Annotation of mandoc/read.c, Revision 1.221
1.221 ! schwarze 1: /* $Id: read.c,v 1.220 2021/06/27 17:57:54 schwarze Exp $ */
1.1 kristaps 2: /*
1.218 schwarze 3: * Copyright (c) 2010-2020 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 4: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.42 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 kristaps 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: *
1.133 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.133 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 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.
1.215 schwarze 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.
1.1 kristaps 24: */
1.11 kristaps 25: #include "config.h"
26:
1.80 schwarze 27: #include <sys/types.h>
1.82 schwarze 28: #include <sys/mman.h>
1.80 schwarze 29: #include <sys/stat.h>
1.1 kristaps 30:
31: #include <assert.h>
32: #include <ctype.h>
1.40 schwarze 33: #include <errno.h>
1.1 kristaps 34: #include <fcntl.h>
1.3 kristaps 35: #include <stdarg.h>
1.1 kristaps 36: #include <stdio.h>
37: #include <stdlib.h>
38: #include <string.h>
39: #include <unistd.h>
1.140 schwarze 40: #include <zlib.h>
1.1 kristaps 41:
1.133 schwarze 42: #include "mandoc_aux.h"
1.1 kristaps 43: #include "mandoc.h"
1.133 schwarze 44: #include "roff.h"
1.1 kristaps 45: #include "mdoc.h"
46: #include "man.h"
1.202 schwarze 47: #include "mandoc_parse.h"
1.133 schwarze 48: #include "libmandoc.h"
1.201 schwarze 49: #include "roff_int.h"
1.218 schwarze 50: #include "tag.h"
1.1 kristaps 51:
52: #define REPARSE_LIMIT 1000
53:
54: struct mparse {
1.160 schwarze 55: struct roff *roff; /* roff parser (!NULL) */
1.134 schwarze 56: struct roff_man *man; /* man parser */
1.83 schwarze 57: struct buf *primary; /* buffer currently being parsed */
1.198 schwarze 58: struct buf *secondary; /* copy of top level input */
1.199 schwarze 59: struct buf *loop; /* open .while request line */
1.179 schwarze 60: const char *os_s; /* default operating system */
1.83 schwarze 61: int options; /* parser options */
1.140 schwarze 62: int gzip; /* current input file is gzipped */
1.93 schwarze 63: int filenc; /* encoding of the current file */
1.1 kristaps 64: int reparse_count; /* finite interp. stack */
1.83 schwarze 65: int line; /* line number in the file */
1.1 kristaps 66: };
67:
1.84 schwarze 68: static void choose_parser(struct mparse *);
1.198 schwarze 69: static void free_buf_list(struct buf *);
1.1 kristaps 70: static void resize_buf(struct buf *, size_t);
1.199 schwarze 71: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.203 schwarze 72: static int read_whole_file(struct mparse *, int, struct buf *, int *);
1.1 kristaps 73: static void mparse_end(struct mparse *);
74:
1.47 schwarze 75:
1.1 kristaps 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
1.198 schwarze 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
1.84 schwarze 98: choose_parser(struct mparse *curp)
1.1 kristaps 99: {
1.83 schwarze 100: char *cp, *ep;
101: int format;
1.1 kristaps 102:
1.83 schwarze 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) {
1.85 schwarze 113: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 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: }
1.1 kristaps 129: }
130:
1.83 schwarze 131: if (format == MPARSE_MDOC) {
1.209 schwarze 132: curp->man->meta.macroset = MACROSET_MDOC;
1.164 schwarze 133: if (curp->man->mdocmac == NULL)
134: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 135: } else {
1.209 schwarze 136: curp->man->meta.macroset = MACROSET_MAN;
1.164 schwarze 137: if (curp->man->manmac == NULL)
138: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 139: }
1.209 schwarze 140: curp->man->meta.first->tok = TOKEN_NONE;
1.1 kristaps 141: }
142:
143: /*
1.95 schwarze 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).
