Annotation of src/usr.bin/mandoc/read.c, Revision 1.185
1.185 ! schwarze 1: /* $OpenBSD: read.c,v 1.184 2019/06/03 19:58:00 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.182 schwarze 4: * Copyright (c) 2010-2019 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 */
1.59 schwarze 48: struct buf *primary; /* buffer currently being parsed */
1.170 schwarze 49: struct buf *secondary; /* copy of top level input */
1.171 schwarze 50: struct buf *loop; /* open .while request line */
1.151 schwarze 51: const char *os_s; /* default operating system */
1.59 schwarze 52: int options; /* parser options */
1.115 schwarze 53: int gzip; /* current input file is gzipped */
1.70 schwarze 54: int filenc; /* encoding of the current file */
1.1 schwarze 55: int reparse_count; /* finite interp. stack */
1.59 schwarze 56: int line; /* line number in the file */
1.1 schwarze 57: };
58:
1.60 schwarze 59: static void choose_parser(struct mparse *);
1.170 schwarze 60: static void free_buf_list(struct buf *);
1.1 schwarze 61: static void resize_buf(struct buf *, size_t);
1.171 schwarze 62: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.175 schwarze 63: static int read_whole_file(struct mparse *, int, struct buf *, int *);
1.1 schwarze 64: static void mparse_end(struct mparse *);
65:
1.25 schwarze 66:
1.1 schwarze 67: static void
68: resize_buf(struct buf *buf, size_t initial)
69: {
70:
71: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
72: buf->buf = mandoc_realloc(buf->buf, buf->sz);
73: }
74:
75: static void
1.170 schwarze 76: free_buf_list(struct buf *buf)
77: {
78: struct buf *tmp;
79:
80: while (buf != NULL) {
81: tmp = buf;
82: buf = tmp->next;
83: free(tmp->buf);
84: free(tmp);
85: }
86: }
87:
88: static void
1.60 schwarze 89: choose_parser(struct mparse *curp)
1.1 schwarze 90: {
1.59 schwarze 91: char *cp, *ep;
92: int format;
1.1 schwarze 93:
1.59 schwarze 94: /*
95: * If neither command line arguments -mdoc or -man select
96: * a parser nor the roff parser found a .Dd or .TH macro
97: * yet, look ahead in the main input buffer.
98: */
99:
100: if ((format = roff_getformat(curp->roff)) == 0) {
101: cp = curp->primary->buf;
102: ep = cp + curp->primary->sz;
103: while (cp < ep) {
1.61 schwarze 104: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 105: cp++;
106: if (cp[0] == 'D' && cp[1] == 'd') {
107: format = MPARSE_MDOC;
108: break;
109: }
110: if (cp[0] == 'T' && cp[1] == 'H') {
111: format = MPARSE_MAN;
112: break;
113: }
114: }
115: cp = memchr(cp, '\n', ep - cp);
116: if (cp == NULL)
117: break;
118: cp++;
119: }
1.1 schwarze 120: }
121:
1.59 schwarze 122: if (format == MPARSE_MDOC) {
1.180 schwarze 123: curp->man->meta.macroset = MACROSET_MDOC;
1.136 schwarze 124: if (curp->man->mdocmac == NULL)
125: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 126: } else {
1.180 schwarze 127: curp->man->meta.macroset = MACROSET_MAN;
1.136 schwarze 128: if (curp->man->manmac == NULL)
129: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 130: }
1.180 schwarze 131: curp->man->meta.first->tok = TOKEN_NONE;
1.1 schwarze 132: }
133:
134: /*
1.71 schwarze 135: * Main parse routine for a buffer.
136: * It assumes encoding and line numbering are already set up.
137: * It can recurse directly (for invocations of user-defined
138: * macros, inline equations, and input line traps)
139: * and indirectly (for .so file inclusion).
1.1 schwarze 140: */
1.171 schwarze 141: static int
1.71 schwarze 142: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 143: {
144: struct buf ln;
1.171 schwarze 145: struct buf *firstln, *lastln, *thisln, *loop;
1.85 schwarze 146: char *cp;
1.71 schwarze 147: size_t pos; /* byte number in the ln buffer */
1.171 schwarze 148: int line_result, result;
1.76 schwarze 149: int of;
1.1 schwarze 150: int lnn; /* line number in the real file */
1.85 schwarze 151: int fd;
1.171 schwarze 152: int inloop; /* Saw .while on this level. */
1.1 schwarze 153: unsigned char c;
154:
1.170 schwarze 155: ln.sz = 256;
156: ln.buf = mandoc_malloc(ln.sz);
157: ln.next = NULL;
1.184 schwarze 158: firstln = lastln = loop = NULL;
1.25 schwarze 159: lnn = curp->line;
160: pos = 0;
1.171 schwarze 161: inloop = 0;
1.170 schwarze 162: result = ROFF_CONT;
1.1 schwarze 163:
1.171 schwarze 164: while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) {
1.1 schwarze 165: if (start) {
166: curp->line = lnn;
167: curp->reparse_count = 0;
1.70 schwarze 168:
169: if (lnn < 3 &&
170: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 171: curp->filenc & MPARSE_LATIN1)
172: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 173: }
174:
1.71 schwarze 175: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 176:
177: /*
178: * When finding an unescaped newline character,
179: * leave the character loop to process the line.
180: * Skip a preceding carriage return, if any.
181: */
182:
1.71 schwarze 183: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 184: '\n' == blk.buf[i + 1])
185: ++i;
186: if ('\n' == blk.buf[i]) {
187: ++i;
188: ++lnn;
189: break;
190: }
191:
1.13 schwarze 192: /*
1.70 schwarze 193: * Make sure we have space for the worst
1.170 schwarze 194: * case of 12 bytes: "\\[u10ffff]\n\0"
1.13 schwarze 195: */
196:
1.170 schwarze 197: if (pos + 12 > ln.sz)
1.13 schwarze 198: resize_buf(&ln, 256);
199:
1.25 schwarze 200: /*
1.70 schwarze 201: * Encode 8-bit input.
1.1 schwarze 202: */
203:
1.70 schwarze 204: c = blk.buf[i];
205: if (c & 0x80) {
1.71 schwarze 206: if ( ! (curp->filenc && preconv_encode(
207: &blk, &i, &ln, &pos, &curp->filenc))) {
1.177 schwarze 208: mandoc_msg(MANDOCERR_CHAR_BAD,
1.89 schwarze 209: curp->line, pos, "0x%x", c);
1.70 schwarze 210: ln.buf[pos++] = '?';
211: i++;
212: }
213: continue;
214: }
215:
216: /*
217: * Exclude control characters.
218: */
1.1 schwarze 219:
1.70 schwarze 220: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.177 schwarze 221: mandoc_msg(c == 0x00 || c == 0x04 ||
1.89 schwarze 222: c > 0x0a ? MANDOCERR_CHAR_BAD :
223: MANDOCERR_CHAR_UNSUPP,
1.177 schwarze 224: curp->line, pos, "0x%x", c);
1.1 schwarze 225: i++;
1.102 schwarze 226: if (c != '\r')
227: ln.buf[pos++] = '?';
1.1 schwarze 228: continue;
229: }
230:
231: ln.buf[pos++] = blk.buf[i++];
232: }
1.170 schwarze 233: ln.buf[pos] = '\0';
234:
235: /*
236: * Maintain a lookaside buffer of all lines.
237: * parsed from this input source.
238: */
239:
240: thisln = mandoc_malloc(sizeof(*thisln));
241: thisln->buf = mandoc_strdup(ln.buf);
242: thisln->sz = strlen(ln.buf) + 1;
243: thisln->next = NULL;
244: if (firstln == NULL) {
245: firstln = lastln = thisln;
246: if (curp->secondary == NULL)
247: curp->secondary = firstln;
248: } else {
249: lastln->next = thisln;
250: lastln = thisln;
251: }
1.1 schwarze 252:
1.170 schwarze 253: /* XXX Ugly hack to mark the end of the input. */
1.1 schwarze 254:
1.170 schwarze 255: if (i == blk.sz || blk.buf[i] == '\0') {
1.183 schwarze 256: if (pos + 2 > ln.sz)
257: resize_buf(&ln, 256);
1.142 schwarze 258: ln.buf[pos++] = '\n';
1.170 schwarze 259: ln.buf[pos] = '\0';
260: }
1.1 schwarze 261:
262: /*
263: * A significant amount of complexity is contained by
264: * the roff preprocessor. It's line-oriented but can be
265: * expressed on one line, so we need at times to
266: * readjust our starting point and re-run it. The roff
267: * preprocessor can also readjust the buffers with new
268: * data, so we pass them in wholesale.
269: */
270:
271: of = 0;
272: rerun:
1.169 schwarze 273: line_result = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 274:
1.171 schwarze 275: /* Process options. */
276:
277: if (line_result & ROFF_APPEND)
278: assert(line_result == (ROFF_IGN | ROFF_APPEND));
279:
280: if (line_result & ROFF_USERCALL)
281: assert((line_result & ROFF_MASK) == ROFF_REPARSE);
282:
283: if (line_result & ROFF_USERRET) {
284: assert(line_result == (ROFF_IGN | ROFF_USERRET));
285: if (start == 0) {
286: /* Return from the current macro. */
287: result = ROFF_USERRET;
288: goto out;
289: }
290: }
291:
292: switch (line_result & ROFF_LOOPMASK) {
293: case ROFF_IGN:
294: break;
295: case ROFF_WHILE:
296: if (curp->loop != NULL) {
297: if (loop == curp->loop)
298: break;
299: mandoc_msg(MANDOCERR_WHILE_NEST,
1.177 schwarze 300: curp->line, pos, NULL);
1.171 schwarze 301: }
302: curp->loop = thisln;
303: loop = NULL;
304: inloop = 1;
305: break;
306: case ROFF_LOOPCONT:
307: case ROFF_LOOPEXIT:
308: if (curp->loop == NULL) {
309: mandoc_msg(MANDOCERR_WHILE_FAIL,
1.177 schwarze 310: curp->line, pos, NULL);
1.171 schwarze 311: break;
312: }
313: if (inloop == 0) {
314: mandoc_msg(MANDOCERR_WHILE_INTO,
1.177 schwarze 315: curp->line, pos, NULL);
1.171 schwarze 316: curp->loop = loop = NULL;
317: break;
318: }
319: if (line_result & ROFF_LOOPCONT)
320: loop = curp->loop;
321: else {
322: curp->loop = loop = NULL;
323: inloop = 0;
324: }
325: break;
326: default:
327: abort();
328: }
329:
330: /* Process the main instruction from the roff parser. */
331:
332: switch (line_result & ROFF_MASK) {
333: case ROFF_IGN:
334: break;
335: case ROFF_CONT:
1.180 schwarze 336: if (curp->man->meta.macroset == MACROSET_NONE)
1.171 schwarze 337: choose_parser(curp);
1.180 schwarze 338: if ((curp->man->meta.macroset == MACROSET_MDOC ?
1.171 schwarze 339: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
340: man_parseln(curp->man, curp->line, ln.buf, of)
341: ) == 2)
342: goto out;
343: break;
344: case ROFF_RERUN:
345: goto rerun;
1.25 schwarze 346: case ROFF_REPARSE:
1.169 schwarze 347: if (++curp->reparse_count > REPARSE_LIMIT) {
1.171 schwarze 348: /* Abort and return to the top level. */
1.170 schwarze 349: result = ROFF_IGN;
1.177 schwarze 350: mandoc_msg(MANDOCERR_ROFFLOOP,
1.25 schwarze 351: curp->line, pos, NULL);
1.171 schwarze 352: goto out;
1.169 schwarze 353: }
1.171 schwarze 354: result = mparse_buf_r(curp, ln, of, 0);
355: if (line_result & ROFF_USERCALL) {
356: roff_userret(curp->roff);
357: /* Continue normally. */
358: if (result & ROFF_USERRET)
359: result = ROFF_CONT;
1.135 schwarze 360: }
1.171 schwarze 361: if (start == 0 && result != ROFF_CONT)
362: goto out;
363: break;
1.25 schwarze 364: case ROFF_SO:
1.71 schwarze 365: if ( ! (curp->options & MPARSE_SO) &&
366: (i >= blk.sz || blk.buf[i] == '\0')) {
1.180 schwarze 367: curp->man->meta.sodest =
368: mandoc_strdup(ln.buf + of);
1.170 schwarze 369: goto out;
1.23 schwarze 370: }
1.122 schwarze 371: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 372: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 373: close(fd);
1.88 schwarze 374: } else {
1.179 schwarze 375: mandoc_msg(MANDOCERR_SO_FAIL,
376: curp->line, of, ".so %s: %s",
377: ln.buf + of, strerror(errno));
1.85 schwarze 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);
1.30 schwarze 385: }
1.171 schwarze 386: break;
1.1 schwarze 387: default:
1.171 schwarze 388: abort();
1.1 schwarze 389: }
390:
1.171 schwarze 391: /* Start the next input line. */
1.1 schwarze 392:
1.171 schwarze 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: }
1.1 schwarze 406:
1.171 schwarze 407: pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
1.1 schwarze 408: }
1.170 schwarze 409: out:
1.171 schwarze 410: if (inloop) {
411: if (result != ROFF_USERRET)
1.177 schwarze 412: mandoc_msg(MANDOCERR_WHILE_OUTOF,
1.171 schwarze 413: curp->line, pos, NULL);
414: curp->loop = NULL;
415: }
1.1 schwarze 416: free(ln.buf);
1.170 schwarze 417: if (firstln != curp->secondary)
418: free_buf_list(firstln);
419: return result;
1.1 schwarze 420: }
421:
422: static int
1.175 schwarze 423: read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
1.1 schwarze 424: {
425: struct stat st;
1.115 schwarze 426: gzFile gz;
1.1 schwarze 427: size_t off;
428: ssize_t ssz;
1.166 schwarze 429: int gzerrnum, retval;
1.1 schwarze 430:
1.164 schwarze 431: if (fstat(fd, &st) == -1) {
1.185 ! schwarze 432: mandoc_msg(MANDOCERR_FSTAT, 0, 0, "%s", strerror(errno));
! 433: return -1;
1.164 schwarze 434: }
1.1 schwarze 435:
436: /*
437: * If we're a regular file, try just reading in the whole entry
438: * via mmap(). This is faster than reading it into blocks, and
439: * since each file is only a few bytes to begin with, I'm not
440: * concerned that this is going to tank any machines.
441: */
442:
1.115 schwarze 443: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 444: if (st.st_size > 0x7fffffff) {
1.177 schwarze 445: mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
1.185 ! schwarze 446: return -1;
1.1 schwarze 447: }
448: *with_mmap = 1;
449: fb->sz = (size_t)st.st_size;
1.15 schwarze 450: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 451: if (fb->buf != MAP_FAILED)
1.185 ! schwarze 452: return 0;
1.1 schwarze 453: }
454:
1.115 schwarze 455: if (curp->gzip) {
1.166 schwarze 456: /*
457: * Duplicating the file descriptor is required
458: * because we will have to call gzclose(3)
459: * to free memory used internally by zlib,
460: * but that will also close the file descriptor,
461: * which this function must not do.
462: */
463: if ((fd = dup(fd)) == -1) {
1.185 ! schwarze 464: mandoc_msg(MANDOCERR_DUP, 0, 0,
! 465: "%s", strerror(errno));
! 466: return -1;
1.166 schwarze 467: }
1.164 schwarze 468: if ((gz = gzdopen(fd, "rb")) == NULL) {
1.185 ! schwarze 469: mandoc_msg(MANDOCERR_GZDOPEN, 0, 0,
! 470: "%s", strerror(errno));
1.166 schwarze 471: close(fd);
1.185 ! schwarze 472: return -1;
1.164 schwarze 473: }
1.115 schwarze 474: } else
475: gz = NULL;
476:
1.1 schwarze 477: /*
478: * If this isn't a regular file (like, say, stdin), then we must
479: * go the old way and just read things in bit by bit.
480: */
481:
482: *with_mmap = 0;
483: off = 0;
1.185 ! schwarze 484: retval = -1;
1.1 schwarze 485: fb->sz = 0;
486: fb->buf = NULL;
487: for (;;) {
488: if (off == fb->sz) {
489: if (fb->sz == (1U << 31)) {
1.177 schwarze 490: mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
1.1 schwarze 491: break;
492: }
493: resize_buf(fb, 65536);
494: }
1.115 schwarze 495: ssz = curp->gzip ?
496: gzread(gz, fb->buf + (int)off, fb->sz - off) :
497: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 498: if (ssz == 0) {
499: fb->sz = off;
1.185 ! schwarze 500: retval = 0;
1.166 schwarze 501: break;
1.1 schwarze 502: }
1.164 schwarze 503: if (ssz == -1) {
1.166 schwarze 504: if (curp->gzip)
505: (void)gzerror(gz, &gzerrnum);
1.185 ! schwarze 506: mandoc_msg(MANDOCERR_READ, 0, 0, "%s",
1.166 schwarze 507: curp->gzip && gzerrnum != Z_ERRNO ?
508: zError(gzerrnum) : strerror(errno));
1.164 schwarze 509: break;
510: }
1.1 schwarze 511: off += (size_t)ssz;
512: }
513:
1.166 schwarze 514: if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
1.185 ! schwarze 515: mandoc_msg(MANDOCERR_GZCLOSE, 0, 0, "%s",
1.166 schwarze 516: gzerrnum == Z_ERRNO ? strerror(errno) :
517: zError(gzerrnum));
1.185 ! schwarze 518: if (retval == -1) {
1.166 schwarze 519: free(fb->buf);
520: fb->buf = NULL;
521: }
522: return retval;
1.1 schwarze 523: }
524:
525: static void
526: mparse_end(struct mparse *curp)
527: {
1.180 schwarze 528: if (curp->man->meta.macroset == MACROSET_NONE)
529: curp->man->meta.macroset = MACROSET_MAN;
530: if (curp->man->meta.macroset == MACROSET_MDOC)
1.110 schwarze 531: mdoc_endparse(curp->man);
532: else
1.86 schwarze 533: man_endparse(curp->man);
1.1 schwarze 534: roff_endparse(curp->roff);
535: }
536:
1.176 schwarze 537: /*
538: * Read the whole file into memory and call the parsers.
539: * Called recursively when an .so request is encountered.
540: */
541: void
542: mparse_readfd(struct mparse *curp, int fd, const char *filename)
1.1 schwarze 543: {
1.176 schwarze 544: static int recursion_depth;
545:
546: struct buf blk;
547: struct buf *save_primary;
548: const char *save_filename;
1.71 schwarze 549: size_t offset;
1.176 schwarze 550: int save_filenc, save_lineno;
551: int with_mmap;
1.14 schwarze 552:
1.176 schwarze 553: if (recursion_depth > 64) {
1.177 schwarze 554: mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, 0, NULL);
1.15 schwarze 555: return;
1.14 schwarze 556: }
1.185 ! schwarze 557: if (read_whole_file(curp, fd, &blk, &with_mmap) == -1)
1.176 schwarze 558: return;
559:
560: /*
561: * Save some properties of the parent file.
562: */
563:
564: save_primary = curp->primary;
565: save_filenc = curp->filenc;
566: save_lineno = curp->line;
567: save_filename = mandoc_msg_getinfilename();
1.1 schwarze 568:
1.59 schwarze 569: curp->primary = &blk;
1.176 schwarze 570: curp->filenc = curp->options & (MPARSE_UTF8 | MPARSE_LATIN1);
1.15 schwarze 571: curp->line = 1;
1.176 schwarze 572: mandoc_msg_setinfilename(filename);
1.15 schwarze 573:
1.70 schwarze 574: /* Skip an UTF-8 byte order mark. */
575: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
576: (unsigned char)blk.buf[0] == 0xef &&
577: (unsigned char)blk.buf[1] == 0xbb &&
578: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 579: offset = 3;
1.70 schwarze 580: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 581: } else
582: offset = 0;
1.70 schwarze 583:
1.176 schwarze 584: recursion_depth++;
1.71 schwarze 585: mparse_buf_r(curp, blk, offset, 1);
1.86 schwarze 586: if (--recursion_depth == 0)
1.15 schwarze 587: mparse_end(curp);
588:
1.176 schwarze 589: /*
590: * Clean up and restore saved parent properties.
591: */
1.15 schwarze 592:
1.176 schwarze 593: if (with_mmap)
594: munmap(blk.buf, blk.sz);
595: else
596: free(blk.buf);
1.1 schwarze 597:
1.176 schwarze 598: curp->primary = save_primary;
599: curp->filenc = save_filenc;
600: curp->line = save_lineno;
601: if (save_filename != NULL)
602: mandoc_msg_setinfilename(save_filename);
1.58 schwarze 603: }
604:
1.122 schwarze 605: int
606: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 607: {
608: char *cp;
1.182 schwarze 609: int fd, save_errno;
1.58 schwarze 610:
1.115 schwarze 611: cp = strrchr(file, '.');
612: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 613:
1.115 schwarze 614: /* First try to use the filename as it is. */
1.74 schwarze 615:
1.122 schwarze 616: if ((fd = open(file, O_RDONLY)) != -1)
617: return fd;
1.74 schwarze 618:
1.115 schwarze 619: /*
620: * If that doesn't work and the filename doesn't
621: * already end in .gz, try appending .gz.
622: */
1.74 schwarze 623:
1.115 schwarze 624: if ( ! curp->gzip) {
1.182 schwarze 625: save_errno = errno;
1.74 schwarze 626: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 627: fd = open(cp, O_RDONLY);
1.83 schwarze 628: free(cp);
1.182 schwarze 629: errno = save_errno;
1.122 schwarze 630: if (fd != -1) {
1.115 schwarze 631: curp->gzip = 1;
1.122 schwarze 632: return fd;
1.58 schwarze 633: }
634: }
635:
1.115 schwarze 636: /* Neither worked, give up. */
1.73 schwarze 637:
1.122 schwarze 638: return -1;
1.1 schwarze 639: }
640:
641: struct mparse *
1.175 schwarze 642: mparse_alloc(int options, enum mandoc_os os_e, const char *os_s)
1.1 schwarze 643: {
644: struct mparse *curp;
645:
646: curp = mandoc_calloc(1, sizeof(struct mparse));
647:
1.22 schwarze 648: curp->options = options;
1.151 schwarze 649: curp->os_s = os_s;
1.1 schwarze 650:
1.178 schwarze 651: curp->roff = roff_alloc(options);
652: curp->man = roff_man_alloc(curp->roff, curp->os_s,
1.112 schwarze 653: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 654: if (curp->options & MPARSE_MDOC) {
1.180 schwarze 655: curp->man->meta.macroset = MACROSET_MDOC;
1.136 schwarze 656: if (curp->man->mdocmac == NULL)
657: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 658: } else if (curp->options & MPARSE_MAN) {
1.180 schwarze 659: curp->man->meta.macroset = MACROSET_MAN;
1.136 schwarze 660: if (curp->man->manmac == NULL)
661: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 662: }
1.180 schwarze 663: curp->man->meta.first->tok = TOKEN_NONE;
1.151 schwarze 664: curp->man->meta.os_e = os_e;
1.117 schwarze 665: return curp;
1.1 schwarze 666: }
667:
668: void
669: mparse_reset(struct mparse *curp)
670: {
671: roff_reset(curp->roff);
1.124 schwarze 672: roff_man_reset(curp->man);
1.170 schwarze 673: free_buf_list(curp->secondary);
674: curp->secondary = NULL;
1.132 schwarze 675: curp->gzip = 0;
1.1 schwarze 676: }
677:
678: void
679: mparse_free(struct mparse *curp)
680: {
1.136 schwarze 681: roffhash_free(curp->man->mdocmac);
682: roffhash_free(curp->man->manmac);
1.112 schwarze 683: roff_man_free(curp->man);
1.133 schwarze 684: roff_free(curp->roff);
1.170 schwarze 685: free_buf_list(curp->secondary);
1.1 schwarze 686: free(curp);
687: }
688:
1.180 schwarze 689: struct roff_meta *
690: mparse_result(struct mparse *curp)
1.1 schwarze 691: {
1.181 schwarze 692: roff_state_reset(curp->man);
1.180 schwarze 693: if (curp->options & MPARSE_VALIDATE) {
694: if (curp->man->meta.macroset == MACROSET_MDOC)
695: mdoc_validate(curp->man);
696: else
697: man_validate(curp->man);
1.23 schwarze 698: }
1.180 schwarze 699: return &curp->man->meta;
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