]> git.cameronkatri.com Git - mandoc.git/blob - read.c
git.ckatri.com / mandoc.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Rename mandoc_getarg() to roff_getarg() and pass it the roff parser
[mandoc.git] / read.c
1 /* $Id: read.c,v 1.208 2018/12/20 21:30:32 schwarze Exp $ */
2 /*
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010-2018 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 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 #include "config.h"
20
21 #include <sys/types.h>
22 #include <sys/mman.h>
23 #include <sys/stat.h>
24
25 #include <assert.h>
26 #include <ctype.h>
27 #include <errno.h>
28 #include <fcntl.h>
29 #include <stdarg.h>
30 #include <stdio.h>
31 #include <stdlib.h>
32 #include <string.h>
33 #include <unistd.h>
34 #include <zlib.h>
35
36 #include "mandoc_aux.h"
37 #include "mandoc.h"
38 #include "roff.h"
39 #include "mdoc.h"
40 #include "man.h"
41 #include "mandoc_parse.h"
42 #include "libmandoc.h"
43 #include "roff_int.h"
44
45 #define REPARSE_LIMIT 1000
46
47 struct mparse {
48 struct roff *roff; /* roff parser (!NULL) */
49 struct roff_man *man; /* man parser */
50 char *sodest; /* filename pointed to by .so */
51 struct buf *primary; /* buffer currently being parsed */
52 struct buf *secondary; /* copy of top level input */
53 struct buf *loop; /* open .while request line */
54 const char *os_s; /* default operating system */
55 int options; /* parser options */
56 int gzip; /* current input file is gzipped */
57 int filenc; /* encoding of the current file */
58 int reparse_count; /* finite interp. stack */
59 int line; /* line number in the file */
60 };
61
62 static void choose_parser(struct mparse *);
63 static void free_buf_list(struct buf *);
64 static void resize_buf(struct buf *, size_t);
65 static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
66 static int read_whole_file(struct mparse *, int, struct buf *, int *);
67 static void mparse_end(struct mparse *);
68
69
70 static void
71 resize_buf(struct buf *buf, size_t initial)
72 {
73
74 buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
75 buf->buf = mandoc_realloc(buf->buf, buf->sz);
76 }
77
78 static void
79 free_buf_list(struct buf *buf)
80 {
81 struct buf *tmp;
82
83 while (buf != NULL) {
84 tmp = buf;
85 buf = tmp->next;
86 free(tmp->buf);
87 free(tmp);
88 }
89 }
90
91 static void
92 choose_parser(struct mparse *curp)
93 {
94 char *cp, *ep;
95 int format;
96
97 /*
98 * If neither command line arguments -mdoc or -man select
99 * a parser nor the roff parser found a .Dd or .TH macro
100 * yet, look ahead in the main input buffer.
101 */
102
103 if ((format = roff_getformat(curp->roff)) == 0) {
104 cp = curp->primary->buf;
105 ep = cp + curp->primary->sz;
106 while (cp < ep) {
107 if (*cp == '.' || *cp == '\'') {
108 cp++;
109 if (cp[0] == 'D' && cp[1] == 'd') {
110 format = MPARSE_MDOC;
111 break;
112 }
113 if (cp[0] == 'T' && cp[1] == 'H') {
114 format = MPARSE_MAN;
115 break;
116 }
117 }
118 cp = memchr(cp, '\n', ep - cp);
119 if (cp == NULL)
120 break;
121 cp++;
122 }
123 }
124
125 if (format == MPARSE_MDOC) {
126 curp->man->macroset = MACROSET_MDOC;
127 if (curp->man->mdocmac == NULL)
128 curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
129 } else {
130 curp->man->macroset = MACROSET_MAN;
131 if (curp->man->manmac == NULL)
132 curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
133 }
134 curp->man->first->tok = TOKEN_NONE;
135 }
136
137 /*
138 * Main parse routine for a buffer.
139 * It assumes encoding and line numbering are already set up.
140 * It can recurse directly (for invocations of user-defined
141 * macros, inline equations, and input line traps)
142 * and indirectly (for .so file inclusion).
