Annotation of mandoc/read.c, Revision 1.199
1.199 ! schwarze 1: /* $Id: read.c,v 1.198 2018/08/23 19:33:27 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"
1.1 kristaps 42:
43: #define REPARSE_LIMIT 1000
44:
45: struct mparse {
1.160 schwarze 46: struct roff *roff; /* roff parser (!NULL) */
1.134 schwarze 47: struct roff_man *man; /* man parser */
1.45 schwarze 48: char *sodest; /* filename pointed to by .so */
1.83 schwarze 49: const char *file; /* filename of current input file */
50: struct buf *primary; /* buffer currently being parsed */
1.198 schwarze 51: struct buf *secondary; /* copy of top level input */
1.199 ! schwarze 52: struct buf *loop; /* open .while request line */
1.179 schwarze 53: const char *os_s; /* default operating system */
1.83 schwarze 54: mandocmsg mmsg; /* warning/error message handler */
55: enum mandoclevel file_status; /* status of current parse */
1.179 schwarze 56: enum mandocerr mmin; /* ignore messages below this */
1.83 schwarze 57: int options; /* parser options */
1.140 schwarze 58: int gzip; /* current input file is gzipped */
1.93 schwarze 59: int filenc; /* encoding of the current file */
1.1 kristaps 60: int reparse_count; /* finite interp. stack */
1.83 schwarze 61: int line; /* line number in the file */
1.1 kristaps 62: };
63:
1.84 schwarze 64: static void choose_parser(struct mparse *);
1.198 schwarze 65: static void free_buf_list(struct buf *);
1.1 kristaps 66: static void resize_buf(struct buf *, size_t);
1.199 ! schwarze 67: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 68: static int read_whole_file(struct mparse *, const char *, int,
69: struct buf *, int *);
1.1 kristaps 70: static void mparse_end(struct mparse *);
1.37 schwarze 71: static void mparse_parse_buffer(struct mparse *, struct buf,
72: const char *);
1.1 kristaps 73:
1.3 kristaps 74: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
75: MANDOCERR_OK,
1.188 schwarze 76: MANDOCERR_OK,
1.3 kristaps 77: MANDOCERR_WARNING,
78: MANDOCERR_ERROR,
1.112 schwarze 79: MANDOCERR_UNSUPP,
1.3 kristaps 80: MANDOCERR_MAX,
81: MANDOCERR_MAX
82: };
83:
1.7 kristaps 84: static const char * const mandocerrs[MANDOCERR_MAX] = {
85: "ok",
86:
1.179 schwarze 87: "base system convention",
1.166 schwarze 88:
1.176 schwarze 89: "Mdocdate found",
90: "Mdocdate missing",
1.181 schwarze 91: "unknown architecture",
1.180 schwarze 92: "operating system explicitly specified",
1.179 schwarze 93: "RCS id missing",
1.184 schwarze 94: "referenced manual not found",
1.179 schwarze 95:
96: "generic style suggestion",
97:
1.176 schwarze 98: "legacy man(7) date format",
1.196 schwarze 99: "normalizing date format to",
1.189 schwarze 100: "lower case character in document title",
1.178 schwarze 101: "duplicate RCS id",
1.193 schwarze 102: "possible typo in section name",
1.189 schwarze 103: "unterminated quoted argument",
1.166 schwarze 104: "useless macro",
1.167 schwarze 105: "consider using OS macro",
1.172 schwarze 106: "errnos out of order",
107: "duplicate errno",
1.187 schwarze 108: "trailing delimiter",
1.174 schwarze 109: "no blank before trailing delimiter",
1.189 schwarze 110: "fill mode already enabled, skipping",
111: "fill mode already disabled, skipping",
1.195 schwarze 112: "verbatim \"--\", maybe consider using \\(em",
1.175 schwarze 113: "function name without markup",
1.189 schwarze 114: "whitespace at end of input line",
115: "bad comment style",
1.166 schwarze 116:
1.7 kristaps 117: "generic warning",
118:
119: /* related to the prologue */
1.79 schwarze 120: "missing manual title, using UNTITLED",
121: "missing manual title, using \"\"",
122: "missing manual section, using \"\"",
1.7 kristaps 123: "unknown manual section",
1.54 schwarze 124: "missing date, using today's date",
1.7 kristaps 125: "cannot parse date, using it verbatim",
1.186 schwarze 126: "date in the future, using it anyway",
1.79 schwarze 127: "missing Os macro, using \"\"",
128: "late prologue macro",
1.7 kristaps 129: "prologue macros out of order",
130:
131: /* related to document structure */
132: ".so is fragile, better use ln(1)",
1.50 schwarze 133: "no document body",
1.54 schwarze 134: "content before first section header",
135: "first section is not \"NAME\"",
1.156 schwarze 136: "NAME section without Nm before Nd",
1.128 schwarze 137: "NAME section without description",
138: "description not at the end of NAME",
139: "bad NAME section content",
1.156 schwarze 140: "missing comma before name",
1.121 schwarze 141: "missing description line, using \"\"",
1.162 schwarze 142: "description line outside NAME section",
1.7 kristaps 143: "sections out of conventional order",
1.54 schwarze 144: "duplicate section title",
145: "unexpected section",
1.185 schwarze 146: "cross reference to self",
1.87 schwarze 147: "unusual Xr order",
148: "unusual Xr punctuation",
1.86 schwarze 149: "AUTHORS section without An macro",
1.7 kristaps 150:
151: /* related to macros and nesting */
1.55 schwarze 152: "obsolete macro",
1.102 schwarze 153: "macro neither callable nor escaped",
1.7 kristaps 154: "skipping paragraph macro",
1.31 schwarze 155: "moving paragraph macro out of list",
1.7 kristaps 156: "skipping no-space macro",
157: "blocks badly nested",
158: "nested displays are not portable",
1.57 schwarze 159: "moving content out of list",
1.183 schwarze 160: "first macro on line",
1.7 kristaps 161: "line scope broken",
1.169 schwarze 162: "skipping blank line in line scope",
1.7 kristaps 163:
164: /* related to missing macro arguments */
1.58 schwarze 165: "skipping empty request",
166: "conditional request controls empty scope",
1.7 kristaps 167: "skipping empty macro",
1.124 schwarze 168: "empty block",
1.62 schwarze 169: "empty argument, using 0n",
1.60 schwarze 170: "missing display type, using -ragged",
171: "list type is not the first argument",
1.152 schwarze 172: "missing -width in -tag list, using 6n",
1.78 schwarze 173: "missing utility name, using \"\"",
1.123 schwarze 174: "missing function name, using \"\"",
1.60 schwarze 175: "empty head in list item",
176: "empty list item",
1.183 schwarze 177: "missing argument, using next line",
1.61 schwarze 178: "missing font type, using \\fR",
179: "unknown font type, using \\fR",
1.103 schwarze 180: "nothing follows prefix",
1.122 schwarze 181: "empty reference block",
1.155 schwarze 182: "missing section argument",
1.60 schwarze 183: "missing -std argument, adding it",
1.125 schwarze 184: "missing option string, using \"\"",
185: "missing resource identifier, using \"\"",
1.90 schwarze 186: "missing eqn box, using \"\"",
1.7 kristaps 187:
188: /* related to bad macro arguments */
189: "duplicate argument",
1.76 schwarze 190: "skipping duplicate argument",
1.63 schwarze 191: "skipping duplicate display type",
192: "skipping duplicate list type",
1.76 schwarze 193: "skipping -width argument",
1.126 schwarze 194: "wrong number of cells",
1.7 kristaps 195: "unknown AT&T UNIX version",
1.88 schwarze 196: "comma in function argument",
1.89 schwarze 197: "parenthesis in function name",
1.173 schwarze 198: "unknown library name",
1.67 schwarze 199: "invalid content in Rs block",
1.63 schwarze 200: "invalid Boolean argument",
201: "unknown font, skipping request",
1.126 schwarze 202: "odd number of characters in request",
1.7 kristaps 203:
204: /* related to plain text */
1.64 schwarze 205: "blank line in fill mode, using .sp",
206: "tab in filled text",
1.158 schwarze 207: "new sentence, new line",
1.64 schwarze 208: "invalid escape sequence",
209: "undefined string, using \"\"",
1.16 kristaps 210:
1.118 schwarze 211: /* related to tables */
212: "tbl line starts with span",
213: "tbl column starts with span",
214: "skipping vertical bar in tbl layout",
215:
1.7 kristaps 216: "generic error",
217:
218: /* related to tables */
1.116 schwarze 219: "non-alphabetic character in tbl options",
220: "skipping unknown tbl option",
221: "missing tbl option argument",
222: "wrong tbl option argument size",
1.118 schwarze 223: "empty tbl layout",
224: "invalid character in tbl layout",
225: "unmatched parenthesis in tbl layout",
1.119 schwarze 226: "tbl without any data cells",
227: "ignoring data in spanned tbl cell",
228: "ignoring extra tbl data cells",
229: "data block open at end of tbl",
1.7 kristaps 230:
1.68 schwarze 231: /* related to document structure and macros */
1.108 schwarze 232: NULL,
1.189 schwarze 233: "duplicate prologue macro",
234: "skipping late title macro",
1.7 kristaps 235: "input stack limit exceeded, infinite loop?",
236: "skipping bad character",
1.68 schwarze 237: "skipping unknown macro",
1.197 schwarze 238: "ignoring request outside macro",
1.112 schwarze 239: "skipping insecure request",
1.73 schwarze 240: "skipping item outside list",
1.68 schwarze 241: "skipping column outside column list",
242: "skipping end of block that is not open",
1.115 schwarze 243: "fewer RS blocks open, skipping",
1.68 schwarze 244: "inserting missing end of block",
245: "appending missing end of block",
246:
247: /* related to request and macro arguments */
1.7 kristaps 248: "escaped character not allowed in a name",
1.197 schwarze 249: "using macro argument outside macro",
250: "argument number is not numeric",
1.99 schwarze 251: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 252: "skipping display without arguments",
1.71 schwarze 253: "missing list type, using -item",
1.171 schwarze 254: "argument is not numeric, using 1",
1.70 schwarze 255: "missing manual name, using \"\"",
1.71 schwarze 256: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 257: "unknown standard specifier",
1.71 schwarze 258: "skipping request without numeric argument",
1.197 schwarze 259: "excessive shift",
1.109 schwarze 260: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 261: ".so request failed",
1.61 schwarze 262: "skipping all arguments",
263: "skipping excess arguments",
1.92 kristaps 264: "divide by zero",
1.112 schwarze 265:
266: "unsupported feature",
267: "input too large",
1.114 schwarze 268: "unsupported control character",
1.112 schwarze 269: "unsupported roff request",
1.199 ! schwarze 270: "nested .while loops",
! 271: "end of scope with open .while loop",
! 272: "end of .while loop in inner scope",
! 273: "cannot continue this .while loop",
1.119 schwarze 274: "eqn delim option in tbl",
1.118 schwarze 275: "unsupported tbl layout modifier",
1.112 schwarze 276: "ignoring macro in table",
1.7 kristaps 277: };
278:
279: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
280: "SUCCESS",
1.165 schwarze 281: "STYLE",
1.7 kristaps 282: "WARNING",
283: "ERROR",
1.112 schwarze 284: "UNSUPP",
1.7 kristaps 285: "BADARG",
286: "SYSERR"
287: };
288:
1.47 schwarze 289:
1.1 kristaps 290: static void
291: resize_buf(struct buf *buf, size_t initial)
292: {
293:
294: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
295: buf->buf = mandoc_realloc(buf->buf, buf->sz);
296: }
297:
298: static void
1.198 schwarze 299: free_buf_list(struct buf *buf)
300: {
301: struct buf *tmp;
302:
303: while (buf != NULL) {
304: tmp = buf;
305: buf = tmp->next;
306: free(tmp->buf);
307: free(tmp);
308: }
309: }
310:
311: static void
1.84 schwarze 312: choose_parser(struct mparse *curp)
1.1 kristaps 313: {
1.83 schwarze 314: char *cp, *ep;
315: int format;
1.1 kristaps 316:
1.83 schwarze 317: /*
318: * If neither command line arguments -mdoc or -man select
319: * a parser nor the roff parser found a .Dd or .TH macro
320: * yet, look ahead in the main input buffer.
321: */
322:
323: if ((format = roff_getformat(curp->roff)) == 0) {
324: cp = curp->primary->buf;
325: ep = cp + curp->primary->sz;
326: while (cp < ep) {
1.85 schwarze 327: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 328: cp++;
329: if (cp[0] == 'D' && cp[1] == 'd') {
330: format = MPARSE_MDOC;
331: break;
332: }
333: if (cp[0] == 'T' && cp[1] == 'H') {
334: format = MPARSE_MAN;
335: break;
336: }
337: }
338: cp = memchr(cp, '\n', ep - cp);
339: if (cp == NULL)
340: break;
341: cp++;
342: }
1.1 kristaps 343: }
344:
1.83 schwarze 345: if (format == MPARSE_MDOC) {
1.137 schwarze 346: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 347: if (curp->man->mdocmac == NULL)
348: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 349: } else {
350: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 351: if (curp->man->manmac == NULL)
352: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 353: }
1.164 schwarze 354: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 355: }
356:
357: /*
1.95 schwarze 358: * Main parse routine for a buffer.
359: * It assumes encoding and line numbering are already set up.
360: * It can recurse directly (for invocations of user-defined
361: * macros, inline equations, and input line traps)
362: * and indirectly (for .so file inclusion).
1.1 kristaps 363: */
1.199 ! schwarze 364: static int
1.95 schwarze 365: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 366: {
367: struct buf ln;
1.199 ! schwarze 368: struct buf *firstln, *lastln, *thisln, *loop;
1.113 schwarze 369: const char *save_file;
1.110 schwarze 370: char *cp;
1.95 schwarze 371: size_t pos; /* byte number in the ln buffer */
1.199 ! schwarze 372: int line_result, result;
1.100 schwarze 373: int of;
1.1 kristaps 374: int lnn; /* line number in the real file */
1.110 schwarze 375: int fd;
1.199 ! schwarze 376: int inloop; /* Saw .while on this level. */
1.1 kristaps 377: unsigned char c;
378:
1.198 schwarze 379: ln.sz = 256;
380: ln.buf = mandoc_malloc(ln.sz);
381: ln.next = NULL;
1.199 ! schwarze 382: firstln = loop = NULL;
1.47 schwarze 383: lnn = curp->line;
384: pos = 0;
1.199 ! schwarze 385: inloop = 0;
1.198 schwarze 386: result = ROFF_CONT;
1.1 kristaps 387:
1.199 ! schwarze 388: while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) {
1.1 kristaps 389: if (start) {
390: curp->line = lnn;
391: curp->reparse_count = 0;
1.93 schwarze 392:
393: if (lnn < 3 &&
394: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 395: curp->filenc & MPARSE_LATIN1)
396: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 397: }
398:
1.95 schwarze 399: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 400:
401: /*
402: * When finding an unescaped newline character,
403: * leave the character loop to process the line.
404: * Skip a preceding carriage return, if any.
405: */
406:
1.95 schwarze 407: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 408: '\n' == blk.buf[i + 1])
409: ++i;
410: if ('\n' == blk.buf[i]) {
411: ++i;
412: ++lnn;
413: break;
414: }
415:
1.35 schwarze 416: /*
1.93 schwarze 417: * Make sure we have space for the worst
1.198 schwarze 418: * case of 12 bytes: "\\[u10ffff]\n\0"
1.35 schwarze 419: */
420:
1.198 schwarze 421: if (pos + 12 > ln.sz)
1.35 schwarze 422: resize_buf(&ln, 256);
423:
1.47 schwarze 424: /*
1.93 schwarze 425: * Encode 8-bit input.
1.1 kristaps 426: */
427:
1.93 schwarze 428: c = blk.buf[i];
429: if (c & 0x80) {
1.95 schwarze 430: if ( ! (curp->filenc && preconv_encode(
431: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 432: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
433: curp->line, pos, "0x%x", c);
1.93 schwarze 434: ln.buf[pos++] = '?';
435: i++;
436: }
437: continue;
438: }
439:
440: /*
441: * Exclude control characters.
442: */
1.1 kristaps 443:
1.93 schwarze 444: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 445: mandoc_vmsg(c == 0x00 || c == 0x04 ||
446: c > 0x0a ? MANDOCERR_CHAR_BAD :
447: MANDOCERR_CHAR_UNSUPP,
448: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 449: i++;
1.127 schwarze 450: if (c != '\r')
451: ln.buf[pos++] = '?';
1.1 kristaps 452: continue;
453: }
454:
455: ln.buf[pos++] = blk.buf[i++];
456: }
1.198 schwarze 457: ln.buf[pos] = '\0';
458:
459: /*
460: * Maintain a lookaside buffer of all lines.
461: * parsed from this input source.
462: */
463:
464: thisln = mandoc_malloc(sizeof(*thisln));
465: thisln->buf = mandoc_strdup(ln.buf);
466: thisln->sz = strlen(ln.buf) + 1;
467: thisln->next = NULL;
468: if (firstln == NULL) {
469: firstln = lastln = thisln;
470: if (curp->secondary == NULL)
471: curp->secondary = firstln;
472: } else {
473: lastln->next = thisln;
474: lastln = thisln;
475: }
1.1 kristaps 476:
1.198 schwarze 477: /* XXX Ugly hack to mark the end of the input. */
1.1 kristaps 478:
1.198 schwarze 479: if (i == blk.sz || blk.buf[i] == '\0') {
1.170 schwarze 480: ln.buf[pos++] = '\n';
1.198 schwarze 481: ln.buf[pos] = '\0';
482: }
1.1 kristaps 483:
484: /*
485: * A significant amount of complexity is contained by
486: * the roff preprocessor. It's line-oriented but can be
487: * expressed on one line, so we need at times to
488: * readjust our starting point and re-run it. The roff
489: * preprocessor can also readjust the buffers with new
490: * data, so we pass them in wholesale.
491: */
492:
493: of = 0;
494: rerun:
1.197 schwarze 495: line_result = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 496:
1.199 ! schwarze 497: /* Process options. */
! 498:
! 499: if (line_result & ROFF_APPEND)
! 500: assert(line_result == (ROFF_IGN | ROFF_APPEND));
! 501:
! 502: if (line_result & ROFF_USERCALL)
! 503: assert((line_result & ROFF_MASK) == ROFF_REPARSE);
! 504:
! 505: if (line_result & ROFF_USERRET) {
! 506: assert(line_result == (ROFF_IGN | ROFF_USERRET));
! 507: if (start == 0) {
! 508: /* Return from the current macro. */
! 509: result = ROFF_USERRET;
! 510: goto out;
! 511: }
! 512: }
! 513:
! 514: switch (line_result & ROFF_LOOPMASK) {
! 515: case ROFF_IGN:
! 516: break;
! 517: case ROFF_WHILE:
! 518: if (curp->loop != NULL) {
! 519: if (loop == curp->loop)
! 520: break;
! 521: mandoc_msg(MANDOCERR_WHILE_NEST,
! 522: curp, curp->line, pos, NULL);
! 523: }
! 524: curp->loop = thisln;
! 525: loop = NULL;
! 526: inloop = 1;
! 527: break;
! 528: case ROFF_LOOPCONT:
! 529: case ROFF_LOOPEXIT:
! 530: if (curp->loop == NULL) {
! 531: mandoc_msg(MANDOCERR_WHILE_FAIL,
! 532: curp, curp->line, pos, NULL);
! 533: break;
! 534: }
! 535: if (inloop == 0) {
! 536: mandoc_msg(MANDOCERR_WHILE_INTO,
! 537: curp, curp->line, pos, NULL);
! 538: curp->loop = loop = NULL;
! 539: break;
! 540: }
! 541: if (line_result & ROFF_LOOPCONT)
! 542: loop = curp->loop;
! 543: else {
! 544: curp->loop = loop = NULL;
! 545: inloop = 0;
! 546: }
! 547: break;
! 548: default:
! 549: abort();
! 550: }
! 551:
! 552: /* Process the main instruction from the roff parser. */
! 553:
! 554: switch (line_result & ROFF_MASK) {
! 555: case ROFF_IGN:
! 556: break;
! 557: case ROFF_CONT:
! 558: if (curp->man->macroset == MACROSET_NONE)
! 559: choose_parser(curp);
! 560: if ((curp->man->macroset == MACROSET_MDOC ?
! 561: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
! 562: man_parseln(curp->man, curp->line, ln.buf, of)
! 563: ) == 2)
! 564: goto out;
! 565: break;
! 566: case ROFF_RERUN:
! 567: goto rerun;
1.47 schwarze 568: case ROFF_REPARSE:
1.197 schwarze 569: if (++curp->reparse_count > REPARSE_LIMIT) {
1.199 ! schwarze 570: /* Abort and return to the top level. */
1.198 schwarze 571: result = ROFF_IGN;
1.3 kristaps 572: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 573: curp->line, pos, NULL);
1.199 ! schwarze 574: goto out;
1.197 schwarze 575: }
1.199 ! schwarze 576: result = mparse_buf_r(curp, ln, of, 0);
! 577: if (line_result & ROFF_USERCALL) {
! 578: roff_userret(curp->roff);
! 579: /* Continue normally. */
! 580: if (result & ROFF_USERRET)
! 581: result = ROFF_CONT;
1.163 schwarze 582: }
1.199 ! schwarze 583: if (start == 0 && result != ROFF_CONT)
! 584: goto out;
! 585: break;
1.47 schwarze 586: case ROFF_SO:
1.95 schwarze 587: if ( ! (curp->options & MPARSE_SO) &&
588: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 589: curp->sodest = mandoc_strdup(ln.buf + of);
1.198 schwarze 590: goto out;
1.45 schwarze 591: }
1.113 schwarze 592: save_file = curp->file;
1.148 schwarze 593: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 594: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 595: close(fd);
1.113 schwarze 596: curp->file = save_file;
597: } else {
598: curp->file = save_file;
1.52 schwarze 599: mandoc_vmsg(MANDOCERR_SO_FAIL,
600: curp, curp->line, pos,
601: ".so %s", ln.buf + of);
1.110 schwarze 602: ln.sz = mandoc_asprintf(&cp,
603: ".sp\nSee the file %s.\n.sp",
604: ln.buf + of);
605: free(ln.buf);
606: ln.buf = cp;
607: of = 0;
608: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 609: }
1.199 ! schwarze 610: break;
1.1 kristaps 611: default:
1.199 ! schwarze 612: abort();
1.1 kristaps 613: }
614:
1.199 ! schwarze 615: /* Start the next input line. */
1.1 kristaps 616:
1.199 ! schwarze 617: if (loop != NULL &&
! 618: (line_result & ROFF_LOOPMASK) == ROFF_IGN)
! 619: loop = loop->next;
! 620:
! 621: if (loop != NULL) {
! 622: if ((line_result & ROFF_APPEND) == 0)
! 623: *ln.buf = '\0';
! 624: if (ln.sz < loop->sz)
! 625: resize_buf(&ln, loop->sz);
! 626: (void)strlcat(ln.buf, loop->buf, ln.sz);
! 627: of = 0;
! 628: goto rerun;
! 629: }
1.1 kristaps 630:
1.199 ! schwarze 631: pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
1.1 kristaps 632: }
1.198 schwarze 633: out:
1.199 ! schwarze 634: if (inloop) {
! 635: if (result != ROFF_USERRET)
! 636: mandoc_msg(MANDOCERR_WHILE_OUTOF, curp,
! 637: curp->line, pos, NULL);
! 638: curp->loop = NULL;
! 639: }
1.1 kristaps 640: free(ln.buf);
1.198 schwarze 641: if (firstln != curp->secondary)
642: free_buf_list(firstln);
643: return result;
1.1 kristaps 644: }
645:
646: static int
1.40 schwarze 647: read_whole_file(struct mparse *curp, const char *file, int fd,
648: struct buf *fb, int *with_mmap)
1.1 kristaps 649: {
1.161 schwarze 650: struct stat st;
1.140 schwarze 651: gzFile gz;
1.1 kristaps 652: size_t off;
653: ssize_t ssz;
1.194 schwarze 654: int gzerrnum, retval;
1.143 schwarze 655:
1.192 schwarze 656: if (fstat(fd, &st) == -1) {
657: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
658: "fstat: %s", strerror(errno));
659: return 0;
660: }
1.1 kristaps 661:
662: /*
663: * If we're a regular file, try just reading in the whole entry
664: * via mmap(). This is faster than reading it into blocks, and
665: * since each file is only a few bytes to begin with, I'm not
666: * concerned that this is going to tank any machines.
667: */
668:
1.140 schwarze 669: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 670: if (st.st_size > 0x7fffffff) {
1.111 schwarze 671: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 672: return 0;
1.1 kristaps 673: }
674: *with_mmap = 1;
675: fb->sz = (size_t)st.st_size;
1.37 schwarze 676: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 677: if (fb->buf != MAP_FAILED)
1.142 schwarze 678: return 1;
1.1 kristaps 679: }
680:
1.140 schwarze 681: if (curp->gzip) {
1.194 schwarze 682: /*
683: * Duplicating the file descriptor is required
684: * because we will have to call gzclose(3)
685: * to free memory used internally by zlib,
686: * but that will also close the file descriptor,
687: * which this function must not do.
688: */
689: if ((fd = dup(fd)) == -1) {
690: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
691: "dup: %s", strerror(errno));
692: return 0;
693: }
1.192 schwarze 694: if ((gz = gzdopen(fd, "rb")) == NULL) {
695: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
696: "gzdopen: %s", strerror(errno));
1.194 schwarze 697: close(fd);
1.192 schwarze 698: return 0;
699: }
1.140 schwarze 700: } else
701: gz = NULL;
702:
1.1 kristaps 703: /*
704: * If this isn't a regular file (like, say, stdin), then we must
705: * go the old way and just read things in bit by bit.
706: */
707:
708: *with_mmap = 0;
709: off = 0;
1.194 schwarze 710: retval = 0;
1.1 kristaps 711: fb->sz = 0;
712: fb->buf = NULL;
713: for (;;) {
714: if (off == fb->sz) {
715: if (fb->sz == (1U << 31)) {
1.111 schwarze 716: mandoc_msg(MANDOCERR_TOOLARGE, curp,
717: 0, 0, NULL);
1.1 kristaps 718: break;
719: }
720: resize_buf(fb, 65536);
721: }
1.140 schwarze 722: ssz = curp->gzip ?
723: gzread(gz, fb->buf + (int)off, fb->sz - off) :
724: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 725: if (ssz == 0) {
726: fb->sz = off;
1.194 schwarze 727: retval = 1;
728: break;
1.1 kristaps 729: }
1.192 schwarze 730: if (ssz == -1) {
1.194 schwarze 731: if (curp->gzip)
732: (void)gzerror(gz, &gzerrnum);
733: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "read: %s",
734: curp->gzip && gzerrnum != Z_ERRNO ?
735: zError(gzerrnum) : strerror(errno));
1.192 schwarze 736: break;
737: }
1.1 kristaps 738: off += (size_t)ssz;
739: }
740:
1.194 schwarze 741: if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
742: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "gzclose: %s",
743: gzerrnum == Z_ERRNO ? strerror(errno) :
744: zError(gzerrnum));
745: if (retval == 0) {
746: free(fb->buf);
747: fb->buf = NULL;
748: }
749: return retval;
1.1 kristaps 750: }
751:
752: static void
753: mparse_end(struct mparse *curp)
754: {
1.135 schwarze 755: if (curp->man->macroset == MACROSET_NONE)
756: curp->man->macroset = MACROSET_MAN;
757: if (curp->man->macroset == MACROSET_MDOC)
758: mdoc_endparse(curp->man);
759: else
1.111 schwarze 760: man_endparse(curp->man);
1.1 kristaps 761: roff_endparse(curp->roff);
762: }
763:
764: static void
1.36 schwarze 765: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 766: {
1.85 schwarze 767: struct buf *svprimary;
1.28 joerg 768: const char *svfile;
1.95 schwarze 769: size_t offset;
1.36 schwarze 770: static int recursion_depth;
771:
772: if (64 < recursion_depth) {
773: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
774: return;
775: }
1.28 joerg 776:
777: /* Line number is per-file. */
778: svfile = curp->file;
779: curp->file = file;
1.85 schwarze 780: svprimary = curp->primary;
1.83 schwarze 781: curp->primary = &blk;
1.28 joerg 782: curp->line = 1;
1.36 schwarze 783: recursion_depth++;
1.28 joerg 784:
1.93 schwarze 785: /* Skip an UTF-8 byte order mark. */
786: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
787: (unsigned char)blk.buf[0] == 0xef &&
788: (unsigned char)blk.buf[1] == 0xbb &&
789: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 790: offset = 3;
1.93 schwarze 791: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 792: } else
793: offset = 0;
1.93 schwarze 794:
1.95 schwarze 795: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 796:
1.111 schwarze 797: if (--recursion_depth == 0)
1.28 joerg 798: mparse_end(curp);
799:
1.85 schwarze 800: curp->primary = svprimary;
1.28 joerg 801: curp->file = svfile;
802: }
803:
804: enum mandoclevel
1.104 schwarze 805: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 806: const char *file)
807: {
808: struct buf blk;
809:
1.104 schwarze 810: blk.buf = buf;
1.28 joerg 811: blk.sz = len;
812:
1.36 schwarze 813: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 814: return curp->file_status;
1.28 joerg 815: }
816:
1.98 schwarze 817: /*
818: * Read the whole file into memory and call the parsers.
819: * Called recursively when an .so request is encountered.
820: */
1.36 schwarze 821: enum mandoclevel
822: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 823: {
1.28 joerg 824: struct buf blk;
825: int with_mmap;
1.93 schwarze 826: int save_filenc;
1.1 kristaps 827:
1.91 schwarze 828: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 829: save_filenc = curp->filenc;
830: curp->filenc = curp->options &
831: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 832: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 833: curp->filenc = save_filenc;
1.91 schwarze 834: if (with_mmap)
835: munmap(blk.buf, blk.sz);
836: else
837: free(blk.buf);
838: }
1.142 schwarze 839: return curp->file_status;
1.82 schwarze 840: }
841:
1.148 schwarze 842: int
843: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 844: {
845: char *cp;
1.148 schwarze 846: int fd;
1.82 schwarze 847:
848: curp->file = file;
1.140 schwarze 849: cp = strrchr(file, '.');
850: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 851:
1.140 schwarze 852: /* First try to use the filename as it is. */
1.98 schwarze 853:
1.148 schwarze 854: if ((fd = open(file, O_RDONLY)) != -1)
855: return fd;
1.98 schwarze 856:
1.140 schwarze 857: /*
858: * If that doesn't work and the filename doesn't
859: * already end in .gz, try appending .gz.
860: */
1.98 schwarze 861:
1.140 schwarze 862: if ( ! curp->gzip) {
1.98 schwarze 863: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 864: fd = open(cp, O_RDONLY);
1.108 schwarze 865: free(cp);
1.148 schwarze 866: if (fd != -1) {
1.140 schwarze 867: curp->gzip = 1;
1.148 schwarze 868: return fd;
1.82 schwarze 869: }
870: }
871:
1.140 schwarze 872: /* Neither worked, give up. */
1.97 schwarze 873:
1.140 schwarze 874: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 875: return -1;
1.1 kristaps 876: }
877:
878: struct mparse *
1.179 schwarze 879: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
880: enum mandoc_os os_e, const char *os_s)
1.1 kristaps 881: {
882: struct mparse *curp;
1.10 kristaps 883:
1.1 kristaps 884: curp = mandoc_calloc(1, sizeof(struct mparse));
885:
1.44 schwarze 886: curp->options = options;
1.179 schwarze 887: curp->mmin = mmin;
1.1 kristaps 888: curp->mmsg = mmsg;
1.179 schwarze 889: curp->os_s = os_s;
1.1 kristaps 890:
1.144 schwarze 891: curp->roff = roff_alloc(curp, options);
1.179 schwarze 892: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.137 schwarze 893: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 894: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 895: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 896: if (curp->man->mdocmac == NULL)
897: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 898: } else if (curp->options & MPARSE_MAN) {
899: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 900: if (curp->man->manmac == NULL)
901: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 902: }
1.138 schwarze 903: curp->man->first->tok = TOKEN_NONE;
1.179 schwarze 904: curp->man->meta.os_e = os_e;
1.142 schwarze 905: return curp;
1.1 kristaps 906: }
907:
908: void
909: mparse_reset(struct mparse *curp)
910: {
911: roff_reset(curp->roff);
1.150 schwarze 912: roff_man_reset(curp->man);
1.198 schwarze 913: free_buf_list(curp->secondary);
914: curp->secondary = NULL;
1.160 schwarze 915:
916: free(curp->sodest);
917: curp->sodest = NULL;
918:
1.1 kristaps 919: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 920: curp->gzip = 0;
1.1 kristaps 921: }
922:
923: void
924: mparse_free(struct mparse *curp)
925: {
1.164 schwarze 926: roffhash_free(curp->man->mdocmac);
927: roffhash_free(curp->man->manmac);
1.137 schwarze 928: roff_man_free(curp->man);
1.160 schwarze 929: roff_free(curp->roff);
1.198 schwarze 930: free_buf_list(curp->secondary);
1.45 schwarze 931: free(curp->sodest);
1.1 kristaps 932: free(curp);
933: }
934:
935: void
1.135 schwarze 936: mparse_result(struct mparse *curp, struct roff_man **man,
937: char **sodest)
1.1 kristaps 938: {
939:
1.45 schwarze 940: if (sodest && NULL != (*sodest = curp->sodest)) {
941: *man = NULL;
942: return;
943: }
1.9 kristaps 944: if (man)
945: *man = curp->man;
1.157 schwarze 946: }
947:
948: void
949: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
950: {
951: if (curp->file_status > *rc)
952: *rc = curp->file_status;
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:
1.179 schwarze 975: if (er < m->mmin && er != MANDOCERR_FILE)
976: return;
977:
1.112 schwarze 978: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 979: while (er < mandoclimits[level])
980: level--;
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:
1.142 schwarze 993: return mandocerrs[er];
1.7 kristaps 994: }
995:
996: const char *
997: mparse_strlevel(enum mandoclevel lvl)
998: {
1.142 schwarze 999: return mandoclevels[lvl];
1.24 kristaps 1000: }
1001:
1002: void
1.198 schwarze 1003: mparse_copy(const struct mparse *p)
1.24 kristaps 1004: {
1.198 schwarze 1005: struct buf *buf;
1.24 kristaps 1006:
1.198 schwarze 1007: for (buf = p->secondary; buf != NULL; buf = buf->next)
1008: puts(buf->buf);
1.1 kristaps 1009: }
CVSweb