Annotation of src/usr.bin/mandoc/read.c, Revision 1.178
1.178 ! schwarze 1: /* $OpenBSD: read.c,v 1.177 2018/12/14 05:17:45 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.168 schwarze 4: * Copyright (c) 2010-2018 Ingo Schwarze <schwarze@openbsd.org>
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.
18: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
22:
23: #include <assert.h>
24: #include <ctype.h>
1.18 schwarze 25: #include <errno.h>
1.1 schwarze 26: #include <fcntl.h>
27: #include <stdarg.h>
28: #include <stdio.h>
29: #include <stdlib.h>
30: #include <string.h>
31: #include <unistd.h>
1.115 schwarze 32: #include <zlib.h>
1.1 schwarze 33:
1.108 schwarze 34: #include "mandoc_aux.h"
1.1 schwarze 35: #include "mandoc.h"
1.108 schwarze 36: #include "roff.h"
1.1 schwarze 37: #include "mdoc.h"
38: #include "man.h"
1.174 schwarze 39: #include "mandoc_parse.h"
1.108 schwarze 40: #include "libmandoc.h"
1.173 schwarze 41: #include "roff_int.h"
1.1 schwarze 42:
43: #define REPARSE_LIMIT 1000
44:
45: struct mparse {
1.133 schwarze 46: struct roff *roff; /* roff parser (!NULL) */
1.109 schwarze 47: struct roff_man *man; /* man parser */
1.23 schwarze 48: char *sodest; /* filename pointed to by .so */
1.59 schwarze 49: struct buf *primary; /* buffer currently being parsed */
1.170 schwarze 50: struct buf *secondary; /* copy of top level input */
1.171 schwarze 51: struct buf *loop; /* open .while request line */
1.151 schwarze 52: const char *os_s; /* default operating system */
1.59 schwarze 53: int options; /* parser options */
1.115 schwarze 54: int gzip; /* current input file is gzipped */
1.70 schwarze 55: int filenc; /* encoding of the current file */
1.1 schwarze 56: int reparse_count; /* finite interp. stack */
1.59 schwarze 57: int line; /* line number in the file */
1.1 schwarze 58: };
59:
1.60 schwarze 60: static void choose_parser(struct mparse *);
1.170 schwarze 61: static void free_buf_list(struct buf *);
1.1 schwarze 62: static void resize_buf(struct buf *, size_t);
1.171 schwarze 63: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.175 schwarze 64: static int read_whole_file(struct mparse *, int, struct buf *, int *);
1.1 schwarze 65: static void mparse_end(struct mparse *);
66:
1.25 schwarze 67:
1.1 schwarze 68: static void
69: resize_buf(struct buf *buf, size_t initial)
70: {
71:
72: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
73: buf->buf = mandoc_realloc(buf->buf, buf->sz);
74: }
75:
76: static void
1.170 schwarze 77: free_buf_list(struct buf *buf)
78: {
79: struct buf *tmp;
80:
81: while (buf != NULL) {
82: tmp = buf;
83: buf = tmp->next;
84: free(tmp->buf);
85: free(tmp);
86: }
87: }
88:
89: static void
1.60 schwarze 90: choose_parser(struct mparse *curp)
1.1 schwarze 91: {
1.59 schwarze 92: char *cp, *ep;
93: int format;
1.1 schwarze 94:
1.59 schwarze 95: /*
96: * If neither command line arguments -mdoc or -man select
97: * a parser nor the roff parser found a .Dd or .TH macro
98: * yet, look ahead in the main input buffer.
99: */
100:
101: if ((format = roff_getformat(curp->roff)) == 0) {
102: cp = curp->primary->buf;
103: ep = cp + curp->primary->sz;
104: while (cp < ep) {
1.61 schwarze 105: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 106: cp++;
107: if (cp[0] == 'D' && cp[1] == 'd') {
108: format = MPARSE_MDOC;
109: break;
110: }
111: if (cp[0] == 'T' && cp[1] == 'H') {
112: format = MPARSE_MAN;
113: break;
114: }
115: }
116: cp = memchr(cp, '\n', ep - cp);
117: if (cp == NULL)
118: break;
119: cp++;
120: }
1.1 schwarze 121: }
122:
1.59 schwarze 123: if (format == MPARSE_MDOC) {
1.112 schwarze 124: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 125: if (curp->man->mdocmac == NULL)
126: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 127: } else {
128: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 129: if (curp->man->manmac == NULL)
130: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 131: }
1.136 schwarze 132: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 133: }
134:
135: /*
1.71 schwarze 136: * Main parse routine for a buffer.
