Annotation of mandoc/read.c, Revision 1.117
1.117 ! schwarze 1: /* $Id: read.c,v 1.116 2015/01/26 00:57:22 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",
113: "sections out of conventional order",
1.54 schwarze 114: "duplicate section title",
115: "unexpected section",
1.87 schwarze 116: "unusual Xr order",
117: "unusual Xr punctuation",
1.86 schwarze 118: "AUTHORS section without An macro",
1.7 kristaps 119:
120: /* related to macros and nesting */
1.55 schwarze 121: "obsolete macro",
1.102 schwarze 122: "macro neither callable nor escaped",
1.7 kristaps 123: "skipping paragraph macro",
1.31 schwarze 124: "moving paragraph macro out of list",
1.7 kristaps 125: "skipping no-space macro",
126: "blocks badly nested",
127: "nested displays are not portable",
1.57 schwarze 128: "moving content out of list",
129: ".Vt block has child macro",
1.78 schwarze 130: "fill mode already enabled, skipping",
131: "fill mode already disabled, skipping",
1.7 kristaps 132: "line scope broken",
133:
134: /* related to missing macro arguments */
1.58 schwarze 135: "skipping empty request",
136: "conditional request controls empty scope",
1.7 kristaps 137: "skipping empty macro",
1.62 schwarze 138: "empty argument, using 0n",
1.7 kristaps 139: "argument count wrong",
1.60 schwarze 140: "missing display type, using -ragged",
141: "list type is not the first argument",
142: "missing -width in -tag list, using 8n",
1.78 schwarze 143: "missing utility name, using \"\"",
1.60 schwarze 144: "empty head in list item",
145: "empty list item",
1.61 schwarze 146: "missing font type, using \\fR",
147: "unknown font type, using \\fR",
1.103 schwarze 148: "nothing follows prefix",
1.60 schwarze 149: "missing -std argument, adding it",
1.90 schwarze 150: "missing eqn box, using \"\"",
1.7 kristaps 151:
152: /* related to bad macro arguments */
1.64 schwarze 153: "unterminated quoted argument",
1.7 kristaps 154: "duplicate argument",
1.76 schwarze 155: "skipping duplicate argument",
1.63 schwarze 156: "skipping duplicate display type",
157: "skipping duplicate list type",
1.76 schwarze 158: "skipping -width argument",
1.7 kristaps 159: "unknown AT&T UNIX version",
1.88 schwarze 160: "comma in function argument",
1.89 schwarze 161: "parenthesis in function name",
1.67 schwarze 162: "invalid content in Rs block",
1.63 schwarze 163: "invalid Boolean argument",
164: "unknown font, skipping request",
1.7 kristaps 165:
166: /* related to plain text */
1.64 schwarze 167: "blank line in fill mode, using .sp",
168: "tab in filled text",
169: "whitespace at end of input line",
1.7 kristaps 170: "bad comment style",
1.64 schwarze 171: "invalid escape sequence",
172: "undefined string, using \"\"",
1.16 kristaps 173:
1.7 kristaps 174: "generic error",
1.17 kristaps 175:
176: /* related to equations */
1.20 kristaps 177: "unexpected equation scope closure",
178: "equation scope open on exit",
1.21 kristaps 179: "overlapping equation scopes",
180: "unexpected end of equation",
1.7 kristaps 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",
1.7 kristaps 187: "no table layout cells specified",
188: "no table data cells specified",
189: "ignore data in cell",
190: "data block still open",
191: "ignoring extra data cells",
192:
1.68 schwarze 193: /* related to document structure and macros */
1.108 schwarze 194: NULL,
1.7 kristaps 195: "input stack limit exceeded, infinite loop?",
196: "skipping bad character",
1.68 schwarze 197: "skipping unknown macro",
1.112 schwarze 198: "skipping insecure request",
1.73 schwarze 199: "skipping item outside list",
1.68 schwarze 200: "skipping column outside column list",
201: "skipping end of block that is not open",
1.115 schwarze 202: "fewer RS blocks open, skipping",
1.68 schwarze 203: "inserting missing end of block",
204: "appending missing end of block",
205:
206: /* related to request and macro arguments */
1.7 kristaps 207: "escaped character not allowed in a name",
208: "argument count wrong",
1.99 schwarze 209: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 210: "missing list type, using -item",
1.70 schwarze 211: "missing manual name, using \"\"",