1.1 kristaps 149: */
1.199 schwarze 150: static int
1.95 schwarze 151: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 152: {
153: struct buf ln;
1.199 schwarze 154: struct buf *firstln, *lastln, *thisln, *loop;
1.110 schwarze 155: char *cp;
1.95 schwarze 156: size_t pos; /* byte number in the ln buffer */
1.220 schwarze 157: size_t spos; /* at the start of the current line parse */
1.199 schwarze 158: int line_result, result;
1.100 schwarze 159: int of;
1.1 kristaps 160: int lnn; /* line number in the real file */
1.110 schwarze 161: int fd;
1.199 schwarze 162: int inloop; /* Saw .while on this level. */
1.1 kristaps 163: unsigned char c;
164:
1.198 schwarze 165: ln.sz = 256;
166: ln.buf = mandoc_malloc(ln.sz);
167: ln.next = NULL;
1.213 schwarze 168: firstln = lastln = loop = NULL;
1.47 schwarze 169: lnn = curp->line;
170: pos = 0;
1.199 schwarze 171: inloop = 0;
1.198 schwarze 172: result = ROFF_CONT;
1.1 kristaps 173:
1.199 schwarze 174: while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) {
1.1 kristaps 175: if (start) {
176: curp->line = lnn;
177: curp->reparse_count = 0;
1.93 schwarze 178:
179: if (lnn < 3 &&
180: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 181: curp->filenc & MPARSE_LATIN1)
182: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 183: }
1.220 schwarze 184: spos = pos;
1.1 kristaps 185:
1.95 schwarze 186: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 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:
1.95 schwarze 194: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 195: '\n' == blk.buf[i + 1])
196: ++i;
197: if ('\n' == blk.buf[i]) {
198: ++i;
199: ++lnn;
200: break;
201: }
202:
1.35 schwarze 203: /*
1.93 schwarze 204: * Make sure we have space for the worst
1.198 schwarze 205: * case of 12 bytes: "\\[u10ffff]\n\0"
1.35 schwarze 206: */
207:
1.198 schwarze 208: if (pos + 12 > ln.sz)
1.35 schwarze 209: resize_buf(&ln, 256);
210:
1.47 schwarze 211: /*
1.93 schwarze 212: * Encode 8-bit input.
1.1 kristaps 213: */
214:
1.93 schwarze 215: c = blk.buf[i];
216: if (c & 0x80) {
1.95 schwarze 217: if ( ! (curp->filenc && preconv_encode(
218: &blk, &i, &ln, &pos, &curp->filenc))) {
1.206 schwarze 219: mandoc_msg(MANDOCERR_CHAR_BAD,
1.114 schwarze 220: curp->line, pos, "0x%x", c);
1.93 schwarze 221: ln.buf[pos++] = '?';
222: i++;
223: }
224: continue;
225: }
226:
227: /*
228: * Exclude control characters.
229: */
1.1 kristaps 230:
1.93 schwarze 231: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.206 schwarze 232: mandoc_msg(c == 0x00 || c == 0x04 ||
1.114 schwarze 233: c > 0x0a ? MANDOCERR_CHAR_BAD :
234: MANDOCERR_CHAR_UNSUPP,
1.206 schwarze 235: curp->line, pos, "0x%x", c);
1.1 kristaps 236: i++;
1.127 schwarze 237: if (c != '\r')
238: ln.buf[pos++] = '?';
1.1 kristaps 239: continue;
240: }
241:
242: ln.buf[pos++] = blk.buf[i++];
243: }
1.198 schwarze 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: }
1.1 kristaps 263:
1.221 ! schwarze 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: */
1.1 kristaps 270:
1.198 schwarze 271: if (i == blk.sz || blk.buf[i] == '\0') {
1.212 schwarze 272: if (pos + 2 > ln.sz)
273: resize_buf(&ln, 256);
1.170 schwarze 274: ln.buf[pos++] = '\n';
1.198 schwarze 275: ln.buf[pos] = '\0';
276: }
1.1 kristaps 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:
1.220 schwarze 289: line_result = roff_parseln(curp->roff, curp->line,
290: &ln, &of, start && spos == 0 ? pos : 0);
1.1 kristaps 291:
1.199 schwarze 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,
1.206 schwarze 317: curp->line, pos, NULL);
1.199 schwarze 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,
1.206 schwarze 327: curp->line, pos, NULL);
1.199 schwarze 328: break;
329: }
330: if (inloop == 0) {
331: mandoc_msg(MANDOCERR_WHILE_INTO,
1.206 schwarze 332: curp->line, pos, NULL);
1.199 schwarze 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:
1.209 schwarze 353: if (curp->man->meta.macroset == MACROSET_NONE)
1.199 schwarze 354: choose_parser(curp);
1.209 schwarze 355: if ((curp->man->meta.macroset == MACROSET_MDOC ?
