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