1.71 schwarze 212: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 213: "unknown standard specifier",
1.71 schwarze 214: "skipping request without numeric argument",
1.109 schwarze 215: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 216: ".so request failed",
1.61 schwarze 217: "skipping all arguments",
218: "skipping excess arguments",
1.92 kristaps 219: "divide by zero",
1.112 schwarze 220:
221: "unsupported feature",
222: "input too large",
1.114 schwarze 223: "unsupported control character",
1.112 schwarze 224: "unsupported roff request",
225: "unsupported table layout",
226: "ignoring macro in table",
1.117 ! schwarze 227: "eqn in tbl",
1.7 kristaps 228: };
229:
230: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
231: "SUCCESS",
232: "RESERVED",
233: "WARNING",
234: "ERROR",
1.112 schwarze 235: "UNSUPP",
1.7 kristaps 236: "BADARG",
237: "SYSERR"
238: };
239:
1.47 schwarze 240:
1.1 kristaps 241: static void
242: resize_buf(struct buf *buf, size_t initial)
243: {
244:
245: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
246: buf->buf = mandoc_realloc(buf->buf, buf->sz);
247: }
248:
249: static void
1.84 schwarze 250: choose_parser(struct mparse *curp)
1.1 kristaps 251: {
1.83 schwarze 252: char *cp, *ep;
253: int format;
1.1 kristaps 254:
1.83 schwarze 255: /*
256: * If neither command line arguments -mdoc or -man select
257: * a parser nor the roff parser found a .Dd or .TH macro
258: * yet, look ahead in the main input buffer.
259: */
260:
261: if ((format = roff_getformat(curp->roff)) == 0) {
262: cp = curp->primary->buf;
263: ep = cp + curp->primary->sz;
264: while (cp < ep) {
1.85 schwarze 265: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 266: cp++;
267: if (cp[0] == 'D' && cp[1] == 'd') {
268: format = MPARSE_MDOC;
269: break;
270: }
271: if (cp[0] == 'T' && cp[1] == 'H') {
272: format = MPARSE_MAN;
273: break;
274: }
275: }
276: cp = memchr(cp, '\n', ep - cp);
277: if (cp == NULL)
278: break;
279: cp++;
280: }
1.1 kristaps 281: }
282:
1.83 schwarze 283: if (format == MPARSE_MDOC) {
1.47 schwarze 284: if (NULL == curp->pmdoc)
1.44 schwarze 285: curp->pmdoc = mdoc_alloc(
286: curp->roff, curp, curp->defos,
287: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 288: assert(curp->pmdoc);
289: curp->mdoc = curp->pmdoc;
290: return;
1.47 schwarze 291: }
1.1 kristaps 292:
1.83 schwarze 293: /* Fall back to man(7) as a last resort. */
294:
1.47 schwarze 295: if (NULL == curp->pman)
1.106 schwarze 296: curp->pman = man_alloc(
297: curp->roff, curp, curp->defos,
1.44 schwarze 298: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 299: assert(curp->pman);
300: curp->man = curp->pman;
301: }
302:
303: /*
1.95 schwarze 304: * Main parse routine for a buffer.
305: * It assumes encoding and line numbering are already set up.
306: * It can recurse directly (for invocations of user-defined
307: * macros, inline equations, and input line traps)
308: * and indirectly (for .so file inclusion).
1.1 kristaps 309: */
310: static void
1.95 schwarze 311: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 312: {
313: const struct tbl_span *span;
314: struct buf ln;
1.113 schwarze 315: const char *save_file;
1.110 schwarze 316: char *cp;
1.95 schwarze 317: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 318: enum rofferr rr;
1.100 schwarze 319: int of;
1.1 kristaps 320: int lnn; /* line number in the real file */
1.110 schwarze 321: int fd;
322: pid_t save_child;
1.1 kristaps 323: unsigned char c;
324:
1.95 schwarze 325: memset(&ln, 0, sizeof(ln));
1.1 kristaps 326:
1.47 schwarze 327: lnn = curp->line;
328: pos = 0;
1.1 kristaps 329:
1.95 schwarze 330: while (i < blk.sz) {
1.1 kristaps 331: if (0 == pos && '\0' == blk.buf[i])
332: break;
333:
334: if (start) {
335: curp->line = lnn;
336: curp->reparse_count = 0;
1.93 schwarze 337:
338: if (lnn < 3 &&
339: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 340: curp->filenc & MPARSE_LATIN1)
341: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 342: }
343:
1.95 schwarze 344: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 345:
346: /*
347: * When finding an unescaped newline character,
348: * leave the character loop to process the line.
349: * Skip a preceding carriage return, if any.
350: */
351:
1.95 schwarze 352: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 353: '\n' == blk.buf[i + 1])
354: ++i;
355: if ('\n' == blk.buf[i]) {
356: ++i;
357: ++lnn;
358: break;
359: }
360:
1.35 schwarze 361: /*
1.93 schwarze 362: * Make sure we have space for the worst
363: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 364: */
365:
1.95 schwarze 366: if (pos + 11 > ln.sz)
1.35 schwarze 367: resize_buf(&ln, 256);
368:
1.47 schwarze 369: /*
1.93 schwarze 370: * Encode 8-bit input.
1.1 kristaps 371: */
372:
1.93 schwarze 373: c = blk.buf[i];
374: if (c & 0x80) {
1.95 schwarze 375: if ( ! (curp->filenc && preconv_encode(
376: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 377: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
378: curp->line, pos, "0x%x", c);
1.93 schwarze 379: ln.buf[pos++] = '?';
380: i++;
381: }
382: continue;
383: }
384:
385: /*
386: * Exclude control characters.
387: */
1.1 kristaps 388:
1.93 schwarze 389: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 390: mandoc_vmsg(c == 0x00 || c == 0x04 ||
391: c > 0x0a ? MANDOCERR_CHAR_BAD :
392: MANDOCERR_CHAR_UNSUPP,
393: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 394: i++;
1.27 joerg 395: ln.buf[pos++] = '?';
1.1 kristaps 396: continue;
397: }
398:
399: /* Trailing backslash = a plain char. */
400:
1.95 schwarze 401: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 402: ln.buf[pos++] = blk.buf[i++];
403: continue;
404: }
405:
406: /*
407: * Found escape and at least one other character.
408: * When it's a newline character, skip it.
409: * When there is a carriage return in between,
410: * skip that one as well.
411: */
412:
1.95 schwarze 413: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 414: '\n' == blk.buf[i + 2])
415: ++i;
416: if ('\n' == blk.buf[i + 1]) {
417: i += 2;
418: ++lnn;
419: continue;
420: }
421:
1.13 kristaps 422: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 423: i += 2;
424: /* Comment, skip to end of line */
1.95 schwarze 425: for (; i < blk.sz; ++i) {
1.1 kristaps 426: if ('\n' == blk.buf[i]) {
427: ++i;
428: ++lnn;
429: break;
430: }
431: }
432:
433: /* Backout trailing whitespaces */
434: for (; pos > 0; --pos) {
435: if (ln.buf[pos - 1] != ' ')
436: break;
437: if (pos > 2 && ln.buf[pos - 2] == '\\')
438: break;
439: }
440: break;
441: }
442:
1.35 schwarze 443: /* Catch escaped bogus characters. */
444:
445: c = (unsigned char) blk.buf[i+1];
446:
1.47 schwarze 447: if ( ! (isascii(c) &&
448: (isgraph(c) || isblank(c)))) {
1.114 schwarze 449: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 450: curp->line, pos, "0x%x", c);
1.35 schwarze 451: i += 2;
452: ln.buf[pos++] = '?';
453: continue;
454: }
455:
1.1 kristaps 456: /* Some other escape sequence, copy & cont. */
457:
458: ln.buf[pos++] = blk.buf[i++];
459: ln.buf[pos++] = blk.buf[i++];
460: }
461:
1.95 schwarze 462: if (pos >= ln.sz)
1.1 kristaps 463: resize_buf(&ln, 256);
464:
465: ln.buf[pos] = '\0';
466:
467: /*
468: * A significant amount of complexity is contained by
469: * the roff preprocessor. It's line-oriented but can be
470: * expressed on one line, so we need at times to
471: * readjust our starting point and re-run it. The roff
472: * preprocessor can also readjust the buffers with new
473: * data, so we pass them in wholesale.
