Annotation of mandoc/read.c, Revision 1.202
1.202 ! schwarze 1: /* $Id: read.c,v 1.201 2018/12/13 06:18:20 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.196 schwarze 4: * Copyright (c) 2010-2018 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: *
1.133 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.133 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 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>
1.82 schwarze 22: #include <sys/mman.h>
1.80 schwarze 23: #include <sys/stat.h>
1.1 kristaps 24:
25: #include <assert.h>
26: #include <ctype.h>
1.40 schwarze 27: #include <errno.h>
1.1 kristaps 28: #include <fcntl.h>
1.3 kristaps 29: #include <stdarg.h>
1.1 kristaps 30: #include <stdio.h>
31: #include <stdlib.h>
32: #include <string.h>
33: #include <unistd.h>
1.140 schwarze 34: #include <zlib.h>
1.1 kristaps 35:
1.133 schwarze 36: #include "mandoc_aux.h"
1.1 kristaps 37: #include "mandoc.h"
1.133 schwarze 38: #include "roff.h"
1.1 kristaps 39: #include "mdoc.h"
40: #include "man.h"
1.202 ! schwarze 41: #include "mandoc_parse.h"
1.133 schwarze 42: #include "libmandoc.h"
1.201 schwarze 43: #include "roff_int.h"
1.1 kristaps 44:
45: #define REPARSE_LIMIT 1000
46:
47: struct mparse {
1.160 schwarze 48: struct roff *roff; /* roff parser (!NULL) */
1.134 schwarze 49: struct roff_man *man; /* man parser */
1.45 schwarze 50: char *sodest; /* filename pointed to by .so */
1.83 schwarze 51: const char *file; /* filename of current input file */
52: struct buf *primary; /* buffer currently being parsed */
1.198 schwarze 53: struct buf *secondary; /* copy of top level input */
1.199 schwarze 54: struct buf *loop; /* open .while request line */
1.179 schwarze 55: const char *os_s; /* default operating system */
1.83 schwarze 56: mandocmsg mmsg; /* warning/error message handler */
57: enum mandoclevel file_status; /* status of current parse */
1.179 schwarze 58: enum mandocerr mmin; /* ignore messages below this */
1.83 schwarze 59: int options; /* parser options */
1.140 schwarze 60: int gzip; /* current input file is gzipped */
1.93 schwarze 61: int filenc; /* encoding of the current file */
1.1 kristaps 62: int reparse_count; /* finite interp. stack */
1.83 schwarze 63: int line; /* line number in the file */
1.1 kristaps 64: };
65:
1.84 schwarze 66: static void choose_parser(struct mparse *);
1.198 schwarze 67: static void free_buf_list(struct buf *);
1.1 kristaps 68: static void resize_buf(struct buf *, size_t);
1.199 schwarze 69: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 70: static int read_whole_file(struct mparse *, const char *, int,
71: struct buf *, int *);
1.1 kristaps 72: static void mparse_end(struct mparse *);
1.37 schwarze 73: static void mparse_parse_buffer(struct mparse *, struct buf,
74: const char *);
1.1 kristaps 75:
1.3 kristaps 76: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
77: MANDOCERR_OK,
1.188 schwarze 78: MANDOCERR_OK,
1.3 kristaps 79: MANDOCERR_WARNING,
80: MANDOCERR_ERROR,
1.112 schwarze 81: MANDOCERR_UNSUPP,
1.3 kristaps 82: MANDOCERR_MAX,
83: MANDOCERR_MAX
84: };
85:
1.7 kristaps 86: static const char * const mandocerrs[MANDOCERR_MAX] = {
87: "ok",
88:
1.179 schwarze 89: "base system convention",
1.166 schwarze 90:
1.176 schwarze 91: "Mdocdate found",
92: "Mdocdate missing",
1.181 schwarze 93: "unknown architecture",
1.180 schwarze 94: "operating system explicitly specified",
1.179 schwarze 95: "RCS id missing",
1.184 schwarze 96: "referenced manual not found",
1.179 schwarze 97:
98: "generic style suggestion",
99:
1.176 schwarze 100: "legacy man(7) date format",
1.196 schwarze 101: "normalizing date format to",
1.189 schwarze 102: "lower case character in document title",
1.178 schwarze 103: "duplicate RCS id",
1.193 schwarze 104: "possible typo in section name",
1.189 schwarze 105: "unterminated quoted argument",
1.166 schwarze 106: "useless macro",
1.167 schwarze 107: "consider using OS macro",
1.172 schwarze 108: "errnos out of order",
109: "duplicate errno",
1.187 schwarze 110: "trailing delimiter",
1.174 schwarze 111: "no blank before trailing delimiter",
1.189 schwarze 112: "fill mode already enabled, skipping",
113: "fill mode already disabled, skipping",
