Annotation of src/usr.bin/mandoc/mdoc.c, Revision 1.149
1.149 ! schwarze 1: /* $OpenBSD: mdoc.c,v 1.148 2017/01/28 23:26:56 schwarze Exp $ */
1.1 kristaps 2: /*
1.77 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.148 schwarze 4: * Copyright (c) 2010, 2012-2017 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.3 schwarze 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.130 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.3 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.130 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.3 schwarze 13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1 kristaps 17: */
1.28 schwarze 18: #include <sys/types.h>
19:
1.1 kristaps 20: #include <assert.h>
1.101 schwarze 21: #include <ctype.h>
1.1 kristaps 22: #include <stdarg.h>
23: #include <stdio.h>
24: #include <stdlib.h>
25: #include <string.h>
1.39 schwarze 26: #include <time.h>
1.1 kristaps 27:
1.128 schwarze 28: #include "mandoc_aux.h"
29: #include "mandoc.h"
30: #include "roff.h"
1.83 schwarze 31: #include "mdoc.h"
1.128 schwarze 32: #include "libmandoc.h"
1.136 schwarze 33: #include "roff_int.h"
1.1 kristaps 34: #include "libmdoc.h"
35:
1.106 schwarze 36: const char *const __mdoc_macronames[MDOC_MAX + 1] = {
1.137 schwarze 37: "text",
1.7 schwarze 38: "Ap", "Dd", "Dt", "Os",
1.1 kristaps 39: "Sh", "Ss", "Pp", "D1",
40: "Dl", "Bd", "Ed", "Bl",
41: "El", "It", "Ad", "An",
42: "Ar", "Cd", "Cm", "Dv",
43: "Er", "Ev", "Ex", "Fa",
44: "Fd", "Fl", "Fn", "Ft",
45: "Ic", "In", "Li", "Nd",
46: "Nm", "Op", "Ot", "Pa",
47: "Rv", "St", "Va", "Vt",
1.33 schwarze 48: "Xr", "%A", "%B", "%D",
49: "%I", "%J", "%N", "%O",
50: "%P", "%R", "%T", "%V",
1.1 kristaps 51: "Ac", "Ao", "Aq", "At",
52: "Bc", "Bf", "Bo", "Bq",
53: "Bsx", "Bx", "Db", "Dc",
54: "Do", "Dq", "Ec", "Ef",
55: "Em", "Eo", "Fx", "Ms",
56: "No", "Ns", "Nx", "Ox",
57: "Pc", "Pf", "Po", "Pq",
58: "Qc", "Ql", "Qo", "Qq",
59: "Re", "Rs", "Sc", "So",
60: "Sq", "Sm", "Sx", "Sy",
61: "Tn", "Ux", "Xc", "Xo",
62: "Fo", "Fc", "Oo", "Oc",
63: "Bk", "Ek", "Bt", "Hf",
1.7 schwarze 64: "Fr", "Ud", "Lb", "Lp",
65: "Lk", "Mt", "Brq", "Bro",
1.33 schwarze 66: "Brc", "%C", "Es", "En",
67: "Dx", "%Q", "br", "sp",
1.137 schwarze 68: "%U", "Ta", "ll",
69: };
1.1 kristaps 70:
1.103 schwarze 71: const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
1.1 kristaps 72: "split", "nosplit", "ragged",
1.103 schwarze 73: "unfilled", "literal", "file",
74: "offset", "bullet", "dash",
75: "hyphen", "item", "enum",
76: "tag", "diag", "hang",
77: "ohang", "inset", "column",
78: "width", "compact", "std",
1.1 kristaps 79: "filled", "words", "emphasis",
1.30 schwarze 80: "symbolic", "nested", "centered"
1.1 kristaps 81: };
82:
1.137 schwarze 83: const char * const *mdoc_macronames = __mdoc_macronames + 1;
1.1 kristaps 84: const char * const *mdoc_argnames = __mdoc_argnames;
85:
1.131 schwarze 86: static int mdoc_ptext(struct roff_man *, int, char *, int);
87: static int mdoc_pmacro(struct roff_man *, int, char *, int);
1.1 kristaps 88:
1.81 schwarze 89:
1.1 kristaps 90: /*
91: * Main parse routine. Parses a single line -- really just hands off to
1.45 schwarze 92: * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
1.1 kristaps 93: */
94: int
1.131 schwarze 95: mdoc_parseln(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 96: {
97:
1.128 schwarze 98: if (mdoc->last->type != ROFFT_EQN || ln > mdoc->last->line)
1.117 schwarze 99: mdoc->flags |= MDOC_NEWLINE;
1.60 schwarze 100:
101: /*
102: * Let the roff nS register switch SYNOPSIS mode early,
103: * such that the parser knows at all times
104: * whether this mode is on or off.
105: * Note that this mode is also switched by the Sh macro.
106: */
1.94 schwarze 107: if (roff_getreg(mdoc->roff, "nS"))
108: mdoc->flags |= MDOC_SYNOPSIS;
109: else
110: mdoc->flags &= ~MDOC_SYNOPSIS;
1.60 schwarze 111:
1.142 schwarze 112: return roff_getcontrol(mdoc->roff, buf, &offs) ?
1.103 schwarze 113: mdoc_pmacro(mdoc, ln, buf, offs) :
1.142 schwarze 114: mdoc_ptext(mdoc, ln, buf, offs);
1.1 kristaps 115: }
116:
1.121 schwarze 117: void
1.58 schwarze 118: mdoc_macro(MACRO_PROT_ARGS)
1.1 kristaps 119: {
1.137 schwarze 120: assert(tok > TOKEN_NONE && tok < MDOC_MAX);
1.46 schwarze 121:
1.121 schwarze 122: (*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf);
1.1 kristaps 123: }
124:
1.120 schwarze 125: void
1.131 schwarze 126: mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, int tok)
1.1 kristaps 127: {
1.129 schwarze 128: struct roff_node *p;
1.1 kristaps 129:
1.136 schwarze 130: p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
131: roff_node_append(mdoc, p);
1.131 schwarze 132: mdoc->next = ROFF_NEXT_CHILD;
1.59 schwarze 133: }
134:
1.129 schwarze 135: struct roff_node *
1.131 schwarze 136: mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos, int tok,
1.149 ! schwarze 137: struct roff_node *body)
1.59 schwarze 138: {
1.129 schwarze 139: struct roff_node *p;
1.59 schwarze 140:
1.147 schwarze 141: body->flags |= NODE_ENDED;
142: body->parent->flags |= NODE_ENDED;
1.136 schwarze 143: p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
1.126 schwarze 144: p->body = body;
1.92 schwarze 145: p->norm = body->norm;
1.149 ! schwarze 146: p->end = ENDBODY_SPACE;
1.136 schwarze 147: roff_node_append(mdoc, p);
1.131 schwarze 148: mdoc->next = ROFF_NEXT_SIBLING;
1.142 schwarze 149: return p;
1.1 kristaps 150: }
151:
1.129 schwarze 152: struct roff_node *
1.131 schwarze 153: mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
1.129 schwarze 154: int tok, struct mdoc_arg *args)
1.1 kristaps 155: {
1.129 schwarze 156: struct roff_node *p;
1.1 kristaps 157:
1.136 schwarze 158: p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
1.4 schwarze 159: p->args = args;
160: if (p->args)
1.1 kristaps 161: (args->refcnt)++;
1.75 schwarze 162:
163: switch (tok) {
1.103 schwarze 164: case MDOC_Bd:
165: case MDOC_Bf:
166: case MDOC_Bl:
1.107 schwarze 167: case MDOC_En:
1.103 schwarze 168: case MDOC_Rs:
1.75 schwarze 169: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
170: break;
171: default:
172: break;
173: }
1.136 schwarze 174: roff_node_append(mdoc, p);
1.131 schwarze 175: mdoc->next = ROFF_NEXT_CHILD;
1.142 schwarze 176: return p;
1.1 kristaps 177: }
178:
1.120 schwarze 179: void
1.131 schwarze 180: mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
1.129 schwarze 181: int tok, struct mdoc_arg *args)
1.1 kristaps 182: {
1.129 schwarze 183: struct roff_node *p;
1.1 kristaps 184:
1.136 schwarze 185: p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
1.4 schwarze 186: p->args = args;
187: if (p->args)
1.1 kristaps 188: (args->refcnt)++;
1.75 schwarze 189:
190: switch (tok) {
1.103 schwarze 191: case MDOC_An:
1.75 schwarze 192: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
193: break;
194: default:
195: break;
196: }
1.136 schwarze 197: roff_node_append(mdoc, p);
1.131 schwarze 198: mdoc->next = ROFF_NEXT_CHILD;
1.1 kristaps 199: }
200:
1.120 schwarze 201: void
1.131 schwarze 202: mdoc_node_relink(struct roff_man *mdoc, struct roff_node *p)
1.91 schwarze 203: {
204:
1.136 schwarze 205: roff_node_unlink(mdoc, p);
1.145 schwarze 206: p->prev = p->next = NULL;
1.136 schwarze 207: roff_node_append(mdoc, p);
1.1 kristaps 208: }
209:
210: /*
211: * Parse free-form text, that is, a line that does not begin with the
212: * control character.
213: */
214: static int
1.131 schwarze 215: mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
1.1 kristaps 216: {
1.129 schwarze 217: struct roff_node *n;
1.56 schwarze 218: char *c, *ws, *end;
1.46 schwarze 219:
1.93 schwarze 220: n = mdoc->last;
1.56 schwarze 221:
222: /*
1.146 schwarze 223: * If a column list contains plain text, assume an implicit item
224: * macro. This can happen one or more times at the beginning
225: * of such a list, intermixed with non-It mdoc macros and with
226: * nodes generated on the roff level, for example by tbl.
1.56 schwarze 227: */
228:
1.146 schwarze 229: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
230: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
231: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
232: n->parent->norm->Bl.type == LIST_column)) {
1.93 schwarze 233: mdoc->flags |= MDOC_FREECOL;
1.121 schwarze 234: mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
1.142 schwarze 235: return 1;
1.56 schwarze 236: }
237:
1.52 schwarze 238: /*
239: * Search for the beginning of unescaped trailing whitespace (ws)
240: * and for the first character not to be output (end).
241: */
1.56 schwarze 242:
243: /* FIXME: replace with strcspn(). */
1.52 schwarze 244: ws = NULL;
1.53 schwarze 245: for (c = end = buf + offs; *c; c++) {
1.52 schwarze 246: switch (*c) {
247: case ' ':
248: if (NULL == ws)
249: ws = c;
250: continue;
251: case '\t':
252: /*
253: * Always warn about trailing tabs,
254: * even outside literal context,
255: * where they should be put on the next line.
256: */
257: if (NULL == ws)
258: ws = c;
259: /*
260: * Strip trailing tabs in literal context only;
261: * outside, they affect the next line.
262: */
1.93 schwarze 263: if (MDOC_LITERAL & mdoc->flags)
1.52 schwarze 264: continue;
265: break;
266: case '\\':
267: /* Skip the escaped character, too, if any. */
268: if (c[1])
269: c++;
270: /* FALLTHROUGH */
271: default:
272: ws = NULL;
273: break;
274: }
275: end = c + 1;
276: }
277: *end = '\0';
1.19 schwarze 278:
1.52 schwarze 279: if (ws)
1.108 schwarze 280: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
281: line, (int)(ws-buf), NULL);
1.34 schwarze 282:
1.120 schwarze 283: if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
1.108 schwarze 284: mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
285: line, (int)(c - buf), NULL);
1.46 schwarze 286:
1.40 schwarze 287: /*
1.66 schwarze 288: * Insert a `sp' in the case of a blank line. Technically,
1.46 schwarze 289: * blank lines aren't allowed, but enough manuals assume this
290: * behaviour that we want to work around it.
1.40 schwarze 291: */
1.139 schwarze 292: roff_elem_alloc(mdoc, line, offs, MDOC_sp);
1.147 schwarze 293: mdoc->last->flags |= NODE_VALID | NODE_ENDED;
1.131 schwarze 294: mdoc->next = ROFF_NEXT_SIBLING;
1.142 schwarze 295: return 1;
1.38 schwarze 296: }
1.1 kristaps 297:
1.138 schwarze 298: roff_word_alloc(mdoc, line, offs, buf+offs);
1.1 kristaps 299:
1.120 schwarze 300: if (mdoc->flags & MDOC_LITERAL)
1.142 schwarze 301: return 1;
1.52 schwarze 302:
1.35 schwarze 303: /*
1.48 schwarze 304: * End-of-sentence check. If the last character is an unescaped
305: * EOS character, then flag the node as being the end of a
306: * sentence. The front-end will know how to interpret this.
1.35 schwarze 307: */
308:
1.52 schwarze 309: assert(buf < end);
1.48 schwarze 310:
1.97 schwarze 311: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
1.147 schwarze 312: mdoc->last->flags |= NODE_EOS;
1.148 schwarze 313:
314: for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
315: if (c - buf < offs + 2)
316: continue;
317: if (end - c < 4)
318: break;
319: if (isalpha((unsigned char)c[-2]) &&
320: isalpha((unsigned char)c[-1]) &&
321: c[1] == ' ' &&
322: isupper((unsigned char)(c[2] == ' ' ? c[3] : c[2])) &&
323: (c[-2] != 'n' || c[-1] != 'c') &&
324: (c[-2] != 'v' || c[-1] != 's'))
325: mandoc_msg(MANDOCERR_EOS, mdoc->parse,
326: line, (int)(c - buf), NULL);
327: }
328:
1.142 schwarze 329: return 1;
1.1 kristaps 330: }
1.19 schwarze 331:
1.1 kristaps 332: /*
333: * Parse a macro line, that is, a line beginning with the control
334: * character.
335: */
1.61 schwarze 336: static int
1.131 schwarze 337: mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 338: {
1.129 schwarze 339: struct roff_node *n;
1.118 schwarze 340: const char *cp;
1.129 schwarze 341: int tok;
1.83 schwarze 342: int i, sv;
1.56 schwarze 343: char mac[5];
1.1 kristaps 344:
1.83 schwarze 345: sv = offs;
1.47 schwarze 346:
1.103 schwarze 347: /*
1.64 schwarze 348: * Copy the first word into a nil-terminated buffer.
1.118 schwarze 349: * Stop when a space, tab, escape, or eoln is encountered.
1.63 schwarze 350: */
1.1 kristaps 351:
1.83 schwarze 352: i = 0;
1.118 schwarze 353: while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
1.83 schwarze 354: mac[i++] = buf[offs++];
355:
356: mac[i] = '\0';
357:
1.137 schwarze 358: tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : TOKEN_NONE;
1.1 kristaps 359:
1.137 schwarze 360: if (tok == TOKEN_NONE) {
1.112 schwarze 361: mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
362: ln, sv, buf + sv - 1);
1.142 schwarze 363: return 1;
1.1 kristaps 364: }
365:
1.118 schwarze 366: /* Skip a leading escape sequence or tab. */
1.63 schwarze 367:
1.118 schwarze 368: switch (buf[offs]) {
369: case '\\':
370: cp = buf + offs + 1;
371: mandoc_escape(&cp, NULL, NULL);
372: offs = cp - buf;
373: break;
374: case '\t':
1.83 schwarze 375: offs++;
1.118 schwarze 376: break;
377: default:
378: break;
379: }
1.63 schwarze 380:
381: /* Jump to the next non-whitespace word. */
1.1 kristaps 382:
1.83 schwarze 383: while (buf[offs] && ' ' == buf[offs])
384: offs++;
1.34 schwarze 385:
1.103 schwarze 386: /*
1.46 schwarze 387: * Trailing whitespace. Note that tabs are allowed to be passed
388: * into the parser as "text", so we only warn about spaces here.
389: */
1.34 schwarze 390:
1.83 schwarze 391: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
1.108 schwarze 392: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
393: ln, offs - 1, NULL);
1.1 kristaps 394:
1.56 schwarze 395: /*
396: * If an initial macro or a list invocation, divert directly
397: * into macro processing.
398: */
399:
1.146 schwarze 400: n = mdoc->last;
401: if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
1.121 schwarze 402: mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
1.142 schwarze 403: return 1;
1.121 schwarze 404: }
1.56 schwarze 405:
406: /*
1.146 schwarze 407: * If a column list contains a non-It macro, assume an implicit
408: * item macro. This can happen one or more times at the
409: * beginning of such a list, intermixed with text lines and
410: * with nodes generated on the roff level, for example by tbl.
1.24 schwarze 411: */
1.56 schwarze 412:
1.146 schwarze 413: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
414: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
415: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
416: n->parent->norm->Bl.type == LIST_column)) {
1.93 schwarze 417: mdoc->flags |= MDOC_FREECOL;
1.121 schwarze 418: mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
1.142 schwarze 419: return 1;
1.56 schwarze 420: }
421:
422: /* Normal processing of a macro. */
423:
1.121 schwarze 424: mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
1.98 schwarze 425:
426: /* In quick mode (for mandocdb), abort after the NAME section. */
427:
428: if (mdoc->quick && MDOC_Sh == tok &&
429: SEC_NAME != mdoc->last->sec)
1.142 schwarze 430: return 2;
1.1 kristaps 431:
1.142 schwarze 432: return 1;
1.1 kristaps 433: }
1.24 schwarze 434:
1.83 schwarze 435: enum mdelim
436: mdoc_isdelim(const char *p)
437: {
438:
439: if ('\0' == p[0])
1.142 schwarze 440: return DELIM_NONE;
1.83 schwarze 441:
442: if ('\0' == p[1])
443: switch (p[0]) {
1.103 schwarze 444: case '(':
445: case '[':
1.142 schwarze 446: return DELIM_OPEN;
1.103 schwarze 447: case '|':
1.142 schwarze 448: return DELIM_MIDDLE;
1.103 schwarze 449: case '.':
450: case ',':
451: case ';':
452: case ':':
453: case '?':
454: case '!':
455: case ')':
456: case ']':
1.142 schwarze 457: return DELIM_CLOSE;
1.83 schwarze 458: default:
1.142 schwarze 459: return DELIM_NONE;
1.83 schwarze 460: }
461:
462: if ('\\' != p[0])
1.142 schwarze 463: return DELIM_NONE;
1.24 schwarze 464:
1.83 schwarze 465: if (0 == strcmp(p + 1, "."))
1.142 schwarze 466: return DELIM_CLOSE;
1.90 schwarze 467: if (0 == strcmp(p + 1, "fR|\\fP"))
1.142 schwarze 468: return DELIM_MIDDLE;
1.83 schwarze 469:
1.142 schwarze 470: return DELIM_NONE;
1.144 schwarze 471: }
472:
473: void
474: mdoc_validate(struct roff_man *mdoc)
475: {
476:
477: mdoc->last = mdoc->first;
478: mdoc_node_validate(mdoc);
479: mdoc_state_reset(mdoc);
1.83 schwarze 480: }
![[BACK]](/icons/back.gif){kind=icon}
Return to mdoc.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc