         /* Trailing backslash = a plain char. */
 366
 367                         if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
 368                                 ln.buf[pos++] = blk.buf[i++];
 369                                 continue;
 370                         }
 371
 372                         /*
 373                          * Found escape and at least one other character.
 374                          * When it's a newline character, skip it.
 375                          * When there is a carriage return in between,
 376                          * skip that one as well.
 377                          */
 378
 379                         if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
 380                             '\n' == blk.buf[i + 2])
 381                                 ++i;
 382                         if ('\n' == blk.buf[i + 1]) {
 383                                 i += 2;
 384                                 ++lnn;
 385                                 continue;
 386                         }
 387
 388                         if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
 389                                 i += 2;
 390                                 /* Comment, skip to end of line */
 391                                 for (; i < (int)blk.sz; ++i) {
 392                                         if ('\n' == blk.buf[i]) {
 393                                                 ++i;
 394                                                 ++lnn;
 395                                                 break;
 396                                         }
 397                                 }
 398
 399                                 /* Backout trailing whitespaces */
 400                                 for (; pos > 0; --pos) {
 401                                         if (ln.buf[pos - 1] != ' ')
 402                                                 break;
 403                                         if (pos > 2 && ln.buf[pos - 2] == '\\')
 404                                                 break;
 405                                 }
 406                                 break;
 407                         }
 408
 409                         /* Catch escaped bogus characters. */
 410
 411                         c = (unsigned char) blk.buf[i+1];
 412
 413                         if ( ! (isascii(c) &&
 414                             (isgraph(c) || isblank(c)))) {
 415                                 mandoc_vmsg(MANDOCERR_BADCHAR, curp,
 416                                     curp->line, pos, "0x%x", c);
 417                                 i += 2;
 418                                 ln.buf[pos++] = '?';
 419                                 continue;
 420                         }
 421
 422                         /* Some other escape sequence, copy & cont. */
 423
 424                         ln.buf[pos++] = blk.buf[i++];
 425                         ln.buf[pos++] = blk.buf[i++];
 426                 }
 427
 428                 if (pos >= (int)ln.sz)
 429                         resize_buf(&ln, 256);
 430
 431                 ln.buf[pos] = '\0';
 432
 433                 /*
 434                  * A significant amount of complexity is contained by
 435                  * the roff preprocessor.  It's line-oriented but can be
 436                  * expressed on one line, so we need at times to
 437                  * readjust our starting point and re-run it.  The roff
 438                  * preprocessor can also readjust the buffers with new
 439                  * data, so we pass them in wholesale.
 440                  */
 441
 442                 of = 0;
 443
 444                 /*
 445                  * Maintain a lookaside buffer of all parsed lines.  We
 446                  * only do this if mparse_keep() has been invoked (the
 447                  * buffer may be accessed with mparse_getkeep()).
 448                  */
 449
 450                 if (curp->secondary) {
 451                         curp->secondary->buf = mandoc_realloc(
 452                             curp->secondary->buf,
 453                             curp->secondary->sz + pos + 2);
 454                         memcpy(curp->secondary->buf +
 455                             curp->secondary->sz,
 456                             ln.buf, pos);
 457                         curp->secondary->sz += pos;
 458                         curp->secondary->buf
 459                                 [curp->secondary->sz] = '\n';
 460                         curp->secondary->sz++;
 461                         curp->secondary->buf
 462                                 [curp->secondary->sz] = '\0';
 463                 }
 464 rerun:
 465                 rr = roff_parseln(curp->roff, curp->line,
 466                     &ln.buf, &ln.sz, of, &of);
 467
 468                 switch (rr) {
 469                 case ROFF_REPARSE:
 470                         if (REPARSE_LIMIT >= ++curp->reparse_count)
 471                                 mparse_buf_r(curp, ln, 0);
 472                         else
 473                                 mandoc_msg(MANDOCERR_ROFFLOOP, curp,
 474                                     curp->line, pos, NULL);
 475                         pos = 0;
 476                         continue;
 477                 case ROFF_APPEND:
 478                         pos = (int)strlen(ln.buf);
 479                         continue;
 480                 case ROFF_RERUN:
 481                         goto rerun;
 482                 case ROFF_IGN:
 483                         pos = 0;
 484                         continue;
 485                 case ROFF_ERR:
 486                         assert(MANDOCLEVEL_FATAL <= curp->file_status);
 487                         break;
 488                 case ROFF_SO:
 489                         if (0 == (MPARSE_SO & curp->options) &&
 490                             (i >= (int)blk.sz || '\0' == blk.buf[i])) {
 491                                 curp->sodest = mandoc_strdup(ln.buf + of);
 492                                 free(ln.buf);
 493                                 return;
 494                         }
 495                         /*
 496                          * We remove `so' clauses from our lookaside
 497                          * buffer because we're going to descend into
 498                          * the file recursively.