137: * It assumes encoding and line numbering are already set up.
138: * It can recurse directly (for invocations of user-defined
139: * macros, inline equations, and input line traps)
140: * and indirectly (for .so file inclusion).
1.1 schwarze 141: */
1.171 schwarze 142: static int
1.71 schwarze 143: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 144: {
145: struct buf ln;
1.171 schwarze 146: struct buf *firstln, *lastln, *thisln, *loop;
1.85 schwarze 147: char *cp;
1.71 schwarze 148: size_t pos; /* byte number in the ln buffer */
1.171 schwarze 149: int line_result, result;
1.76 schwarze 150: int of;
1.1 schwarze 151: int lnn; /* line number in the real file */
1.85 schwarze 152: int fd;
1.171 schwarze 153: int inloop; /* Saw .while on this level. */
1.1 schwarze 154: unsigned char c;
155:
1.170 schwarze 156: ln.sz = 256;
157: ln.buf = mandoc_malloc(ln.sz);
158: ln.next = NULL;
1.171 schwarze 159: firstln = loop = NULL;
1.25 schwarze 160: lnn = curp->line;
161: pos = 0;
1.171 schwarze 162: inloop = 0;
1.170 schwarze 163: result = ROFF_CONT;
1.1 schwarze 164:
1.171 schwarze 165: while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) {
1.1 schwarze 166: if (start) {
167: curp->line = lnn;
168: curp->reparse_count = 0;
1.70 schwarze 169:
170: if (lnn < 3 &&
171: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 172: curp->filenc & MPARSE_LATIN1)
173: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 174: }
175:
1.71 schwarze 176: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 177:
178: /*
179: * When finding an unescaped newline character,
180: * leave the character loop to process the line.
181: * Skip a preceding carriage return, if any.
182: */
183:
1.71 schwarze 184: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 185: '\n' == blk.buf[i + 1])
186: ++i;
187: if ('\n' == blk.buf[i]) {
188: ++i;
189: ++lnn;
190: break;
191: }
192:
1.13 schwarze 193: /*
1.70 schwarze 194: * Make sure we have space for the worst
1.170 schwarze 195: * case of 12 bytes: "\\[u10ffff]\n\0"
1.13 schwarze 196: */
197:
1.170 schwarze 198: if (pos + 12 > ln.sz)
1.13 schwarze 199: resize_buf(&ln, 256);
200:
1.25 schwarze 201: /*
1.70 schwarze 202: * Encode 8-bit input.
1.1 schwarze 203: */
204:
1.70 schwarze 205: c = blk.buf[i];
206: if (c & 0x80) {
1.71 schwarze 207: if ( ! (curp->filenc && preconv_encode(
208: &blk, &i, &ln, &pos, &curp->filenc))) {
1.177 schwarze 209: mandoc_msg(MANDOCERR_CHAR_BAD,
1.89 schwarze 210: curp->line, pos, "0x%x", c);
1.70 schwarze 211: ln.buf[pos++] = '?';
212: i++;
213: }
214: continue;
215: }
216:
217: /*
218: * Exclude control characters.
219: */
1.1 schwarze 220:
1.70 schwarze 221: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.177 schwarze 222: mandoc_msg(c == 0x00 || c == 0x04 ||
1.89 schwarze 223: c > 0x0a ? MANDOCERR_CHAR_BAD :
224: MANDOCERR_CHAR_UNSUPP,
1.177 schwarze 225: curp->line, pos, "0x%x", c);
1.1 schwarze 226: i++;
1.102 schwarze 227: if (c != '\r')
228: ln.buf[pos++] = '?';
1.1 schwarze 229: continue;
230: }
231:
232: ln.buf[pos++] = blk.buf[i++];
233: }
1.170 schwarze 234: ln.buf[pos] = '\0';
235:
236: /*
237: * Maintain a lookaside buffer of all lines.
238: * parsed from this input source.
