Annotation of src/usr.bin/mandoc/read.c, Revision 1.192
1.192 ! schwarze 1: /* $OpenBSD: read.c,v 1.191 2021/06/27 17:57:13 schwarze Exp $ */
1.1 schwarze 2: /*
1.189 schwarze 3: * Copyright (c) 2010-2020 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 4: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.20 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 schwarze 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: *
1.108 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.108 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 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.
1.186 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 schwarze 24: */
1.58 schwarze 25: #include <sys/types.h>
26: #include <sys/mman.h>
1.1 schwarze 27: #include <sys/stat.h>
28:
29: #include <assert.h>
30: #include <ctype.h>
1.18 schwarze 31: #include <errno.h>
1.1 schwarze 32: #include <fcntl.h>
33: #include <stdarg.h>
34: #include <stdio.h>
35: #include <stdlib.h>
36: #include <string.h>
37: #include <unistd.h>
1.115 schwarze 38: #include <zlib.h>
1.1 schwarze 39:
1.108 schwarze 40: #include "mandoc_aux.h"
1.1 schwarze 41: #include "mandoc.h"
1.108 schwarze 42: #include "roff.h"
1.1 schwarze 43: #include "mdoc.h"
44: #include "man.h"
1.174 schwarze 45: #include "mandoc_parse.h"
1.108 schwarze 46: #include "libmandoc.h"
1.173 schwarze 47: #include "roff_int.h"
1.189 schwarze 48: #include "tag.h"
1.1 schwarze 49:
50: #define REPARSE_LIMIT 1000
51:
52: struct mparse {
1.133 schwarze 53: struct roff *roff; /* roff parser (!NULL) */
1.109 schwarze 54: struct roff_man *man; /* man parser */
1.59 schwarze 55: struct buf *primary; /* buffer currently being parsed */
1.170 schwarze 56: struct buf *secondary; /* copy of top level input */
1.171 schwarze 57: struct buf *loop; /* open .while request line */
1.151 schwarze 58: const char *os_s; /* default operating system */
1.59 schwarze 59: int options; /* parser options */
1.115 schwarze 60: int gzip; /* current input file is gzipped */
1.70 schwarze 61: int filenc; /* encoding of the current file */
1.1 schwarze 62: int reparse_count; /* finite interp. stack */
1.59 schwarze 63: int line; /* line number in the file */
1.1 schwarze 64: };
65:
1.60 schwarze 66: static void choose_parser(struct mparse *);
1.170 schwarze 67: static void free_buf_list(struct buf *);
1.1 schwarze 68: static void resize_buf(struct buf *, size_t);
1.171 schwarze 69: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.175 schwarze 70: static int read_whole_file(struct mparse *, int, struct buf *, int *);
1.1 schwarze 71: static void mparse_end(struct mparse *);
72:
1.25 schwarze 73:
1.1 schwarze 74: static void
75: resize_buf(struct buf *buf, size_t initial)
76: {
77:
78: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
79: buf->buf = mandoc_realloc(buf->buf, buf->sz);
80: }
81:
82: static void
1.170 schwarze 83: free_buf_list(struct buf *buf)
84: {
85: struct buf *tmp;
86:
87: while (buf != NULL) {
88: tmp = buf;
89: buf = tmp->next;
90: free(tmp->buf);
91: free(tmp);
92: }
93: }
94:
95: static void
1.60 schwarze 96: choose_parser(struct mparse *curp)
1.1 schwarze 97: {
1.59 schwarze 98: char *cp, *ep;
99: int format;
1.1 schwarze 100:
1.59 schwarze 101: /*
102: * If neither command line arguments -mdoc or -man select
103: * a parser nor the roff parser found a .Dd or .TH macro
104: * yet, look ahead in the main input buffer.
