read.c

   1 /* $Id: read.c,v 1.221 2022/05/19 14:48:56 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2010-2020 Ingo Schwarze <schwarze@openbsd.org>
   4  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   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 AUTHORS DISCLAIM ALL WARRANTIES
  12  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  13  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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  * Top-level functions of the mandoc(3) parser:
  20  * Parser and input encoding selection, decompression,
  21  * handling of input bytes, characters, lines, and files,
  22  * handling of roff(7) loops and file inclusion,
  23  * and steering of the various parsers.
  24  */
  25 #include "config.h"
  26
  27 #include <sys/types.h>
  28 #include <sys/mman.h>
  29 #include <sys/stat.h>
  30
  31 #include <assert.h>
  32 #include <ctype.h>
  33 #include <errno.h>
  34 #include <fcntl.h>
  35 #include <stdarg.h>
  36 #include <stdio.h>
  37 #include <stdlib.h>
  38 #include <string.h>
  39 #include <unistd.h>
  40 #include <zlib.h>
  41
  42 #include "mandoc_aux.h"
  43 #include "mandoc.h"
  44 #include "roff.h"
  45 #include "mdoc.h"
  46 #include "man.h"
  47 #include "mandoc_parse.h"
  48 #include "libmandoc.h"
  49 #include "roff_int.h"
  50 #include "tag.h"
  51
  52 #define REPARSE_LIMIT   1000
  53
  54 struct  mparse {
  55         struct roff      *roff; /* roff parser (!NULL) */
  56         struct roff_man  *man; /* man parser */
  57         struct buf       *primary; /* buffer currently being parsed */
  58         struct buf       *secondary; /* copy of top level input */
  59         struct buf       *loop; /* open .while request line */
  60         const char       *os_s; /* default operating system */
  61         int               options; /* parser options */
  62         int               gzip; /* current input file is gzipped */
  63         int               filenc; /* encoding of the current file */
  64         int               reparse_count; /* finite interp. stack */
  65         int               line; /* line number in the file */
  66 };
  67
  68 static  void      choose_parser(struct mparse *);
  69 static  void      free_buf_list(struct buf *);
  70 static  void      resize_buf(struct buf *, size_t);
  71 static  int       mparse_buf_r(struct mparse *, struct buf, size_t, int);
  72 static  int       read_whole_file(struct mparse *, int, struct buf *, int *);
  73 static  void      mparse_end(struct mparse *);
  74
  75
  76 static void
  77 resize_buf(struct buf *buf, size_t initial)
  78 {
  79
  80         buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
  81         buf->buf = mandoc_realloc(buf->buf, buf->sz);
  82 }
  83
  84 static void
  85 free_buf_list(struct buf *buf)
  86 {
  87         struct buf *tmp;
  88
  89         while (buf != NULL) {
  90                 tmp = buf;
  91                 buf = tmp->next;
  92                 free(tmp->buf);
  93                 free(tmp);
  94         }
  95 }
  96
  97 static void
  98 choose_parser(struct mparse *curp)
  99 {
 100         char            *cp, *ep;
 101         int              format;
 102
 103         /*
 104          * If neither command line arguments -mdoc or -man select
 105          * a parser nor the roff parser found a .Dd or .TH macro
 106          * yet, look ahead in the main input buffer.
 107          */
 108
 109         if ((format = roff_getformat(curp->roff)) == 0) {
 110                 cp = curp->primary->buf;
 111                 ep = cp + curp->primary->sz;
 112                 while (cp < ep) {
 113                         if (*cp == '.' || *cp == '\'') {
 114                                 cp++;
 115                                 if (cp[0] == 'D' && cp[1] == 'd') {
 116                                         format = MPARSE_MDOC;
 117                                         break;
 118                                 }
 119                                 if (cp[0] == 'T' && cp[1] == 'H') {
 120                                         format = MPARSE_MAN;
 121                                         break;
 122                                 }
 123                         }
 124                         cp = memchr(cp, '\n', ep - cp);
 125                         if (cp == NULL)
 126                                 break;
 127                         cp++;
 128                 }
 129         }
 130
 131         if (format == MPARSE_MDOC) {
 132                 curp->man->meta.macroset = MACROSET_MDOC;
 133                 if (curp->man->mdocmac == NULL)
 134                         curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
 135         } else {
 136                 curp->man->meta.macroset = MACROSET_MAN;
 137                 if (curp->man->manmac == NULL)
 138                         curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
 139         }
 140         curp->man->meta.first->tok = TOKEN_NONE;
 141 }
 142
 143 /*
 144  * Main parse routine for a buffer.
 145  * It assumes encoding and line numbering are already set up.
 146  * It can recurse directly (for invocations of user-defined
 147  * macros, inline equations, and input line traps)
 148  * and indirectly (for .so file inclusion).
 149  */
 150 static int
 151 mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
 152 {
 153         struct buf       ln;
 154         struct buf      *firstln, *lastln, *thisln, *loop;
 155         char            *cp;
 156         size_t           pos; /* byte number in the ln buffer */
 157         size_t           spos; /* at the start of the current line parse */
 158         int              line_result, result;
 159         int              of;
 160         int              lnn; /* line number in the real file */
 161         int              fd;
 162         int              inloop; /* Saw .while on this level. */
 163         unsigned char    c;
 164
 165         ln.sz = 256;
 166         ln.buf = mandoc_malloc(ln.sz);
 167         ln.next = NULL;
 168         firstln = lastln = loop = NULL;
 169         lnn = curp->line;
 170         pos = 0;
 171         inloop = 0;
 172         result = ROFF_CONT;
 173
 174         while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) {
 175                 if (start) {
 176                         curp->line = lnn;
 177                         curp->reparse_count = 0;
 178
 179                         if (lnn < 3 &&
 180                             curp->filenc & MPARSE_UTF8 &&
 181                             curp->filenc & MPARSE_LATIN1)
 182                                 curp->filenc = preconv_cue(&blk, i);
 183                 }
 184                 spos = pos;
 185
 186                 while (i < blk.sz && (start || blk.buf[i] != '\0')) {
 187
 188                         /*
 189                          * When finding an unescaped newline character,
 190                          * leave the character loop to process the line.
 191                          * Skip a preceding carriage return, if any.
 192                          */
 193
 194                         if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
 195                             '\n' == blk.buf[i + 1])
 196                                 ++i;
 197                         if ('\n' == blk.buf[i]) {
 198                                 ++i;
 199                                 ++lnn;
 200                                 break;
 201                         }
 202
 203                         /*
 204                          * Make sure we have space for the worst
 205                          * case of 12 bytes: "\\[u10ffff]\n\0"
 206                          */
 207
 208                         if (pos + 12 > ln.sz)
 209                                 resize_buf(&ln, 256);
 210
 211                         /*
 212                          * Encode 8-bit input.
 213                          */
 214
 215                         c = blk.buf[i];
 216                         if (c & 0x80) {
 217                                 if ( ! (curp->filenc && preconv_encode(
 218                                     &blk, &i, &ln, &pos, &curp->filenc))) {
 219                                         mandoc_msg(MANDOCERR_CHAR_BAD,
 220                                             curp->line, pos, "0x%x", c);
 221                                         ln.buf[pos++] = '?';
 222                                         i++;
 223                                 }
 224                                 continue;
 225                         }
 226
 227                         /*
 228                          * Exclude control characters.
 229                          */
 230
 231                         if (c == 0x7f || (c < 0x20 && c != 0x09)) {
 232                                 mandoc_msg(c == 0x00 || c == 0x04 ||
 233                                     c > 0x0a ? MANDOCERR_CHAR_BAD :
 234                                     MANDOCERR_CHAR_UNSUPP,
 235                                     curp->line, pos, "0x%x", c);
 236                                 i++;
 237                                 if (c != '\r')
 238                                         ln.buf[pos++] = '?';
 239                                 continue;
 240                         }
 241
 242                         ln.buf[pos++] = blk.buf[i++];
 243                 }
 244                 ln.buf[pos] = '\0';
 245
 246                 /*
 247                  * Maintain a lookaside buffer of all lines.
 248                  * parsed from this input source.
 249                  */
 250
 251                 thisln = mandoc_malloc(sizeof(*thisln));
 252                 thisln->buf = mandoc_strdup(ln.buf);
 253                 thisln->sz = strlen(ln.buf) + 1;
 254                 thisln->next = NULL;
 255                 if (firstln == NULL) {
 256                         firstln = lastln = thisln;
 257                         if (curp->secondary == NULL)
 258                                 curp->secondary = firstln;
 259                 } else {
 260                         lastln->next = thisln;
 261                         lastln = thisln;
 262                 }
 263
 264                 /*
 265                  * XXX Ugly hack to mark the end of the input,
 266                  * such that the function roff_parse_comment()
 267                  * doesn't attempt to append another line if the
 268                  * last input line ends with an escape character.
 269                  */
 270
 271                 if (i == blk.sz || blk.buf[i] == '\0') {
 272                         if (pos + 2 > ln.sz)
 273                                 resize_buf(&ln, 256);
 274                         ln.buf[pos++] = '\n';
 275                         ln.buf[pos] = '\0';
 276                 }
 277
 278                 /*
 279                  * A significant amount of complexity is contained by
 280                  * the roff preprocessor.  It's line-oriented but can be
 281                  * expressed on one line, so we need at times to
 282                  * readjust our starting point and re-run it.  The roff
 283                  * preprocessor can also readjust the buffers with new
 284                  * data, so we pass them in wholesale.
 285                  */
 286
 287                 of = 0;
 288 rerun:
 289                 line_result = roff_parseln(curp->roff, curp->line,
 290                     &ln, &of, start && spos == 0 ? pos : 0);
 291
 292                 /* Process options. */
 293
 294                 if (line_result & ROFF_APPEND)
 295                         assert(line_result == (ROFF_IGN | ROFF_APPEND));
 296
 297                 if (line_result & ROFF_USERCALL)
 298                         assert((line_result & ROFF_MASK) == ROFF_REPARSE);
 299
 300                 if (line_result & ROFF_USERRET) {
 301                         assert(line_result == (ROFF_IGN | ROFF_USERRET));
 302                         if (start == 0) {
 303                                 /* Return from the current macro. */
 304                                 result = ROFF_USERRET;
 305                                 goto out;
 306                         }
 307                 }
 308
 309                 switch (line_result & ROFF_LOOPMASK) {
 310                 case ROFF_IGN:
 311                         break;
 312                 case ROFF_WHILE:
 313                         if (curp->loop != NULL) {
 314                                 if (loop == curp->loop)
 315                                         break;
 316                                 mandoc_msg(MANDOCERR_WHILE_NEST,
 317                                     curp->line, pos, NULL);
 318                         }
 319                         curp->loop = thisln;
 320                         loop = NULL;
 321                         inloop = 1;
 322                         break;
 323                 case ROFF_LOOPCONT:
 324                 case ROFF_LOOPEXIT:
 325                         if (curp->loop == NULL) {
 326                                 mandoc_msg(MANDOCERR_WHILE_FAIL,
 327                                     curp->line, pos, NULL);
 328                                 break;
 329                         }
 330                         if (inloop == 0) {
 331                                 mandoc_msg(MANDOCERR_WHILE_INTO,
 332                                     curp->line, pos, NULL);
 333                                 curp->loop = loop = NULL;
 334                                 break;
 335                         }
 336                         if (line_result & ROFF_LOOPCONT)
 337                                 loop = curp->loop;
 338                         else {
 339                                 curp->loop = loop = NULL;
 340                                 inloop = 0;
 341                         }
 342                         break;
 343                 default:
 344                         abort();
 345                 }
 346
 347                 /* Process the main instruction from the roff parser. */
 348
 349                 switch (line_result & ROFF_MASK) {
 350                 case ROFF_IGN:
 351                         break;
 352                 case ROFF_CONT:
 353                         if (curp->man->meta.macroset == MACROSET_NONE)
 354                                 choose_parser(curp);
 355                         if ((curp->man->meta.macroset == MACROSET_MDOC ?
 356                              mdoc_parseln(curp->man, curp->line, ln.buf, of) :
 357                              man_parseln(curp->man, curp->line, ln.buf, of)
 358                             ) == 2)
 359                                 goto out;
 360                         break;
 361                 case ROFF_RERUN:
 362                         goto rerun;
 363                 case ROFF_REPARSE:
 364                         if (++curp->reparse_count > REPARSE_LIMIT) {
 365                                 /* Abort and return to the top level. */
 366                                 result = ROFF_IGN;
 367                                 mandoc_msg(MANDOCERR_ROFFLOOP,
 368                                     curp->line, pos, NULL);
 369                                 goto out;
 370                         }
 371                         result = mparse_buf_r(curp, ln, of, 0);
 372                         if (line_result & ROFF_USERCALL) {
 373                                 roff_userret(curp->roff);
 374                                 /* Continue normally. */
 375                                 if (result & ROFF_USERRET)
 376                                         result = ROFF_CONT;
 377                         }
 378                         if (start == 0 && result != ROFF_CONT)
 379                                 goto out;
 380                         break;
 381                 case ROFF_SO:
 382                         if ( ! (curp->options & MPARSE_SO) &&
 383                             (i >= blk.sz || blk.buf[i] == '\0')) {
 384                                 curp->man->meta.sodest =
 385                                     mandoc_strdup(ln.buf + of);
 386                                 goto out;
 387                         }
 388                         if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
 389                                 mparse_readfd(curp, fd, ln.buf + of);
 390                                 close(fd);
 391                         } else {
 392                                 mandoc_msg(MANDOCERR_SO_FAIL,
 393                                     curp->line, of, ".so %s: %s",
 394                                     ln.buf + of, strerror(errno));
 395                                 ln.sz = mandoc_asprintf(&cp,
 396                                     ".sp\nSee the file %s.\n.sp",
 397                                     ln.buf + of);
 398                                 free(ln.buf);
 399                                 ln.buf = cp;
 400                                 of = 0;
 401                                 mparse_buf_r(curp, ln, of, 0);
 402                         }
 403                         break;
 404                 default:
 405                         abort();
 406                 }
 407
 408                 /* Start the next input line. */
 409
 410                 if (loop != NULL &&
 411                     (line_result & ROFF_LOOPMASK) == ROFF_IGN)
 412                         loop = loop->next;
 413
 414                 if (loop != NULL) {
 415                         if ((line_result & ROFF_APPEND) == 0)
 416                                 *ln.buf = '\0';
 417                         if (ln.sz < loop->sz)
 418                                 resize_buf(&ln, loop->sz);
 419                         (void)strlcat(ln.buf, loop->buf, ln.sz);
 420                         of = 0;
 421                         goto rerun;
 422                 }
 423
 424                 pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
 425         }
 426 out:
 427         if (inloop) {
 428                 if (result != ROFF_USERRET)
 429                         mandoc_msg(MANDOCERR_WHILE_OUTOF,
 430                             curp->line, pos, NULL);
 431                 curp->loop = NULL;
 432         }
 433         free(ln.buf);
 434         if (firstln != curp->secondary)
 435                 free_buf_list(firstln);
 436         return result;
 437 }
 438
 439 static int
 440 read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
 441 {
 442         struct stat      st;
 443         gzFile           gz;
 444         size_t           off;
 445         ssize_t          ssz;
 446         int              gzerrnum, retval;
 447
 448         if (fstat(fd, &st) == -1) {
 449                 mandoc_msg(MANDOCERR_FSTAT, 0, 0, "%s", strerror(errno));
 450                 return -1;
 451         }
 452
 453         /*
 454          * If we're a regular file, try just reading in the whole entry
 455          * via mmap().  This is faster than reading it into blocks, and
 456          * since each file is only a few bytes to begin with, I'm not
 457          * concerned that this is going to tank any machines.
 458          */
 459
 460         if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
 461                 if (st.st_size > 0x7fffffff) {
 462                         mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
 463                         return -1;
 464                 }
 465                 *with_mmap = 1;
 466                 fb->sz = (size_t)st.st_size;
 467                 fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
 468                 if (fb->buf != MAP_FAILED)
 469                         return 0;
 470         }
 471
 472         if (curp->gzip) {
 473                 /*
 474                  * Duplicating the file descriptor is required
 475                  * because we will have to call gzclose(3)
 476                  * to free memory used internally by zlib,
 477                  * but that will also close the file descriptor,
 478                  * which this function must not do.
 479                  */
 480                 if ((fd = dup(fd)) == -1) {
 481                         mandoc_msg(MANDOCERR_DUP, 0, 0,
 482                             "%s", strerror(errno));
 483                         return -1;
 484                 }
 485                 if ((gz = gzdopen(fd, "rb")) == NULL) {
 486                         mandoc_msg(MANDOCERR_GZDOPEN, 0, 0,
 487                             "%s", strerror(errno));
 488                         close(fd);
 489                         return -1;
 490                 }
 491         } else
 492                 gz = NULL;
 493
 494         /*
 495          * If this isn't a regular file (like, say, stdin), then we must
 496          * go the old way and just read things in bit by bit.
 497          */
 498
 499         *with_mmap = 0;
 500         off = 0;
 501         retval = -1;
 502         fb->sz = 0;
 503         fb->buf = NULL;
 504         for (;;) {
 505                 if (off == fb->sz) {
 506                         if (fb->sz == (1U << 31)) {
 507                                 mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
 508                                 break;
 509                         }
 510                         resize_buf(fb, 65536);
 511                 }
 512                 ssz = curp->gzip ?
 513                     gzread(gz, fb->buf + (int)off, fb->sz - off) :
 514                     read(fd, fb->buf + (int)off, fb->sz - off);
 515                 if (ssz == 0) {
 516                         fb->sz = off;
 517                         retval = 0;
 518                         break;
 519                 }
 520                 if (ssz == -1) {
 521                         if (curp->gzip)
 522                                 (void)gzerror(gz, &gzerrnum);
 523                         mandoc_msg(MANDOCERR_READ, 0, 0, "%s",
 524                             curp->gzip && gzerrnum != Z_ERRNO ?
 525                             zError(gzerrnum) : strerror(errno));
 526                         break;
 527                 }
 528                 off += (size_t)ssz;
 529         }
 530
 531         if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
 532                 mandoc_msg(MANDOCERR_GZCLOSE, 0, 0, "%s",
 533                     gzerrnum == Z_ERRNO ? strerror(errno) :
 534                     zError(gzerrnum));
 535         if (retval == -1) {
 536                 free(fb->buf);
 537                 fb->buf = NULL;
 538         }
 539         return retval;
 540 }
 541
 542 static void
 543 mparse_end(struct mparse *curp)
 544 {
 545         if (curp->man->meta.macroset == MACROSET_NONE)
 546                 curp->man->meta.macroset = MACROSET_MAN;
 547         if (curp->man->meta.macroset == MACROSET_MDOC)
 548                 mdoc_endparse(curp->man);
 549         else
 550                 man_endparse(curp->man);
 551         roff_endparse(curp->roff);
 552 }
 553
 554 /*
 555  * Read the whole file into memory and call the parsers.
 556  * Called recursively when an .so request is encountered.
 557  */
 558 void
 559 mparse_readfd(struct mparse *curp, int fd, const char *filename)
 560 {
 561         static int       recursion_depth;
 562
 563         struct buf       blk;
 564         struct buf      *save_primary;
 565         const char      *save_filename, *cp;
 566         size_t           offset;
 567         int              save_filenc, save_lineno;
 568         int              with_mmap;
 569
 570         if (recursion_depth > 64) {
 571                 mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, 0, NULL);
 572                 return;
 573         } else if (recursion_depth == 0 &&
 574             (cp = strrchr(filename, '.')) != NULL &&
 575             cp[1] >= '1' && cp[1] <= '9')
 576                 curp->man->filesec = cp[1];
 577         else
 578                 curp->man->filesec = '\0';
 579
 580         if (read_whole_file(curp, fd, &blk, &with_mmap) == -1)
 581                 return;
 582
 583         /*
 584          * Save some properties of the parent file.
 585          */
 586
 587         save_primary = curp->primary;
 588         save_filenc = curp->filenc;
 589         save_lineno = curp->line;
 590         save_filename = mandoc_msg_getinfilename();
 591
 592         curp->primary = &blk;
 593         curp->filenc = curp->options & (MPARSE_UTF8 | MPARSE_LATIN1);
 594         curp->line = 1;
 595         mandoc_msg_setinfilename(filename);
 596
 597         /* Skip an UTF-8 byte order mark. */
 598         if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
 599             (unsigned char)blk.buf[0] == 0xef &&
 600             (unsigned char)blk.buf[1] == 0xbb &&
 601             (unsigned char)blk.buf[2] == 0xbf) {
 602                 offset = 3;
 603                 curp->filenc &= ~MPARSE_LATIN1;
 604         } else
 605                 offset = 0;
 606
 607         recursion_depth++;
 608         mparse_buf_r(curp, blk, offset, 1);
 609         if (--recursion_depth == 0)
 610                 mparse_end(curp);
 611
 612         /*
 613          * Clean up and restore saved parent properties.
 614          */
 615
 616         if (with_mmap)
 617                 munmap(blk.buf, blk.sz);
 618         else
 619                 free(blk.buf);
 620
 621         curp->primary = save_primary;
 622         curp->filenc = save_filenc;
 623         curp->line = save_lineno;
 624         if (save_filename != NULL)
 625                 mandoc_msg_setinfilename(save_filename);
 626 }
 627
 628 int
 629 mparse_open(struct mparse *curp, const char *file)
 630 {
 631         char             *cp;
 632         int               fd, save_errno;
 633
 634         cp = strrchr(file, '.');
 635         curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
 636
 637         /* First try to use the filename as it is. */
 638
 639         if ((fd = open(file, O_RDONLY)) != -1)
 640                 return fd;
 641
 642         /*
 643          * If that doesn't work and the filename doesn't
 644          * already  end in .gz, try appending .gz.
 645          */
 646
 647         if ( ! curp->gzip) {
 648                 save_errno = errno;
 649                 mandoc_asprintf(&cp, "%s.gz", file);
 650                 fd = open(cp, O_RDONLY);
 651                 free(cp);
 652                 errno = save_errno;
 653                 if (fd != -1) {
 654                         curp->gzip = 1;
 655                         return fd;
 656                 }
 657         }
 658
 659         /* Neither worked, give up. */
 660
 661         return -1;
 662 }
 663
 664 struct mparse *
 665 mparse_alloc(int options, enum mandoc_os os_e, const char *os_s)
 666 {
 667         struct mparse   *curp;
 668
 669         curp = mandoc_calloc(1, sizeof(struct mparse));
 670
 671         curp->options = options;
 672         curp->os_s = os_s;
 673
 674         curp->roff = roff_alloc(options);
 675         curp->man = roff_man_alloc(curp->roff, curp->os_s,
 676                 curp->options & MPARSE_QUICK ? 1 : 0);
 677         if (curp->options & MPARSE_MDOC) {
 678                 curp->man->meta.macroset = MACROSET_MDOC;
 679                 if (curp->man->mdocmac == NULL)
 680                         curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
 681         } else if (curp->options & MPARSE_MAN) {
 682                 curp->man->meta.macroset = MACROSET_MAN;
 683                 if (curp->man->manmac == NULL)
 684                         curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
 685         }
 686         curp->man->meta.first->tok = TOKEN_NONE;
 687         curp->man->meta.os_e = os_e;
 688         tag_alloc();
 689         return curp;
 690 }
 691
 692 void
 693 mparse_reset(struct mparse *curp)
 694 {
 695         tag_free();
 696         roff_reset(curp->roff);
 697         roff_man_reset(curp->man);
 698         free_buf_list(curp->secondary);
 699         curp->secondary = NULL;
 700         curp->gzip = 0;
 701         tag_alloc();
 702 }
 703
 704 void
 705 mparse_free(struct mparse *curp)
 706 {
 707         tag_free();
 708         roffhash_free(curp->man->mdocmac);
 709         roffhash_free(curp->man->manmac);
 710         roff_man_free(curp->man);
 711         roff_free(curp->roff);
 712         free_buf_list(curp->secondary);
 713         free(curp);
 714 }
 715
 716 struct roff_meta *
 717 mparse_result(struct mparse *curp)
 718 {
 719         roff_state_reset(curp->man);
 720         if (curp->options & MPARSE_VALIDATE) {
 721                 if (curp->man->meta.macroset == MACROSET_MDOC)
 722                         mdoc_validate(curp->man);
 723                 else
 724                         man_validate(curp->man);
 725                 tag_postprocess(curp->man, curp->man->meta.first);
 726         }
 727         return &curp->man->meta;
 728 }
 729
 730 void
 731 mparse_copy(const struct mparse *p)
 732 {
 733         struct buf      *buf;
 734
 735         for (buf = p->secondary; buf != NULL; buf = buf->next)
 736                 puts(buf->buf);
 737 }