1.199 schwarze 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;
1.47 schwarze 363: case ROFF_REPARSE:
1.197 schwarze 364: if (++curp->reparse_count > REPARSE_LIMIT) {
1.199 schwarze 365: /* Abort and return to the top level. */
1.198 schwarze 366: result = ROFF_IGN;
1.206 schwarze 367: mandoc_msg(MANDOCERR_ROFFLOOP,
1.47 schwarze 368: curp->line, pos, NULL);
1.199 schwarze 369: goto out;
1.197 schwarze 370: }
1.199 schwarze 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;
1.163 schwarze 377: }
1.199 schwarze 378: if (start == 0 && result != ROFF_CONT)
379: goto out;
380: break;
1.47 schwarze 381: case ROFF_SO:
1.95 schwarze 382: if ( ! (curp->options & MPARSE_SO) &&
383: (i >= blk.sz || blk.buf[i] == '\0')) {
1.209 schwarze 384: curp->man->meta.sodest =
385: mandoc_strdup(ln.buf + of);
1.198 schwarze 386: goto out;
1.45 schwarze 387: }
1.148 schwarze 388: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 389: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 390: close(fd);
1.113 schwarze 391: } else {
1.208 schwarze 392: mandoc_msg(MANDOCERR_SO_FAIL,
393: curp->line, of, ".so %s: %s",
394: ln.buf + of, strerror(errno));
1.110 schwarze 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);
1.52 schwarze 402: }
1.199 schwarze 403: break;
1.1 kristaps 404: default:
1.199 schwarze 405: abort();
1.1 kristaps 406: }
407:
1.199 schwarze 408: /* Start the next input line. */
1.1 kristaps 409:
1.199 schwarze 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: }
1.1 kristaps 423:
1.199 schwarze 424: pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
1.1 kristaps 425: }
1.198 schwarze 426: out:
1.199 schwarze 427: if (inloop) {
428: if (result != ROFF_USERRET)
1.206 schwarze 429: mandoc_msg(MANDOCERR_WHILE_OUTOF,
1.199 schwarze 430: curp->line, pos, NULL);
431: curp->loop = NULL;
432: }
1.1 kristaps 433: free(ln.buf);
1.198 schwarze 434: if (firstln != curp->secondary)
435: free_buf_list(firstln);
436: return result;
1.1 kristaps 437: }
438:
439: static int
1.203 schwarze 440: read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
1.1 kristaps 441: {
1.161 schwarze 442: struct stat st;
1.140 schwarze 443: gzFile gz;
1.1 kristaps 444: size_t off;
445: ssize_t ssz;
1.194 schwarze 446: int gzerrnum, retval;
1.143 schwarze 447:
1.192 schwarze 448: if (fstat(fd, &st) == -1) {
1.214 schwarze 449: mandoc_msg(MANDOCERR_FSTAT, 0, 0, "%s", strerror(errno));
450: return -1;
1.192 schwarze 451: }
1.1 kristaps 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:
1.140 schwarze 460: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 461: if (st.st_size > 0x7fffffff) {
1.206 schwarze 462: mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
1.214 schwarze 463: return -1;
1.1 kristaps 464: }
465: *with_mmap = 1;
466: fb->sz = (size_t)st.st_size;
1.37 schwarze 467: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 468: if (fb->buf != MAP_FAILED)
1.214 schwarze 469: return 0;
1.1 kristaps 470: }
471:
1.140 schwarze 472: if (curp->gzip) {
1.194 schwarze 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) {
1.214 schwarze 481: mandoc_msg(MANDOCERR_DUP, 0, 0,
482: "%s", strerror(errno));
483: return -1;
1.194 schwarze 484: }
1.192 schwarze 485: if ((gz = gzdopen(fd, "rb")) == NULL) {
1.214 schwarze 486: mandoc_msg(MANDOCERR_GZDOPEN, 0, 0,
487: "%s", strerror(errno));
1.194 schwarze 488: close(fd);
1.214 schwarze 489: return -1;
1.192 schwarze 490: }
1.140 schwarze 491: } else
492: gz = NULL;
493:
1.1 kristaps 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;
1.214 schwarze 501: retval = -1;
1.1 kristaps 502: fb->sz = 0;
503: fb->buf = NULL;
504: for (;;) {
505: if (off == fb->sz) {
506: if (fb->sz == (1U << 31)) {
1.206 schwarze 507: mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
1.1 kristaps 508: break;
509: }
510: resize_buf(fb, 65536);
511: }
1.140 schwarze 512: ssz = curp->gzip ?
513: gzread(gz, fb->buf + (int)off, fb->sz - off) :
514: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 515: if (ssz == 0) {
516: fb->sz = off;
1.214 schwarze 517: retval = 0;
1.194 schwarze 518: break;
1.1 kristaps 519: }
1.192 schwarze 520: if (ssz == -1) {
1.194 schwarze 521: if (curp->gzip)
522: (void)gzerror(gz, &gzerrnum);
1.214 schwarze 523: mandoc_msg(MANDOCERR_READ, 0, 0, "%s",
1.194 schwarze 524: curp->gzip && gzerrnum != Z_ERRNO ?
525: zError(gzerrnum) : strerror(errno));
1.192 schwarze 526: break;
527: }
1.1 kristaps 528: off += (size_t)ssz;
529: }
530:
1.194 schwarze 531: if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
1.214 schwarze 532: mandoc_msg(MANDOCERR_GZCLOSE, 0, 0, "%s",
1.194 schwarze 533: gzerrnum == Z_ERRNO ? strerror(errno) :
534: zError(gzerrnum));
1.214 schwarze 535: if (retval == -1) {
1.194 schwarze 536: free(fb->buf);
537: fb->buf = NULL;
538: }
539: return retval;
1.1 kristaps 540: }
541:
542: static void
543: mparse_end(struct mparse *curp)
544: {
1.209 schwarze 545: if (curp->man->meta.macroset == MACROSET_NONE)
546: curp->man->meta.macroset = MACROSET_MAN;
547: if (curp->man->meta.macroset == MACROSET_MDOC)
1.135 schwarze 548: mdoc_endparse(curp->man);
549: else
1.111 schwarze 550: man_endparse(curp->man);
1.1 kristaps 551: roff_endparse(curp->roff);
552: }
553:
1.205 schwarze 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)
1.28 joerg 560: {
1.205 schwarze 561: static int recursion_depth;
562:
563: struct buf blk;
564: struct buf *save_primary;
1.219 schwarze 565: const char *save_filename, *cp;
1.95 schwarze 566: size_t offset;
1.205 schwarze 567: int save_filenc, save_lineno;
568: int with_mmap;
1.36 schwarze 569:
1.205 schwarze 570: if (recursion_depth > 64) {
1.206 schwarze 571: mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, 0, NULL);
1.36 schwarze 572: return;
1.219 schwarze 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:
1.214 schwarze 580: if (read_whole_file(curp, fd, &blk, &with_mmap) == -1)
1.205 schwarze 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();
1.28 joerg 591:
1.83 schwarze 592: curp->primary = &blk;
1.205 schwarze 593: curp->filenc = curp->options & (MPARSE_UTF8 | MPARSE_LATIN1);
1.28 joerg 594: curp->line = 1;
1.205 schwarze 595: mandoc_msg_setinfilename(filename);
1.28 joerg 596:
1.93 schwarze 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) {
1.95 schwarze 602: offset = 3;
1.93 schwarze 603: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 604: } else
605: offset = 0;
1.93 schwarze 606:
1.205 schwarze 607: recursion_depth++;
1.95 schwarze 608: mparse_buf_r(curp, blk, offset, 1);
1.111 schwarze 609: if (--recursion_depth == 0)
1.28 joerg 610: mparse_end(curp);
611:
1.205 schwarze 612: /*
613: * Clean up and restore saved parent properties.
614: */
1.28 joerg 615:
1.205 schwarze 616: if (with_mmap)
617: munmap(blk.buf, blk.sz);
618: else
619: free(blk.buf);
1.1 kristaps 620:
1.205 schwarze 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);
1.82 schwarze 626: }
627:
1.148 schwarze 628: int
629: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 630: {
631: char *cp;
1.211 schwarze 632: int fd, save_errno;
1.82 schwarze 633:
1.140 schwarze 634: cp = strrchr(file, '.');
635: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 636:
1.140 schwarze 637: /* First try to use the filename as it is. */
1.98 schwarze 638:
1.148 schwarze 639: if ((fd = open(file, O_RDONLY)) != -1)
640: return fd;
1.98 schwarze 641:
1.140 schwarze 642: /*
643: * If that doesn't work and the filename doesn't
644: * already end in .gz, try appending .gz.
645: */
1.98 schwarze 646:
1.140 schwarze 647: if ( ! curp->gzip) {
1.211 schwarze 648: save_errno = errno;
1.98 schwarze 649: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 650: fd = open(cp, O_RDONLY);
1.108 schwarze 651: free(cp);
1.211 schwarze 652: errno = save_errno;
1.148 schwarze 653: if (fd != -1) {
1.140 schwarze 654: curp->gzip = 1;
1.148 schwarze 655: return fd;
1.82 schwarze 656: }
657: }
658:
1.140 schwarze 659: /* Neither worked, give up. */
1.97 schwarze 660:
1.148 schwarze 661: return -1;
1.1 kristaps 662: }
663:
664: struct mparse *
1.203 schwarze 665: mparse_alloc(int options, enum mandoc_os os_e, const char *os_s)
1.1 kristaps 666: {
667: struct mparse *curp;
1.10 kristaps 668:
1.1 kristaps 669: curp = mandoc_calloc(1, sizeof(struct mparse));
670:
1.44 schwarze 671: curp->options = options;
1.179 schwarze 672: curp->os_s = os_s;
1.1 kristaps 673:
1.207 schwarze 674: curp->roff = roff_alloc(options);
675: curp->man = roff_man_alloc(curp->roff, curp->os_s,
1.137 schwarze 676: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 677: if (curp->options & MPARSE_MDOC) {
1.209 schwarze 678: curp->man->meta.macroset = MACROSET_MDOC;
1.164 schwarze 679: if (curp->man->mdocmac == NULL)
680: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 681: } else if (curp->options & MPARSE_MAN) {
1.209 schwarze 682: curp->man->meta.macroset = MACROSET_MAN;
1.164 schwarze 683: if (curp->man->manmac == NULL)
684: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 685: }
1.209 schwarze 686: curp->man->meta.first->tok = TOKEN_NONE;
1.179 schwarze 687: curp->man->meta.os_e = os_e;
1.215 schwarze 688: tag_alloc();
1.142 schwarze 689: return curp;
1.1 kristaps 690: }
691:
692: void
693: mparse_reset(struct mparse *curp)
694: {
1.216 schwarze 695: tag_free();
1.1 kristaps 696: roff_reset(curp->roff);
1.150 schwarze 697: roff_man_reset(curp->man);
1.198 schwarze 698: free_buf_list(curp->secondary);
699: curp->secondary = NULL;
1.159 schwarze 700: curp->gzip = 0;
1.216 schwarze 701: tag_alloc();
1.1 kristaps 702: }
703:
704: void
705: mparse_free(struct mparse *curp)
706: {
1.215 schwarze 707: tag_free();
1.164 schwarze 708: roffhash_free(curp->man->mdocmac);
709: roffhash_free(curp->man->manmac);
1.137 schwarze 710: roff_man_free(curp->man);
1.160 schwarze 711: roff_free(curp->roff);
1.198 schwarze 712: free_buf_list(curp->secondary);
1.1 kristaps 713: free(curp);
714: }
715:
1.209 schwarze 716: struct roff_meta *
717: mparse_result(struct mparse *curp)
1.1 kristaps 718: {
1.210 schwarze 719: roff_state_reset(curp->man);
1.209 schwarze 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);
1.218 schwarze 725: tag_postprocess(curp->man, curp->man->meta.first);
1.45 schwarze 726: }
1.209 schwarze 727: return &curp->man->meta;
1.24 kristaps 728: }
729:
730: void
1.198 schwarze 731: mparse_copy(const struct mparse *p)
1.24 kristaps 732: {
1.198 schwarze 733: struct buf *buf;
1.24 kristaps 734:
1.198 schwarze 735: for (buf = p->secondary; buf != NULL; buf = buf->next)
736: puts(buf->buf);
1.1 kristaps 737: }
CVSweb