Annotation of mandoc/read.c, Revision 1.201
1.201 ! schwarze 1: /* $Id: read.c,v 1.200 2018/08/25 16:53:39 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"
1.133 schwarze 41: #include "libmandoc.h"
1.201 ! schwarze 42: #include "roff_int.h"
1.1 kristaps 43:
44: #define REPARSE_LIMIT 1000
45:
46: struct mparse {
1.160 schwarze 47: struct roff *roff; /* roff parser (!NULL) */
1.134 schwarze 48: struct roff_man *man; /* man parser */
1.45 schwarze 49: char *sodest; /* filename pointed to by .so */
1.83 schwarze 50: const char *file; /* filename of current input file */
51: struct buf *primary; /* buffer currently being parsed */
1.198 schwarze 52: struct buf *secondary; /* copy of top level input */
1.199 schwarze 53: struct buf *loop; /* open .while request line */
1.179 schwarze 54: const char *os_s; /* default operating system */
1.83 schwarze 55: mandocmsg mmsg; /* warning/error message handler */
56: enum mandoclevel file_status; /* status of current parse */
1.179 schwarze 57: enum mandocerr mmin; /* ignore messages below this */
1.83 schwarze 58: int options; /* parser options */
1.140 schwarze 59: int gzip; /* current input file is gzipped */
1.93 schwarze 60: int filenc; /* encoding of the current file */
1.1 kristaps 61: int reparse_count; /* finite interp. stack */
1.83 schwarze 62: int line; /* line number in the file */
1.1 kristaps 63: };
64:
1.84 schwarze 65: static void choose_parser(struct mparse *);
1.198 schwarze 66: static void free_buf_list(struct buf *);
1.1 kristaps 67: static void resize_buf(struct buf *, size_t);
1.199 schwarze 68: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 69: static int read_whole_file(struct mparse *, const char *, int,
70: struct buf *, int *);
1.1 kristaps 71: static void mparse_end(struct mparse *);
1.37 schwarze 72: static void mparse_parse_buffer(struct mparse *, struct buf,
73: const char *);
1.1 kristaps 74:
1.3 kristaps 75: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
76: MANDOCERR_OK,
1.188 schwarze 77: MANDOCERR_OK,
1.3 kristaps 78: MANDOCERR_WARNING,
79: MANDOCERR_ERROR,
1.112 schwarze 80: MANDOCERR_UNSUPP,
1.3 kristaps 81: MANDOCERR_MAX,
82: MANDOCERR_MAX
83: };
84:
1.7 kristaps 85: static const char * const mandocerrs[MANDOCERR_MAX] = {
86: "ok",
87:
1.179 schwarze 88: "base system convention",
1.166 schwarze 89:
1.176 schwarze 90: "Mdocdate found",
91: "Mdocdate missing",
1.181 schwarze 92: "unknown architecture",
1.180 schwarze 93: "operating system explicitly specified",
1.179 schwarze 94: "RCS id missing",
1.184 schwarze 95: "referenced manual not found",
1.179 schwarze 96:
97: "generic style suggestion",
98:
1.176 schwarze 99: "legacy man(7) date format",
1.196 schwarze 100: "normalizing date format to",
1.189 schwarze 101: "lower case character in document title",
1.178 schwarze 102: "duplicate RCS id",
1.193 schwarze 103: "possible typo in section name",
1.189 schwarze 104: "unterminated quoted argument",
1.166 schwarze 105: "useless macro",
1.167 schwarze 106: "consider using OS macro",
1.172 schwarze 107: "errnos out of order",
108: "duplicate errno",
1.187 schwarze 109: "trailing delimiter",
1.174 schwarze 110: "no blank before trailing delimiter",
1.189 schwarze 111: "fill mode already enabled, skipping",
112: "fill mode already disabled, skipping",
1.195 schwarze 113: "verbatim \"--\", maybe consider using \\(em",
1.175 schwarze 114: "function name without markup",
1.189 schwarze 115: "whitespace at end of input line",
116: "bad comment style",
1.166 schwarze 117:
1.7 kristaps 118: "generic warning",
119:
120: /* related to the prologue */
1.79 schwarze 121: "missing manual title, using UNTITLED",
122: "missing manual title, using \"\"",
123: "missing manual section, using \"\"",
1.7 kristaps 124: "unknown manual section",
1.54 schwarze 125: "missing date, using today's date",
1.7 kristaps 126: "cannot parse date, using it verbatim",
1.186 schwarze 127: "date in the future, using it anyway",
1.79 schwarze 128: "missing Os macro, using \"\"",
129: "late prologue macro",
1.7 kristaps 130: "prologue macros out of order",
131:
132: /* related to document structure */
133: ".so is fragile, better use ln(1)",
1.50 schwarze 134: "no document body",
1.54 schwarze 135: "content before first section header",
136: "first section is not \"NAME\"",
1.156 schwarze 137: "NAME section without Nm before Nd",
1.128 schwarze 138: "NAME section without description",
139: "description not at the end of NAME",
140: "bad NAME section content",
1.156 schwarze 141: "missing comma before name",
1.121 schwarze 142: "missing description line, using \"\"",
1.162 schwarze 143: "description line outside NAME section",
1.7 kristaps 144: "sections out of conventional order",
1.54 schwarze 145: "duplicate section title",
146: "unexpected section",
1.185 schwarze 147: "cross reference to self",
1.87 schwarze 148: "unusual Xr order",
149: "unusual Xr punctuation",
1.86 schwarze 150: "AUTHORS section without An macro",
1.7 kristaps 151:
152: /* related to macros and nesting */
1.55 schwarze 153: "obsolete macro",
1.102 schwarze 154: "macro neither callable nor escaped",
1.7 kristaps 155: "skipping paragraph macro",
1.31 schwarze 156: "moving paragraph macro out of list",
1.7 kristaps 157: "skipping no-space macro",
158: "blocks badly nested",
159: "nested displays are not portable",
1.57 schwarze 160: "moving content out of list",
1.183 schwarze 161: "first macro on line",
1.7 kristaps 162: "line scope broken",
1.169 schwarze 163: "skipping blank line in line scope",
1.7 kristaps 164:
165: /* related to missing macro arguments */
1.58 schwarze 166: "skipping empty request",
167: "conditional request controls empty scope",
1.7 kristaps 168: "skipping empty macro",
1.124 schwarze 169: "empty block",
1.62 schwarze 170: "empty argument, using 0n",
1.60 schwarze 171: "missing display type, using -ragged",
172: "list type is not the first argument",
1.152 schwarze 173: "missing -width in -tag list, using 6n",
1.78 schwarze 174: "missing utility name, using \"\"",
1.123 schwarze 175: "missing function name, using \"\"",
1.60 schwarze 176: "empty head in list item",
177: "empty list item",
1.183 schwarze 178: "missing argument, using next line",
1.61 schwarze 179: "missing font type, using \\fR",
180: "unknown font type, using \\fR",
1.103 schwarze 181: "nothing follows prefix",
1.122 schwarze 182: "empty reference block",
1.155 schwarze 183: "missing section argument",
1.60 schwarze 184: "missing -std argument, adding it",
1.125 schwarze 185: "missing option string, using \"\"",
186: "missing resource identifier, using \"\"",
1.90 schwarze 187: "missing eqn box, using \"\"",
1.7 kristaps 188:
189: /* related to bad macro arguments */
190: "duplicate argument",
1.76 schwarze 191: "skipping duplicate argument",
1.63 schwarze 192: "skipping duplicate display type",
193: "skipping duplicate list type",
1.76 schwarze 194: "skipping -width argument",
1.126 schwarze 195: "wrong number of cells",
1.7 kristaps 196: "unknown AT&T UNIX version",
1.88 schwarze 197: "comma in function argument",
1.89 schwarze 198: "parenthesis in function name",
1.173 schwarze 199: "unknown library name",
1.67 schwarze 200: "invalid content in Rs block",
1.63 schwarze 201: "invalid Boolean argument",
1.200 schwarze 202: "argument contains two font escapes",
1.63 schwarze 203: "unknown font, skipping request",
1.126 schwarze 204: "odd number of characters in request",
1.7 kristaps 205:
206: /* related to plain text */
1.64 schwarze 207: "blank line in fill mode, using .sp",
208: "tab in filled text",
1.158 schwarze 209: "new sentence, new line",
1.64 schwarze 210: "invalid escape sequence",
211: "undefined string, using \"\"",
1.16 kristaps 212:
1.118 schwarze 213: /* related to tables */
214: "tbl line starts with span",
215: "tbl column starts with span",
216: "skipping vertical bar in tbl layout",
217:
1.7 kristaps 218: "generic error",
219:
220: /* related to tables */
1.116 schwarze 221: "non-alphabetic character in tbl options",
222: "skipping unknown tbl option",
223: "missing tbl option argument",
224: "wrong tbl option argument size",
1.118 schwarze 225: "empty tbl layout",
226: "invalid character in tbl layout",
227: "unmatched parenthesis in tbl layout",
1.119 schwarze 228: "tbl without any data cells",
229: "ignoring data in spanned tbl cell",
230: "ignoring extra tbl data cells",
231: "data block open at end of tbl",
1.7 kristaps 232:
1.68 schwarze 233: /* related to document structure and macros */
1.108 schwarze 234: NULL,
1.189 schwarze 235: "duplicate prologue macro",
236: "skipping late title macro",
1.7 kristaps 237: "input stack limit exceeded, infinite loop?",
238: "skipping bad character",
1.68 schwarze 239: "skipping unknown macro",
1.197 schwarze 240: "ignoring request outside macro",
1.112 schwarze 241: "skipping insecure request",
1.73 schwarze 242: "skipping item outside list",
1.68 schwarze 243: "skipping column outside column list",
244: "skipping end of block that is not open",
1.115 schwarze 245: "fewer RS blocks open, skipping",
1.68 schwarze 246: "inserting missing end of block",
247: "appending missing end of block",
248:
249: /* related to request and macro arguments */
1.7 kristaps 250: "escaped character not allowed in a name",
1.197 schwarze 251: "using macro argument outside macro",
252: "argument number is not numeric",
1.99 schwarze 253: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 254: "skipping display without arguments",
1.71 schwarze 255: "missing list type, using -item",
1.171 schwarze 256: "argument is not numeric, using 1",
1.200 schwarze 257: "argument is not a character",
1.70 schwarze 258: "missing manual name, using \"\"",
1.71 schwarze 259: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 260: "unknown standard specifier",
1.71 schwarze 261: "skipping request without numeric argument",
1.197 schwarze 262: "excessive shift",
1.109 schwarze 263: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 264: ".so request failed",
1.61 schwarze 265: "skipping all arguments",
266: "skipping excess arguments",
1.92 kristaps 267: "divide by zero",
1.112 schwarze 268:
269: "unsupported feature",
270: "input too large",
1.114 schwarze 271: "unsupported control character",
1.112 schwarze 272: "unsupported roff request",
1.199 schwarze 273: "nested .while loops",
274: "end of scope with open .while loop",
275: "end of .while loop in inner scope",
276: "cannot continue this .while loop",
1.119 schwarze 277: "eqn delim option in tbl",
1.118 schwarze 278: "unsupported tbl layout modifier",
1.112 schwarze 279: "ignoring macro in table",
1.7 kristaps 280: };
281:
282: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
283: "SUCCESS",
1.165 schwarze 284: "STYLE",
1.7 kristaps 285: "WARNING",
286: "ERROR",
1.112 schwarze 287: "UNSUPP",
1.7 kristaps 288: "BADARG",
289: "SYSERR"
290: };
291:
1.47 schwarze 292:
1.1 kristaps 293: static void
294: resize_buf(struct buf *buf, size_t initial)
295: {
296:
297: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
298: buf->buf = mandoc_realloc(buf->buf, buf->sz);
299: }
300:
301: static void
1.198 schwarze 302: free_buf_list(struct buf *buf)
303: {
304: struct buf *tmp;
305:
306: while (buf != NULL) {
307: tmp = buf;
308: buf = tmp->next;
309: free(tmp->buf);
310: free(tmp);
311: }
312: }
313:
314: static void
1.84 schwarze 315: choose_parser(struct mparse *curp)
1.1 kristaps 316: {
1.83 schwarze 317: char *cp, *ep;
318: int format;
1.1 kristaps 319:
1.83 schwarze 320: /*
321: * If neither command line arguments -mdoc or -man select
322: * a parser nor the roff parser found a .Dd or .TH macro
323: * yet, look ahead in the main input buffer.
324: */
325:
326: if ((format = roff_getformat(curp->roff)) == 0) {
327: cp = curp->primary->buf;
328: ep = cp + curp->primary->sz;
329: while (cp < ep) {
1.85 schwarze 330: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 331: cp++;
332: if (cp[0] == 'D' && cp[1] == 'd') {
333: format = MPARSE_MDOC;
334: break;
335: }
336: if (cp[0] == 'T' && cp[1] == 'H') {
337: format = MPARSE_MAN;
338: break;
339: }
340: }
341: cp = memchr(cp, '\n', ep - cp);
342: if (cp == NULL)
343: break;
344: cp++;
345: }
1.1 kristaps 346: }
347:
1.83 schwarze 348: if (format == MPARSE_MDOC) {
1.137 schwarze 349: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 350: if (curp->man->mdocmac == NULL)
351: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 352: } else {
353: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 354: if (curp->man->manmac == NULL)
355: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 356: }
1.164 schwarze 357: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 358: }
359:
360: /*
1.95 schwarze 361: * Main parse routine for a buffer.
362: * It assumes encoding and line numbering are already set up.
363: * It can recurse directly (for invocations of user-defined
364: * macros, inline equations, and input line traps)
365: * and indirectly (for .so file inclusion).
1.1 kristaps 366: */
1.199 schwarze 367: static int
1.95 schwarze 368: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 369: {
370: struct buf ln;
1.199 schwarze 371: struct buf *firstln, *lastln, *thisln, *loop;
1.113 schwarze 372: const char *save_file;
1.110 schwarze 373: char *cp;
1.95 schwarze 374: size_t pos; /* byte number in the ln buffer */
1.199 schwarze 375: int line_result, result;
1.100 schwarze 376: int of;
1.1 kristaps 377: int lnn; /* line number in the real file */
1.110 schwarze 378: int fd;
1.199 schwarze 379: int inloop; /* Saw .while on this level. */
1.1 kristaps 380: unsigned char c;
381:
1.198 schwarze 382: ln.sz = 256;
383: ln.buf = mandoc_malloc(ln.sz);
384: ln.next = NULL;
1.199 schwarze 385: firstln = loop = NULL;
1.47 schwarze 386: lnn = curp->line;
387: pos = 0;
1.199 schwarze 388: inloop = 0;
1.198 schwarze 389: result = ROFF_CONT;
1.1 kristaps 390:
1.199 schwarze 391: while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) {
1.1 kristaps 392: if (start) {
393: curp->line = lnn;
394: curp->reparse_count = 0;
1.93 schwarze 395:
396: if (lnn < 3 &&
397: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 398: curp->filenc & MPARSE_LATIN1)
399: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 400: }
401:
1.95 schwarze 402: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 403:
404: /*
405: * When finding an unescaped newline character,
406: * leave the character loop to process the line.
407: * Skip a preceding carriage return, if any.
408: */
409:
1.95 schwarze 410: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 411: '\n' == blk.buf[i + 1])
412: ++i;
413: if ('\n' == blk.buf[i]) {
414: ++i;
415: ++lnn;
416: break;
417: }
418:
1.35 schwarze 419: /*
1.93 schwarze 420: * Make sure we have space for the worst
1.198 schwarze 421: * case of 12 bytes: "\\[u10ffff]\n\0"
1.35 schwarze 422: */
423:
1.198 schwarze 424: if (pos + 12 > ln.sz)
1.35 schwarze 425: resize_buf(&ln, 256);
426:
1.47 schwarze 427: /*
1.93 schwarze 428: * Encode 8-bit input.
1.1 kristaps 429: */
430:
1.93 schwarze 431: c = blk.buf[i];
432: if (c & 0x80) {
1.95 schwarze 433: if ( ! (curp->filenc && preconv_encode(
434: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 435: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
436: curp->line, pos, "0x%x", c);
1.93 schwarze 437: ln.buf[pos++] = '?';
438: i++;
439: }
440: continue;
441: }
442:
443: /*
444: * Exclude control characters.
445: */
1.1 kristaps 446:
1.93 schwarze 447: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 448: mandoc_vmsg(c == 0x00 || c == 0x04 ||
449: c > 0x0a ? MANDOCERR_CHAR_BAD :
450: MANDOCERR_CHAR_UNSUPP,
451: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 452: i++;
1.127 schwarze 453: if (c != '\r')
454: ln.buf[pos++] = '?';
1.1 kristaps 455: continue;
456: }
457:
458: ln.buf[pos++] = blk.buf[i++];
459: }
1.198 schwarze 460: ln.buf[pos] = '\0';
461:
462: /*
463: * Maintain a lookaside buffer of all lines.
464: * parsed from this input source.
465: */
466:
467: thisln = mandoc_malloc(sizeof(*thisln));
468: thisln->buf = mandoc_strdup(ln.buf);
469: thisln->sz = strlen(ln.buf) + 1;
470: thisln->next = NULL;
471: if (firstln == NULL) {
472: firstln = lastln = thisln;
473: if (curp->secondary == NULL)
474: curp->secondary = firstln;
475: } else {
476: lastln->next = thisln;
477: lastln = thisln;
478: }
1.1 kristaps 479:
1.198 schwarze 480: /* XXX Ugly hack to mark the end of the input. */
1.1 kristaps 481:
1.198 schwarze 482: if (i == blk.sz || blk.buf[i] == '\0') {
1.170 schwarze 483: ln.buf[pos++] = '\n';
1.198 schwarze 484: ln.buf[pos] = '\0';
485: }
1.1 kristaps 486:
487: /*
488: * A significant amount of complexity is contained by
489: * the roff preprocessor. It's line-oriented but can be
490: * expressed on one line, so we need at times to
491: * readjust our starting point and re-run it. The roff
492: * preprocessor can also readjust the buffers with new
493: * data, so we pass them in wholesale.
494: */
495:
496: of = 0;
497: rerun:
1.197 schwarze 498: line_result = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 499:
1.199 schwarze 500: /* Process options. */
501:
502: if (line_result & ROFF_APPEND)
503: assert(line_result == (ROFF_IGN | ROFF_APPEND));
504:
505: if (line_result & ROFF_USERCALL)
506: assert((line_result & ROFF_MASK) == ROFF_REPARSE);
507:
508: if (line_result & ROFF_USERRET) {
509: assert(line_result == (ROFF_IGN | ROFF_USERRET));
510: if (start == 0) {
511: /* Return from the current macro. */
512: result = ROFF_USERRET;
513: goto out;
514: }
515: }
516:
517: switch (line_result & ROFF_LOOPMASK) {
518: case ROFF_IGN:
519: break;
520: case ROFF_WHILE:
521: if (curp->loop != NULL) {
522: if (loop == curp->loop)
523: break;
524: mandoc_msg(MANDOCERR_WHILE_NEST,
525: curp, curp->line, pos, NULL);
526: }
527: curp->loop = thisln;
528: loop = NULL;
529: inloop = 1;
530: break;
531: case ROFF_LOOPCONT:
532: case ROFF_LOOPEXIT:
533: if (curp->loop == NULL) {
534: mandoc_msg(MANDOCERR_WHILE_FAIL,
535: curp, curp->line, pos, NULL);
536: break;
537: }
538: if (inloop == 0) {
539: mandoc_msg(MANDOCERR_WHILE_INTO,
540: curp, curp->line, pos, NULL);
541: curp->loop = loop = NULL;
542: break;
543: }
544: if (line_result & ROFF_LOOPCONT)
545: loop = curp->loop;
546: else {
547: curp->loop = loop = NULL;
548: inloop = 0;
549: }
550: break;
551: default:
552: abort();
553: }
554:
555: /* Process the main instruction from the roff parser. */
556:
557: switch (line_result & ROFF_MASK) {
558: case ROFF_IGN:
559: break;
560: case ROFF_CONT:
561: if (curp->man->macroset == MACROSET_NONE)
562: choose_parser(curp);
563: if ((curp->man->macroset == MACROSET_MDOC ?
564: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
565: man_parseln(curp->man, curp->line, ln.buf, of)
566: ) == 2)
567: goto out;
568: break;
569: case ROFF_RERUN:
570: goto rerun;
1.47 schwarze 571: case ROFF_REPARSE:
1.197 schwarze 572: if (++curp->reparse_count > REPARSE_LIMIT) {
1.199 schwarze 573: /* Abort and return to the top level. */
1.198 schwarze 574: result = ROFF_IGN;
1.3 kristaps 575: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 576: curp->line, pos, NULL);
1.199 schwarze 577: goto out;
1.197 schwarze 578: }
1.199 schwarze 579: result = mparse_buf_r(curp, ln, of, 0);
580: if (line_result & ROFF_USERCALL) {
581: roff_userret(curp->roff);
582: /* Continue normally. */
583: if (result & ROFF_USERRET)
584: result = ROFF_CONT;
1.163 schwarze 585: }
1.199 schwarze 586: if (start == 0 && result != ROFF_CONT)
587: goto out;
588: break;
1.47 schwarze 589: case ROFF_SO:
1.95 schwarze 590: if ( ! (curp->options & MPARSE_SO) &&
591: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 592: curp->sodest = mandoc_strdup(ln.buf + of);
1.198 schwarze 593: goto out;
1.45 schwarze 594: }
1.113 schwarze 595: save_file = curp->file;
1.148 schwarze 596: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 597: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 598: close(fd);
1.113 schwarze 599: curp->file = save_file;
600: } else {
601: curp->file = save_file;
1.52 schwarze 602: mandoc_vmsg(MANDOCERR_SO_FAIL,
603: curp, curp->line, pos,
604: ".so %s", ln.buf + of);
1.110 schwarze 605: ln.sz = mandoc_asprintf(&cp,
606: ".sp\nSee the file %s.\n.sp",
607: ln.buf + of);
608: free(ln.buf);
609: ln.buf = cp;
610: of = 0;
611: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 612: }
1.199 schwarze 613: break;
1.1 kristaps 614: default:
1.199 schwarze 615: abort();
1.1 kristaps 616: }
617:
1.199 schwarze 618: /* Start the next input line. */
1.1 kristaps 619:
1.199 schwarze 620: if (loop != NULL &&
621: (line_result & ROFF_LOOPMASK) == ROFF_IGN)
622: loop = loop->next;
623:
624: if (loop != NULL) {
625: if ((line_result & ROFF_APPEND) == 0)
626: *ln.buf = '\0';
627: if (ln.sz < loop->sz)
628: resize_buf(&ln, loop->sz);
629: (void)strlcat(ln.buf, loop->buf, ln.sz);
630: of = 0;
631: goto rerun;
632: }
1.1 kristaps 633:
1.199 schwarze 634: pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
1.1 kristaps 635: }
1.198 schwarze 636: out:
1.199 schwarze 637: if (inloop) {
638: if (result != ROFF_USERRET)
639: mandoc_msg(MANDOCERR_WHILE_OUTOF, curp,
640: curp->line, pos, NULL);
641: curp->loop = NULL;
642: }
1.1 kristaps 643: free(ln.buf);
1.198 schwarze 644: if (firstln != curp->secondary)
645: free_buf_list(firstln);
646: return result;
1.1 kristaps 647: }
648:
649: static int
1.40 schwarze 650: read_whole_file(struct mparse *curp, const char *file, int fd,
651: struct buf *fb, int *with_mmap)
1.1 kristaps 652: {
1.161 schwarze 653: struct stat st;
1.140 schwarze 654: gzFile gz;
1.1 kristaps 655: size_t off;
656: ssize_t ssz;
1.194 schwarze 657: int gzerrnum, retval;
1.143 schwarze 658:
1.192 schwarze 659: if (fstat(fd, &st) == -1) {
660: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
661: "fstat: %s", strerror(errno));
662: return 0;
663: }
1.1 kristaps 664:
665: /*
666: * If we're a regular file, try just reading in the whole entry
667: * via mmap(). This is faster than reading it into blocks, and
668: * since each file is only a few bytes to begin with, I'm not
669: * concerned that this is going to tank any machines.
670: */
671:
1.140 schwarze 672: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 673: if (st.st_size > 0x7fffffff) {
1.111 schwarze 674: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 675: return 0;
1.1 kristaps 676: }
677: *with_mmap = 1;
678: fb->sz = (size_t)st.st_size;
1.37 schwarze 679: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 680: if (fb->buf != MAP_FAILED)
1.142 schwarze 681: return 1;
1.1 kristaps 682: }
683:
1.140 schwarze 684: if (curp->gzip) {
1.194 schwarze 685: /*
686: * Duplicating the file descriptor is required
687: * because we will have to call gzclose(3)
688: * to free memory used internally by zlib,
689: * but that will also close the file descriptor,
690: * which this function must not do.
691: */
692: if ((fd = dup(fd)) == -1) {
693: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
694: "dup: %s", strerror(errno));
695: return 0;
696: }
1.192 schwarze 697: if ((gz = gzdopen(fd, "rb")) == NULL) {
698: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
699: "gzdopen: %s", strerror(errno));
1.194 schwarze 700: close(fd);
1.192 schwarze 701: return 0;
702: }
1.140 schwarze 703: } else
704: gz = NULL;
705:
1.1 kristaps 706: /*
707: * If this isn't a regular file (like, say, stdin), then we must
708: * go the old way and just read things in bit by bit.
709: */
710:
711: *with_mmap = 0;
712: off = 0;
1.194 schwarze 713: retval = 0;
1.1 kristaps 714: fb->sz = 0;
715: fb->buf = NULL;
716: for (;;) {
717: if (off == fb->sz) {
718: if (fb->sz == (1U << 31)) {
1.111 schwarze 719: mandoc_msg(MANDOCERR_TOOLARGE, curp,
720: 0, 0, NULL);
1.1 kristaps 721: break;
722: }
723: resize_buf(fb, 65536);
724: }
1.140 schwarze 725: ssz = curp->gzip ?
726: gzread(gz, fb->buf + (int)off, fb->sz - off) :
727: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 728: if (ssz == 0) {
729: fb->sz = off;
1.194 schwarze 730: retval = 1;
731: break;
1.1 kristaps 732: }
1.192 schwarze 733: if (ssz == -1) {
1.194 schwarze 734: if (curp->gzip)
735: (void)gzerror(gz, &gzerrnum);
736: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "read: %s",
737: curp->gzip && gzerrnum != Z_ERRNO ?
738: zError(gzerrnum) : strerror(errno));
1.192 schwarze 739: break;
740: }
1.1 kristaps 741: off += (size_t)ssz;
742: }
743:
1.194 schwarze 744: if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
745: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "gzclose: %s",
746: gzerrnum == Z_ERRNO ? strerror(errno) :
747: zError(gzerrnum));
748: if (retval == 0) {
749: free(fb->buf);
750: fb->buf = NULL;
751: }
752: return retval;
1.1 kristaps 753: }
754:
755: static void
756: mparse_end(struct mparse *curp)
757: {
1.135 schwarze 758: if (curp->man->macroset == MACROSET_NONE)
759: curp->man->macroset = MACROSET_MAN;
760: if (curp->man->macroset == MACROSET_MDOC)
761: mdoc_endparse(curp->man);
762: else
1.111 schwarze 763: man_endparse(curp->man);
1.1 kristaps 764: roff_endparse(curp->roff);
765: }
766:
767: static void
1.36 schwarze 768: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 769: {
1.85 schwarze 770: struct buf *svprimary;
1.28 joerg 771: const char *svfile;
1.95 schwarze 772: size_t offset;
1.36 schwarze 773: static int recursion_depth;
774:
775: if (64 < recursion_depth) {
776: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
777: return;
778: }
1.28 joerg 779:
780: /* Line number is per-file. */
781: svfile = curp->file;
782: curp->file = file;
1.85 schwarze 783: svprimary = curp->primary;
1.83 schwarze 784: curp->primary = &blk;
1.28 joerg 785: curp->line = 1;
1.36 schwarze 786: recursion_depth++;
1.28 joerg 787:
1.93 schwarze 788: /* Skip an UTF-8 byte order mark. */
789: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
790: (unsigned char)blk.buf[0] == 0xef &&
791: (unsigned char)blk.buf[1] == 0xbb &&
792: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 793: offset = 3;
1.93 schwarze 794: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 795: } else
796: offset = 0;
1.93 schwarze 797:
1.95 schwarze 798: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 799:
1.111 schwarze 800: if (--recursion_depth == 0)
1.28 joerg 801: mparse_end(curp);
802:
1.85 schwarze 803: curp->primary = svprimary;
1.28 joerg 804: curp->file = svfile;
805: }
806:
807: enum mandoclevel
1.104 schwarze 808: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 809: const char *file)
810: {
811: struct buf blk;
812:
1.104 schwarze 813: blk.buf = buf;
1.28 joerg 814: blk.sz = len;
815:
1.36 schwarze 816: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 817: return curp->file_status;
1.28 joerg 818: }
819:
1.98 schwarze 820: /*
821: * Read the whole file into memory and call the parsers.
822: * Called recursively when an .so request is encountered.
823: */
1.36 schwarze 824: enum mandoclevel
825: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 826: {
1.28 joerg 827: struct buf blk;
828: int with_mmap;
1.93 schwarze 829: int save_filenc;
1.1 kristaps 830:
1.91 schwarze 831: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 832: save_filenc = curp->filenc;
833: curp->filenc = curp->options &
834: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 835: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 836: curp->filenc = save_filenc;
1.91 schwarze 837: if (with_mmap)
838: munmap(blk.buf, blk.sz);
839: else
840: free(blk.buf);
841: }
1.142 schwarze 842: return curp->file_status;
1.82 schwarze 843: }
844:
1.148 schwarze 845: int
846: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 847: {
848: char *cp;
1.148 schwarze 849: int fd;
1.82 schwarze 850:
851: curp->file = file;
1.140 schwarze 852: cp = strrchr(file, '.');
853: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 854:
1.140 schwarze 855: /* First try to use the filename as it is. */
1.98 schwarze 856:
1.148 schwarze 857: if ((fd = open(file, O_RDONLY)) != -1)
858: return fd;
1.98 schwarze 859:
1.140 schwarze 860: /*
861: * If that doesn't work and the filename doesn't
862: * already end in .gz, try appending .gz.
863: */
1.98 schwarze 864:
1.140 schwarze 865: if ( ! curp->gzip) {
1.98 schwarze 866: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 867: fd = open(cp, O_RDONLY);
1.108 schwarze 868: free(cp);
1.148 schwarze 869: if (fd != -1) {
1.140 schwarze 870: curp->gzip = 1;
1.148 schwarze 871: return fd;
1.82 schwarze 872: }
873: }
874:
1.140 schwarze 875: /* Neither worked, give up. */
1.97 schwarze 876:
1.140 schwarze 877: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 878: return -1;
1.1 kristaps 879: }
880:
881: struct mparse *
1.179 schwarze 882: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
883: enum mandoc_os os_e, const char *os_s)
1.1 kristaps 884: {
885: struct mparse *curp;
1.10 kristaps 886:
1.1 kristaps 887: curp = mandoc_calloc(1, sizeof(struct mparse));
888:
1.44 schwarze 889: curp->options = options;
1.179 schwarze 890: curp->mmin = mmin;
1.1 kristaps 891: curp->mmsg = mmsg;
1.179 schwarze 892: curp->os_s = os_s;
1.1 kristaps 893:
1.144 schwarze 894: curp->roff = roff_alloc(curp, options);
1.179 schwarze 895: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.137 schwarze 896: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 897: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 898: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 899: if (curp->man->mdocmac == NULL)
900: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 901: } else if (curp->options & MPARSE_MAN) {
902: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 903: if (curp->man->manmac == NULL)
904: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 905: }
1.138 schwarze 906: curp->man->first->tok = TOKEN_NONE;
1.179 schwarze 907: curp->man->meta.os_e = os_e;
1.142 schwarze 908: return curp;
1.1 kristaps 909: }
910:
911: void
912: mparse_reset(struct mparse *curp)
913: {
914: roff_reset(curp->roff);
1.150 schwarze 915: roff_man_reset(curp->man);
1.198 schwarze 916: free_buf_list(curp->secondary);
917: curp->secondary = NULL;
1.160 schwarze 918:
919: free(curp->sodest);
920: curp->sodest = NULL;
921:
1.1 kristaps 922: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 923: curp->gzip = 0;
1.1 kristaps 924: }
925:
926: void
927: mparse_free(struct mparse *curp)
928: {
1.164 schwarze 929: roffhash_free(curp->man->mdocmac);
930: roffhash_free(curp->man->manmac);
1.137 schwarze 931: roff_man_free(curp->man);
1.160 schwarze 932: roff_free(curp->roff);
1.198 schwarze 933: free_buf_list(curp->secondary);
1.45 schwarze 934: free(curp->sodest);
1.1 kristaps 935: free(curp);
936: }
937:
938: void
1.135 schwarze 939: mparse_result(struct mparse *curp, struct roff_man **man,
940: char **sodest)
1.1 kristaps 941: {
942:
1.45 schwarze 943: if (sodest && NULL != (*sodest = curp->sodest)) {
944: *man = NULL;
945: return;
946: }
1.9 kristaps 947: if (man)
948: *man = curp->man;
1.157 schwarze 949: }
950:
951: void
952: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
953: {
954: if (curp->file_status > *rc)
955: *rc = curp->file_status;
1.3 kristaps 956: }
957:
958: void
959: mandoc_vmsg(enum mandocerr t, struct mparse *m,
960: int ln, int pos, const char *fmt, ...)
961: {
962: char buf[256];
963: va_list ap;
964:
965: va_start(ap, fmt);
1.48 schwarze 966: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 967: va_end(ap);
968:
969: mandoc_msg(t, m, ln, pos, buf);
970: }
971:
972: void
1.47 schwarze 973: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 974: int ln, int col, const char *msg)
975: {
976: enum mandoclevel level;
977:
1.179 schwarze 978: if (er < m->mmin && er != MANDOCERR_FILE)
979: return;
980:
1.112 schwarze 981: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 982: while (er < mandoclimits[level])
983: level--;
984:
1.8 kristaps 985: if (m->mmsg)
986: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 987:
988: if (m->file_status < level)
989: m->file_status = level;
1.7 kristaps 990: }
991:
992: const char *
993: mparse_strerror(enum mandocerr er)
994: {
995:
1.142 schwarze 996: return mandocerrs[er];
1.7 kristaps 997: }
998:
999: const char *
1000: mparse_strlevel(enum mandoclevel lvl)
1001: {
1.142 schwarze 1002: return mandoclevels[lvl];
1.24 kristaps 1003: }
1004:
1005: void
1.198 schwarze 1006: mparse_copy(const struct mparse *p)
1.24 kristaps 1007: {
1.198 schwarze 1008: struct buf *buf;
1.24 kristaps 1009:
1.198 schwarze 1010: for (buf = p->secondary; buf != NULL; buf = buf->next)
1011: puts(buf->buf);
1.1 kristaps 1012: }
CVSweb