1.195 schwarze 114: "verbatim \"--\", maybe consider using \\(em",
1.175 schwarze 115: "function name without markup",
1.189 schwarze 116: "whitespace at end of input line",
117: "bad comment style",
1.166 schwarze 118:
1.7 kristaps 119: "generic warning",
120:
121: /* related to the prologue */
1.79 schwarze 122: "missing manual title, using UNTITLED",
123: "missing manual title, using \"\"",
124: "missing manual section, using \"\"",
1.7 kristaps 125: "unknown manual section",
1.54 schwarze 126: "missing date, using today's date",
1.7 kristaps 127: "cannot parse date, using it verbatim",
1.186 schwarze 128: "date in the future, using it anyway",
1.79 schwarze 129: "missing Os macro, using \"\"",
130: "late prologue macro",
1.7 kristaps 131: "prologue macros out of order",
132:
133: /* related to document structure */
134: ".so is fragile, better use ln(1)",
1.50 schwarze 135: "no document body",
1.54 schwarze 136: "content before first section header",
137: "first section is not \"NAME\"",
1.156 schwarze 138: "NAME section without Nm before Nd",
1.128 schwarze 139: "NAME section without description",
140: "description not at the end of NAME",
141: "bad NAME section content",
1.156 schwarze 142: "missing comma before name",
1.121 schwarze 143: "missing description line, using \"\"",
1.162 schwarze 144: "description line outside NAME section",
1.7 kristaps 145: "sections out of conventional order",
1.54 schwarze 146: "duplicate section title",
147: "unexpected section",
1.185 schwarze 148: "cross reference to self",
1.87 schwarze 149: "unusual Xr order",
150: "unusual Xr punctuation",
1.86 schwarze 151: "AUTHORS section without An macro",
1.7 kristaps 152:
153: /* related to macros and nesting */
1.55 schwarze 154: "obsolete macro",
1.102 schwarze 155: "macro neither callable nor escaped",
1.7 kristaps 156: "skipping paragraph macro",
1.31 schwarze 157: "moving paragraph macro out of list",
1.7 kristaps 158: "skipping no-space macro",
159: "blocks badly nested",
160: "nested displays are not portable",
1.57 schwarze 161: "moving content out of list",
1.183 schwarze 162: "first macro on line",
1.7 kristaps 163: "line scope broken",
1.169 schwarze 164: "skipping blank line in line scope",
1.7 kristaps 165:
166: /* related to missing macro arguments */
1.58 schwarze 167: "skipping empty request",
168: "conditional request controls empty scope",
1.7 kristaps 169: "skipping empty macro",
1.124 schwarze 170: "empty block",
1.62 schwarze 171: "empty argument, using 0n",
1.60 schwarze 172: "missing display type, using -ragged",
173: "list type is not the first argument",
1.152 schwarze 174: "missing -width in -tag list, using 6n",
1.78 schwarze 175: "missing utility name, using \"\"",
1.123 schwarze 176: "missing function name, using \"\"",
1.60 schwarze 177: "empty head in list item",
178: "empty list item",
1.183 schwarze 179: "missing argument, using next line",
1.61 schwarze 180: "missing font type, using \\fR",
181: "unknown font type, using \\fR",
1.103 schwarze 182: "nothing follows prefix",
1.122 schwarze 183: "empty reference block",
1.155 schwarze 184: "missing section argument",
1.60 schwarze 185: "missing -std argument, adding it",
1.125 schwarze 186: "missing option string, using \"\"",
187: "missing resource identifier, using \"\"",
1.90 schwarze 188: "missing eqn box, using \"\"",
1.7 kristaps 189:
190: /* related to bad macro arguments */
191: "duplicate argument",
1.76 schwarze 192: "skipping duplicate argument",
1.63 schwarze 193: "skipping duplicate display type",
194: "skipping duplicate list type",
1.76 schwarze 195: "skipping -width argument",
1.126 schwarze 196: "wrong number of cells",
1.7 kristaps 197: "unknown AT&T UNIX version",
1.88 schwarze 198: "comma in function argument",
1.89 schwarze 199: "parenthesis in function name",
1.173 schwarze 200: "unknown library name",
1.67 schwarze 201: "invalid content in Rs block",
1.63 schwarze 202: "invalid Boolean argument",
1.200 schwarze 203: "argument contains two font escapes",
1.63 schwarze 204: "unknown font, skipping request",
1.126 schwarze 205: "odd number of characters in request",
1.7 kristaps 206:
207: /* related to plain text */
1.64 schwarze 208: "blank line in fill mode, using .sp",
209: "tab in filled text",
1.158 schwarze 210: "new sentence, new line",
1.64 schwarze 211: "invalid escape sequence",
212: "undefined string, using \"\"",
1.16 kristaps 213:
1.118 schwarze 214: /* related to tables */
215: "tbl line starts with span",
216: "tbl column starts with span",
217: "skipping vertical bar in tbl layout",
218:
1.7 kristaps 219: "generic error",
220:
221: /* related to tables */
1.116 schwarze 222: "non-alphabetic character in tbl options",
223: "skipping unknown tbl option",
224: "missing tbl option argument",
225: "wrong tbl option argument size",
1.118 schwarze 226: "empty tbl layout",
227: "invalid character in tbl layout",
228: "unmatched parenthesis in tbl layout",
1.119 schwarze 229: "tbl without any data cells",
230: "ignoring data in spanned tbl cell",
231: "ignoring extra tbl data cells",
232: "data block open at end of tbl",
1.7 kristaps 233:
1.68 schwarze 234: /* related to document structure and macros */
1.108 schwarze 235: NULL,
1.189 schwarze 236: "duplicate prologue macro",
237: "skipping late title macro",
1.7 kristaps 238: "input stack limit exceeded, infinite loop?",
239: "skipping bad character",
1.68 schwarze 240: "skipping unknown macro",
1.197 schwarze 241: "ignoring request outside macro",
1.112 schwarze 242: "skipping insecure request",
1.73 schwarze 243: "skipping item outside list",
1.68 schwarze 244: "skipping column outside column list",
245: "skipping end of block that is not open",
1.115 schwarze 246: "fewer RS blocks open, skipping",
1.68 schwarze 247: "inserting missing end of block",
248: "appending missing end of block",
249:
250: /* related to request and macro arguments */
1.7 kristaps 251: "escaped character not allowed in a name",
1.197 schwarze 252: "using macro argument outside macro",
253: "argument number is not numeric",
1.99 schwarze 254: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 255: "skipping display without arguments",
1.71 schwarze 256: "missing list type, using -item",
1.171 schwarze 257: "argument is not numeric, using 1",
1.200 schwarze 258: "argument is not a character",
1.70 schwarze 259: "missing manual name, using \"\"",
1.71 schwarze 260: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 261: "unknown standard specifier",
1.71 schwarze 262: "skipping request without numeric argument",
1.197 schwarze 263: "excessive shift",
1.109 schwarze 264: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 265: ".so request failed",
1.61 schwarze 266: "skipping all arguments",
267: "skipping excess arguments",
1.92 kristaps 268: "divide by zero",
1.112 schwarze 269:
270: "unsupported feature",
271: "input too large",
1.114 schwarze 272: "unsupported control character",
1.112 schwarze 273: "unsupported roff request",
1.199 schwarze 274: "nested .while loops",
275: "end of scope with open .while loop",
276: "end of .while loop in inner scope",
277: "cannot continue this .while loop",
1.119 schwarze 278: "eqn delim option in tbl",
1.118 schwarze 279: "unsupported tbl layout modifier",
1.112 schwarze 280: "ignoring macro in table",
1.7 kristaps 281: };
282:
283: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
284: "SUCCESS",
1.165 schwarze 285: "STYLE",
1.7 kristaps 286: "WARNING",
287: "ERROR",
1.112 schwarze 288: "UNSUPP",
1.7 kristaps 289: "BADARG",
290: "SYSERR"
291: };
292:
1.47 schwarze 293:
1.1 kristaps 294: static void
295: resize_buf(struct buf *buf, size_t initial)
296: {
297:
298: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
299: buf->buf = mandoc_realloc(buf->buf, buf->sz);
300: }
301:
302: static void
1.198 schwarze 303: free_buf_list(struct buf *buf)
304: {
305: struct buf *tmp;
306:
307: while (buf != NULL) {
308: tmp = buf;
309: buf = tmp->next;
310: free(tmp->buf);
311: free(tmp);
312: }
313: }
314:
315: static void
1.84 schwarze 316: choose_parser(struct mparse *curp)
1.1 kristaps 317: {
1.83 schwarze 318: char *cp, *ep;
319: int format;
1.1 kristaps 320:
1.83 schwarze 321: /*
322: * If neither command line arguments -mdoc or -man select
323: * a parser nor the roff parser found a .Dd or .TH macro
324: * yet, look ahead in the main input buffer.
325: */
326:
327: if ((format = roff_getformat(curp->roff)) == 0) {
328: cp = curp->primary->buf;
329: ep = cp + curp->primary->sz;
330: while (cp < ep) {
1.85 schwarze 331: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 332: cp++;
333: if (cp[0] == 'D' && cp[1] == 'd') {
334: format = MPARSE_MDOC;
335: break;
336: }
337: if (cp[0] == 'T' && cp[1] == 'H') {
338: format = MPARSE_MAN;
339: break;
340: }
341: }
342: cp = memchr(cp, '\n', ep - cp);
343: if (cp == NULL)
344: break;
345: cp++;
346: }
1.1 kristaps 347: }
348:
1.83 schwarze 349: if (format == MPARSE_MDOC) {
1.137 schwarze 350: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 351: if (curp->man->mdocmac == NULL)
352: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 353: } else {
354: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 355: if (curp->man->manmac == NULL)
356: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 357: }
1.164 schwarze 358: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 359: }
360:
361: /*
1.95 schwarze 362: * Main parse routine for a buffer.
363: * It assumes encoding and line numbering are already set up.
364: * It can recurse directly (for invocations of user-defined
365: * macros, inline equations, and input line traps)
366: * and indirectly (for .so file inclusion).
1.1 kristaps 367: */
1.199 schwarze 368: static int
1.95 schwarze 369: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 370: {
371: struct buf ln;
1.199 schwarze 372: struct buf *firstln, *lastln, *thisln, *loop;
1.113 schwarze 373: const char *save_file;
1.110 schwarze 374: char *cp;
1.95 schwarze 375: size_t pos; /* byte number in the ln buffer */
1.199 schwarze 376: int line_result, result;
1.100 schwarze 377: int of;
1.1 kristaps 378: int lnn; /* line number in the real file */
1.110 schwarze 379: int fd;
1.199 schwarze 380: int inloop; /* Saw .while on this level. */
1.1 kristaps 381: unsigned char c;
382:
1.198 schwarze 383: ln.sz = 256;
384: ln.buf = mandoc_malloc(ln.sz);
385: ln.next = NULL;
1.199 schwarze 386: firstln = loop = NULL;
1.47 schwarze 387: lnn = curp->line;
388: pos = 0;
1.199 schwarze 389: inloop = 0;
1.198 schwarze 390: result = ROFF_CONT;
1.1 kristaps 391:
1.199 schwarze 392: while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) {
1.1 kristaps 393: if (start) {
394: curp->line = lnn;
395: curp->reparse_count = 0;
1.93 schwarze 396:
397: if (lnn < 3 &&
398: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 399: curp->filenc & MPARSE_LATIN1)
400: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 401: }
402:
1.95 schwarze 403: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 404:
405: /*
406: * When finding an unescaped newline character,
407: * leave the character loop to process the line.
408: * Skip a preceding carriage return, if any.
409: */
410:
1.95 schwarze 411: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 412: '\n' == blk.buf[i + 1])
413: ++i;
414: if ('\n' == blk.buf[i]) {
415: ++i;
416: ++lnn;
417: break;
418: }
419:
1.35 schwarze 420: /*
1.93 schwarze 421: * Make sure we have space for the worst
1.198 schwarze 422: * case of 12 bytes: "\\[u10ffff]\n\0"
1.35 schwarze 423: */
424:
1.198 schwarze 425: if (pos + 12 > ln.sz)
1.35 schwarze 426: resize_buf(&ln, 256);
427:
1.47 schwarze 428: /*
1.93 schwarze 429: * Encode 8-bit input.
1.1 kristaps 430: */
431:
1.93 schwarze 432: c = blk.buf[i];
433: if (c & 0x80) {
1.95 schwarze 434: if ( ! (curp->filenc && preconv_encode(
435: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 436: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
437: curp->line, pos, "0x%x", c);
1.93 schwarze 438: ln.buf[pos++] = '?';
439: i++;
440: }
441: continue;
442: }
443:
444: /*
445: * Exclude control characters.
446: */
1.1 kristaps 447:
1.93 schwarze 448: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 449: mandoc_vmsg(c == 0x00 || c == 0x04 ||
450: c > 0x0a ? MANDOCERR_CHAR_BAD :
451: MANDOCERR_CHAR_UNSUPP,
452: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 453: i++;
1.127 schwarze 454: if (c != '\r')
455: ln.buf[pos++] = '?';
1.1 kristaps 456: continue;
457: }
458:
459: ln.buf[pos++] = blk.buf[i++];
460: }
1.198 schwarze 461: ln.buf[pos] = '\0';
462:
463: /*
464: * Maintain a lookaside buffer of all lines.
465: * parsed from this input source.
466: */
467:
468: thisln = mandoc_malloc(sizeof(*thisln));
469: thisln->buf = mandoc_strdup(ln.buf);
470: thisln->sz = strlen(ln.buf) + 1;
471: thisln->next = NULL;
472: if (firstln == NULL) {
473: firstln = lastln = thisln;
474: if (curp->secondary == NULL)
475: curp->secondary = firstln;
476: } else {
477: lastln->next = thisln;
478: lastln = thisln;
479: }
1.1 kristaps 480:
1.198 schwarze 481: /* XXX Ugly hack to mark the end of the input. */
1.1 kristaps 482:
1.198 schwarze 483: if (i == blk.sz || blk.buf[i] == '\0') {
1.170 schwarze 484: ln.buf[pos++] = '\n';
1.198 schwarze 485: ln.buf[pos] = '\0';
486: }
1.1 kristaps 487:
488: /*
489: * A significant amount of complexity is contained by
490: * the roff preprocessor. It's line-oriented but can be
491: * expressed on one line, so we need at times to
492: * readjust our starting point and re-run it. The roff
493: * preprocessor can also readjust the buffers with new
494: * data, so we pass them in wholesale.
495: */
496:
497: of = 0;
498: rerun:
1.197 schwarze 499: line_result = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 500:
1.199 schwarze 501: /* Process options. */
502:
503: if (line_result & ROFF_APPEND)
504: assert(line_result == (ROFF_IGN | ROFF_APPEND));
505:
506: if (line_result & ROFF_USERCALL)
507: assert((line_result & ROFF_MASK) == ROFF_REPARSE);
508:
509: if (line_result & ROFF_USERRET) {
510: assert(line_result == (ROFF_IGN | ROFF_USERRET));
511: if (start == 0) {
512: /* Return from the current macro. */
513: result = ROFF_USERRET;
514: goto out;
515: }
516: }
517:
518: switch (line_result & ROFF_LOOPMASK) {
519: case ROFF_IGN:
520: break;
521: case ROFF_WHILE:
522: if (curp->loop != NULL) {
523: if (loop == curp->loop)
524: break;
525: mandoc_msg(MANDOCERR_WHILE_NEST,
526: curp, curp->line, pos, NULL);
527: }
528: curp->loop = thisln;
529: loop = NULL;
530: inloop = 1;
531: break;
532: case ROFF_LOOPCONT:
533: case ROFF_LOOPEXIT:
534: if (curp->loop == NULL) {
535: mandoc_msg(MANDOCERR_WHILE_FAIL,
536: curp, curp->line, pos, NULL);
537: break;
538: }
539: if (inloop == 0) {
540: mandoc_msg(MANDOCERR_WHILE_INTO,
541: curp, curp->line, pos, NULL);
542: curp->loop = loop = NULL;
543: break;
544: }
545: if (line_result & ROFF_LOOPCONT)
546: loop = curp->loop;
547: else {
548: curp->loop = loop = NULL;
549: inloop = 0;
550: }
551: break;
552: default:
553: abort();
554: }
555:
556: /* Process the main instruction from the roff parser. */
557:
558: switch (line_result & ROFF_MASK) {
559: case ROFF_IGN:
560: break;
561: case ROFF_CONT:
562: if (curp->man->macroset == MACROSET_NONE)
563: choose_parser(curp);
564: if ((curp->man->macroset == MACROSET_MDOC ?
565: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
566: man_parseln(curp->man, curp->line, ln.buf, of)
567: ) == 2)
568: goto out;
569: break;
570: case ROFF_RERUN:
571: goto rerun;
1.47 schwarze 572: case ROFF_REPARSE:
1.197 schwarze 573: if (++curp->reparse_count > REPARSE_LIMIT) {
1.199 schwarze 574: /* Abort and return to the top level. */
1.198 schwarze 575: result = ROFF_IGN;
1.3 kristaps 576: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 577: curp->line, pos, NULL);
1.199 schwarze 578: goto out;
1.197 schwarze 579: }
1.199 schwarze 580: result = mparse_buf_r(curp, ln, of, 0);
581: if (line_result & ROFF_USERCALL) {
582: roff_userret(curp->roff);
583: /* Continue normally. */
584: if (result & ROFF_USERRET)
585: result = ROFF_CONT;
1.163 schwarze 586: }
1.199 schwarze 587: if (start == 0 && result != ROFF_CONT)
588: goto out;
589: break;
1.47 schwarze 590: case ROFF_SO:
1.95 schwarze 591: if ( ! (curp->options & MPARSE_SO) &&
592: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 593: curp->sodest = mandoc_strdup(ln.buf + of);
1.198 schwarze 594: goto out;
1.45 schwarze 595: }
1.113 schwarze 596: save_file = curp->file;
1.148 schwarze 597: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 598: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 599: close(fd);
1.113 schwarze 600: curp->file = save_file;
601: } else {
602: curp->file = save_file;
1.52 schwarze 603: mandoc_vmsg(MANDOCERR_SO_FAIL,
604: curp, curp->line, pos,
605: ".so %s", ln.buf + of);
1.110 schwarze 606: ln.sz = mandoc_asprintf(&cp,
607: ".sp\nSee the file %s.\n.sp",
608: ln.buf + of);
609: free(ln.buf);
610: ln.buf = cp;
611: of = 0;
612: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 613: }
1.199 schwarze 614: break;
1.1 kristaps 615: default:
1.199 schwarze 616: abort();
1.1 kristaps 617: }
618:
1.199 schwarze 619: /* Start the next input line. */
1.1 kristaps 620:
1.199 schwarze 621: if (loop != NULL &&
622: (line_result & ROFF_LOOPMASK) == ROFF_IGN)
623: loop = loop->next;
624:
625: if (loop != NULL) {
626: if ((line_result & ROFF_APPEND) == 0)
627: *ln.buf = '\0';
628: if (ln.sz < loop->sz)
629: resize_buf(&ln, loop->sz);
630: (void)strlcat(ln.buf, loop->buf, ln.sz);
631: of = 0;
632: goto rerun;
633: }
1.1 kristaps 634:
1.199 schwarze 635: pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
1.1 kristaps 636: }
1.198 schwarze 637: out:
1.199 schwarze 638: if (inloop) {
639: if (result != ROFF_USERRET)
640: mandoc_msg(MANDOCERR_WHILE_OUTOF, curp,
641: curp->line, pos, NULL);
642: curp->loop = NULL;
643: }
1.1 kristaps 644: free(ln.buf);
1.198 schwarze 645: if (firstln != curp->secondary)
646: free_buf_list(firstln);
647: return result;
1.1 kristaps 648: }
649:
650: static int
1.40 schwarze 651: read_whole_file(struct mparse *curp, const char *file, int fd,
652: struct buf *fb, int *with_mmap)
1.1 kristaps 653: {
1.161 schwarze 654: struct stat st;
1.140 schwarze 655: gzFile gz;
1.1 kristaps 656: size_t off;
657: ssize_t ssz;
1.194 schwarze 658: int gzerrnum, retval;
1.143 schwarze 659:
1.192 schwarze 660: if (fstat(fd, &st) == -1) {
661: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
662: "fstat: %s", strerror(errno));
663: return 0;
664: }
1.1 kristaps 665:
666: /*
667: * If we're a regular file, try just reading in the whole entry
668: * via mmap(). This is faster than reading it into blocks, and
669: * since each file is only a few bytes to begin with, I'm not
670: * concerned that this is going to tank any machines.
671: */
672:
1.140 schwarze 673: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 674: if (st.st_size > 0x7fffffff) {
1.111 schwarze 675: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 676: return 0;
1.1 kristaps 677: }
678: *with_mmap = 1;
679: fb->sz = (size_t)st.st_size;
1.37 schwarze 680: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 681: if (fb->buf != MAP_FAILED)
1.142 schwarze 682: return 1;
1.1 kristaps 683: }
684:
1.140 schwarze 685: if (curp->gzip) {
1.194 schwarze 686: /*
687: * Duplicating the file descriptor is required
688: * because we will have to call gzclose(3)
689: * to free memory used internally by zlib,
690: * but that will also close the file descriptor,
691: * which this function must not do.
692: */
693: if ((fd = dup(fd)) == -1) {
694: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
695: "dup: %s", strerror(errno));
696: return 0;
697: }
1.192 schwarze 698: if ((gz = gzdopen(fd, "rb")) == NULL) {
699: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
700: "gzdopen: %s", strerror(errno));
1.194 schwarze 701: close(fd);
1.192 schwarze 702: return 0;
703: }
1.140 schwarze 704: } else
705: gz = NULL;
706:
1.1 kristaps 707: /*
708: * If this isn't a regular file (like, say, stdin), then we must
709: * go the old way and just read things in bit by bit.
710: */
711:
712: *with_mmap = 0;
713: off = 0;
1.194 schwarze 714: retval = 0;
1.1 kristaps 715: fb->sz = 0;
716: fb->buf = NULL;
717: for (;;) {
718: if (off == fb->sz) {
719: if (fb->sz == (1U << 31)) {
1.111 schwarze 720: mandoc_msg(MANDOCERR_TOOLARGE, curp,
721: 0, 0, NULL);
1.1 kristaps 722: break;
723: }
724: resize_buf(fb, 65536);
725: }
1.140 schwarze 726: ssz = curp->gzip ?
727: gzread(gz, fb->buf + (int)off, fb->sz - off) :
728: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 729: if (ssz == 0) {
730: fb->sz = off;
1.194 schwarze 731: retval = 1;
732: break;
1.1 kristaps 733: }
1.192 schwarze 734: if (ssz == -1) {
1.194 schwarze 735: if (curp->gzip)
736: (void)gzerror(gz, &gzerrnum);
737: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "read: %s",
738: curp->gzip && gzerrnum != Z_ERRNO ?
739: zError(gzerrnum) : strerror(errno));
1.192 schwarze 740: break;
741: }
1.1 kristaps 742: off += (size_t)ssz;
743: }
744:
1.194 schwarze 745: if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
746: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "gzclose: %s",
747: gzerrnum == Z_ERRNO ? strerror(errno) :
748: zError(gzerrnum));
749: if (retval == 0) {
750: free(fb->buf);
751: fb->buf = NULL;
752: }
753: return retval;
1.1 kristaps 754: }
755:
756: static void
757: mparse_end(struct mparse *curp)
758: {
1.135 schwarze 759: if (curp->man->macroset == MACROSET_NONE)
760: curp->man->macroset = MACROSET_MAN;
761: if (curp->man->macroset == MACROSET_MDOC)
762: mdoc_endparse(curp->man);
763: else
1.111 schwarze 764: man_endparse(curp->man);
1.1 kristaps 765: roff_endparse(curp->roff);
766: }
767:
768: static void
1.36 schwarze 769: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 770: {
1.85 schwarze 771: struct buf *svprimary;
1.28 joerg 772: const char *svfile;
1.95 schwarze 773: size_t offset;
1.36 schwarze 774: static int recursion_depth;
775:
776: if (64 < recursion_depth) {
777: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
778: return;
779: }
1.28 joerg 780:
781: /* Line number is per-file. */
782: svfile = curp->file;
783: curp->file = file;
1.85 schwarze 784: svprimary = curp->primary;
1.83 schwarze 785: curp->primary = &blk;
1.28 joerg 786: curp->line = 1;
1.36 schwarze 787: recursion_depth++;
1.28 joerg 788:
1.93 schwarze 789: /* Skip an UTF-8 byte order mark. */
790: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
791: (unsigned char)blk.buf[0] == 0xef &&
792: (unsigned char)blk.buf[1] == 0xbb &&
793: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 794: offset = 3;
1.93 schwarze 795: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 796: } else
797: offset = 0;
1.93 schwarze 798:
1.95 schwarze 799: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 800:
1.111 schwarze 801: if (--recursion_depth == 0)
1.28 joerg 802: mparse_end(curp);
803:
1.85 schwarze 804: curp->primary = svprimary;
1.28 joerg 805: curp->file = svfile;
806: }
807:
808: enum mandoclevel
1.104 schwarze 809: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 810: const char *file)
811: {
812: struct buf blk;
813:
1.104 schwarze 814: blk.buf = buf;
1.28 joerg 815: blk.sz = len;
816:
1.36 schwarze 817: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 818: return curp->file_status;
1.28 joerg 819: }
820:
1.98 schwarze 821: /*
822: * Read the whole file into memory and call the parsers.
823: * Called recursively when an .so request is encountered.
824: */
1.36 schwarze 825: enum mandoclevel
826: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 827: {
1.28 joerg 828: struct buf blk;
829: int with_mmap;
1.93 schwarze 830: int save_filenc;
1.1 kristaps 831:
1.91 schwarze 832: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 833: save_filenc = curp->filenc;
834: curp->filenc = curp->options &
835: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 836: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 837: curp->filenc = save_filenc;
1.91 schwarze 838: if (with_mmap)
839: munmap(blk.buf, blk.sz);
840: else
841: free(blk.buf);
842: }
1.142 schwarze 843: return curp->file_status;
1.82 schwarze 844: }
845:
1.148 schwarze 846: int
847: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 848: {
849: char *cp;
1.148 schwarze 850: int fd;
1.82 schwarze 851:
852: curp->file = file;
1.140 schwarze 853: cp = strrchr(file, '.');
854: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 855:
1.140 schwarze 856: /* First try to use the filename as it is. */
1.98 schwarze 857:
1.148 schwarze 858: if ((fd = open(file, O_RDONLY)) != -1)
859: return fd;
1.98 schwarze 860:
1.140 schwarze 861: /*
862: * If that doesn't work and the filename doesn't
863: * already end in .gz, try appending .gz.
864: */
1.98 schwarze 865:
1.140 schwarze 866: if ( ! curp->gzip) {
1.98 schwarze 867: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 868: fd = open(cp, O_RDONLY);
1.108 schwarze 869: free(cp);
1.148 schwarze 870: if (fd != -1) {
1.140 schwarze 871: curp->gzip = 1;
1.148 schwarze 872: return fd;
1.82 schwarze 873: }
874: }
875:
1.140 schwarze 876: /* Neither worked, give up. */
1.97 schwarze 877:
1.140 schwarze 878: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 879: return -1;
1.1 kristaps 880: }
881:
882: struct mparse *
1.179 schwarze 883: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
884: enum mandoc_os os_e, const char *os_s)
1.1 kristaps 885: {
886: struct mparse *curp;
1.10 kristaps 887:
1.1 kristaps 888: curp = mandoc_calloc(1, sizeof(struct mparse));
889:
1.44 schwarze 890: curp->options = options;
1.179 schwarze 891: curp->mmin = mmin;
1.1 kristaps 892: curp->mmsg = mmsg;
1.179 schwarze 893: curp->os_s = os_s;
1.1 kristaps 894:
1.144 schwarze 895: curp->roff = roff_alloc(curp, options);
1.179 schwarze 896: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.137 schwarze 897: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 898: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 899: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 900: if (curp->man->mdocmac == NULL)
901: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 902: } else if (curp->options & MPARSE_MAN) {
903: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 904: if (curp->man->manmac == NULL)
905: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 906: }
1.138 schwarze 907: curp->man->first->tok = TOKEN_NONE;
1.179 schwarze 908: curp->man->meta.os_e = os_e;
1.142 schwarze 909: return curp;
1.1 kristaps 910: }
911:
912: void
913: mparse_reset(struct mparse *curp)
914: {
915: roff_reset(curp->roff);
1.150 schwarze 916: roff_man_reset(curp->man);
1.198 schwarze 917: free_buf_list(curp->secondary);
918: curp->secondary = NULL;
1.160 schwarze 919:
920: free(curp->sodest);
921: curp->sodest = NULL;
922:
1.1 kristaps 923: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 924: curp->gzip = 0;
1.1 kristaps 925: }
926:
927: void
928: mparse_free(struct mparse *curp)
929: {
1.164 schwarze 930: roffhash_free(curp->man->mdocmac);
931: roffhash_free(curp->man->manmac);
1.137 schwarze 932: roff_man_free(curp->man);
1.160 schwarze 933: roff_free(curp->roff);
1.198 schwarze 934: free_buf_list(curp->secondary);
1.45 schwarze 935: free(curp->sodest);
1.1 kristaps 936: free(curp);
937: }
938:
939: void
1.135 schwarze 940: mparse_result(struct mparse *curp, struct roff_man **man,
941: char **sodest)
1.1 kristaps 942: {
943:
1.45 schwarze 944: if (sodest && NULL != (*sodest = curp->sodest)) {
945: *man = NULL;
946: return;
947: }
1.9 kristaps 948: if (man)
949: *man = curp->man;
1.157 schwarze 950: }
951:
952: void
953: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
954: {
955: if (curp->file_status > *rc)
956: *rc = curp->file_status;
1.3 kristaps 957: }
958:
959: void
960: mandoc_vmsg(enum mandocerr t, struct mparse *m,
961: int ln, int pos, const char *fmt, ...)
962: {
963: char buf[256];
964: va_list ap;
965:
966: va_start(ap, fmt);
1.48 schwarze 967: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 968: va_end(ap);
969:
970: mandoc_msg(t, m, ln, pos, buf);
971: }
972:
973: void
1.47 schwarze 974: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 975: int ln, int col, const char *msg)
976: {
977: enum mandoclevel level;
978:
1.179 schwarze 979: if (er < m->mmin && er != MANDOCERR_FILE)
980: return;
981:
1.112 schwarze 982: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 983: while (er < mandoclimits[level])
984: level--;
985:
1.8 kristaps 986: if (m->mmsg)
987: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 988:
989: if (m->file_status < level)
990: m->file_status = level;
1.7 kristaps 991: }
992:
993: const char *
994: mparse_strerror(enum mandocerr er)
995: {
996:
1.142 schwarze 997: return mandocerrs[er];
1.7 kristaps 998: }
999:
1000: const char *
1001: mparse_strlevel(enum mandoclevel lvl)
1002: {
1.142 schwarze 1003: return mandoclevels[lvl];
1.24 kristaps 1004: }
1005:
1006: void
1.198 schwarze 1007: mparse_copy(const struct mparse *p)
1.24 kristaps 1008: {
1.198 schwarze 1009: struct buf *buf;
1.24 kristaps 1010:
1.198 schwarze 1011: for (buf = p->secondary; buf != NULL; buf = buf->next)
1012: puts(buf->buf);
1.1 kristaps 1013: }
CVSweb