474: */
475:
476: of = 0;
477:
1.24 kristaps 478: /*
479: * Maintain a lookaside buffer of all parsed lines. We
480: * only do this if mparse_keep() has been invoked (the
481: * buffer may be accessed with mparse_getkeep()).
482: */
483:
484: if (curp->secondary) {
1.47 schwarze 485: curp->secondary->buf = mandoc_realloc(
486: curp->secondary->buf,
487: curp->secondary->sz + pos + 2);
488: memcpy(curp->secondary->buf +
489: curp->secondary->sz,
490: ln.buf, pos);
1.24 kristaps 491: curp->secondary->sz += pos;
492: curp->secondary->buf
493: [curp->secondary->sz] = '\n';
494: curp->secondary->sz++;
495: curp->secondary->buf
496: [curp->secondary->sz] = '\0';
497: }
1.1 kristaps 498: rerun:
1.96 schwarze 499: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 500:
501: switch (rr) {
1.47 schwarze 502: case ROFF_REPARSE:
1.1 kristaps 503: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 504: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 505: else
1.3 kristaps 506: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 507: curp->line, pos, NULL);
1.1 kristaps 508: pos = 0;
509: continue;
1.47 schwarze 510: case ROFF_APPEND:
1.95 schwarze 511: pos = strlen(ln.buf);
1.1 kristaps 512: continue;
1.47 schwarze 513: case ROFF_RERUN:
1.1 kristaps 514: goto rerun;
1.47 schwarze 515: case ROFF_IGN:
1.1 kristaps 516: pos = 0;
517: continue;
1.47 schwarze 518: case ROFF_SO:
1.95 schwarze 519: if ( ! (curp->options & MPARSE_SO) &&
520: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 521: curp->sodest = mandoc_strdup(ln.buf + of);
522: free(ln.buf);
523: return;
524: }
1.24 kristaps 525: /*
526: * We remove `so' clauses from our lookaside
527: * buffer because we're going to descend into
528: * the file recursively.
529: */
1.47 schwarze 530: if (curp->secondary)
1.25 kristaps 531: curp->secondary->sz -= pos + 1;
1.113 schwarze 532: save_file = curp->file;
1.110 schwarze 533: save_child = curp->child;
534: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 535: MANDOCLEVEL_OK) {
1.110 schwarze 536: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 537: curp->file = save_file;
538: } else {
539: curp->file = save_file;
1.52 schwarze 540: mandoc_vmsg(MANDOCERR_SO_FAIL,
541: curp, curp->line, pos,
542: ".so %s", ln.buf + of);
1.110 schwarze 543: ln.sz = mandoc_asprintf(&cp,
544: ".sp\nSee the file %s.\n.sp",
545: ln.buf + of);
546: free(ln.buf);
547: ln.buf = cp;
548: of = 0;
549: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 550: }
1.110 schwarze 551: curp->child = save_child;
1.1 kristaps 552: pos = 0;
553: continue;
554: default:
555: break;
556: }
557:
558: /*
559: * If input parsers have not been allocated, do so now.
1.14 kristaps 560: * We keep these instanced between parsers, but set them
1.1 kristaps 561: * locally per parse routine since we can use different
562: * parsers with each one.
563: */
564:
565: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 566: choose_parser(curp);
1.1 kristaps 567:
1.47 schwarze 568: /*
1.84 schwarze 569: * Lastly, push down into the parsers themselves.
1.1 kristaps 570: * If libroff returns ROFF_TBL, then add it to the
571: * currently open parse. Since we only get here if
572: * there does exist data (see tbl_data.c), we're
573: * guaranteed that something's been allocated.
574: * Do the same for ROFF_EQN.
575: */
576:
1.100 schwarze 577: if (rr == ROFF_TBL) {
578: while ((span = roff_span(curp->roff)) != NULL)
579: if (curp->man == NULL)
580: mdoc_addspan(curp->mdoc, span);
581: else
582: man_addspan(curp->man, span);
583: } else if (rr == ROFF_EQN) {
584: if (curp->man == NULL)
585: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
586: else
587: man_addeqn(curp->man, roff_eqn(curp->roff));
588: } else if ((curp->man == NULL ?
589: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
590: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
591: break;
1.1 kristaps 592:
593: /* Temporary buffers typically are not full. */
594:
595: if (0 == start && '\0' == blk.buf[i])
596: break;
597:
598: /* Start the next input line. */
599:
600: pos = 0;
601: }
602:
603: free(ln.buf);
604: }
605:
606: static int
1.40 schwarze 607: read_whole_file(struct mparse *curp, const char *file, int fd,
608: struct buf *fb, int *with_mmap)
1.1 kristaps 609: {
610: size_t off;
611: ssize_t ssz;
612:
1.81 schwarze 613: #if HAVE_MMAP
1.15 kristaps 614: struct stat st;
1.1 kristaps 615: if (-1 == fstat(fd, &st)) {
1.107 schwarze 616: perror(file);
617: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 618: }
619:
620: /*
621: * If we're a regular file, try just reading in the whole entry
622: * via mmap(). This is faster than reading it into blocks, and
623: * since each file is only a few bytes to begin with, I'm not
624: * concerned that this is going to tank any machines.
625: */
626:
627: if (S_ISREG(st.st_mode)) {
628: if (st.st_size >= (1U << 31)) {
1.111 schwarze 629: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 630: return(0);
631: }
632: *with_mmap = 1;
633: fb->sz = (size_t)st.st_size;
1.37 schwarze 634: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 635: if (fb->buf != MAP_FAILED)
636: return(1);
637: }
1.15 kristaps 638: #endif
1.1 kristaps 639:
640: /*
641: * If this isn't a regular file (like, say, stdin), then we must
642: * go the old way and just read things in bit by bit.
643: */
644:
645: *with_mmap = 0;
646: off = 0;
647: fb->sz = 0;
648: fb->buf = NULL;
649: for (;;) {
650: if (off == fb->sz) {
651: if (fb->sz == (1U << 31)) {
1.111 schwarze 652: mandoc_msg(MANDOCERR_TOOLARGE, curp,
653: 0, 0, NULL);
1.1 kristaps 654: break;
655: }
656: resize_buf(fb, 65536);
657: }
658: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
659: if (ssz == 0) {
660: fb->sz = off;
661: return(1);
662: }
663: if (ssz == -1) {
1.107 schwarze 664: perror(file);
665: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 666: }
667: off += (size_t)ssz;
668: }
669:
670: free(fb->buf);
671: fb->buf = NULL;
672: return(0);
673: }
674:
675: static void
676: mparse_end(struct mparse *curp)
677: {
678:
1.72 schwarze 679: if (curp->mdoc == NULL &&
680: curp->man == NULL &&
681: curp->sodest == NULL) {
682: if (curp->options & MPARSE_MDOC)
683: curp->mdoc = curp->pmdoc;
684: else {
685: if (curp->pman == NULL)
1.106 schwarze 686: curp->pman = man_alloc(
687: curp->roff, curp, curp->defos,
1.72 schwarze 688: curp->options & MPARSE_QUICK ? 1 : 0);
689: curp->man = curp->pman;
690: }
691: }
1.111 schwarze 692: if (curp->mdoc)
693: mdoc_endparse(curp->mdoc);
694: if (curp->man)
695: man_endparse(curp->man);
1.1 kristaps 696: roff_endparse(curp->roff);
697: }
698:
699: static void
1.36 schwarze 700: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 701: {
1.85 schwarze 702: struct buf *svprimary;
1.28 joerg 703: const char *svfile;
1.95 schwarze 704: size_t offset;
1.36 schwarze 705: static int recursion_depth;
706:
707: if (64 < recursion_depth) {
708: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
709: return;
710: }
1.28 joerg 711:
712: /* Line number is per-file. */
713: svfile = curp->file;
714: curp->file = file;
1.85 schwarze 715: svprimary = curp->primary;
1.83 schwarze 716: curp->primary = &blk;
1.28 joerg 717: curp->line = 1;
1.36 schwarze 718: recursion_depth++;
1.28 joerg 719:
1.93 schwarze 720: /* Skip an UTF-8 byte order mark. */
721: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
722: (unsigned char)blk.buf[0] == 0xef &&
723: (unsigned char)blk.buf[1] == 0xbb &&
724: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 725: offset = 3;
1.93 schwarze 726: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 727: } else
728: offset = 0;
1.93 schwarze 729:
1.95 schwarze 730: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 731:
1.111 schwarze 732: if (--recursion_depth == 0)
1.28 joerg 733: mparse_end(curp);
734:
1.85 schwarze 735: curp->primary = svprimary;
1.28 joerg 736: curp->file = svfile;
737: }
738:
739: enum mandoclevel
1.104 schwarze 740: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 741: const char *file)
742: {
743: struct buf blk;
744:
1.104 schwarze 745: blk.buf = buf;
1.28 joerg 746: blk.sz = len;
747:
1.36 schwarze 748: mparse_parse_buffer(curp, blk, file);
1.28 joerg 749: return(curp->file_status);
750: }
751:
1.98 schwarze 752: /*
753: * Read the whole file into memory and call the parsers.
754: * Called recursively when an .so request is encountered.
755: */
1.36 schwarze 756: enum mandoclevel
757: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 758: {
1.28 joerg 759: struct buf blk;
760: int with_mmap;
1.93 schwarze 761: int save_filenc;
1.1 kristaps 762:
1.91 schwarze 763: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 764: save_filenc = curp->filenc;
765: curp->filenc = curp->options &
766: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 767: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 768: curp->filenc = save_filenc;
1.81 schwarze 769: #if HAVE_MMAP
1.91 schwarze 770: if (with_mmap)
771: munmap(blk.buf, blk.sz);
772: else
1.28 joerg 773: #endif
1.91 schwarze 774: free(blk.buf);
775: }
1.1 kristaps 776:
1.98 schwarze 777: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 778: perror(file);
1.91 schwarze 779:
1.98 schwarze 780: mparse_wait(curp);
1.1 kristaps 781: return(curp->file_status);
1.82 schwarze 782: }
783:
784: enum mandoclevel
1.97 schwarze 785: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 786: {
787: int pfd[2];
1.98 schwarze 788: int save_errno;
1.82 schwarze 789: char *cp;
790:
791: curp->file = file;
1.98 schwarze 792:
793: /* Unless zipped, try to just open the file. */
794:
1.82 schwarze 795: if ((cp = strrchr(file, '.')) == NULL ||
796: strcmp(cp + 1, "gz")) {
1.97 schwarze 797: curp->child = 0;
1.98 schwarze 798: if ((*fd = open(file, O_RDONLY)) != -1)
799: return(MANDOCLEVEL_OK);
800:
801: /* Open failed; try to append ".gz". */
802:
803: mandoc_asprintf(&cp, "%s.gz", file);
804: file = cp;
805: } else
806: cp = NULL;
807:
808: /* Before forking, make sure the file can be read. */
809:
810: save_errno = errno;
811: if (access(file, R_OK) == -1) {
812: if (cp != NULL)
813: errno = save_errno;
1.108 schwarze 814: free(cp);
815: *fd = -1;
816: curp->child = 0;
817: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
818: return(MANDOCLEVEL_ERROR);
1.82 schwarze 819: }
820:
1.98 schwarze 821: /* Run gunzip(1). */
822:
1.82 schwarze 823: if (pipe(pfd) == -1) {
1.107 schwarze 824: perror("pipe");
825: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 826: }
827:
1.97 schwarze 828: switch (curp->child = fork()) {
1.82 schwarze 829: case -1:
1.107 schwarze 830: perror("fork");
831: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 832: case 0:
833: close(pfd[0]);
834: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 835: perror("dup");
836: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 837: }
838: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 839: perror("exec");
840: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 841: default:
842: close(pfd[1]);
843: *fd = pfd[0];
844: return(MANDOCLEVEL_OK);
845: }
846: }
847:
848: enum mandoclevel
1.97 schwarze 849: mparse_wait(struct mparse *curp)
1.82 schwarze 850: {
851: int status;
852:
1.97 schwarze 853: if (curp->child == 0)
854: return(MANDOCLEVEL_OK);
855:
856: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 857: perror("wait");
858: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 859: }
860: if (WIFSIGNALED(status)) {
1.108 schwarze 861: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
862: "gunzip died from signal %d", WTERMSIG(status));
863: return(MANDOCLEVEL_ERROR);
1.82 schwarze 864: }
865: if (WEXITSTATUS(status)) {
1.108 schwarze 866: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
867: "gunzip failed with code %d", WEXITSTATUS(status));
868: return(MANDOCLEVEL_ERROR);
1.82 schwarze 869: }
870: return(MANDOCLEVEL_OK);
1.1 kristaps 871: }
872:
873: struct mparse *
1.94 schwarze 874: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
875: const struct mchars *mchars, const char *defos)
1.1 kristaps 876: {
877: struct mparse *curp;
1.10 kristaps 878:
1.1 kristaps 879: curp = mandoc_calloc(1, sizeof(struct mparse));
880:
1.44 schwarze 881: curp->options = options;
1.3 kristaps 882: curp->wlevel = wlevel;
1.1 kristaps 883: curp->mmsg = mmsg;
1.29 schwarze 884: curp->defos = defos;
1.1 kristaps 885:
1.94 schwarze 886: curp->mchars = mchars;
887: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 888: if (curp->options & MPARSE_MDOC)
889: curp->pmdoc = mdoc_alloc(
890: curp->roff, curp, curp->defos,
891: curp->options & MPARSE_QUICK ? 1 : 0);
892: if (curp->options & MPARSE_MAN)
1.106 schwarze 893: curp->pman = man_alloc(
894: curp->roff, curp, curp->defos,
1.72 schwarze 895: curp->options & MPARSE_QUICK ? 1 : 0);
896:
1.1 kristaps 897: return(curp);
898: }
899:
900: void
901: mparse_reset(struct mparse *curp)
902: {
903:
904: roff_reset(curp->roff);
905:
906: if (curp->mdoc)
907: mdoc_reset(curp->mdoc);
908: if (curp->man)
909: man_reset(curp->man);
1.24 kristaps 910: if (curp->secondary)
911: curp->secondary->sz = 0;
1.1 kristaps 912:
913: curp->file_status = MANDOCLEVEL_OK;
914: curp->mdoc = NULL;
915: curp->man = NULL;
1.45 schwarze 916:
917: free(curp->sodest);
918: curp->sodest = NULL;
1.1 kristaps 919: }
920:
921: void
922: mparse_free(struct mparse *curp)
923: {
924:
925: if (curp->pmdoc)
926: mdoc_free(curp->pmdoc);
927: if (curp->pman)
928: man_free(curp->pman);
929: if (curp->roff)
930: roff_free(curp->roff);
1.24 kristaps 931: if (curp->secondary)
932: free(curp->secondary->buf);
1.1 kristaps 933:
1.24 kristaps 934: free(curp->secondary);
1.45 schwarze 935: free(curp->sodest);
1.1 kristaps 936: free(curp);
937: }
938:
939: void
1.45 schwarze 940: mparse_result(struct mparse *curp,
941: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 942: {
943:
1.45 schwarze 944: if (sodest && NULL != (*sodest = curp->sodest)) {
945: *mdoc = NULL;
946: *man = NULL;
947: return;
948: }
1.9 kristaps 949: if (mdoc)
950: *mdoc = curp->mdoc;
951: if (man)
952: *man = curp->man;
1.3 kristaps 953: }
954:
955: void
956: mandoc_vmsg(enum mandocerr t, struct mparse *m,
957: int ln, int pos, const char *fmt, ...)
958: {
959: char buf[256];
960: va_list ap;
961:
962: va_start(ap, fmt);
1.48 schwarze 963: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 964: va_end(ap);
965:
966: mandoc_msg(t, m, ln, pos, buf);
967: }
968:
969: void
1.47 schwarze 970: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 971: int ln, int col, const char *msg)
972: {
973: enum mandoclevel level;
974:
1.112 schwarze 975: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 976: while (er < mandoclimits[level])
977: level--;
978:
1.108 schwarze 979: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 980: return;
981:
1.8 kristaps 982: if (m->mmsg)
983: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 984:
985: if (m->file_status < level)
986: m->file_status = level;
1.7 kristaps 987: }
988:
989: const char *
990: mparse_strerror(enum mandocerr er)
991: {
992:
993: return(mandocerrs[er]);
994: }
995:
996: const char *
997: mparse_strlevel(enum mandoclevel lvl)
998: {
999: return(mandoclevels[lvl]);
1.24 kristaps 1000: }
1001:
1002: void
1003: mparse_keep(struct mparse *p)
1004: {
1005:
1006: assert(NULL == p->secondary);
1007: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1008: }
1009:
1010: const char *
1011: mparse_getkeep(const struct mparse *p)
1012: {
1013:
1014: assert(p->secondary);
1015: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1016: }
CVSweb