Annotation of src/usr.bin/mandoc/read.c, Revision 1.96
1.96 ! schwarze 1: /* $OpenBSD: read.c,v 1.95 2015/01/28 21:10:28 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.82 schwarze 4: * Copyright (c) 2010-2015 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: *
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: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
1.58 schwarze 22: #include <sys/wait.h>
1.1 schwarze 23:
24: #include <assert.h>
25: #include <ctype.h>
1.18 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
28: #include <stdarg.h>
29: #include <stdio.h>
30: #include <stdlib.h>
31: #include <string.h>
32: #include <unistd.h>
33:
34: #include "mandoc.h"
1.24 schwarze 35: #include "mandoc_aux.h"
1.1 schwarze 36: #include "libmandoc.h"
37: #include "mdoc.h"
38: #include "man.h"
39:
40: #define REPARSE_LIMIT 1000
41:
42: struct mparse {
43: struct man *pman; /* persistent man parser */
44: struct mdoc *pmdoc; /* persistent mdoc parser */
45: struct man *man; /* man parser */
46: struct mdoc *mdoc; /* mdoc parser */
47: struct roff *roff; /* roff parser (!NULL) */
1.69 schwarze 48: const struct mchars *mchars; /* character table */
1.23 schwarze 49: char *sodest; /* filename pointed to by .so */
1.59 schwarze 50: const char *file; /* filename of current input file */
51: struct buf *primary; /* buffer currently being parsed */
52: struct buf *secondary; /* preprocessed copy of input */
53: const char *defos; /* default operating system */
54: mandocmsg mmsg; /* warning/error message handler */
55: enum mandoclevel file_status; /* status of current parse */
56: enum mandoclevel wlevel; /* ignore messages below this */
57: int options; /* parser options */
1.70 schwarze 58: int filenc; /* encoding of the current file */
1.1 schwarze 59: int reparse_count; /* finite interp. stack */
1.59 schwarze 60: int line; /* line number in the file */
1.73 schwarze 61: pid_t child; /* the gunzip(1) process */
1.1 schwarze 62: };
63:
1.60 schwarze 64: static void choose_parser(struct mparse *);
1.1 schwarze 65: static void resize_buf(struct buf *, size_t);
1.71 schwarze 66: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 67: static int read_whole_file(struct mparse *, const char *, int,
68: struct buf *, int *);
1.1 schwarze 69: static void mparse_end(struct mparse *);
1.15 schwarze 70: static void mparse_parse_buffer(struct mparse *, struct buf,
71: const char *);
1.1 schwarze 72:
73: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
74: MANDOCERR_OK,
75: MANDOCERR_WARNING,
76: MANDOCERR_WARNING,
77: MANDOCERR_ERROR,
1.87 schwarze 78: MANDOCERR_UNSUPP,
1.1 schwarze 79: MANDOCERR_MAX,
80: MANDOCERR_MAX
81: };
82:
83: static const char * const mandocerrs[MANDOCERR_MAX] = {
84: "ok",
85:
86: "generic warning",
87:
88: /* related to the prologue */
1.57 schwarze 89: "missing manual title, using UNTITLED",
90: "missing manual title, using \"\"",
1.32 schwarze 91: "lower case character in document title",
1.57 schwarze 92: "missing manual section, using \"\"",
1.1 schwarze 93: "unknown manual section",
1.32 schwarze 94: "missing date, using today's date",
1.1 schwarze 95: "cannot parse date, using it verbatim",
1.57 schwarze 96: "missing Os macro, using \"\"",
97: "duplicate prologue macro",
98: "late prologue macro",
99: "skipping late title macro",
1.1 schwarze 100: "prologue macros out of order",
101:
102: /* related to document structure */
103: ".so is fragile, better use ln(1)",
1.28 schwarze 104: "no document body",
1.32 schwarze 105: "content before first section header",
106: "first section is not \"NAME\"",
1.1 schwarze 107: "bad NAME section contents",
1.96 ! schwarze 108: "missing description line, using \"\"",
1.1 schwarze 109: "sections out of conventional order",
1.32 schwarze 110: "duplicate section title",
111: "unexpected section",
1.63 schwarze 112: "unusual Xr order",
113: "unusual Xr punctuation",
1.62 schwarze 114: "AUTHORS section without An macro",
1.1 schwarze 115:
116: /* related to macros and nesting */
1.33 schwarze 117: "obsolete macro",
1.78 schwarze 118: "macro neither callable nor escaped",
1.1 schwarze 119: "skipping paragraph macro",
1.10 schwarze 120: "moving paragraph macro out of list",
1.1 schwarze 121: "skipping no-space macro",
122: "blocks badly nested",
123: "nested displays are not portable",
1.35 schwarze 124: "moving content out of list",
125: ".Vt block has child macro",
1.56 schwarze 126: "fill mode already enabled, skipping",
127: "fill mode already disabled, skipping",
1.1 schwarze 128: "line scope broken",
129:
130: /* related to missing macro arguments */
1.36 schwarze 131: "skipping empty request",
132: "conditional request controls empty scope",
1.1 schwarze 133: "skipping empty macro",
1.40 schwarze 134: "empty argument, using 0n",
1.1 schwarze 135: "argument count wrong",
1.38 schwarze 136: "missing display type, using -ragged",
137: "list type is not the first argument",
138: "missing -width in -tag list, using 8n",
1.56 schwarze 139: "missing utility name, using \"\"",
1.38 schwarze 140: "empty head in list item",
141: "empty list item",
1.39 schwarze 142: "missing font type, using \\fR",
143: "unknown font type, using \\fR",
1.79 schwarze 144: "nothing follows prefix",
1.38 schwarze 145: "missing -std argument, adding it",
1.66 schwarze 146: "missing eqn box, using \"\"",
1.1 schwarze 147:
148: /* related to bad macro arguments */
1.42 schwarze 149: "unterminated quoted argument",
1.1 schwarze 150: "duplicate argument",
1.54 schwarze 151: "skipping duplicate argument",
1.41 schwarze 152: "skipping duplicate display type",
153: "skipping duplicate list type",
1.54 schwarze 154: "skipping -width argument",
1.1 schwarze 155: "unknown AT&T UNIX version",
1.64 schwarze 156: "comma in function argument",
1.65 schwarze 157: "parenthesis in function name",
1.45 schwarze 158: "invalid content in Rs block",
1.41 schwarze 159: "invalid Boolean argument",
160: "unknown font, skipping request",
1.1 schwarze 161:
162: /* related to plain text */
1.42 schwarze 163: "blank line in fill mode, using .sp",
164: "tab in filled text",
165: "whitespace at end of input line",
1.1 schwarze 166: "bad comment style",
1.42 schwarze 167: "invalid escape sequence",
168: "undefined string, using \"\"",
1.3 schwarze 169:
1.93 schwarze 170: /* related to tables */
171: "tbl line starts with span",
172: "tbl column starts with span",
173: "skipping vertical bar in tbl layout",
174:
1.1 schwarze 175: "generic error",
1.3 schwarze 176:
1.1 schwarze 177: /* related to tables */
1.91 schwarze 178: "non-alphabetic character in tbl options",
179: "skipping unknown tbl option",
180: "missing tbl option argument",
181: "wrong tbl option argument size",
1.93 schwarze 182: "empty tbl layout",
183: "invalid character in tbl layout",
184: "unmatched parenthesis in tbl layout",
1.94 schwarze 185: "tbl without any data cells",
186: "ignoring data in spanned tbl cell",
187: "ignoring extra tbl data cells",
188: "data block open at end of tbl",
1.1 schwarze 189:
1.46 schwarze 190: /* related to document structure and macros */
1.83 schwarze 191: NULL,
1.1 schwarze 192: "input stack limit exceeded, infinite loop?",
193: "skipping bad character",
1.46 schwarze 194: "skipping unknown macro",
1.87 schwarze 195: "skipping insecure request",
1.51 schwarze 196: "skipping item outside list",
1.46 schwarze 197: "skipping column outside column list",
198: "skipping end of block that is not open",
1.90 schwarze 199: "fewer RS blocks open, skipping",
1.46 schwarze 200: "inserting missing end of block",
201: "appending missing end of block",
202:
203: /* related to request and macro arguments */
1.1 schwarze 204: "escaped character not allowed in a name",
205: "argument count wrong",
1.75 schwarze 206: "NOT IMPLEMENTED: Bd -file",
1.49 schwarze 207: "missing list type, using -item",
1.48 schwarze 208: "missing manual name, using \"\"",
1.49 schwarze 209: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 210: "unknown standard specifier",
1.49 schwarze 211: "skipping request without numeric argument",
1.84 schwarze 212: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 213: ".so request failed",
1.39 schwarze 214: "skipping all arguments",
215: "skipping excess arguments",
1.68 schwarze 216: "divide by zero",
1.87 schwarze 217:
218: "unsupported feature",
219: "input too large",
1.89 schwarze 220: "unsupported control character",
1.87 schwarze 221: "unsupported roff request",
1.94 schwarze 222: "eqn delim option in tbl",
1.93 schwarze 223: "unsupported tbl layout modifier",
1.87 schwarze 224: "ignoring macro in table",
1.1 schwarze 225: };
226:
227: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
228: "SUCCESS",
229: "RESERVED",
230: "WARNING",
231: "ERROR",
1.87 schwarze 232: "UNSUPP",
1.1 schwarze 233: "BADARG",
234: "SYSERR"
235: };
236:
1.25 schwarze 237:
1.1 schwarze 238: static void
239: resize_buf(struct buf *buf, size_t initial)
240: {
241:
242: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
243: buf->buf = mandoc_realloc(buf->buf, buf->sz);
244: }
245:
246: static void
1.60 schwarze 247: choose_parser(struct mparse *curp)
1.1 schwarze 248: {
1.59 schwarze 249: char *cp, *ep;
250: int format;
1.1 schwarze 251:
1.59 schwarze 252: /*
253: * If neither command line arguments -mdoc or -man select
254: * a parser nor the roff parser found a .Dd or .TH macro
255: * yet, look ahead in the main input buffer.
256: */
257:
258: if ((format = roff_getformat(curp->roff)) == 0) {
259: cp = curp->primary->buf;
260: ep = cp + curp->primary->sz;
261: while (cp < ep) {
1.61 schwarze 262: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 263: cp++;
264: if (cp[0] == 'D' && cp[1] == 'd') {
265: format = MPARSE_MDOC;
266: break;
267: }
268: if (cp[0] == 'T' && cp[1] == 'H') {
269: format = MPARSE_MAN;
270: break;
271: }
272: }
273: cp = memchr(cp, '\n', ep - cp);
274: if (cp == NULL)
275: break;
276: cp++;
277: }
1.1 schwarze 278: }
279:
1.59 schwarze 280: if (format == MPARSE_MDOC) {
1.25 schwarze 281: if (NULL == curp->pmdoc)
1.22 schwarze 282: curp->pmdoc = mdoc_alloc(
283: curp->roff, curp, curp->defos,
284: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 285: assert(curp->pmdoc);
286: curp->mdoc = curp->pmdoc;
287: return;
1.25 schwarze 288: }
1.1 schwarze 289:
1.59 schwarze 290: /* Fall back to man(7) as a last resort. */
291:
1.25 schwarze 292: if (NULL == curp->pman)
1.81 schwarze 293: curp->pman = man_alloc(
294: curp->roff, curp, curp->defos,
1.22 schwarze 295: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 296: assert(curp->pman);
297: curp->man = curp->pman;
298: }
299:
300: /*
1.71 schwarze 301: * Main parse routine for a buffer.
302: * It assumes encoding and line numbering are already set up.
303: * It can recurse directly (for invocations of user-defined
304: * macros, inline equations, and input line traps)
305: * and indirectly (for .so file inclusion).
1.1 schwarze 306: */
307: static void
1.71 schwarze 308: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 309: {
310: const struct tbl_span *span;
311: struct buf ln;
1.88 schwarze 312: const char *save_file;
1.85 schwarze 313: char *cp;
1.71 schwarze 314: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 315: enum rofferr rr;
1.76 schwarze 316: int of;
1.1 schwarze 317: int lnn; /* line number in the real file */
1.85 schwarze 318: int fd;
319: pid_t save_child;
1.1 schwarze 320: unsigned char c;
321:
1.71 schwarze 322: memset(&ln, 0, sizeof(ln));
1.1 schwarze 323:
1.25 schwarze 324: lnn = curp->line;
325: pos = 0;
1.1 schwarze 326:
1.71 schwarze 327: while (i < blk.sz) {
1.1 schwarze 328: if (0 == pos && '\0' == blk.buf[i])
329: break;
330:
331: if (start) {
332: curp->line = lnn;
333: curp->reparse_count = 0;
1.70 schwarze 334:
335: if (lnn < 3 &&
336: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 337: curp->filenc & MPARSE_LATIN1)
338: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 339: }
340:
1.71 schwarze 341: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 342:
343: /*
344: * When finding an unescaped newline character,
345: * leave the character loop to process the line.
346: * Skip a preceding carriage return, if any.
347: */
348:
1.71 schwarze 349: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 350: '\n' == blk.buf[i + 1])
351: ++i;
352: if ('\n' == blk.buf[i]) {
353: ++i;
354: ++lnn;
355: break;
356: }
357:
1.13 schwarze 358: /*
1.70 schwarze 359: * Make sure we have space for the worst
360: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 361: */
362:
1.71 schwarze 363: if (pos + 11 > ln.sz)
1.13 schwarze 364: resize_buf(&ln, 256);
365:
1.25 schwarze 366: /*
1.70 schwarze 367: * Encode 8-bit input.
1.1 schwarze 368: */
369:
1.70 schwarze 370: c = blk.buf[i];
371: if (c & 0x80) {
1.71 schwarze 372: if ( ! (curp->filenc && preconv_encode(
373: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 374: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
375: curp->line, pos, "0x%x", c);
1.70 schwarze 376: ln.buf[pos++] = '?';
377: i++;
378: }
379: continue;
380: }
381:
382: /*
383: * Exclude control characters.
384: */
1.1 schwarze 385:
1.70 schwarze 386: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 387: mandoc_vmsg(c == 0x00 || c == 0x04 ||
388: c > 0x0a ? MANDOCERR_CHAR_BAD :
389: MANDOCERR_CHAR_UNSUPP,
390: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 391: i++;
1.6 schwarze 392: ln.buf[pos++] = '?';
1.1 schwarze 393: continue;
394: }
395:
396: /* Trailing backslash = a plain char. */
397:
1.71 schwarze 398: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 399: ln.buf[pos++] = blk.buf[i++];
400: continue;
401: }
402:
403: /*
404: * Found escape and at least one other character.
405: * When it's a newline character, skip it.
406: * When there is a carriage return in between,
407: * skip that one as well.
408: */
409:
1.71 schwarze 410: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 411: '\n' == blk.buf[i + 2])
412: ++i;
413: if ('\n' == blk.buf[i + 1]) {
414: i += 2;
415: ++lnn;
416: continue;
417: }
418:
419: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
420: i += 2;
421: /* Comment, skip to end of line */
1.71 schwarze 422: for (; i < blk.sz; ++i) {
1.1 schwarze 423: if ('\n' == blk.buf[i]) {
424: ++i;
425: ++lnn;
426: break;
427: }
428: }
429:
430: /* Backout trailing whitespaces */
431: for (; pos > 0; --pos) {
432: if (ln.buf[pos - 1] != ' ')
433: break;
434: if (pos > 2 && ln.buf[pos - 2] == '\\')
435: break;
436: }
437: break;
438: }
439:
1.13 schwarze 440: /* Catch escaped bogus characters. */
441:
442: c = (unsigned char) blk.buf[i+1];
443:
1.25 schwarze 444: if ( ! (isascii(c) &&
445: (isgraph(c) || isblank(c)))) {
1.89 schwarze 446: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 447: curp->line, pos, "0x%x", c);
1.13 schwarze 448: i += 2;
449: ln.buf[pos++] = '?';
450: continue;
451: }
452:
1.1 schwarze 453: /* Some other escape sequence, copy & cont. */
454:
455: ln.buf[pos++] = blk.buf[i++];
456: ln.buf[pos++] = blk.buf[i++];
457: }
458:
1.71 schwarze 459: if (pos >= ln.sz)
1.1 schwarze 460: resize_buf(&ln, 256);
461:
462: ln.buf[pos] = '\0';
463:
464: /*
465: * A significant amount of complexity is contained by
466: * the roff preprocessor. It's line-oriented but can be
467: * expressed on one line, so we need at times to
468: * readjust our starting point and re-run it. The roff
469: * preprocessor can also readjust the buffers with new
470: * data, so we pass them in wholesale.
471: */
472:
473: of = 0;
474:
1.4 schwarze 475: /*
476: * Maintain a lookaside buffer of all parsed lines. We
477: * only do this if mparse_keep() has been invoked (the
478: * buffer may be accessed with mparse_getkeep()).
479: */
480:
481: if (curp->secondary) {
1.25 schwarze 482: curp->secondary->buf = mandoc_realloc(
483: curp->secondary->buf,
484: curp->secondary->sz + pos + 2);
485: memcpy(curp->secondary->buf +
486: curp->secondary->sz,
487: ln.buf, pos);
1.4 schwarze 488: curp->secondary->sz += pos;
489: curp->secondary->buf
490: [curp->secondary->sz] = '\n';
491: curp->secondary->sz++;
492: curp->secondary->buf
493: [curp->secondary->sz] = '\0';
494: }
1.1 schwarze 495: rerun:
1.72 schwarze 496: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 497:
498: switch (rr) {
1.25 schwarze 499: case ROFF_REPARSE:
1.1 schwarze 500: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.71 schwarze 501: mparse_buf_r(curp, ln, of, 0);
1.1 schwarze 502: else
503: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 504: curp->line, pos, NULL);
1.1 schwarze 505: pos = 0;
506: continue;
1.25 schwarze 507: case ROFF_APPEND:
1.71 schwarze 508: pos = strlen(ln.buf);
1.1 schwarze 509: continue;
1.25 schwarze 510: case ROFF_RERUN:
1.1 schwarze 511: goto rerun;
1.25 schwarze 512: case ROFF_IGN:
1.1 schwarze 513: pos = 0;
514: continue;
1.25 schwarze 515: case ROFF_SO:
1.71 schwarze 516: if ( ! (curp->options & MPARSE_SO) &&
517: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 518: curp->sodest = mandoc_strdup(ln.buf + of);
519: free(ln.buf);
520: return;
521: }
1.4 schwarze 522: /*
523: * We remove `so' clauses from our lookaside
524: * buffer because we're going to descend into
525: * the file recursively.
526: */
1.25 schwarze 527: if (curp->secondary)
1.4 schwarze 528: curp->secondary->sz -= pos + 1;
1.88 schwarze 529: save_file = curp->file;
1.85 schwarze 530: save_child = curp->child;
531: if (mparse_open(curp, &fd, ln.buf + of) ==
1.88 schwarze 532: MANDOCLEVEL_OK) {
1.85 schwarze 533: mparse_readfd(curp, fd, ln.buf + of);
1.88 schwarze 534: curp->file = save_file;
535: } else {
536: curp->file = save_file;
1.30 schwarze 537: mandoc_vmsg(MANDOCERR_SO_FAIL,
538: curp, curp->line, pos,
539: ".so %s", ln.buf + of);
1.85 schwarze 540: ln.sz = mandoc_asprintf(&cp,
541: ".sp\nSee the file %s.\n.sp",
542: ln.buf + of);
543: free(ln.buf);
544: ln.buf = cp;
545: of = 0;
546: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 547: }
1.85 schwarze 548: curp->child = save_child;
1.1 schwarze 549: pos = 0;
550: continue;
551: default:
552: break;
553: }
554:
555: /*
556: * If input parsers have not been allocated, do so now.
1.2 schwarze 557: * We keep these instanced between parsers, but set them
1.1 schwarze 558: * locally per parse routine since we can use different
559: * parsers with each one.
560: */
561:
562: if ( ! (curp->man || curp->mdoc))
1.60 schwarze 563: choose_parser(curp);
1.1 schwarze 564:
1.25 schwarze 565: /*
1.60 schwarze 566: * Lastly, push down into the parsers themselves.
1.1 schwarze 567: * If libroff returns ROFF_TBL, then add it to the
568: * currently open parse. Since we only get here if
569: * there does exist data (see tbl_data.c), we're
570: * guaranteed that something's been allocated.
571: * Do the same for ROFF_EQN.
572: */
573:
1.76 schwarze 574: if (rr == ROFF_TBL) {
575: while ((span = roff_span(curp->roff)) != NULL)
576: if (curp->man == NULL)
577: mdoc_addspan(curp->mdoc, span);
578: else
579: man_addspan(curp->man, span);
580: } else if (rr == ROFF_EQN) {
581: if (curp->man == NULL)
582: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
583: else
584: man_addeqn(curp->man, roff_eqn(curp->roff));
585: } else if ((curp->man == NULL ?
586: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
587: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
588: break;
1.1 schwarze 589:
590: /* Temporary buffers typically are not full. */
591:
592: if (0 == start && '\0' == blk.buf[i])
593: break;
594:
595: /* Start the next input line. */
596:
597: pos = 0;
598: }
599:
600: free(ln.buf);
601: }
602:
603: static int
1.18 schwarze 604: read_whole_file(struct mparse *curp, const char *file, int fd,
605: struct buf *fb, int *with_mmap)
1.1 schwarze 606: {
607: struct stat st;
608: size_t off;
609: ssize_t ssz;
610:
611: if (-1 == fstat(fd, &st)) {
1.82 schwarze 612: perror(file);
613: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 614: }
615:
616: /*
617: * If we're a regular file, try just reading in the whole entry
618: * via mmap(). This is faster than reading it into blocks, and
619: * since each file is only a few bytes to begin with, I'm not
620: * concerned that this is going to tank any machines.
621: */
622:
623: if (S_ISREG(st.st_mode)) {
624: if (st.st_size >= (1U << 31)) {
1.86 schwarze 625: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 schwarze 626: return(0);
627: }
628: *with_mmap = 1;
629: fb->sz = (size_t)st.st_size;
1.15 schwarze 630: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 631: if (fb->buf != MAP_FAILED)
632: return(1);
633: }
634:
635: /*
636: * If this isn't a regular file (like, say, stdin), then we must
637: * go the old way and just read things in bit by bit.
638: */
639:
640: *with_mmap = 0;
641: off = 0;
642: fb->sz = 0;
643: fb->buf = NULL;
644: for (;;) {
645: if (off == fb->sz) {
646: if (fb->sz == (1U << 31)) {
1.86 schwarze 647: mandoc_msg(MANDOCERR_TOOLARGE, curp,
648: 0, 0, NULL);
1.1 schwarze 649: break;
650: }
651: resize_buf(fb, 65536);
652: }
653: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
654: if (ssz == 0) {
655: fb->sz = off;
656: return(1);
657: }
658: if (ssz == -1) {
1.82 schwarze 659: perror(file);
660: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 661: }
662: off += (size_t)ssz;
663: }
664:
665: free(fb->buf);
666: fb->buf = NULL;
667: return(0);
668: }
669:
670: static void
671: mparse_end(struct mparse *curp)
672: {
673:
1.50 schwarze 674: if (curp->mdoc == NULL &&
675: curp->man == NULL &&
676: curp->sodest == NULL) {
677: if (curp->options & MPARSE_MDOC)
678: curp->mdoc = curp->pmdoc;
679: else {
680: if (curp->pman == NULL)
1.81 schwarze 681: curp->pman = man_alloc(
682: curp->roff, curp, curp->defos,
1.50 schwarze 683: curp->options & MPARSE_QUICK ? 1 : 0);
684: curp->man = curp->pman;
685: }
686: }
1.86 schwarze 687: if (curp->mdoc)
688: mdoc_endparse(curp->mdoc);
689: if (curp->man)
690: man_endparse(curp->man);
1.1 schwarze 691: roff_endparse(curp->roff);
692: }
693:
1.15 schwarze 694: static void
695: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 696: {
1.61 schwarze 697: struct buf *svprimary;
1.1 schwarze 698: const char *svfile;
1.71 schwarze 699: size_t offset;
1.14 schwarze 700: static int recursion_depth;
701:
702: if (64 < recursion_depth) {
703: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 704: return;
1.14 schwarze 705: }
1.1 schwarze 706:
1.15 schwarze 707: /* Line number is per-file. */
708: svfile = curp->file;
709: curp->file = file;
1.61 schwarze 710: svprimary = curp->primary;
1.59 schwarze 711: curp->primary = &blk;
1.15 schwarze 712: curp->line = 1;
713: recursion_depth++;
714:
1.70 schwarze 715: /* Skip an UTF-8 byte order mark. */
716: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
717: (unsigned char)blk.buf[0] == 0xef &&
718: (unsigned char)blk.buf[1] == 0xbb &&
719: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 720: offset = 3;
1.70 schwarze 721: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 722: } else
723: offset = 0;
1.70 schwarze 724:
1.71 schwarze 725: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 726:
1.86 schwarze 727: if (--recursion_depth == 0)
1.15 schwarze 728: mparse_end(curp);
729:
1.61 schwarze 730: curp->primary = svprimary;
1.15 schwarze 731: curp->file = svfile;
732: }
733:
1.74 schwarze 734: /*
735: * Read the whole file into memory and call the parsers.
736: * Called recursively when an .so request is encountered.
737: */
1.15 schwarze 738: enum mandoclevel
739: mparse_readfd(struct mparse *curp, int fd, const char *file)
740: {
741: struct buf blk;
742: int with_mmap;
1.70 schwarze 743: int save_filenc;
1.1 schwarze 744:
1.67 schwarze 745: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 746: save_filenc = curp->filenc;
747: curp->filenc = curp->options &
748: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 749: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 750: curp->filenc = save_filenc;
1.67 schwarze 751: if (with_mmap)
752: munmap(blk.buf, blk.sz);
753: else
754: free(blk.buf);
755: }
1.1 schwarze 756:
1.74 schwarze 757: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 schwarze 758: perror(file);
1.67 schwarze 759:
1.74 schwarze 760: mparse_wait(curp);
1.1 schwarze 761: return(curp->file_status);
1.58 schwarze 762: }
763:
764: enum mandoclevel
1.73 schwarze 765: mparse_open(struct mparse *curp, int *fd, const char *file)
1.58 schwarze 766: {
767: int pfd[2];
1.74 schwarze 768: int save_errno;
1.58 schwarze 769: char *cp;
770:
771: curp->file = file;
1.74 schwarze 772:
773: /* Unless zipped, try to just open the file. */
774:
1.58 schwarze 775: if ((cp = strrchr(file, '.')) == NULL ||
776: strcmp(cp + 1, "gz")) {
1.73 schwarze 777: curp->child = 0;
1.74 schwarze 778: if ((*fd = open(file, O_RDONLY)) != -1)
779: return(MANDOCLEVEL_OK);
780:
781: /* Open failed; try to append ".gz". */
782:
783: mandoc_asprintf(&cp, "%s.gz", file);
784: file = cp;
785: } else
786: cp = NULL;
787:
788: /* Before forking, make sure the file can be read. */
789:
790: save_errno = errno;
791: if (access(file, R_OK) == -1) {
792: if (cp != NULL)
793: errno = save_errno;
1.83 schwarze 794: free(cp);
795: *fd = -1;
796: curp->child = 0;
797: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
798: return(MANDOCLEVEL_ERROR);
1.58 schwarze 799: }
800:
1.74 schwarze 801: /* Run gunzip(1). */
802:
1.58 schwarze 803: if (pipe(pfd) == -1) {
1.82 schwarze 804: perror("pipe");
805: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 806: }
807:
1.73 schwarze 808: switch (curp->child = fork()) {
1.58 schwarze 809: case -1:
1.82 schwarze 810: perror("fork");
811: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 812: case 0:
813: close(pfd[0]);
814: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.82 schwarze 815: perror("dup");
816: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 817: }
818: execlp("gunzip", "gunzip", "-c", file, NULL);
1.82 schwarze 819: perror("exec");
820: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 821: default:
822: close(pfd[1]);
823: *fd = pfd[0];
824: return(MANDOCLEVEL_OK);
825: }
826: }
827:
828: enum mandoclevel
1.73 schwarze 829: mparse_wait(struct mparse *curp)
1.58 schwarze 830: {
831: int status;
832:
1.73 schwarze 833: if (curp->child == 0)
834: return(MANDOCLEVEL_OK);
835:
836: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 837: perror("wait");
838: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 839: }
840: if (WIFSIGNALED(status)) {
1.83 schwarze 841: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
842: "gunzip died from signal %d", WTERMSIG(status));
843: return(MANDOCLEVEL_ERROR);
1.58 schwarze 844: }
845: if (WEXITSTATUS(status)) {
1.83 schwarze 846: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
847: "gunzip failed with code %d", WEXITSTATUS(status));
848: return(MANDOCLEVEL_ERROR);
1.58 schwarze 849: }
850: return(MANDOCLEVEL_OK);
1.1 schwarze 851: }
852:
853: struct mparse *
1.69 schwarze 854: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
855: const struct mchars *mchars, const char *defos)
1.1 schwarze 856: {
857: struct mparse *curp;
858:
859: curp = mandoc_calloc(1, sizeof(struct mparse));
860:
1.22 schwarze 861: curp->options = options;
1.1 schwarze 862: curp->wlevel = wlevel;
863: curp->mmsg = mmsg;
1.7 schwarze 864: curp->defos = defos;
1.1 schwarze 865:
1.69 schwarze 866: curp->mchars = mchars;
867: curp->roff = roff_alloc(curp, curp->mchars, options);
1.50 schwarze 868: if (curp->options & MPARSE_MDOC)
869: curp->pmdoc = mdoc_alloc(
870: curp->roff, curp, curp->defos,
871: curp->options & MPARSE_QUICK ? 1 : 0);
872: if (curp->options & MPARSE_MAN)
1.81 schwarze 873: curp->pman = man_alloc(
874: curp->roff, curp, curp->defos,
1.50 schwarze 875: curp->options & MPARSE_QUICK ? 1 : 0);
876:
1.1 schwarze 877: return(curp);
878: }
879:
880: void
881: mparse_reset(struct mparse *curp)
882: {
883:
884: roff_reset(curp->roff);
885:
886: if (curp->mdoc)
887: mdoc_reset(curp->mdoc);
888: if (curp->man)
889: man_reset(curp->man);
1.4 schwarze 890: if (curp->secondary)
891: curp->secondary->sz = 0;
1.1 schwarze 892:
893: curp->file_status = MANDOCLEVEL_OK;
894: curp->mdoc = NULL;
895: curp->man = NULL;
1.23 schwarze 896:
897: free(curp->sodest);
898: curp->sodest = NULL;
1.1 schwarze 899: }
900:
901: void
902: mparse_free(struct mparse *curp)
903: {
904:
905: if (curp->pmdoc)
906: mdoc_free(curp->pmdoc);
907: if (curp->pman)
908: man_free(curp->pman);
909: if (curp->roff)
910: roff_free(curp->roff);
1.4 schwarze 911: if (curp->secondary)
912: free(curp->secondary->buf);
1.1 schwarze 913:
1.4 schwarze 914: free(curp->secondary);
1.23 schwarze 915: free(curp->sodest);
1.1 schwarze 916: free(curp);
917: }
918:
919: void
1.23 schwarze 920: mparse_result(struct mparse *curp,
921: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 schwarze 922: {
923:
1.23 schwarze 924: if (sodest && NULL != (*sodest = curp->sodest)) {
925: *mdoc = NULL;
926: *man = NULL;
927: return;
928: }
1.1 schwarze 929: if (mdoc)
930: *mdoc = curp->mdoc;
931: if (man)
932: *man = curp->man;
933: }
934:
935: void
936: mandoc_vmsg(enum mandocerr t, struct mparse *m,
937: int ln, int pos, const char *fmt, ...)
938: {
939: char buf[256];
940: va_list ap;
941:
942: va_start(ap, fmt);
1.26 schwarze 943: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 944: va_end(ap);
945:
946: mandoc_msg(t, m, ln, pos, buf);
947: }
948:
949: void
1.25 schwarze 950: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 951: int ln, int col, const char *msg)
952: {
953: enum mandoclevel level;
954:
1.87 schwarze 955: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 956: while (er < mandoclimits[level])
957: level--;
958:
1.83 schwarze 959: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 960: return;
961:
962: if (m->mmsg)
963: (*m->mmsg)(er, level, m->file, ln, col, msg);
964:
965: if (m->file_status < level)
966: m->file_status = level;
967: }
968:
969: const char *
970: mparse_strerror(enum mandocerr er)
971: {
972:
973: return(mandocerrs[er]);
974: }
975:
976: const char *
977: mparse_strlevel(enum mandoclevel lvl)
978: {
979: return(mandoclevels[lvl]);
1.4 schwarze 980: }
981:
982: void
983: mparse_keep(struct mparse *p)
984: {
985:
986: assert(NULL == p->secondary);
987: p->secondary = mandoc_calloc(1, sizeof(struct buf));
988: }
989:
990: const char *
991: mparse_getkeep(const struct mparse *p)
992: {
993:
994: assert(p->secondary);
995: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 schwarze 996: }
![[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