 499                          */
 500                         if (curp->secondary)
 501                                 curp->secondary->sz -= pos + 1;
 502                         mparse_readfd(curp, -1, ln.buf + of);
 503                         if (MANDOCLEVEL_FATAL <= curp->file_status) {
 504                                 mandoc_vmsg(MANDOCERR_SO_FAIL,
 505                                     curp, curp->line, pos,
 506                                     ".so %s", ln.buf + of);
 507                                 break;
 508                         }
 509                         pos = 0;
 510                         continue;
 511                 default:
 512                         break;
 513                 }
 514
 515                 /*
 516                  * If we encounter errors in the recursive parse, make
 517                  * sure we don't continue parsing.
 518                  */
 519
 520                 if (MANDOCLEVEL_FATAL <= curp->file_status)
 521                         break;
 522
 523                 /*
 524                  * If input parsers have not been allocated, do so now.
 525                  * We keep these instanced between parsers, but set them
 526                  * locally per parse routine since we can use different
 527                  * parsers with each one.
 528                  */
 529
 530                 if ( ! (curp->man || curp->mdoc))
 531                         pset(ln.buf + of, pos - of, curp);
 532
 533                 /*
 534                  * Lastly, push down into the parsers themselves.  One
 535                  * of these will have already been set in the pset()
 536                  * routine.
 537                  * If libroff returns ROFF_TBL, then add it to the
 538                  * currently open parse.  Since we only get here if
 539                  * there does exist data (see tbl_data.c), we're
 540                  * guaranteed that something's been allocated.
 541                  * Do the same for ROFF_EQN.
 542                  */
 543
 544                 rc = -1;
 545
 546                 if (ROFF_TBL == rr)
 547                         while (NULL != (span = roff_span(curp->roff))) {
 548                                 rc = curp->man ?
 549                                     man_addspan(curp->man, span) :
 550                                     mdoc_addspan(curp->mdoc, span);
 551                                 if (0 == rc)
 552                                         break;
 553                         }
 554                 else if (ROFF_EQN == rr)
 555                         rc = curp->mdoc ?
 556                             mdoc_addeqn(curp->mdoc,
 557                                 roff_eqn(curp->roff)) :
 558                             man_addeqn(curp->man,
 559                                 roff_eqn(curp->roff));
 560                 else if (curp->man || curp->mdoc)
 561                         rc = curp->man ?
 562                             man_parseln(curp->man,
 563                                 curp->line, ln.buf, of) :
 564                             mdoc_parseln(curp->mdoc,
 565                                 curp->line, ln.buf, of);
 566
 567                 if (0 == rc) {
 568                         assert(MANDOCLEVEL_FATAL <= curp->file_status);
 569                         break;
 570                 } else if (2 == rc)
 571                         break;
 572
 573                 /* Temporary buffers typically are not full. */
 574
 575                 if (0 == start && '\0' == blk.buf[i])
 576                         break;
 577
 578                 /* Start the next input line. */
 579
 580                 pos = 0;
 581         }
 582
 583         free(ln.buf);
 584 }
 585
 586 static int
 587 read_whole_file(struct mparse *curp, const char *file, int fd,
 588                 struct buf *fb, int *with_mmap)
 589 {
 590         size_t           off;
 591         ssize_t          ssz;
 592
 593 #if HAVE_MMAP
 594         struct stat      st;
 595         if (-1 == fstat(fd, &st)) {
 596                 curp->file_status = MANDOCLEVEL_SYSERR;
 597                 if (curp->mmsg)
 598                         (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
 599                             file, 0, 0, strerror(errno));
 600                 return(0);
 601         }
 602
 603         /*
 604          * If we're a regular file, try just reading in the whole entry
 605          * via mmap().  This is faster than reading it into blocks, and
 606          * since each file is only a few bytes to begin with, I'm not
 607          * concerned that this is going to tank any machines.
 608          */
 609
 610         if (S_ISREG(st.st_mode)) {
 611                 if (st.st_size >= (1U << 31)) {
 612                         curp->file_status = MANDOCLEVEL_FATAL;
 613                         if (curp->mmsg)
 614                                 (*curp->mmsg)(MANDOCERR_TOOLARGE,
 615                                     curp->file_status, file, 0, 0, NULL);
 616                         return(0);
 617                 }
 618                 *with_mmap = 1;
 619                 fb->sz = (size_t)st.st_size;
 620                 fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
 621                 if (fb->buf != MAP_FAILED)
 622                         return(1);
 623         }
 624 #endif
 625
 626         /*
 627          * If this isn't a regular file (like, say, stdin), then we must
 628          * go the old way and just read things in bit by bit.