105: */
106:
107: if ((format = roff_getformat(curp->roff)) == 0) {
108: cp = curp->primary->buf;
109: ep = cp + curp->primary->sz;
110: while (cp < ep) {
1.61 schwarze 111: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 112: cp++;
113: if (cp[0] == 'D' && cp[1] == 'd') {
114: format = MPARSE_MDOC;
115: break;
116: }
117: if (cp[0] == 'T' && cp[1] == 'H') {
118: format = MPARSE_MAN;
119: break;
120: }
121: }
122: cp = memchr(cp, '\n', ep - cp);
123: if (cp == NULL)
124: break;
125: cp++;
126: }
1.1 schwarze 127: }
128:
1.59 schwarze 129: if (format == MPARSE_MDOC) {
1.180 schwarze 130: curp->man->meta.macroset = MACROSET_MDOC;
1.136 schwarze 131: if (curp->man->mdocmac == NULL)
132: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 133: } else {
1.180 schwarze 134: curp->man->meta.macroset = MACROSET_MAN;
1.136 schwarze 135: if (curp->man->manmac == NULL)
136: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 137: }
1.180 schwarze 138: curp->man->meta.first->tok = TOKEN_NONE;
1.1 schwarze 139: }
140:
141: /*
1.71 schwarze 142: * Main parse routine for a buffer.
143: * It assumes encoding and line numbering are already set up.
144: * It can recurse directly (for invocations of user-defined
145: * macros, inline equations, and input line traps)
146: * and indirectly (for .so file inclusion).
1.1 schwarze 147: */
1.171 schwarze 148: static int
1.71 schwarze 149: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 150: {
151: struct buf ln;
1.171 schwarze 152: struct buf *firstln, *lastln, *thisln, *loop;
1.85 schwarze 153: char *cp;
1.71 schwarze 154: size_t pos; /* byte number in the ln buffer */
1.191 schwarze 155: size_t spos; /* at the start of the current line parse */
1.171 schwarze 156: int line_result, result;
1.76 schwarze 157: int of;
1.1 schwarze 158: int lnn; /* line number in the real file */
1.85 schwarze 159: int fd;
1.171 schwarze 160: int inloop; /* Saw .while on this level. */
1.1 schwarze 161: unsigned char c;
162:
1.170 schwarze 163: ln.sz = 256;
164: ln.buf = mandoc_malloc(ln.sz);
165: ln.next = NULL;
1.184 schwarze 166: firstln = lastln = loop = NULL;
1.25 schwarze 167: lnn = curp->line;
168: pos = 0;
1.171 schwarze 169: inloop = 0;
1.170 schwarze 170: result = ROFF_CONT;
1.1 schwarze 171:
1.171 schwarze 172: while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) {
1.1 schwarze 173: if (start) {
174: curp->line = lnn;
175: curp->reparse_count = 0;
1.70 schwarze 176:
177: if (lnn < 3 &&
178: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 179: curp->filenc & MPARSE_LATIN1)
180: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 181: }
1.191 schwarze 182: spos = pos;
1.1 schwarze 183:
1.71 schwarze 184: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 185:
186: /*
187: * When finding an unescaped newline character,
188: * leave the character loop to process the line.
189: * Skip a preceding carriage return, if any.
190: */
191:
1.71 schwarze 192: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 193: '\n' == blk.buf[i + 1])
194: ++i;
195: if ('\n' == blk.buf[i]) {
196: ++i;
197: ++lnn;
198: break;
199: }
200:
1.13 schwarze 201: /*
1.70 schwarze 202: * Make sure we have space for the worst
1.170 schwarze 203: * case of 12 bytes: "\\[u10ffff]\n\0"
1.13 schwarze 204: */
205:
1.170 schwarze 206: if (pos + 12 > ln.sz)
1.13 schwarze 207: resize_buf(&ln, 256);
208:
1.25 schwarze 209: /*
1.70 schwarze 210: * Encode 8-bit input.
1.1 schwarze 211: */
212:
1.70 schwarze 213: c = blk.buf[i];
214: if (c & 0x80) {
1.71 schwarze 215: if ( ! (curp->filenc && preconv_encode(
216: &blk, &i, &ln, &pos, &curp->filenc))) {
1.177 schwarze 217: mandoc_msg(MANDOCERR_CHAR_BAD,
1.89 schwarze 218: curp->line, pos, "0x%x", c);
1.70 schwarze 219: ln.buf[pos++] = '?';
220: i++;
221: }
222: continue;
223: }
224:
225: /*
226: * Exclude control characters.
227: */
1.1 schwarze 228:
1.70 schwarze 229: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.177 schwarze 230: mandoc_msg(c == 0x00 || c == 0x04 ||
1.89 schwarze 231: c > 0x0a ? MANDOCERR_CHAR_BAD :
232: MANDOCERR_CHAR_UNSUPP,
1.177 schwarze 233: curp->line, pos, "0x%x", c);
1.1 schwarze 234: i++;
1.102 schwarze 235: if (c != '\r')
236: ln.buf[pos++] = '?';
1.1 schwarze 237: continue;
238: }
239:
240: ln.buf[pos++] = blk.buf[i++];
241: }
1.170 schwarze 242: ln.buf[pos] = '\0';
243:
244: /*
245: * Maintain a lookaside buffer of all lines.
246: * parsed from this input source.
247: */
248:
249: thisln = mandoc_malloc(sizeof(*thisln));
250: thisln->buf = mandoc_strdup(ln.buf);
251: thisln->sz = strlen(ln.buf) + 1;
252: thisln->next = NULL;
253: if (firstln == NULL) {
254: firstln = lastln = thisln;
255: if (curp->secondary == NULL)
256: curp->secondary = firstln;
257: } else {
258: lastln->next = thisln;
259: lastln = thisln;
260: }
1.1 schwarze 261:
1.192 ! schwarze 262: /*
! 263: * XXX Ugly hack to mark the end of the input,
! 264: * such that the function roff_parse_comment()
! 265: * doesn't attempt to append another line if the
! 266: * last input line ends with an escape character.
! 267: */
1.1 schwarze 268:
1.170 schwarze 269: if (i == blk.sz || blk.buf[i] == '\0') {
1.183 schwarze 270: if (pos + 2 > ln.sz)
271: resize_buf(&ln, 256);
1.142 schwarze 272: ln.buf[pos++] = '\n';
1.170 schwarze 273: ln.buf[pos] = '\0';
274: }
1.1 schwarze 275:
276: /*
277: * A significant amount of complexity is contained by
278: * the roff preprocessor. It's line-oriented but can be
279: * expressed on one line, so we need at times to
280: * readjust our starting point and re-run it. The roff
281: * preprocessor can also readjust the buffers with new
282: * data, so we pass them in wholesale.
283: */
284:
285: of = 0;
286: rerun:
1.191 schwarze 287: line_result = roff_parseln(curp->roff, curp->line,
288: &ln, &of, start && spos == 0 ? pos : 0);
1.1 schwarze 289:
1.171 schwarze 290: /* Process options. */
291:
292: if (line_result & ROFF_APPEND)
293: assert(line_result == (ROFF_IGN | ROFF_APPEND));
294:
295: if (line_result & ROFF_USERCALL)
296: assert((line_result & ROFF_MASK) == ROFF_REPARSE);
297:
298: if (line_result & ROFF_USERRET) {
299: assert(line_result == (ROFF_IGN | ROFF_USERRET));
300: if (start == 0) {
301: /* Return from the current macro. */
302: result = ROFF_USERRET;
303: goto out;
304: }
305: }
306:
307: switch (line_result & ROFF_LOOPMASK) {
308: case ROFF_IGN:
309: break;
310: case ROFF_WHILE:
311: if (curp->loop != NULL) {
312: if (loop == curp->loop)
313: break;
314: mandoc_msg(MANDOCERR_WHILE_NEST,
1.177 schwarze 315: curp->line, pos, NULL);
1.171 schwarze 316: }
317: curp->loop = thisln;
318: loop = NULL;
319: inloop = 1;
320: break;
321: case ROFF_LOOPCONT:
322: case ROFF_LOOPEXIT:
323: if (curp->loop == NULL) {
324: mandoc_msg(MANDOCERR_WHILE_FAIL,
1.177 schwarze 325: curp->line, pos, NULL);
1.171 schwarze 326: break;
327: }
328: if (inloop == 0) {
329: mandoc_msg(MANDOCERR_WHILE_INTO,
1.177 schwarze 330: curp->line, pos, NULL);
1.171 schwarze 331: curp->loop = loop = NULL;
332: break;
333: }
334: if (line_result & ROFF_LOOPCONT)
335: loop = curp->loop;
336: else {
337: curp->loop = loop = NULL;
338: inloop = 0;
339: }
340: break;
341: default:
342: abort();
343: }
344:
345: /* Process the main instruction from the roff parser. */
346:
347: switch (line_result & ROFF_MASK) {
348: case ROFF_IGN:
349: break;
350: case ROFF_CONT:
1.180 schwarze 351: if (curp->man->meta.macroset == MACROSET_NONE)
1.171 schwarze 352: choose_parser(curp);
1.180 schwarze 353: if ((curp->man->meta.macroset == MACROSET_MDOC ?
1.171 schwarze 354: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
355: man_parseln(curp->man, curp->line, ln.buf, of)
356: ) == 2)
357: goto out;
358: break;
359: case ROFF_RERUN:
360: goto rerun;
1.25 schwarze 361: case ROFF_REPARSE:
1.169 schwarze 362: if (++curp->reparse_count > REPARSE_LIMIT) {
1.171 schwarze 363: /* Abort and return to the top level. */
1.170 schwarze 364: result = ROFF_IGN;
1.177 schwarze 365: mandoc_msg(MANDOCERR_ROFFLOOP,
1.25 schwarze 366: curp->line, pos, NULL);
1.171 schwarze 367: goto out;
1.169 schwarze 368: }
1.171 schwarze 369: result = mparse_buf_r(curp, ln, of, 0);
370: if (line_result & ROFF_USERCALL) {
371: roff_userret(curp->roff);
372: /* Continue normally. */
373: if (result & ROFF_USERRET)
374: result = ROFF_CONT;
1.135 schwarze 375: }
1.171 schwarze 376: if (start == 0 && result != ROFF_CONT)
377: goto out;
378: break;
1.25 schwarze 379: case ROFF_SO:
1.71 schwarze 380: if ( ! (curp->options & MPARSE_SO) &&
381: (i >= blk.sz || blk.buf[i] == '\0')) {
1.180 schwarze 382: curp->man->meta.sodest =
383: mandoc_strdup(ln.buf + of);
1.170 schwarze 384: goto out;
1.23 schwarze 385: }
1.122 schwarze 386: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 387: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 388: close(fd);
1.88 schwarze 389: } else {
1.179 schwarze 390: mandoc_msg(MANDOCERR_SO_FAIL,
391: curp->line, of, ".so %s: %s",
392: ln.buf + of, strerror(errno));
1.85 schwarze 393: ln.sz = mandoc_asprintf(&cp,
394: ".sp\nSee the file %s.\n.sp",
395: ln.buf + of);
396: free(ln.buf);
397: ln.buf = cp;
398: of = 0;
399: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 400: }
1.171 schwarze 401: break;
1.1 schwarze 402: default:
1.171 schwarze 403: abort();
1.1 schwarze 404: }
405:
1.171 schwarze 406: /* Start the next input line. */
1.1 schwarze 407:
1.171 schwarze 408: if (loop != NULL &&
409: (line_result & ROFF_LOOPMASK) == ROFF_IGN)
410: loop = loop->next;
411:
412: if (loop != NULL) {
413: if ((line_result & ROFF_APPEND) == 0)
414: *ln.buf = '\0';
415: if (ln.sz < loop->sz)
416: resize_buf(&ln, loop->sz);
417: (void)strlcat(ln.buf, loop->buf, ln.sz);
418: of = 0;
419: goto rerun;
420: }
1.1 schwarze 421:
1.171 schwarze 422: pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
1.1 schwarze 423: }
1.170 schwarze 424: out:
1.171 schwarze 425: if (inloop) {
426: if (result != ROFF_USERRET)
1.177 schwarze 427: mandoc_msg(MANDOCERR_WHILE_OUTOF,
1.171 schwarze 428: curp->line, pos, NULL);
429: curp->loop = NULL;
430: }
1.1 schwarze 431: free(ln.buf);
1.170 schwarze 432: if (firstln != curp->secondary)
433: free_buf_list(firstln);
434: return result;
1.1 schwarze 435: }
436:
437: static int
1.175 schwarze 438: read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
1.1 schwarze 439: {
440: struct stat st;
1.115 schwarze 441: gzFile gz;
1.1 schwarze 442: size_t off;
443: ssize_t ssz;
1.166 schwarze 444: int gzerrnum, retval;
1.1 schwarze 445:
1.164 schwarze 446: if (fstat(fd, &st) == -1) {
1.185 schwarze 447: mandoc_msg(MANDOCERR_FSTAT, 0, 0, "%s", strerror(errno));
448: return -1;
1.164 schwarze 449: }
1.1 schwarze 450:
451: /*
452: * If we're a regular file, try just reading in the whole entry
453: * via mmap(). This is faster than reading it into blocks, and
454: * since each file is only a few bytes to begin with, I'm not
455: * concerned that this is going to tank any machines.
456: */
457:
1.115 schwarze 458: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 459: if (st.st_size > 0x7fffffff) {
1.177 schwarze 460: mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
1.185 schwarze 461: return -1;
1.1 schwarze 462: }
463: *with_mmap = 1;
464: fb->sz = (size_t)st.st_size;
1.15 schwarze 465: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 466: if (fb->buf != MAP_FAILED)
1.185 schwarze 467: return 0;
1.1 schwarze 468: }
469:
1.115 schwarze 470: if (curp->gzip) {
1.166 schwarze 471: /*
472: * Duplicating the file descriptor is required
473: * because we will have to call gzclose(3)
474: * to free memory used internally by zlib,
475: * but that will also close the file descriptor,
476: * which this function must not do.
477: */
478: if ((fd = dup(fd)) == -1) {
1.185 schwarze 479: mandoc_msg(MANDOCERR_DUP, 0, 0,
480: "%s", strerror(errno));
481: return -1;
1.166 schwarze 482: }
1.164 schwarze 483: if ((gz = gzdopen(fd, "rb")) == NULL) {
1.185 schwarze 484: mandoc_msg(MANDOCERR_GZDOPEN, 0, 0,
485: "%s", strerror(errno));
1.166 schwarze 486: close(fd);
1.185 schwarze 487: return -1;
1.164 schwarze 488: }
1.115 schwarze 489: } else
490: gz = NULL;
491:
1.1 schwarze 492: /*
493: * If this isn't a regular file (like, say, stdin), then we must
494: * go the old way and just read things in bit by bit.
495: */
496:
497: *with_mmap = 0;
498: off = 0;
1.185 schwarze 499: retval = -1;
1.1 schwarze 500: fb->sz = 0;
501: fb->buf = NULL;
502: for (;;) {
503: if (off == fb->sz) {
504: if (fb->sz == (1U << 31)) {
1.177 schwarze 505: mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
1.1 schwarze 506: break;
507: }
508: resize_buf(fb, 65536);
509: }
1.115 schwarze 510: ssz = curp->gzip ?
511: gzread(gz, fb->buf + (int)off, fb->sz - off) :
512: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 513: if (ssz == 0) {
514: fb->sz = off;
1.185 schwarze 515: retval = 0;
1.166 schwarze 516: break;
1.1 schwarze 517: }
1.164 schwarze 518: if (ssz == -1) {
1.166 schwarze 519: if (curp->gzip)
520: (void)gzerror(gz, &gzerrnum);
1.185 schwarze 521: mandoc_msg(MANDOCERR_READ, 0, 0, "%s",
1.166 schwarze 522: curp->gzip && gzerrnum != Z_ERRNO ?
