Annotation of mandoc/read.c, Revision 1.121
1.121 ! schwarze 1: /* $Id: read.c,v 1.120 2015/01/28 21:11:54 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.107 schwarze 4: * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
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: *
11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR 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: */
1.11 kristaps 19: #include "config.h"
20:
1.80 schwarze 21: #include <sys/types.h>
1.81 schwarze 22: #if HAVE_MMAP
1.82 schwarze 23: #include <sys/mman.h>
1.80 schwarze 24: #include <sys/stat.h>
1.15 kristaps 25: #endif
1.82 schwarze 26: #include <sys/wait.h>
1.1 kristaps 27:
28: #include <assert.h>
29: #include <ctype.h>
1.40 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.3 kristaps 32: #include <stdarg.h>
1.28 joerg 33: #include <stdint.h>
1.1 kristaps 34: #include <stdio.h>
35: #include <stdlib.h>
36: #include <string.h>
37: #include <unistd.h>
38:
39: #include "mandoc.h"
1.46 schwarze 40: #include "mandoc_aux.h"
1.3 kristaps 41: #include "libmandoc.h"
1.1 kristaps 42: #include "mdoc.h"
43: #include "man.h"
44:
45: #define REPARSE_LIMIT 1000
46:
47: struct mparse {
48: struct man *pman; /* persistent man parser */
49: struct mdoc *pmdoc; /* persistent mdoc parser */
50: struct man *man; /* man parser */
51: struct mdoc *mdoc; /* mdoc parser */
52: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 53: const struct mchars *mchars; /* character table */
1.45 schwarze 54: char *sodest; /* filename pointed to by .so */
1.83 schwarze 55: const char *file; /* filename of current input file */
56: struct buf *primary; /* buffer currently being parsed */
57: struct buf *secondary; /* preprocessed copy of input */
58: const char *defos; /* default operating system */
59: mandocmsg mmsg; /* warning/error message handler */
60: enum mandoclevel file_status; /* status of current parse */
61: enum mandoclevel wlevel; /* ignore messages below this */
62: int options; /* parser options */
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 */
1.97 schwarze 66: pid_t child; /* the gunzip(1) process */
1.1 kristaps 67: };
68:
1.84 schwarze 69: static void choose_parser(struct mparse *);
1.1 kristaps 70: static void resize_buf(struct buf *, size_t);
1.95 schwarze 71: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 72: static int read_whole_file(struct mparse *, const char *, int,
73: struct buf *, int *);
1.1 kristaps 74: static void mparse_end(struct mparse *);
1.37 schwarze 75: static void mparse_parse_buffer(struct mparse *, struct buf,
76: const char *);
1.1 kristaps 77:
1.3 kristaps 78: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
79: MANDOCERR_OK,
80: MANDOCERR_WARNING,
81: MANDOCERR_WARNING,
82: MANDOCERR_ERROR,
1.112 schwarze 83: MANDOCERR_UNSUPP,
1.3 kristaps 84: MANDOCERR_MAX,
85: MANDOCERR_MAX
86: };
87:
1.7 kristaps 88: static const char * const mandocerrs[MANDOCERR_MAX] = {
89: "ok",
90:
91: "generic warning",
92:
93: /* related to the prologue */
1.79 schwarze 94: "missing manual title, using UNTITLED",
95: "missing manual title, using \"\"",
1.54 schwarze 96: "lower case character in document title",
1.79 schwarze 97: "missing manual section, using \"\"",
1.7 kristaps 98: "unknown manual section",
1.54 schwarze 99: "missing date, using today's date",
1.7 kristaps 100: "cannot parse date, using it verbatim",
1.79 schwarze 101: "missing Os macro, using \"\"",
102: "duplicate prologue macro",
103: "late prologue macro",
104: "skipping late title macro",
1.7 kristaps 105: "prologue macros out of order",
106:
107: /* related to document structure */
108: ".so is fragile, better use ln(1)",
1.50 schwarze 109: "no document body",
1.54 schwarze 110: "content before first section header",
111: "first section is not \"NAME\"",
1.7 kristaps 112: "bad NAME section contents",
1.121 ! schwarze 113: "missing description line, using \"\"",
1.7 kristaps 114: "sections out of conventional order",
1.54 schwarze 115: "duplicate section title",
116: "unexpected section",
1.87 schwarze 117: "unusual Xr order",
118: "unusual Xr punctuation",
1.86 schwarze 119: "AUTHORS section without An macro",
1.7 kristaps 120:
121: /* related to macros and nesting */
1.55 schwarze 122: "obsolete macro",
1.102 schwarze 123: "macro neither callable nor escaped",
1.7 kristaps 124: "skipping paragraph macro",
1.31 schwarze 125: "moving paragraph macro out of list",
1.7 kristaps 126: "skipping no-space macro",
127: "blocks badly nested",
128: "nested displays are not portable",
1.57 schwarze 129: "moving content out of list",
130: ".Vt block has child macro",
1.78 schwarze 131: "fill mode already enabled, skipping",
132: "fill mode already disabled, skipping",
1.7 kristaps 133: "line scope broken",
134:
135: /* related to missing macro arguments */
1.58 schwarze 136: "skipping empty request",
137: "conditional request controls empty scope",
1.7 kristaps 138: "skipping empty macro",
1.62 schwarze 139: "empty argument, using 0n",
1.7 kristaps 140: "argument count wrong",
1.60 schwarze 141: "missing display type, using -ragged",
142: "list type is not the first argument",
143: "missing -width in -tag list, using 8n",
1.78 schwarze 144: "missing utility name, using \"\"",
1.60 schwarze 145: "empty head in list item",
146: "empty list item",
1.61 schwarze 147: "missing font type, using \\fR",
148: "unknown font type, using \\fR",
1.103 schwarze 149: "nothing follows prefix",
1.60 schwarze 150: "missing -std argument, adding it",
1.90 schwarze 151: "missing eqn box, using \"\"",
1.7 kristaps 152:
153: /* related to bad macro arguments */
1.64 schwarze 154: "unterminated quoted argument",
1.7 kristaps 155: "duplicate argument",
1.76 schwarze 156: "skipping duplicate argument",
1.63 schwarze 157: "skipping duplicate display type",
158: "skipping duplicate list type",
1.76 schwarze 159: "skipping -width argument",
1.7 kristaps 160: "unknown AT&T UNIX version",
1.88 schwarze 161: "comma in function argument",
1.89 schwarze 162: "parenthesis in function name",
1.67 schwarze 163: "invalid content in Rs block",
1.63 schwarze 164: "invalid Boolean argument",
165: "unknown font, skipping request",
1.7 kristaps 166:
167: /* related to plain text */
1.64 schwarze 168: "blank line in fill mode, using .sp",
169: "tab in filled text",
170: "whitespace at end of input line",
1.7 kristaps 171: "bad comment style",
1.64 schwarze 172: "invalid escape sequence",
173: "undefined string, using \"\"",
1.16 kristaps 174:
1.118 schwarze 175: /* related to tables */
176: "tbl line starts with span",
177: "tbl column starts with span",
178: "skipping vertical bar in tbl layout",
179:
1.7 kristaps 180: "generic error",
181:
182: /* related to tables */
1.116 schwarze 183: "non-alphabetic character in tbl options",
184: "skipping unknown tbl option",
185: "missing tbl option argument",
186: "wrong tbl option argument size",
1.118 schwarze 187: "empty tbl layout",
188: "invalid character in tbl layout",
189: "unmatched parenthesis in tbl layout",
1.119 schwarze 190: "tbl without any data cells",
191: "ignoring data in spanned tbl cell",
192: "ignoring extra tbl data cells",
193: "data block open at end of tbl",
1.7 kristaps 194:
1.68 schwarze 195: /* related to document structure and macros */
1.108 schwarze 196: NULL,
1.7 kristaps 197: "input stack limit exceeded, infinite loop?",
198: "skipping bad character",
1.68 schwarze 199: "skipping unknown macro",
1.112 schwarze 200: "skipping insecure request",
1.73 schwarze 201: "skipping item outside list",
1.68 schwarze 202: "skipping column outside column list",
203: "skipping end of block that is not open",
1.115 schwarze 204: "fewer RS blocks open, skipping",
1.68 schwarze 205: "inserting missing end of block",
206: "appending missing end of block",
207:
208: /* related to request and macro arguments */
1.7 kristaps 209: "escaped character not allowed in a name",
210: "argument count wrong",
1.99 schwarze 211: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 212: "missing list type, using -item",
1.70 schwarze 213: "missing manual name, using \"\"",
1.71 schwarze 214: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 215: "unknown standard specifier",
1.71 schwarze 216: "skipping request without numeric argument",
1.109 schwarze 217: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 218: ".so request failed",
1.61 schwarze 219: "skipping all arguments",
220: "skipping excess arguments",
1.92 kristaps 221: "divide by zero",
1.112 schwarze 222:
223: "unsupported feature",
224: "input too large",
1.114 schwarze 225: "unsupported control character",
1.112 schwarze 226: "unsupported roff request",
1.119 schwarze 227: "eqn delim option in tbl",
1.118 schwarze 228: "unsupported tbl layout modifier",
1.112 schwarze 229: "ignoring macro in table",
1.7 kristaps 230: };
231:
232: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
233: "SUCCESS",
234: "RESERVED",
235: "WARNING",
236: "ERROR",
1.112 schwarze 237: "UNSUPP",
1.7 kristaps 238: "BADARG",
239: "SYSERR"
240: };
241:
1.47 schwarze 242:
1.1 kristaps 243: static void
244: resize_buf(struct buf *buf, size_t initial)
245: {
246:
247: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
248: buf->buf = mandoc_realloc(buf->buf, buf->sz);
249: }
250:
251: static void
1.84 schwarze 252: choose_parser(struct mparse *curp)
1.1 kristaps 253: {
1.83 schwarze 254: char *cp, *ep;
255: int format;
1.1 kristaps 256:
1.83 schwarze 257: /*
258: * If neither command line arguments -mdoc or -man select
259: * a parser nor the roff parser found a .Dd or .TH macro
260: * yet, look ahead in the main input buffer.
261: */
262:
263: if ((format = roff_getformat(curp->roff)) == 0) {
264: cp = curp->primary->buf;
265: ep = cp + curp->primary->sz;
266: while (cp < ep) {
1.85 schwarze 267: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 268: cp++;
269: if (cp[0] == 'D' && cp[1] == 'd') {
270: format = MPARSE_MDOC;
271: break;
272: }
273: if (cp[0] == 'T' && cp[1] == 'H') {
274: format = MPARSE_MAN;
275: break;
276: }
277: }
278: cp = memchr(cp, '\n', ep - cp);
279: if (cp == NULL)
280: break;
281: cp++;
282: }
1.1 kristaps 283: }
284:
1.83 schwarze 285: if (format == MPARSE_MDOC) {
1.47 schwarze 286: if (NULL == curp->pmdoc)
1.44 schwarze 287: curp->pmdoc = mdoc_alloc(
288: curp->roff, curp, curp->defos,
289: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 290: assert(curp->pmdoc);
291: curp->mdoc = curp->pmdoc;
292: return;
1.47 schwarze 293: }
1.1 kristaps 294:
1.83 schwarze 295: /* Fall back to man(7) as a last resort. */
296:
1.47 schwarze 297: if (NULL == curp->pman)
1.106 schwarze 298: curp->pman = man_alloc(
299: curp->roff, curp, curp->defos,
1.44 schwarze 300: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 301: assert(curp->pman);
302: curp->man = curp->pman;
303: }
304:
305: /*
1.95 schwarze 306: * Main parse routine for a buffer.
307: * It assumes encoding and line numbering are already set up.
308: * It can recurse directly (for invocations of user-defined
309: * macros, inline equations, and input line traps)
310: * and indirectly (for .so file inclusion).
1.1 kristaps 311: */
312: static void
1.95 schwarze 313: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 314: {
315: const struct tbl_span *span;
316: struct buf ln;
1.113 schwarze 317: const char *save_file;
1.110 schwarze 318: char *cp;
1.95 schwarze 319: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 320: enum rofferr rr;
1.100 schwarze 321: int of;
1.1 kristaps 322: int lnn; /* line number in the real file */
1.110 schwarze 323: int fd;
324: pid_t save_child;
1.1 kristaps 325: unsigned char c;
326:
1.95 schwarze 327: memset(&ln, 0, sizeof(ln));
1.1 kristaps 328:
1.47 schwarze 329: lnn = curp->line;
330: pos = 0;
1.1 kristaps 331:
1.95 schwarze 332: while (i < blk.sz) {
1.1 kristaps 333: if (0 == pos && '\0' == blk.buf[i])
334: break;
335:
336: if (start) {
337: curp->line = lnn;
338: curp->reparse_count = 0;
1.93 schwarze 339:
340: if (lnn < 3 &&
341: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 342: curp->filenc & MPARSE_LATIN1)
343: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 344: }
345:
1.95 schwarze 346: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 347:
348: /*
349: * When finding an unescaped newline character,
350: * leave the character loop to process the line.
351: * Skip a preceding carriage return, if any.
352: */
353:
1.95 schwarze 354: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 355: '\n' == blk.buf[i + 1])
356: ++i;
357: if ('\n' == blk.buf[i]) {
358: ++i;
359: ++lnn;
360: break;
361: }
362:
1.35 schwarze 363: /*
1.93 schwarze 364: * Make sure we have space for the worst
365: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 366: */
367:
1.95 schwarze 368: if (pos + 11 > ln.sz)
1.35 schwarze 369: resize_buf(&ln, 256);
370:
1.47 schwarze 371: /*
1.93 schwarze 372: * Encode 8-bit input.
1.1 kristaps 373: */
374:
1.93 schwarze 375: c = blk.buf[i];
376: if (c & 0x80) {
1.95 schwarze 377: if ( ! (curp->filenc && preconv_encode(
378: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 379: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
380: curp->line, pos, "0x%x", c);
1.93 schwarze 381: ln.buf[pos++] = '?';
382: i++;
383: }
384: continue;
385: }
386:
387: /*
388: * Exclude control characters.
389: */
1.1 kristaps 390:
1.93 schwarze 391: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 392: mandoc_vmsg(c == 0x00 || c == 0x04 ||
393: c > 0x0a ? MANDOCERR_CHAR_BAD :
394: MANDOCERR_CHAR_UNSUPP,
395: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 396: i++;
1.27 joerg 397: ln.buf[pos++] = '?';
1.1 kristaps 398: continue;
399: }
400:
401: /* Trailing backslash = a plain char. */
402:
1.95 schwarze 403: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 404: ln.buf[pos++] = blk.buf[i++];
405: continue;
406: }
407:
408: /*
409: * Found escape and at least one other character.
410: * When it's a newline character, skip it.
411: * When there is a carriage return in between,
412: * skip that one as well.
413: */
414:
1.95 schwarze 415: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 416: '\n' == blk.buf[i + 2])
417: ++i;
418: if ('\n' == blk.buf[i + 1]) {
419: i += 2;
420: ++lnn;
421: continue;
422: }
423:
1.13 kristaps 424: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 425: i += 2;
426: /* Comment, skip to end of line */
1.95 schwarze 427: for (; i < blk.sz; ++i) {
1.1 kristaps 428: if ('\n' == blk.buf[i]) {
429: ++i;
430: ++lnn;
431: break;
432: }
433: }
434:
435: /* Backout trailing whitespaces */
436: for (; pos > 0; --pos) {
437: if (ln.buf[pos - 1] != ' ')
438: break;
439: if (pos > 2 && ln.buf[pos - 2] == '\\')
440: break;
441: }
442: break;
443: }
444:
1.35 schwarze 445: /* Catch escaped bogus characters. */
446:
447: c = (unsigned char) blk.buf[i+1];
448:
1.47 schwarze 449: if ( ! (isascii(c) &&
450: (isgraph(c) || isblank(c)))) {
1.114 schwarze 451: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 452: curp->line, pos, "0x%x", c);
1.35 schwarze 453: i += 2;
454: ln.buf[pos++] = '?';
455: continue;
456: }
457:
1.1 kristaps 458: /* Some other escape sequence, copy & cont. */
459:
460: ln.buf[pos++] = blk.buf[i++];
461: ln.buf[pos++] = blk.buf[i++];
462: }
463:
1.95 schwarze 464: if (pos >= ln.sz)
1.1 kristaps 465: resize_buf(&ln, 256);
466:
467: ln.buf[pos] = '\0';
468:
469: /*
470: * A significant amount of complexity is contained by
471: * the roff preprocessor. It's line-oriented but can be
472: * expressed on one line, so we need at times to
473: * readjust our starting point and re-run it. The roff
474: * preprocessor can also readjust the buffers with new
475: * data, so we pass them in wholesale.
476: */
477:
478: of = 0;
479:
1.24 kristaps 480: /*
481: * Maintain a lookaside buffer of all parsed lines. We
482: * only do this if mparse_keep() has been invoked (the
483: * buffer may be accessed with mparse_getkeep()).
484: */
485:
486: if (curp->secondary) {
1.47 schwarze 487: curp->secondary->buf = mandoc_realloc(
488: curp->secondary->buf,
489: curp->secondary->sz + pos + 2);
490: memcpy(curp->secondary->buf +
491: curp->secondary->sz,
492: ln.buf, pos);
1.24 kristaps 493: curp->secondary->sz += pos;
494: curp->secondary->buf
495: [curp->secondary->sz] = '\n';
496: curp->secondary->sz++;
497: curp->secondary->buf
498: [curp->secondary->sz] = '\0';
499: }
1.1 kristaps 500: rerun:
1.96 schwarze 501: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 502:
503: switch (rr) {
1.47 schwarze 504: case ROFF_REPARSE:
1.1 kristaps 505: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 506: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 507: else
1.3 kristaps 508: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 509: curp->line, pos, NULL);
1.1 kristaps 510: pos = 0;
511: continue;
1.47 schwarze 512: case ROFF_APPEND:
1.95 schwarze 513: pos = strlen(ln.buf);
1.1 kristaps 514: continue;
1.47 schwarze 515: case ROFF_RERUN:
1.1 kristaps 516: goto rerun;
1.47 schwarze 517: case ROFF_IGN:
1.1 kristaps 518: pos = 0;
519: continue;
1.47 schwarze 520: case ROFF_SO:
1.95 schwarze 521: if ( ! (curp->options & MPARSE_SO) &&
522: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 523: curp->sodest = mandoc_strdup(ln.buf + of);
524: free(ln.buf);
525: return;
526: }
1.24 kristaps 527: /*
528: * We remove `so' clauses from our lookaside
529: * buffer because we're going to descend into
530: * the file recursively.
531: */
1.47 schwarze 532: if (curp->secondary)
1.25 kristaps 533: curp->secondary->sz -= pos + 1;
1.113 schwarze 534: save_file = curp->file;
1.110 schwarze 535: save_child = curp->child;
536: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 537: MANDOCLEVEL_OK) {
1.110 schwarze 538: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 539: curp->file = save_file;
540: } else {
541: curp->file = save_file;
1.52 schwarze 542: mandoc_vmsg(MANDOCERR_SO_FAIL,
543: curp, curp->line, pos,
544: ".so %s", ln.buf + of);
1.110 schwarze 545: ln.sz = mandoc_asprintf(&cp,
546: ".sp\nSee the file %s.\n.sp",
547: ln.buf + of);
548: free(ln.buf);
549: ln.buf = cp;
550: of = 0;
551: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 552: }
1.110 schwarze 553: curp->child = save_child;
1.1 kristaps 554: pos = 0;
555: continue;
556: default:
557: break;
558: }
559:
560: /*
561: * If input parsers have not been allocated, do so now.
1.14 kristaps 562: * We keep these instanced between parsers, but set them
1.1 kristaps 563: * locally per parse routine since we can use different
564: * parsers with each one.
565: */
566:
567: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 568: choose_parser(curp);
1.1 kristaps 569:
1.47 schwarze 570: /*
1.84 schwarze 571: * Lastly, push down into the parsers themselves.
1.1 kristaps 572: * If libroff returns ROFF_TBL, then add it to the
573: * currently open parse. Since we only get here if
574: * there does exist data (see tbl_data.c), we're
575: * guaranteed that something's been allocated.
576: * Do the same for ROFF_EQN.
577: */
578:
1.100 schwarze 579: if (rr == ROFF_TBL) {
580: while ((span = roff_span(curp->roff)) != NULL)
581: if (curp->man == NULL)
582: mdoc_addspan(curp->mdoc, span);
583: else
584: man_addspan(curp->man, span);
585: } else if (rr == ROFF_EQN) {
586: if (curp->man == NULL)
587: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
588: else
589: man_addeqn(curp->man, roff_eqn(curp->roff));
590: } else if ((curp->man == NULL ?
591: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
592: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
593: break;
1.1 kristaps 594:
595: /* Temporary buffers typically are not full. */
596:
597: if (0 == start && '\0' == blk.buf[i])
598: break;
599:
600: /* Start the next input line. */
601:
602: pos = 0;
603: }
604:
605: free(ln.buf);
606: }
607:
608: static int
1.40 schwarze 609: read_whole_file(struct mparse *curp, const char *file, int fd,
610: struct buf *fb, int *with_mmap)
1.1 kristaps 611: {
612: size_t off;
613: ssize_t ssz;
614:
1.81 schwarze 615: #if HAVE_MMAP
1.15 kristaps 616: struct stat st;
1.1 kristaps 617: if (-1 == fstat(fd, &st)) {
1.107 schwarze 618: perror(file);
619: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 620: }
621:
622: /*
623: * If we're a regular file, try just reading in the whole entry
624: * via mmap(). This is faster than reading it into blocks, and
625: * since each file is only a few bytes to begin with, I'm not
626: * concerned that this is going to tank any machines.
627: */
628:
629: if (S_ISREG(st.st_mode)) {
630: if (st.st_size >= (1U << 31)) {
1.111 schwarze 631: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 632: return(0);
633: }
634: *with_mmap = 1;
635: fb->sz = (size_t)st.st_size;
1.37 schwarze 636: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 637: if (fb->buf != MAP_FAILED)
638: return(1);
639: }
1.15 kristaps 640: #endif
1.1 kristaps 641:
642: /*
643: * If this isn't a regular file (like, say, stdin), then we must
644: * go the old way and just read things in bit by bit.
645: */
646:
647: *with_mmap = 0;
648: off = 0;
649: fb->sz = 0;
650: fb->buf = NULL;
651: for (;;) {
652: if (off == fb->sz) {
653: if (fb->sz == (1U << 31)) {
1.111 schwarze 654: mandoc_msg(MANDOCERR_TOOLARGE, curp,
655: 0, 0, NULL);
1.1 kristaps 656: break;
657: }
658: resize_buf(fb, 65536);
659: }
660: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
661: if (ssz == 0) {
662: fb->sz = off;
663: return(1);
664: }
665: if (ssz == -1) {
1.107 schwarze 666: perror(file);
667: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 668: }
669: off += (size_t)ssz;
670: }
671:
672: free(fb->buf);
673: fb->buf = NULL;
674: return(0);
675: }
676:
677: static void
678: mparse_end(struct mparse *curp)
679: {
680:
1.72 schwarze 681: if (curp->mdoc == NULL &&
682: curp->man == NULL &&
683: curp->sodest == NULL) {
684: if (curp->options & MPARSE_MDOC)
685: curp->mdoc = curp->pmdoc;
686: else {
687: if (curp->pman == NULL)
1.106 schwarze 688: curp->pman = man_alloc(
689: curp->roff, curp, curp->defos,
1.72 schwarze 690: curp->options & MPARSE_QUICK ? 1 : 0);
691: curp->man = curp->pman;
692: }
693: }
1.111 schwarze 694: if (curp->mdoc)
695: mdoc_endparse(curp->mdoc);
696: if (curp->man)
697: man_endparse(curp->man);
1.1 kristaps 698: roff_endparse(curp->roff);
699: }
700:
701: static void
1.36 schwarze 702: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 703: {
1.85 schwarze 704: struct buf *svprimary;
1.28 joerg 705: const char *svfile;
1.95 schwarze 706: size_t offset;
1.36 schwarze 707: static int recursion_depth;
708:
709: if (64 < recursion_depth) {
710: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
711: return;
712: }
1.28 joerg 713:
714: /* Line number is per-file. */
715: svfile = curp->file;
716: curp->file = file;
1.85 schwarze 717: svprimary = curp->primary;
1.83 schwarze 718: curp->primary = &blk;
1.28 joerg 719: curp->line = 1;
1.36 schwarze 720: recursion_depth++;
1.28 joerg 721:
1.93 schwarze 722: /* Skip an UTF-8 byte order mark. */
723: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
724: (unsigned char)blk.buf[0] == 0xef &&
725: (unsigned char)blk.buf[1] == 0xbb &&
726: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 727: offset = 3;
1.93 schwarze 728: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 729: } else
730: offset = 0;
1.93 schwarze 731:
1.95 schwarze 732: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 733:
1.111 schwarze 734: if (--recursion_depth == 0)
1.28 joerg 735: mparse_end(curp);
736:
1.85 schwarze 737: curp->primary = svprimary;
1.28 joerg 738: curp->file = svfile;
739: }
740:
741: enum mandoclevel
1.104 schwarze 742: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 743: const char *file)
744: {
745: struct buf blk;
746:
1.104 schwarze 747: blk.buf = buf;
1.28 joerg 748: blk.sz = len;
749:
1.36 schwarze 750: mparse_parse_buffer(curp, blk, file);
1.28 joerg 751: return(curp->file_status);
752: }
753:
1.98 schwarze 754: /*
755: * Read the whole file into memory and call the parsers.
756: * Called recursively when an .so request is encountered.
757: */
1.36 schwarze 758: enum mandoclevel
759: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 760: {
1.28 joerg 761: struct buf blk;
762: int with_mmap;
1.93 schwarze 763: int save_filenc;
1.1 kristaps 764:
1.91 schwarze 765: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 766: save_filenc = curp->filenc;
767: curp->filenc = curp->options &
768: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 769: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 770: curp->filenc = save_filenc;
1.81 schwarze 771: #if HAVE_MMAP
1.91 schwarze 772: if (with_mmap)
773: munmap(blk.buf, blk.sz);
774: else
1.28 joerg 775: #endif
1.91 schwarze 776: free(blk.buf);
777: }
1.1 kristaps 778:
1.98 schwarze 779: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 780: perror(file);
1.91 schwarze 781:
1.98 schwarze 782: mparse_wait(curp);
1.1 kristaps 783: return(curp->file_status);
1.82 schwarze 784: }
785:
786: enum mandoclevel
1.97 schwarze 787: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 788: {
789: int pfd[2];
1.98 schwarze 790: int save_errno;
1.82 schwarze 791: char *cp;
792:
793: curp->file = file;
1.98 schwarze 794:
795: /* Unless zipped, try to just open the file. */
796:
1.82 schwarze 797: if ((cp = strrchr(file, '.')) == NULL ||
798: strcmp(cp + 1, "gz")) {
1.97 schwarze 799: curp->child = 0;
1.98 schwarze 800: if ((*fd = open(file, O_RDONLY)) != -1)
801: return(MANDOCLEVEL_OK);
802:
803: /* Open failed; try to append ".gz". */
804:
805: mandoc_asprintf(&cp, "%s.gz", file);
806: file = cp;
807: } else
808: cp = NULL;
809:
810: /* Before forking, make sure the file can be read. */
811:
812: save_errno = errno;
813: if (access(file, R_OK) == -1) {
814: if (cp != NULL)
815: errno = save_errno;
1.108 schwarze 816: free(cp);
817: *fd = -1;
818: curp->child = 0;
819: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
820: return(MANDOCLEVEL_ERROR);
1.82 schwarze 821: }
822:
1.98 schwarze 823: /* Run gunzip(1). */
824:
1.82 schwarze 825: if (pipe(pfd) == -1) {
1.107 schwarze 826: perror("pipe");
827: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 828: }
829:
1.97 schwarze 830: switch (curp->child = fork()) {
1.82 schwarze 831: case -1:
1.107 schwarze 832: perror("fork");
833: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 834: case 0:
835: close(pfd[0]);
836: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 837: perror("dup");
838: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 839: }
840: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 841: perror("exec");
842: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 843: default:
844: close(pfd[1]);
845: *fd = pfd[0];
846: return(MANDOCLEVEL_OK);
847: }
848: }
849:
850: enum mandoclevel
1.97 schwarze 851: mparse_wait(struct mparse *curp)
1.82 schwarze 852: {
853: int status;
854:
1.97 schwarze 855: if (curp->child == 0)
856: return(MANDOCLEVEL_OK);
857:
858: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 859: perror("wait");
860: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 861: }
862: if (WIFSIGNALED(status)) {
1.108 schwarze 863: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
864: "gunzip died from signal %d", WTERMSIG(status));
865: return(MANDOCLEVEL_ERROR);
1.82 schwarze 866: }
867: if (WEXITSTATUS(status)) {
1.108 schwarze 868: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
869: "gunzip failed with code %d", WEXITSTATUS(status));
870: return(MANDOCLEVEL_ERROR);
1.82 schwarze 871: }
872: return(MANDOCLEVEL_OK);
1.1 kristaps 873: }
874:
875: struct mparse *
1.94 schwarze 876: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
877: const struct mchars *mchars, const char *defos)
1.1 kristaps 878: {
879: struct mparse *curp;
1.10 kristaps 880:
1.1 kristaps 881: curp = mandoc_calloc(1, sizeof(struct mparse));
882:
1.44 schwarze 883: curp->options = options;
1.3 kristaps 884: curp->wlevel = wlevel;
1.1 kristaps 885: curp->mmsg = mmsg;
1.29 schwarze 886: curp->defos = defos;
1.1 kristaps 887:
1.94 schwarze 888: curp->mchars = mchars;
889: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 890: if (curp->options & MPARSE_MDOC)
891: curp->pmdoc = mdoc_alloc(
892: curp->roff, curp, curp->defos,
893: curp->options & MPARSE_QUICK ? 1 : 0);
894: if (curp->options & MPARSE_MAN)
1.106 schwarze 895: curp->pman = man_alloc(
896: curp->roff, curp, curp->defos,
1.72 schwarze 897: curp->options & MPARSE_QUICK ? 1 : 0);
898:
1.1 kristaps 899: return(curp);
900: }
901:
902: void
903: mparse_reset(struct mparse *curp)
904: {
905:
906: roff_reset(curp->roff);
907:
908: if (curp->mdoc)
909: mdoc_reset(curp->mdoc);
910: if (curp->man)
911: man_reset(curp->man);
1.24 kristaps 912: if (curp->secondary)
913: curp->secondary->sz = 0;
1.1 kristaps 914:
915: curp->file_status = MANDOCLEVEL_OK;
916: curp->mdoc = NULL;
917: curp->man = NULL;
1.45 schwarze 918:
919: free(curp->sodest);
920: curp->sodest = NULL;
1.1 kristaps 921: }
922:
923: void
924: mparse_free(struct mparse *curp)
925: {
926:
927: if (curp->pmdoc)
928: mdoc_free(curp->pmdoc);
929: if (curp->pman)
930: man_free(curp->pman);
931: if (curp->roff)
932: roff_free(curp->roff);
1.24 kristaps 933: if (curp->secondary)
934: free(curp->secondary->buf);
1.1 kristaps 935:
1.24 kristaps 936: free(curp->secondary);
1.45 schwarze 937: free(curp->sodest);
1.1 kristaps 938: free(curp);
939: }
940:
941: void
1.45 schwarze 942: mparse_result(struct mparse *curp,
943: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 944: {
945:
1.45 schwarze 946: if (sodest && NULL != (*sodest = curp->sodest)) {
947: *mdoc = NULL;
948: *man = NULL;
949: return;
950: }
1.9 kristaps 951: if (mdoc)
952: *mdoc = curp->mdoc;
953: if (man)
954: *man = curp->man;
1.3 kristaps 955: }
956:
957: void
958: mandoc_vmsg(enum mandocerr t, struct mparse *m,
959: int ln, int pos, const char *fmt, ...)
960: {
961: char buf[256];
962: va_list ap;
963:
964: va_start(ap, fmt);
1.48 schwarze 965: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 966: va_end(ap);
967:
968: mandoc_msg(t, m, ln, pos, buf);
969: }
970:
971: void
1.47 schwarze 972: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 973: int ln, int col, const char *msg)
974: {
975: enum mandoclevel level;
976:
1.112 schwarze 977: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 978: while (er < mandoclimits[level])
979: level--;
980:
1.108 schwarze 981: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 982: return;
983:
1.8 kristaps 984: if (m->mmsg)
985: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 986:
987: if (m->file_status < level)
988: m->file_status = level;
1.7 kristaps 989: }
990:
991: const char *
992: mparse_strerror(enum mandocerr er)
993: {
994:
995: return(mandocerrs[er]);
996: }
997:
998: const char *
999: mparse_strlevel(enum mandoclevel lvl)
1000: {
1001: return(mandoclevels[lvl]);
1.24 kristaps 1002: }
1003:
1004: void
1005: mparse_keep(struct mparse *p)
1006: {
1007:
1008: assert(NULL == p->secondary);
1009: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1010: }
1011:
1012: const char *
1013: mparse_getkeep(const struct mparse *p)
1014: {
1015:
1016: assert(p->secondary);
1017: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1018: }
CVSweb