143 */
144 static int
145 mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
146 {
147 struct buf ln;
148 struct buf *firstln, *lastln, *thisln, *loop;
149 char *cp;
150 size_t pos; /* byte number in the ln buffer */
151 int line_result, result;
152 int of;
153 int lnn; /* line number in the real file */
154 int fd;
155 int inloop; /* Saw .while on this level. */
156 unsigned char c;
157
158 ln.sz = 256;
159 ln.buf = mandoc_malloc(ln.sz);
160 ln.next = NULL;
161 firstln = loop = NULL;
162 lnn = curp->line;
163 pos = 0;
164 inloop = 0;
165 result = ROFF_CONT;
166
167 while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) {
168 if (start) {
169 curp->line = lnn;
170 curp->reparse_count = 0;
171
172 if (lnn < 3 &&
173 curp->filenc & MPARSE_UTF8 &&
174 curp->filenc & MPARSE_LATIN1)
175 curp->filenc = preconv_cue(&blk, i);
176 }
177
178 while (i < blk.sz && (start || blk.buf[i] != '\0')) {
179
180 /*
181 * When finding an unescaped newline character,
182 * leave the character loop to process the line.
183 * Skip a preceding carriage return, if any.
184 */
185
186 if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
187 '\n' == blk.buf[i + 1])
188 ++i;
189 if ('\n' == blk.buf[i]) {
190 ++i;
191 ++lnn;
192 break;
193 }
194
195 /*
196 * Make sure we have space for the worst
197 * case of 12 bytes: "\\[u10ffff]\n\0"
198 */
199
200 if (pos + 12 > ln.sz)
201 resize_buf(&ln, 256);
202
203 /*
204 * Encode 8-bit input.
205 */
206
207 c = blk.buf[i];
208 if (c & 0x80) {
209 if ( ! (curp->filenc && preconv_encode(
210 &blk, &i, &ln, &pos, &curp->filenc))) {
211 mandoc_msg(MANDOCERR_CHAR_BAD,
212 curp->line, pos, "0x%x", c);
213 ln.buf[pos++] = '?';
214 i++;
215 }
216 continue;
217 }
218
219 /*
220 * Exclude control characters.
221 */
222
223 if (c == 0x7f || (c < 0x20 && c != 0x09)) {
224 mandoc_msg(c == 0x00 || c == 0x04 ||
225 c > 0x0a ? MANDOCERR_CHAR_BAD :
226 MANDOCERR_CHAR_UNSUPP,
227 curp->line, pos, "0x%x", c);
228 i++;
229 if (c != '\r')
230 ln.buf[pos++] = '?';
231 continue;
232 }
233
234 ln.buf[pos++] = blk.buf[i++];
235 }
236 ln.buf[pos] = '\0';
237
238 /*
239 * Maintain a lookaside buffer of all lines.
240 * parsed from this input source.
241 */
242
243 thisln = mandoc_malloc(sizeof(*thisln));
244 thisln->buf = mandoc_strdup(ln.buf);
245 thisln->sz = strlen(ln.buf) + 1;
246 thisln->next = NULL;
247 if (firstln == NULL) {
248 firstln = lastln = thisln;
249 if (curp->secondary == NULL)
250 curp->secondary = firstln;
251 } else {
252 lastln->next = thisln;
253 lastln = thisln;
254 }
255
256 /* XXX Ugly hack to mark the end of the input. */
257
258 if (i == blk.sz || blk.buf[i] == '\0') {
259 ln.buf[pos++] = '\n';
260 ln.buf[pos] = '\0';
261 }
262
263 /*
264 * A significant amount of complexity is contained by
265 * the roff preprocessor. It's line-oriented but can be
266 * expressed on one line, so we need at times to
267 * readjust our starting point and re-run it. The roff
268 * preprocessor can also readjust the buffers with new
269 * data, so we pass them in wholesale.
270 */
271
272 of = 0;
273 rerun:
274 line_result = roff_parseln(curp->roff, curp->line, &ln, &of);
275
276 /* Process options. */
277
278 if (line_result & ROFF_APPEND)
279 assert(line_result == (ROFF_IGN | ROFF_APPEND));
280
281 if (line_result & ROFF_USERCALL)
282 assert((line_result & ROFF_MASK) == ROFF_REPARSE);
283
284 if (line_result & ROFF_USERRET) {
285 assert(line_result == (ROFF_IGN | ROFF_USERRET));
286 if (start == 0) {
287 /* Return from the current macro. */
288 result = ROFF_USERRET;
289 goto out;
290 }
291 }
292
293 switch (line_result & ROFF_LOOPMASK) {
294 case ROFF_IGN:
295 break;
296 case ROFF_WHILE:
297 if (curp->loop != NULL) {
298 if (loop == curp->loop)
299 break;
300 mandoc_msg(MANDOCERR_WHILE_NEST,
301 curp->line, pos, NULL);
302 }
303 curp->loop = thisln;
304 loop = NULL;
305 inloop = 1;
306 break;
307 case ROFF_LOOPCONT:
308 case ROFF_LOOPEXIT:
309 if (curp->loop == NULL) {
310 mandoc_msg(MANDOCERR_WHILE_FAIL,
311 curp->line, pos, NULL);
312 break;
313 }
314 if (inloop == 0) {
315 mandoc_msg(MANDOCERR_WHILE_INTO,
316 curp->line, pos, NULL);
317 curp->loop = loop = NULL;
318 break;
319 }
320 if (line_result & ROFF_LOOPCONT)
321 loop = curp->loop;
322 else {
323 curp->loop = loop = NULL;
324 inloop = 0;
325 }
326 break;
327 default:
328 abort();
329 }
330
331 /* Process the main instruction from the roff parser. */
332
333 switch (line_result & ROFF_MASK) {
334 case ROFF_IGN:
335 break;
336 case ROFF_CONT:
337 if (curp->man->macroset == MACROSET_NONE)
338 choose_parser(curp);
339 if ((curp->man->macroset == MACROSET_MDOC ?
340 mdoc_parseln(curp->man, curp->line, ln.buf, of) :
341 man_parseln(curp->man, curp->line, ln.buf, of)
342 ) == 2)
343 goto out;
344 break;
345 case ROFF_RERUN:
346 goto rerun;
347 case ROFF_REPARSE:
348 if (++curp->reparse_count > REPARSE_LIMIT) {
349 /* Abort and return to the top level. */
350 result = ROFF_IGN;
351 mandoc_msg(MANDOCERR_ROFFLOOP,
352 curp->line, pos, NULL);
353 goto out;
354 }
355 result = mparse_buf_r(curp, ln, of, 0);
356 if (line_result & ROFF_USERCALL) {
357 roff_userret(curp->roff);
358 /* Continue normally. */
359 if (result & ROFF_USERRET)
360 result = ROFF_CONT;
361 }
362 if (start == 0 && result != ROFF_CONT)
363 goto out;
364 break;
365 case ROFF_SO:
366 if ( ! (curp->options & MPARSE_SO) &&
367 (i >= blk.sz || blk.buf[i] == '\0')) {
368 curp->sodest = mandoc_strdup(ln.buf + of);
369 goto out;
370 }
371 if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
372 mparse_readfd(curp, fd, ln.buf + of);
373 close(fd);
374 } else {
375 mandoc_msg(MANDOCERR_SO_FAIL,
376 curp->line, of, ".so %s: %s",
377 ln.buf + of, strerror(errno));
378 ln.sz = mandoc_asprintf(&cp,
379 ".sp\nSee the file %s.\n.sp",
380 ln.buf + of);
381 free(ln.buf);
382 ln.buf = cp;
383 of = 0;
384 mparse_buf_r(curp, ln, of, 0);
385 }
386 break;
387 default:
388 abort();
389 }
390
391 /* Start the next input line. */
392
393 if (loop != NULL &&
394 (line_result & ROFF_LOOPMASK) == ROFF_IGN)
395 loop = loop->next;
396
397 if (loop != NULL) {
398 if ((line_result & ROFF_APPEND) == 0)
399 *ln.buf = '\0';
400 if (ln.sz < loop->sz)
401 resize_buf(&ln, loop->sz);
402 (void)strlcat(ln.buf, loop->buf, ln.sz);
403 of = 0;
404 goto rerun;
405 }
406
407 pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
408 }
409 out:
410 if (inloop) {
411 if (result != ROFF_USERRET)
412 mandoc_msg(MANDOCERR_WHILE_OUTOF,
413 curp->line, pos, NULL);
414 curp->loop = NULL;
415 }
416 free(ln.buf);
417 if (firstln != curp->secondary)
418 free_buf_list(firstln);
419 return result;
420 }
421
422 static int
423 read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
424 {
425 struct stat st;
426 gzFile gz;
427 size_t off;
428 ssize_t ssz;
429 int gzerrnum, retval;
430
431 if (fstat(fd, &st) == -1) {
432 mandoc_msg(MANDOCERR_FILE, 0, 0,
433 "fstat: %s", strerror(errno));
434 return 0;
435 }
436
437 /*
438 * If we're a regular file, try just reading in the whole entry
439 * via mmap(). This is faster than reading it into blocks, and
440 * since each file is only a few bytes to begin with, I'm not
441 * concerned that this is going to tank any machines.
442 */
443
444 if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
445 if (st.st_size > 0x7fffffff) {
446 mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
447 return 0;
448 }
449 *with_mmap = 1;
450 fb->sz = (size_t)st.st_size;
451 fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
452 if (fb->buf != MAP_FAILED)
453 return 1;
454 }
455
456 if (curp->gzip) {
457 /*
458 * Duplicating the file descriptor is required
459 * because we will have to call gzclose(3)
460 * to free memory used internally by zlib,
461 * but that will also close the file descriptor,
462 * which this function must not do.
463 */
464 if ((fd = dup(fd)) == -1) {
465 mandoc_msg(MANDOCERR_FILE, 0, 0,
466 "dup: %s", strerror(errno));
467 return 0;
468 }
469 if ((gz = gzdopen(fd, "rb")) == NULL) {
470 mandoc_msg(MANDOCERR_FILE, 0, 0,
471 "gzdopen: %s", strerror(errno));
472 close(fd);
473 return 0;
474 }
475 } else
476 gz = NULL;
477
478 /*
479 * If this isn't a regular file (like, say, stdin), then we must
480 * go the old way and just read things in bit by bit.
481 */
482
483 *with_mmap = 0;
484 off = 0;
485 retval = 0;
486 fb->sz = 0;
487 fb->buf = NULL;
488 for (;;) {
489 if (off == fb->sz) {
490 if (fb->sz == (1U << 31)) {
491 mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
492 break;
493 }
494 resize_buf(fb, 65536);
495 }
496 ssz = curp->gzip ?
497 gzread(gz, fb->buf + (int)off, fb->sz - off) :
498 read(fd, fb->buf + (int)off, fb->sz - off);
499 if (ssz == 0) {
500 fb->sz = off;
501 retval = 1;
502 break;
503 }
504 if (ssz == -1) {
505 if (curp->gzip)
506 (void)gzerror(gz, &gzerrnum);
507 mandoc_msg(MANDOCERR_FILE, 0, 0, "read: %s",
508 curp->gzip && gzerrnum != Z_ERRNO ?
509 zError(gzerrnum) : strerror(errno));
510 break;
511 }
512 off += (size_t)ssz;
513 }
514
515 if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
516 mandoc_msg(MANDOCERR_FILE, 0, 0, "gzclose: %s",
517 gzerrnum == Z_ERRNO ? strerror(errno) :
518 zError(gzerrnum));
519 if (retval == 0) {
520 free(fb->buf);
521 fb->buf = NULL;
522 }
523 return retval;
524 }
525
526 static void
527 mparse_end(struct mparse *curp)
528 {
529 if (curp->man->macroset == MACROSET_NONE)
530 curp->man->macroset = MACROSET_MAN;
531 if (curp->man->macroset == MACROSET_MDOC)
532 mdoc_endparse(curp->man);
533 else
534 man_endparse(curp->man);
535 roff_endparse(curp->roff);
536 }
537
538 /*
539 * Read the whole file into memory and call the parsers.
540 * Called recursively when an .so request is encountered.
541 */
542 void
543 mparse_readfd(struct mparse *curp, int fd, const char *filename)
544 {
545 static int recursion_depth;
546
547 struct buf blk;
548 struct buf *save_primary;
549 const char *save_filename;
550 size_t offset;
551 int save_filenc, save_lineno;
552 int with_mmap;
553
554 if (recursion_depth > 64) {
555 mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, 0, NULL);
556 return;
557 }
558 if (read_whole_file(curp, fd, &blk, &with_mmap) == 0)
559 return;
560
561 /*
562 * Save some properties of the parent file.
563 */
564
565 save_primary = curp->primary;
566 save_filenc = curp->filenc;
567 save_lineno = curp->line;
568 save_filename = mandoc_msg_getinfilename();
569
570 curp->primary = &blk;
571 curp->filenc = curp->options & (MPARSE_UTF8 | MPARSE_LATIN1);
572 curp->line = 1;
573 mandoc_msg_setinfilename(filename);
574
575 /* Skip an UTF-8 byte order mark. */
576 if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
577 (unsigned char)blk.buf[0] == 0xef &&
578 (unsigned char)blk.buf[1] == 0xbb &&
579 (unsigned char)blk.buf[2] == 0xbf) {
580 offset = 3;
581 curp->filenc &= ~MPARSE_LATIN1;
582 } else
583 offset = 0;
584
585 recursion_depth++;
586 mparse_buf_r(curp, blk, offset, 1);
587 if (--recursion_depth == 0)
588 mparse_end(curp);
589
590 /*
591 * Clean up and restore saved parent properties.
592 */
593
594 if (with_mmap)
595 munmap(blk.buf, blk.sz);
596 else
597 free(blk.buf);
598
599 curp->primary = save_primary;
600 curp->filenc = save_filenc;
601 curp->line = save_lineno;
602 if (save_filename != NULL)
603 mandoc_msg_setinfilename(save_filename);
604 }
605
606 int
607 mparse_open(struct mparse *curp, const char *file)
608 {
609 char *cp;
610 int fd;
611
612 cp = strrchr(file, '.');
613 curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
614
615 /* First try to use the filename as it is. */
616
617 if ((fd = open(file, O_RDONLY)) != -1)
618 return fd;
619
620 /*
621 * If that doesn't work and the filename doesn't
622 * already end in .gz, try appending .gz.
623 */
624
625 if ( ! curp->gzip) {
626 mandoc_asprintf(&cp, "%s.gz", file);
627 fd = open(cp, O_RDONLY);
628 free(cp);
629 if (fd != -1) {
630 curp->gzip = 1;
631 return fd;
632 }
633 }
634
635 /* Neither worked, give up. */
636
637 return -1;
638 }
639
640 struct mparse *
641 mparse_alloc(int options, enum mandoc_os os_e, const char *os_s)
642 {
643 struct mparse *curp;
644
645 curp = mandoc_calloc(1, sizeof(struct mparse));
646
647 curp->options = options;
648 curp->os_s = os_s;
649
650 curp->roff = roff_alloc(options);
651 curp->man = roff_man_alloc(curp->roff, curp->os_s,
652 curp->options & MPARSE_QUICK ? 1 : 0);
653 if (curp->options & MPARSE_MDOC) {
654 curp->man->macroset = MACROSET_MDOC;
655 if (curp->man->mdocmac == NULL)
656 curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
657 } else if (curp->options & MPARSE_MAN) {
658 curp->man->macroset = MACROSET_MAN;
659 if (curp->man->manmac == NULL)
660 curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
661 }
662 curp->man->first->tok = TOKEN_NONE;
663 curp->man->meta.os_e = os_e;
664 return curp;
665 }
666
667 void
668 mparse_reset(struct mparse *curp)
669 {
670 roff_reset(curp->roff);
671 roff_man_reset(curp->man);
672 free_buf_list(curp->secondary);
673 curp->secondary = NULL;
674 free(curp->sodest);
675 curp->sodest = NULL;
676 curp->gzip = 0;
677 }
678
679 void
680 mparse_free(struct mparse *curp)
681 {
682 roffhash_free(curp->man->mdocmac);
683 roffhash_free(curp->man->manmac);
684 roff_man_free(curp->man);
685 roff_free(curp->roff);
686 free_buf_list(curp->secondary);
687 free(curp->sodest);
688 free(curp);
689 }
690
691 void
692 mparse_result(struct mparse *curp, struct roff_man **man,
693 char **sodest)
694 {
695
696 if (sodest && NULL != (*sodest = curp->sodest)) {
697 *man = NULL;
698 return;
699 }
700 if (man)
701 *man = curp->man;
702 }
703
704 void
705 mparse_copy(const struct mparse *p)
706 {
707 struct buf *buf;
708
709 for (buf = p->secondary; buf != NULL; buf = buf->next)
710 puts(buf->buf);
711 }
A git conversion mirror of mandoc.bsd.lv