239: */
240:
241: thisln = mandoc_malloc(sizeof(*thisln));
242: thisln->buf = mandoc_strdup(ln.buf);
243: thisln->sz = strlen(ln.buf) + 1;
244: thisln->next = NULL;
245: if (firstln == NULL) {
246: firstln = lastln = thisln;
247: if (curp->secondary == NULL)
248: curp->secondary = firstln;
249: } else {
250: lastln->next = thisln;
251: lastln = thisln;
252: }
1.1 schwarze 253:
1.170 schwarze 254: /* XXX Ugly hack to mark the end of the input. */
1.1 schwarze 255:
1.170 schwarze 256: if (i == blk.sz || blk.buf[i] == '\0') {
1.142 schwarze 257: ln.buf[pos++] = '\n';
1.170 schwarze 258: ln.buf[pos] = '\0';
259: }
1.1 schwarze 260:
261: /*
262: * A significant amount of complexity is contained by
263: * the roff preprocessor. It's line-oriented but can be
264: * expressed on one line, so we need at times to
265: * readjust our starting point and re-run it. The roff
266: * preprocessor can also readjust the buffers with new
267: * data, so we pass them in wholesale.
268: */
269:
270: of = 0;
271: rerun:
1.169 schwarze 272: line_result = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 273:
1.171 schwarze 274: /* Process options. */
275:
276: if (line_result & ROFF_APPEND)
277: assert(line_result == (ROFF_IGN | ROFF_APPEND));
278:
279: if (line_result & ROFF_USERCALL)
280: assert((line_result & ROFF_MASK) == ROFF_REPARSE);
281:
282: if (line_result & ROFF_USERRET) {
283: assert(line_result == (ROFF_IGN | ROFF_USERRET));
284: if (start == 0) {
285: /* Return from the current macro. */
286: result = ROFF_USERRET;
287: goto out;
288: }
289: }
290:
291: switch (line_result & ROFF_LOOPMASK) {
292: case ROFF_IGN:
293: break;
294: case ROFF_WHILE:
295: if (curp->loop != NULL) {
296: if (loop == curp->loop)
297: break;
298: mandoc_msg(MANDOCERR_WHILE_NEST,
1.177 schwarze 299: curp->line, pos, NULL);
1.171 schwarze 300: }
301: curp->loop = thisln;
302: loop = NULL;
303: inloop = 1;
304: break;
305: case ROFF_LOOPCONT:
306: case ROFF_LOOPEXIT:
307: if (curp->loop == NULL) {
308: mandoc_msg(MANDOCERR_WHILE_FAIL,
1.177 schwarze 309: curp->line, pos, NULL);
1.171 schwarze 310: break;
311: }
312: if (inloop == 0) {
313: mandoc_msg(MANDOCERR_WHILE_INTO,
1.177 schwarze 314: curp->line, pos, NULL);
1.171 schwarze 315: curp->loop = loop = NULL;
316: break;
317: }
318: if (line_result & ROFF_LOOPCONT)
319: loop = curp->loop;
320: else {
321: curp->loop = loop = NULL;
322: inloop = 0;
323: }
324: break;
325: default:
326: abort();
327: }
328:
329: /* Process the main instruction from the roff parser. */
330:
331: switch (line_result & ROFF_MASK) {
332: case ROFF_IGN:
333: break;
334: case ROFF_CONT:
335: if (curp->man->macroset == MACROSET_NONE)
336: choose_parser(curp);
337: if ((curp->man->macroset == MACROSET_MDOC ?
338: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
339: man_parseln(curp->man, curp->line, ln.buf, of)
340: ) == 2)
341: goto out;
342: break;
343: case ROFF_RERUN:
344: goto rerun;
1.25 schwarze 345: case ROFF_REPARSE:
1.169 schwarze 346: if (++curp->reparse_count > REPARSE_LIMIT) {
1.171 schwarze 347: /* Abort and return to the top level. */
1.170 schwarze 348: result = ROFF_IGN;
1.177 schwarze 349: mandoc_msg(MANDOCERR_ROFFLOOP,
1.25 schwarze 350: curp->line, pos, NULL);
1.171 schwarze 351: goto out;
1.169 schwarze 352: }
1.171 schwarze 353: result = mparse_buf_r(curp, ln, of, 0);
354: if (line_result & ROFF_USERCALL) {
355: roff_userret(curp->roff);
356: /* Continue normally. */
357: if (result & ROFF_USERRET)
358: result = ROFF_CONT;
1.135 schwarze 359: }
1.171 schwarze 360: if (start == 0 && result != ROFF_CONT)
361: goto out;
362: break;
1.25 schwarze 363: case ROFF_SO:
1.71 schwarze 364: if ( ! (curp->options & MPARSE_SO) &&
365: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 366: curp->sodest = mandoc_strdup(ln.buf + of);
1.170 schwarze 367: goto out;
1.23 schwarze 368: }
1.122 schwarze 369: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 370: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 371: close(fd);
1.88 schwarze 372: } else {
1.177 schwarze 373: mandoc_msg(MANDOCERR_SO_FAIL, curp->line,
374: pos, ".so %s", ln.buf + of);
1.85 schwarze 375: ln.sz = mandoc_asprintf(&cp,
376: ".sp\nSee the file %s.\n.sp",
377: ln.buf + of);
378: free(ln.buf);
379: ln.buf = cp;
380: of = 0;
381: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 382: }
1.171 schwarze 383: break;
1.1 schwarze 384: default:
1.171 schwarze 385: abort();
1.1 schwarze 386: }
387:
1.171 schwarze 388: /* Start the next input line. */
1.1 schwarze 389:
1.171 schwarze 390: if (loop != NULL &&
391: (line_result & ROFF_LOOPMASK) == ROFF_IGN)
392: loop = loop->next;
393:
394: if (loop != NULL) {
395: if ((line_result & ROFF_APPEND) == 0)
396: *ln.buf = '\0';
397: if (ln.sz < loop->sz)
398: resize_buf(&ln, loop->sz);
399: (void)strlcat(ln.buf, loop->buf, ln.sz);
400: of = 0;
401: goto rerun;
402: }
1.1 schwarze 403:
1.171 schwarze 404: pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
1.1 schwarze 405: }
1.170 schwarze 406: out:
1.171 schwarze 407: if (inloop) {
408: if (result != ROFF_USERRET)
1.177 schwarze 409: mandoc_msg(MANDOCERR_WHILE_OUTOF,
1.171 schwarze 410: curp->line, pos, NULL);
411: curp->loop = NULL;
412: }
1.1 schwarze 413: free(ln.buf);
1.170 schwarze 414: if (firstln != curp->secondary)
415: free_buf_list(firstln);
416: return result;
1.1 schwarze 417: }
418:
419: static int
1.175 schwarze 420: read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
1.1 schwarze 421: {
422: struct stat st;
1.115 schwarze 423: gzFile gz;
1.1 schwarze 424: size_t off;
425: ssize_t ssz;
1.166 schwarze 426: int gzerrnum, retval;
1.1 schwarze 427:
1.164 schwarze 428: if (fstat(fd, &st) == -1) {
1.177 schwarze 429: mandoc_msg(MANDOCERR_FILE, 0, 0,
1.164 schwarze 430: "fstat: %s", strerror(errno));
431: return 0;
432: }
1.1 schwarze 433:
434: /*
435: * If we're a regular file, try just reading in the whole entry
436: * via mmap(). This is faster than reading it into blocks, and
437: * since each file is only a few bytes to begin with, I'm not
438: * concerned that this is going to tank any machines.
439: */
440:
1.115 schwarze 441: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 442: if (st.st_size > 0x7fffffff) {
1.177 schwarze 443: mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
1.117 schwarze 444: return 0;
1.1 schwarze 445: }
446: *with_mmap = 1;
447: fb->sz = (size_t)st.st_size;
1.15 schwarze 448: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 449: if (fb->buf != MAP_FAILED)
1.117 schwarze 450: return 1;
1.1 schwarze 451: }
452:
1.115 schwarze 453: if (curp->gzip) {
1.166 schwarze 454: /*
455: * Duplicating the file descriptor is required
456: * because we will have to call gzclose(3)
457: * to free memory used internally by zlib,
458: * but that will also close the file descriptor,
459: * which this function must not do.
460: */
461: if ((fd = dup(fd)) == -1) {
1.177 schwarze 462: mandoc_msg(MANDOCERR_FILE, 0, 0,
1.166 schwarze 463: "dup: %s", strerror(errno));
464: return 0;
465: }
1.164 schwarze 466: if ((gz = gzdopen(fd, "rb")) == NULL) {
1.177 schwarze 467: mandoc_msg(MANDOCERR_FILE, 0, 0,
1.164 schwarze 468: "gzdopen: %s", strerror(errno));
1.166 schwarze 469: close(fd);
1.164 schwarze 470: return 0;
471: }
1.115 schwarze 472: } else
473: gz = NULL;
474:
1.1 schwarze 475: /*
476: * If this isn't a regular file (like, say, stdin), then we must
477: * go the old way and just read things in bit by bit.
478: */
479:
480: *with_mmap = 0;
481: off = 0;
1.166 schwarze 482: retval = 0;
1.1 schwarze 483: fb->sz = 0;
484: fb->buf = NULL;
485: for (;;) {
486: if (off == fb->sz) {
487: if (fb->sz == (1U << 31)) {
1.177 schwarze 488: mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
1.1 schwarze 489: break;
490: }
491: resize_buf(fb, 65536);
492: }
1.115 schwarze 493: ssz = curp->gzip ?
494: gzread(gz, fb->buf + (int)off, fb->sz - off) :
495: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 496: if (ssz == 0) {
497: fb->sz = off;
1.166 schwarze 498: retval = 1;
499: break;
1.1 schwarze 500: }
1.164 schwarze 501: if (ssz == -1) {
1.166 schwarze 502: if (curp->gzip)
503: (void)gzerror(gz, &gzerrnum);
1.177 schwarze 504: mandoc_msg(MANDOCERR_FILE, 0, 0, "read: %s",
1.166 schwarze 505: curp->gzip && gzerrnum != Z_ERRNO ?
506: zError(gzerrnum) : strerror(errno));
1.164 schwarze 507: break;
508: }
1.1 schwarze 509: off += (size_t)ssz;
510: }
511:
1.166 schwarze 512: if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
1.177 schwarze 513: mandoc_msg(MANDOCERR_FILE, 0, 0, "gzclose: %s",
1.166 schwarze 514: gzerrnum == Z_ERRNO ? strerror(errno) :
515: zError(gzerrnum));
516: if (retval == 0) {
517: free(fb->buf);
518: fb->buf = NULL;
519: }
520: return retval;
1.1 schwarze 521: }
522:
523: static void
524: mparse_end(struct mparse *curp)
525: {
1.110 schwarze 526: if (curp->man->macroset == MACROSET_NONE)
527: curp->man->macroset = MACROSET_MAN;
528: if (curp->man->macroset == MACROSET_MDOC)
529: mdoc_endparse(curp->man);
530: else
1.86 schwarze 531: man_endparse(curp->man);
1.1 schwarze 532: roff_endparse(curp->roff);
533: }
534:
1.176 schwarze 535: /*
536: * Read the whole file into memory and call the parsers.
537: * Called recursively when an .so request is encountered.
538: */
539: void
540: mparse_readfd(struct mparse *curp, int fd, const char *filename)
1.1 schwarze 541: {
1.176 schwarze 542: static int recursion_depth;
543:
544: struct buf blk;
545: struct buf *save_primary;
546: const char *save_filename;
1.71 schwarze 547: size_t offset;
1.176 schwarze 548: int save_filenc, save_lineno;
549: int with_mmap;
1.14 schwarze 550:
1.176 schwarze 551: if (recursion_depth > 64) {
1.177 schwarze 552: mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, 0, NULL);
1.15 schwarze 553: return;
1.14 schwarze 554: }
1.176 schwarze 555: if (read_whole_file(curp, fd, &blk, &with_mmap) == 0)
556: return;
557:
558: /*
559: * Save some properties of the parent file.
560: */
561:
562: save_primary = curp->primary;
563: save_filenc = curp->filenc;
564: save_lineno = curp->line;
565: save_filename = mandoc_msg_getinfilename();
1.1 schwarze 566:
1.59 schwarze 567: curp->primary = &blk;
1.176 schwarze 568: curp->filenc = curp->options & (MPARSE_UTF8 | MPARSE_LATIN1);
1.15 schwarze 569: curp->line = 1;
1.176 schwarze 570: mandoc_msg_setinfilename(filename);
1.15 schwarze 571:
1.70 schwarze 572: /* Skip an UTF-8 byte order mark. */
573: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
574: (unsigned char)blk.buf[0] == 0xef &&
575: (unsigned char)blk.buf[1] == 0xbb &&
576: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 577: offset = 3;
1.70 schwarze 578: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 579: } else
580: offset = 0;
1.70 schwarze 581:
1.176 schwarze 582: recursion_depth++;
1.71 schwarze 583: mparse_buf_r(curp, blk, offset, 1);
1.86 schwarze 584: if (--recursion_depth == 0)
1.15 schwarze 585: mparse_end(curp);
586:
1.176 schwarze 587: /*
588: * Clean up and restore saved parent properties.
589: */
1.15 schwarze 590:
1.176 schwarze 591: if (with_mmap)
592: munmap(blk.buf, blk.sz);
593: else
594: free(blk.buf);
1.1 schwarze 595:
1.176 schwarze 596: curp->primary = save_primary;
597: curp->filenc = save_filenc;
598: curp->line = save_lineno;
599: if (save_filename != NULL)
600: mandoc_msg_setinfilename(save_filename);
1.58 schwarze 601: }
602:
1.122 schwarze 603: int
604: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 605: {
606: char *cp;
1.122 schwarze 607: int fd;
1.58 schwarze 608:
1.115 schwarze 609: cp = strrchr(file, '.');
610: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 611:
1.115 schwarze 612: /* First try to use the filename as it is. */
1.74 schwarze 613:
1.122 schwarze 614: if ((fd = open(file, O_RDONLY)) != -1)
615: return fd;
1.74 schwarze 616:
1.115 schwarze 617: /*
618: * If that doesn't work and the filename doesn't
619: * already end in .gz, try appending .gz.
620: */
1.74 schwarze 621:
1.115 schwarze 622: if ( ! curp->gzip) {
1.74 schwarze 623: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 624: fd = open(cp, O_RDONLY);
1.83 schwarze 625: free(cp);
1.122 schwarze 626: if (fd != -1) {
1.115 schwarze 627: curp->gzip = 1;
1.122 schwarze 628: return fd;
1.58 schwarze 629: }
630: }
631:
1.115 schwarze 632: /* Neither worked, give up. */
1.73 schwarze 633:
1.177 schwarze 634: mandoc_msg(MANDOCERR_FILE, 0, 0, "%s", strerror(errno));
1.122 schwarze 635: return -1;
1.1 schwarze 636: }
637:
638: struct mparse *
1.175 schwarze 639: mparse_alloc(int options, enum mandoc_os os_e, const char *os_s)
1.1 schwarze 640: {
641: struct mparse *curp;
642:
643: curp = mandoc_calloc(1, sizeof(struct mparse));
644:
1.22 schwarze 645: curp->options = options;
1.151 schwarze 646: curp->os_s = os_s;
1.1 schwarze 647:
1.178 ! schwarze 648: curp->roff = roff_alloc(options);
! 649: curp->man = roff_man_alloc(curp->roff, curp->os_s,
1.112 schwarze 650: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 651: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 652: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 653: if (curp->man->mdocmac == NULL)
654: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 655: } else if (curp->options & MPARSE_MAN) {
656: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 657: if (curp->man->manmac == NULL)
658: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 659: }
1.113 schwarze 660: curp->man->first->tok = TOKEN_NONE;
1.151 schwarze 661: curp->man->meta.os_e = os_e;
1.117 schwarze 662: return curp;
1.1 schwarze 663: }
664:
665: void
666: mparse_reset(struct mparse *curp)
667: {
668: roff_reset(curp->roff);
1.124 schwarze 669: roff_man_reset(curp->man);
1.170 schwarze 670: free_buf_list(curp->secondary);
671: curp->secondary = NULL;
1.133 schwarze 672: free(curp->sodest);
673: curp->sodest = NULL;
1.132 schwarze 674: curp->gzip = 0;
1.1 schwarze 675: }
676:
677: void
678: mparse_free(struct mparse *curp)
679: {
1.136 schwarze 680: roffhash_free(curp->man->mdocmac);
681: roffhash_free(curp->man->manmac);
1.112 schwarze 682: roff_man_free(curp->man);
1.133 schwarze 683: roff_free(curp->roff);
1.170 schwarze 684: free_buf_list(curp->secondary);
1.23 schwarze 685: free(curp->sodest);
1.1 schwarze 686: free(curp);
687: }
688:
689: void
1.110 schwarze 690: mparse_result(struct mparse *curp, struct roff_man **man,
691: char **sodest)
1.1 schwarze 692: {
693:
1.23 schwarze 694: if (sodest && NULL != (*sodest = curp->sodest)) {
695: *man = NULL;
696: return;
697: }
1.1 schwarze 698: if (man)
699: *man = curp->man;
1.4 schwarze 700: }
701:
702: void
1.170 schwarze 703: mparse_copy(const struct mparse *p)
1.4 schwarze 704: {
1.170 schwarze 705: struct buf *buf;
1.4 schwarze 706:
1.170 schwarze 707: for (buf = p->secondary; buf != NULL; buf = buf->next)
708: puts(buf->buf);
1.1 schwarze 709: }
![[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