 629          */
 630
 631         *with_mmap = 0;
 632         off = 0;
 633         fb->sz = 0;
 634         fb->buf = NULL;
 635         for (;;) {
 636                 if (off == fb->sz) {
 637                         if (fb->sz == (1U << 31)) {
 638                                 curp->file_status = MANDOCLEVEL_FATAL;
 639                                 if (curp->mmsg)
 640                                         (*curp->mmsg)(MANDOCERR_TOOLARGE,
 641                                             curp->file_status,
 642                                             file, 0, 0, NULL);
 643                                 break;
 644                         }
 645                         resize_buf(fb, 65536);
 646                 }
 647                 ssz = read(fd, fb->buf + (int)off, fb->sz - off);
 648                 if (ssz == 0) {
 649                         fb->sz = off;
 650                         return(1);
 651                 }
 652                 if (ssz == -1) {
 653                         curp->file_status = MANDOCLEVEL_SYSERR;
 654                         if (curp->mmsg)
 655                                 (*curp->mmsg)(MANDOCERR_SYSREAD,
 656                                     curp->file_status, file, 0, 0,
 657                                     strerror(errno));
 658                         break;
 659                 }
 660                 off += (size_t)ssz;
 661         }
 662
 663         free(fb->buf);
 664         fb->buf = NULL;
 665         return(0);
 666 }
 667
 668 static void
 669 mparse_end(struct mparse *curp)
 670 {
 671
 672         if (MANDOCLEVEL_FATAL <= curp->file_status)
 673                 return;
 674
 675         if (curp->mdoc == NULL &&
 676             curp->man == NULL &&
 677             curp->sodest == NULL) {
 678                 if (curp->options & MPARSE_MDOC)
 679                         curp->mdoc = curp->pmdoc;
 680                 else {
 681                         if (curp->pman == NULL)
 682                                 curp->pman = man_alloc(curp->roff, curp,
 683                                     curp->options & MPARSE_QUICK ? 1 : 0);
 684                         curp->man = curp->pman;
 685                 }
 686         }
 687
 688         if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
 689                 assert(MANDOCLEVEL_FATAL <= curp->file_status);
 690                 return;
 691         }
 692
 693         if (curp->man && ! man_endparse(curp->man)) {
 694                 assert(MANDOCLEVEL_FATAL <= curp->file_status);
 695                 return;
 696         }
 697
 698         roff_endparse(curp->roff);
 699 }
 700
 701 static void
 702 mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
 703 {
 704         const char      *svfile;
 705         static int       recursion_depth;
 706
 707         if (64 < recursion_depth) {
 708                 mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
 709                 return;
 710         }
 711
 712         /* Line number is per-file. */
 713         svfile = curp->file;
 714         curp->file = file;
 715         curp->line = 1;
 716         recursion_depth++;
 717
 718         mparse_buf_r(curp, blk, 1);
 719
 720         if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
 721                 mparse_end(curp);
 722
 723         curp->file = svfile;
 724 }
 725
 726 enum mandoclevel
 727 mparse_readmem(struct mparse *curp, const void *buf, size_t len,
 728                 const char *file)
 729 {
 730         struct buf blk;
 731
 732         blk.buf = UNCONST(buf);
 733         blk.sz = len;
 734
 735         mparse_parse_buffer(curp, blk, file);
 736         return(curp->file_status);
 737 }
 738
 739 enum mandoclevel
 740 mparse_readfd(struct mparse *curp, int fd, const char *file)
 741 {
 742         struct buf       blk;
 743         int              with_mmap;
 744
 745         if (-1 == fd && -1 == (fd = open(file, O_RDONLY, 0))) {
 746                 curp->file_status = MANDOCLEVEL_SYSERR;
 747                 if (curp->mmsg)
 748                         (*curp->mmsg)(MANDOCERR_SYSOPEN,
 749                             curp->file_status,
 750                             file, 0, 0, strerror(errno));
 751                 goto out;
 752         }
 753
 754         /*
 755          * Run for each opened file; may be called more than once for
 756          * each full parse sequence if the opened file is nested (i.e.,
 757          * from `so').  Simply sucks in the whole file and moves into
 758          * the parse phase for the file.
 759          */
 760
 761         if ( ! read_whole_file(curp, file, fd, &blk, &with_mmap))
 762                 goto out;
 763
 764         mparse_parse_buffer(curp, blk, file);
 765
 766 #if HAVE_MMAP
 767         if (with_mmap)
 768                 munmap(blk.buf, blk.sz);
 769         else
 770 #endif
 771                 free(blk.buf);
 772
 773         if (STDIN_FILENO != fd && -1 == close(fd))
 774                 perror(file);
 775 out:
 776         return(curp->file_status);
 777 }
 778
 779 struct mparse *
 780 mparse_alloc(int options, enum mandoclevel wlevel,
 781                 mandocmsg mmsg, const char *defos)
 782 {
 783         struct mparse   *curp;
 784
 785         assert(wlevel <= MANDOCLEVEL_FATAL);
 786
 787         curp = mandoc_calloc(1, sizeof(struct mparse));
 788
 789         curp->options = options;
 790         curp->wlevel = wlevel;
 791         curp->mmsg = mmsg;
 792         curp->defos = defos;
 793
 794         curp->roff = roff_alloc(curp, options);
 795         if (curp->options & MPARSE_MDOC)
 796                 curp->pmdoc = mdoc_alloc(
 797                     curp->roff, curp, curp->defos,
 798                     curp->options & MPARSE_QUICK ? 1 : 0);
 799         if (curp->options & MPARSE_MAN)
 800                 curp->pman = man_alloc(curp->roff, curp,
 801                     curp->options & MPARSE_QUICK ? 1 : 0);
 802
 803         return(curp);
 804 }
 805
 806 void
 807 mparse_reset(struct mparse *curp)
 808 {
 809
 810         roff_reset(curp->roff);
 811
 812         if (curp->mdoc)
 813                 mdoc_reset(curp->mdoc);
 814         if (curp->man)
 815                 man_reset(curp->man);
 816         if (curp->secondary)
 817                 curp->secondary->sz = 0;
 818
 819         curp->file_status = MANDOCLEVEL_OK;
 820         curp->mdoc = NULL;
 821         curp->man = NULL;
 822
 823         free(curp->sodest);
 824         curp->sodest = NULL;
 825 }
 826
 827 void
 828 mparse_free(struct mparse *curp)
 829 {
 830
 831         if (curp->pmdoc)
 832                 mdoc_free(curp->pmdoc);
 833         if (curp->pman)
 834                 man_free(curp->pman);
 835         if (curp->roff)
 836                 roff_free(curp->roff);
 837         if (curp->secondary)
 838                 free(curp->secondary->buf);
 839
 840         free(curp->secondary);
 841         free(curp->sodest);
 842         free(curp);
 843 }
 844
 845 void
 846 mparse_result(struct mparse *curp,
 847         struct mdoc **mdoc, struct man **man, char **sodest)
 848 {
 849
 850         if (sodest && NULL != (*sodest = curp->sodest)) {
 851                 *mdoc = NULL;
 852                 *man = NULL;
 853                 return;
 854         }
 855         if (mdoc)
 856                 *mdoc = curp->mdoc;
 857         if (man)
 858                 *man = curp->man;
 859 }
 860
 861 void
 862 mandoc_vmsg(enum mandocerr t, struct mparse *m,
 863                 int ln, int pos, const char *fmt, ...)
 864 {
 865         char             buf[256];
 866         va_list          ap;
 867
 868         va_start(ap, fmt);
 869         (void)vsnprintf(buf, sizeof(buf), fmt, ap);
 870         va_end(ap);
 871
 872         mandoc_msg(t, m, ln, pos, buf);
 873 }
 874
 875 void
 876 mandoc_msg(enum mandocerr er, struct mparse *m,
 877                 int ln, int col, const char *msg)
 878 {
 879         enum mandoclevel level;
 880
 881         level = MANDOCLEVEL_FATAL;
 882         while (er < mandoclimits[level])
 883                 level--;
 884
 885         if (level < m->wlevel)
 886                 return;
 887
 888         if (m->mmsg)
 889                 (*m->mmsg)(er, level, m->file, ln, col, msg);
 890
 891         if (m->file_status < level)
 892                 m->file_status = level;
 893 }
 894
 895 const char *
 896 mparse_strerror(enum mandocerr er)
 897 {
 898
 899         return(mandocerrs[er]);
 900 }
 901
 902 const char *
 903 mparse_strlevel(enum mandoclevel lvl)
 904 {
 905         return(mandoclevels[lvl]);
 906 }
 907
 908 void
 909 mparse_keep(struct mparse *p)
 910 {
 911
 912         assert(NULL == p->secondary);
 913         p->secondary = mandoc_calloc(1, sizeof(struct buf));
 914 }
 915
 916 const char *
 917 mparse_getkeep(const struct mparse *p)
 918 {
 919
 920         assert(p->secondary);
 921         return(p->secondary->sz ? p->secondary->buf : NULL);
 922 }