523: zError(gzerrnum) : strerror(errno));
1.164 schwarze 524: break;
525: }
1.1 schwarze 526: off += (size_t)ssz;
527: }
528:
1.166 schwarze 529: if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
1.185 schwarze 530: mandoc_msg(MANDOCERR_GZCLOSE, 0, 0, "%s",
1.166 schwarze 531: gzerrnum == Z_ERRNO ? strerror(errno) :
532: zError(gzerrnum));
1.185 schwarze 533: if (retval == -1) {
1.166 schwarze 534: free(fb->buf);
535: fb->buf = NULL;
536: }
537: return retval;
1.1 schwarze 538: }
539:
540: static void
541: mparse_end(struct mparse *curp)
542: {
1.180 schwarze 543: if (curp->man->meta.macroset == MACROSET_NONE)
544: curp->man->meta.macroset = MACROSET_MAN;
545: if (curp->man->meta.macroset == MACROSET_MDOC)
1.110 schwarze 546: mdoc_endparse(curp->man);
547: else
1.86 schwarze 548: man_endparse(curp->man);
1.1 schwarze 549: roff_endparse(curp->roff);
550: }
551:
1.176 schwarze 552: /*
553: * Read the whole file into memory and call the parsers.
554: * Called recursively when an .so request is encountered.
555: */
556: void
557: mparse_readfd(struct mparse *curp, int fd, const char *filename)
1.1 schwarze 558: {
1.176 schwarze 559: static int recursion_depth;
560:
561: struct buf blk;
562: struct buf *save_primary;
1.190 schwarze 563: const char *save_filename, *cp;
1.71 schwarze 564: size_t offset;
1.176 schwarze 565: int save_filenc, save_lineno;
566: int with_mmap;
1.14 schwarze 567:
1.176 schwarze 568: if (recursion_depth > 64) {
1.177 schwarze 569: mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, 0, NULL);
1.15 schwarze 570: return;
1.190 schwarze 571: } else if (recursion_depth == 0 &&
572: (cp = strrchr(filename, '.')) != NULL &&
573: cp[1] >= '1' && cp[1] <= '9')
574: curp->man->filesec = cp[1];
575: else
576: curp->man->filesec = '\0';
577:
1.185 schwarze 578: if (read_whole_file(curp, fd, &blk, &with_mmap) == -1)
1.176 schwarze 579: return;
580:
581: /*
582: * Save some properties of the parent file.
583: */
584:
585: save_primary = curp->primary;
586: save_filenc = curp->filenc;
587: save_lineno = curp->line;
588: save_filename = mandoc_msg_getinfilename();
1.1 schwarze 589:
1.59 schwarze 590: curp->primary = &blk;
1.176 schwarze 591: curp->filenc = curp->options & (MPARSE_UTF8 | MPARSE_LATIN1);
1.15 schwarze 592: curp->line = 1;
1.176 schwarze 593: mandoc_msg_setinfilename(filename);
1.15 schwarze 594:
1.70 schwarze 595: /* Skip an UTF-8 byte order mark. */
596: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
597: (unsigned char)blk.buf[0] == 0xef &&
598: (unsigned char)blk.buf[1] == 0xbb &&
599: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 600: offset = 3;
1.70 schwarze 601: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 602: } else
603: offset = 0;
1.70 schwarze 604:
1.176 schwarze 605: recursion_depth++;
1.71 schwarze 606: mparse_buf_r(curp, blk, offset, 1);
1.86 schwarze 607: if (--recursion_depth == 0)
1.15 schwarze 608: mparse_end(curp);
609:
1.176 schwarze 610: /*
611: * Clean up and restore saved parent properties.
612: */
1.15 schwarze 613:
1.176 schwarze 614: if (with_mmap)
615: munmap(blk.buf, blk.sz);
616: else
617: free(blk.buf);
1.1 schwarze 618:
1.176 schwarze 619: curp->primary = save_primary;
620: curp->filenc = save_filenc;
621: curp->line = save_lineno;
622: if (save_filename != NULL)
623: mandoc_msg_setinfilename(save_filename);
1.58 schwarze 624: }
625:
1.122 schwarze 626: int
627: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 628: {
629: char *cp;
1.182 schwarze 630: int fd, save_errno;
1.58 schwarze 631:
1.115 schwarze 632: cp = strrchr(file, '.');
633: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 634:
1.115 schwarze 635: /* First try to use the filename as it is. */
1.74 schwarze 636:
1.122 schwarze 637: if ((fd = open(file, O_RDONLY)) != -1)
638: return fd;
1.74 schwarze 639:
1.115 schwarze 640: /*
641: * If that doesn't work and the filename doesn't
642: * already end in .gz, try appending .gz.
643: */
1.74 schwarze 644:
1.115 schwarze 645: if ( ! curp->gzip) {
1.182 schwarze 646: save_errno = errno;
1.74 schwarze 647: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 648: fd = open(cp, O_RDONLY);
1.83 schwarze 649: free(cp);
1.182 schwarze 650: errno = save_errno;
1.122 schwarze 651: if (fd != -1) {
1.115 schwarze 652: curp->gzip = 1;
1.122 schwarze 653: return fd;
1.58 schwarze 654: }
655: }
656:
1.115 schwarze 657: /* Neither worked, give up. */
1.73 schwarze 658:
1.122 schwarze 659: return -1;
1.1 schwarze 660: }
661:
662: struct mparse *
1.175 schwarze 663: mparse_alloc(int options, enum mandoc_os os_e, const char *os_s)
1.1 schwarze 664: {
665: struct mparse *curp;
666:
667: curp = mandoc_calloc(1, sizeof(struct mparse));
668:
1.22 schwarze 669: curp->options = options;
1.151 schwarze 670: curp->os_s = os_s;
1.1 schwarze 671:
1.178 schwarze 672: curp->roff = roff_alloc(options);
673: curp->man = roff_man_alloc(curp->roff, curp->os_s,
1.112 schwarze 674: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 675: if (curp->options & MPARSE_MDOC) {
1.180 schwarze 676: curp->man->meta.macroset = MACROSET_MDOC;
1.136 schwarze 677: if (curp->man->mdocmac == NULL)
678: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 679: } else if (curp->options & MPARSE_MAN) {
1.180 schwarze 680: curp->man->meta.macroset = MACROSET_MAN;
1.136 schwarze 681: if (curp->man->manmac == NULL)
682: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 683: }
1.180 schwarze 684: curp->man->meta.first->tok = TOKEN_NONE;
1.151 schwarze 685: curp->man->meta.os_e = os_e;
1.186 schwarze 686: tag_alloc();
1.117 schwarze 687: return curp;
1.1 schwarze 688: }
689:
690: void
691: mparse_reset(struct mparse *curp)
692: {
1.187 schwarze 693: tag_free();
1.1 schwarze 694: roff_reset(curp->roff);
1.124 schwarze 695: roff_man_reset(curp->man);
1.170 schwarze 696: free_buf_list(curp->secondary);
697: curp->secondary = NULL;
1.132 schwarze 698: curp->gzip = 0;
1.187 schwarze 699: tag_alloc();
1.1 schwarze 700: }
701:
702: void
703: mparse_free(struct mparse *curp)
704: {
1.186 schwarze 705: tag_free();
1.136 schwarze 706: roffhash_free(curp->man->mdocmac);
707: roffhash_free(curp->man->manmac);
1.112 schwarze 708: roff_man_free(curp->man);
1.133 schwarze 709: roff_free(curp->roff);
1.170 schwarze 710: free_buf_list(curp->secondary);
1.1 schwarze 711: free(curp);
712: }
713:
1.180 schwarze 714: struct roff_meta *
715: mparse_result(struct mparse *curp)
1.1 schwarze 716: {
1.181 schwarze 717: roff_state_reset(curp->man);
1.180 schwarze 718: if (curp->options & MPARSE_VALIDATE) {
719: if (curp->man->meta.macroset == MACROSET_MDOC)
720: mdoc_validate(curp->man);
721: else
722: man_validate(curp->man);
1.189 schwarze 723: tag_postprocess(curp->man, curp->man->meta.first);
1.23 schwarze 724: }
1.180 schwarze 725: return &curp->man->meta;
1.4 schwarze 726: }
727:
728: void
1.170 schwarze 729: mparse_copy(const struct mparse *p)
1.4 schwarze 730: {
1.170 schwarze 731: struct buf *buf;
1.4 schwarze 732:
1.170 schwarze 733: for (buf = p->secondary; buf != NULL; buf = buf->next)
734: puts(buf->buf);
1.1 schwarze 735: }
![[BACK]](/icons/back.gif){kind=icon}
Return to read.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc