Print this page
9718 update mandoc to 1.14.4
| Split |
Close |
| Expand all |
| Collapse all |
--- old/usr/src/cmd/mandoc/mdoc.c
+++ new/usr/src/cmd/mandoc/mdoc.c
1 -/* $Id: mdoc.c,v 1.267 2017/06/17 13:06:16 schwarze Exp $ */
1 +/* $Id: mdoc.c,v 1.268 2017/08/11 16:56:21 schwarze Exp $ */
2 2 /*
3 3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 4 * Copyright (c) 2010, 2012-2017 Ingo Schwarze <schwarze@openbsd.org>
5 5 *
6 6 * Permission to use, copy, modify, and distribute this software for any
7 7 * purpose with or without fee is hereby granted, provided that the above
8 8 * copyright notice and this permission notice appear in all copies.
9 9 *
10 10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
11 11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
13 13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14 14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15 15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16 16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 17 */
18 18 #include "config.h"
19 19
20 20 #include <sys/types.h>
21 21
22 22 #include <assert.h>
23 23 #include <ctype.h>
24 24 #include <stdarg.h>
25 25 #include <stdio.h>
26 26 #include <stdlib.h>
27 27 #include <string.h>
28 28 #include <time.h>
29 29
30 30 #include "mandoc_aux.h"
31 31 #include "mandoc.h"
32 32 #include "roff.h"
33 33 #include "mdoc.h"
34 34 #include "libmandoc.h"
35 35 #include "roff_int.h"
36 36 #include "libmdoc.h"
37 37
38 38 const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
39 39 "split", "nosplit", "ragged",
40 40 "unfilled", "literal", "file",
41 41 "offset", "bullet", "dash",
42 42 "hyphen", "item", "enum",
43 43 "tag", "diag", "hang",
44 44 "ohang", "inset", "column",
45 45 "width", "compact", "std",
46 46 "filled", "words", "emphasis",
47 47 "symbolic", "nested", "centered"
48 48 };
49 49 const char * const *mdoc_argnames = __mdoc_argnames;
50 50
51 51 static int mdoc_ptext(struct roff_man *, int, char *, int);
52 52 static int mdoc_pmacro(struct roff_man *, int, char *, int);
53 53
54 54
55 55 /*
56 56 * Main parse routine. Parses a single line -- really just hands off to
57 57 * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
58 58 */
59 59 int
60 60 mdoc_parseln(struct roff_man *mdoc, int ln, char *buf, int offs)
61 61 {
62 62
63 63 if (mdoc->last->type != ROFFT_EQN || ln > mdoc->last->line)
64 64 mdoc->flags |= MDOC_NEWLINE;
65 65
66 66 /*
67 67 * Let the roff nS register switch SYNOPSIS mode early,
68 68 * such that the parser knows at all times
69 69 * whether this mode is on or off.
70 70 * Note that this mode is also switched by the Sh macro.
71 71 */
72 72 if (roff_getreg(mdoc->roff, "nS"))
73 73 mdoc->flags |= MDOC_SYNOPSIS;
74 74 else
75 75 mdoc->flags &= ~MDOC_SYNOPSIS;
76 76
77 77 return roff_getcontrol(mdoc->roff, buf, &offs) ?
78 78 mdoc_pmacro(mdoc, ln, buf, offs) :
79 79 mdoc_ptext(mdoc, ln, buf, offs);
80 80 }
81 81
82 82 void
83 83 mdoc_macro(MACRO_PROT_ARGS)
84 84 {
85 85 assert(tok >= MDOC_Dd && tok < MDOC_MAX);
86 86 (*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf);
87 87 }
88 88
89 89 void
90 90 mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, enum roff_tok tok)
91 91 {
92 92 struct roff_node *p;
93 93
94 94 p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
95 95 roff_node_append(mdoc, p);
96 96 mdoc->next = ROFF_NEXT_CHILD;
97 97 }
98 98
99 99 struct roff_node *
100 100 mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos,
101 101 enum roff_tok tok, struct roff_node *body)
102 102 {
103 103 struct roff_node *p;
104 104
105 105 body->flags |= NODE_ENDED;
106 106 body->parent->flags |= NODE_ENDED;
107 107 p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
108 108 p->body = body;
109 109 p->norm = body->norm;
110 110 p->end = ENDBODY_SPACE;
111 111 roff_node_append(mdoc, p);
112 112 mdoc->next = ROFF_NEXT_SIBLING;
113 113 return p;
114 114 }
115 115
116 116 struct roff_node *
117 117 mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
118 118 enum roff_tok tok, struct mdoc_arg *args)
119 119 {
120 120 struct roff_node *p;
121 121
122 122 p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
123 123 p->args = args;
124 124 if (p->args)
125 125 (args->refcnt)++;
126 126
127 127 switch (tok) {
128 128 case MDOC_Bd:
129 129 case MDOC_Bf:
130 130 case MDOC_Bl:
131 131 case MDOC_En:
132 132 case MDOC_Rs:
133 133 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
134 134 break;
135 135 default:
136 136 break;
137 137 }
138 138 roff_node_append(mdoc, p);
139 139 mdoc->next = ROFF_NEXT_CHILD;
140 140 return p;
141 141 }
142 142
143 143 void
144 144 mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
145 145 enum roff_tok tok, struct mdoc_arg *args)
146 146 {
147 147 struct roff_node *p;
148 148
149 149 p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
150 150 p->args = args;
151 151 if (p->args)
152 152 (args->refcnt)++;
153 153
154 154 switch (tok) {
155 155 case MDOC_An:
156 156 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
157 157 break;
158 158 default:
159 159 break;
160 160 }
161 161 roff_node_append(mdoc, p);
162 162 mdoc->next = ROFF_NEXT_CHILD;
163 163 }
164 164
165 165 void
166 166 mdoc_node_relink(struct roff_man *mdoc, struct roff_node *p)
167 167 {
168 168
169 169 roff_node_unlink(mdoc, p);
170 170 p->prev = p->next = NULL;
171 171 roff_node_append(mdoc, p);
172 172 }
173 173
174 174 /*
175 175 * Parse free-form text, that is, a line that does not begin with the
176 176 * control character.
177 177 */
178 178 static int
179 179 mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
180 180 {
181 181 struct roff_node *n;
182 182 const char *cp, *sp;
183 183 char *c, *ws, *end;
184 184
185 185 n = mdoc->last;
186 186
187 187 /*
188 188 * If a column list contains plain text, assume an implicit item
189 189 * macro. This can happen one or more times at the beginning
190 190 * of such a list, intermixed with non-It mdoc macros and with
191 191 * nodes generated on the roff level, for example by tbl.
192 192 */
193 193
194 194 if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
195 195 n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
196 196 (n->parent != NULL && n->parent->tok == MDOC_Bl &&
197 197 n->parent->norm->Bl.type == LIST_column)) {
198 198 mdoc->flags |= MDOC_FREECOL;
199 199 mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
200 200 return 1;
201 201 }
202 202
203 203 /*
204 204 * Search for the beginning of unescaped trailing whitespace (ws)
205 205 * and for the first character not to be output (end).
206 206 */
207 207
208 208 /* FIXME: replace with strcspn(). */
209 209 ws = NULL;
210 210 for (c = end = buf + offs; *c; c++) {
211 211 switch (*c) {
212 212 case ' ':
213 213 if (NULL == ws)
214 214 ws = c;
215 215 continue;
216 216 case '\t':
217 217 /*
218 218 * Always warn about trailing tabs,
219 219 * even outside literal context,
220 220 * where they should be put on the next line.
221 221 */
222 222 if (NULL == ws)
223 223 ws = c;
224 224 /*
225 225 * Strip trailing tabs in literal context only;
226 226 * outside, they affect the next line.
227 227 */
228 228 if (MDOC_LITERAL & mdoc->flags)
229 229 continue;
230 230 break;
231 231 case '\\':
232 232 /* Skip the escaped character, too, if any. */
233 233 if (c[1])
234 234 c++;
235 235 /* FALLTHROUGH */
236 236 default:
237 237 ws = NULL;
238 238 break;
239 239 }
240 240 end = c + 1;
241 241 }
242 242 *end = '\0';
243 243
244 244 if (ws)
245 245 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
246 246 line, (int)(ws-buf), NULL);
247 247
248 248 /*
249 249 * Blank lines are allowed in no-fill mode
250 250 * and cancel preceding \c,
251 251 * but add a single vertical space elsewhere.
252 252 */
253 253
254 254 if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
255 255 switch (mdoc->last->type) {
256 256 case ROFFT_TEXT:
257 257 sp = mdoc->last->string;
258 258 cp = end = strchr(sp, '\0') - 2;
259 259 if (cp < sp || cp[0] != '\\' || cp[1] != 'c')
260 260 break;
261 261 while (cp > sp && cp[-1] == '\\')
262 262 cp--;
263 263 if ((end - cp) % 2)
264 264 break;
265 265 *end = '\0';
266 266 return 1;
267 267 default:
268 268 break;
269 269 }
270 270 mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
271 271 line, (int)(c - buf), NULL);
272 272 roff_elem_alloc(mdoc, line, offs, ROFF_sp);
273 273 mdoc->last->flags |= NODE_VALID | NODE_ENDED;
274 274 mdoc->next = ROFF_NEXT_SIBLING;
275 275 return 1;
276 276 }
277 277
278 278 roff_word_alloc(mdoc, line, offs, buf+offs);
279 279
280 280 if (mdoc->flags & MDOC_LITERAL)
281 281 return 1;
282 282
283 283 /*
284 284 * End-of-sentence check. If the last character is an unescaped
285 285 * EOS character, then flag the node as being the end of a
286 286 * sentence. The front-end will know how to interpret this.
287 287 */
288 288
289 289 assert(buf < end);
|
↓ open down ↓ |
278 lines elided |
↑ open up ↑ |
290 290
291 291 if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
292 292 mdoc->last->flags |= NODE_EOS;
293 293
294 294 for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
295 295 if (c - buf < offs + 2)
296 296 continue;
297 297 if (end - c < 3)
298 298 break;
299 299 if (c[1] != ' ' ||
300 - isalpha((unsigned char)c[-2]) == 0 ||
301 - isalpha((unsigned char)c[-1]) == 0 ||
300 + isalnum((unsigned char)c[-2]) == 0 ||
301 + isalnum((unsigned char)c[-1]) == 0 ||
302 302 (c[-2] == 'n' && c[-1] == 'c') ||
303 303 (c[-2] == 'v' && c[-1] == 's'))
304 304 continue;
305 305 c += 2;
306 306 if (*c == ' ')
307 307 c++;
308 308 if (*c == ' ')
309 309 c++;
310 310 if (isupper((unsigned char)(*c)))
311 311 mandoc_msg(MANDOCERR_EOS, mdoc->parse,
312 312 line, (int)(c - buf), NULL);
313 313 }
314 314
315 315 return 1;
316 316 }
317 317
318 318 /*
319 319 * Parse a macro line, that is, a line beginning with the control
320 320 * character.
321 321 */
322 322 static int
323 323 mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
324 324 {
325 325 struct roff_node *n;
326 326 const char *cp;
327 327 size_t sz;
328 328 enum roff_tok tok;
329 329 int sv;
330 330
331 331 /* Determine the line macro. */
332 332
333 333 sv = offs;
334 334 tok = TOKEN_NONE;
335 335 for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++)
336 336 offs++;
337 337 if (sz == 2 || sz == 3)
338 338 tok = roffhash_find(mdoc->mdocmac, buf + sv, sz);
339 339 if (tok == TOKEN_NONE) {
340 340 mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
341 341 ln, sv, buf + sv - 1);
342 342 return 1;
343 343 }
344 344
345 345 /* Skip a leading escape sequence or tab. */
346 346
347 347 switch (buf[offs]) {
348 348 case '\\':
349 349 cp = buf + offs + 1;
350 350 mandoc_escape(&cp, NULL, NULL);
351 351 offs = cp - buf;
352 352 break;
353 353 case '\t':
354 354 offs++;
355 355 break;
356 356 default:
357 357 break;
358 358 }
359 359
360 360 /* Jump to the next non-whitespace word. */
361 361
362 362 while (buf[offs] == ' ')
363 363 offs++;
364 364
365 365 /*
366 366 * Trailing whitespace. Note that tabs are allowed to be passed
367 367 * into the parser as "text", so we only warn about spaces here.
368 368 */
369 369
370 370 if ('\0' == buf[offs] && ' ' == buf[offs - 1])
371 371 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
372 372 ln, offs - 1, NULL);
373 373
374 374 /*
375 375 * If an initial macro or a list invocation, divert directly
376 376 * into macro processing.
377 377 */
378 378
379 379 n = mdoc->last;
380 380 if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
381 381 mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
382 382 return 1;
383 383 }
384 384
385 385 /*
386 386 * If a column list contains a non-It macro, assume an implicit
387 387 * item macro. This can happen one or more times at the
388 388 * beginning of such a list, intermixed with text lines and
389 389 * with nodes generated on the roff level, for example by tbl.
390 390 */
391 391
392 392 if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
393 393 n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
394 394 (n->parent != NULL && n->parent->tok == MDOC_Bl &&
395 395 n->parent->norm->Bl.type == LIST_column)) {
396 396 mdoc->flags |= MDOC_FREECOL;
397 397 mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
398 398 return 1;
399 399 }
400 400
401 401 /* Normal processing of a macro. */
402 402
403 403 mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
404 404
405 405 /* In quick mode (for mandocdb), abort after the NAME section. */
406 406
407 407 if (mdoc->quick && MDOC_Sh == tok &&
408 408 SEC_NAME != mdoc->last->sec)
409 409 return 2;
410 410
411 411 return 1;
412 412 }
413 413
414 414 enum mdelim
415 415 mdoc_isdelim(const char *p)
416 416 {
417 417
418 418 if ('\0' == p[0])
419 419 return DELIM_NONE;
420 420
421 421 if ('\0' == p[1])
422 422 switch (p[0]) {
423 423 case '(':
424 424 case '[':
425 425 return DELIM_OPEN;
426 426 case '|':
427 427 return DELIM_MIDDLE;
428 428 case '.':
429 429 case ',':
430 430 case ';':
431 431 case ':':
432 432 case '?':
433 433 case '!':
434 434 case ')':
435 435 case ']':
436 436 return DELIM_CLOSE;
437 437 default:
438 438 return DELIM_NONE;
439 439 }
440 440
441 441 if ('\\' != p[0])
442 442 return DELIM_NONE;
443 443
444 444 if (0 == strcmp(p + 1, "."))
445 445 return DELIM_CLOSE;
446 446 if (0 == strcmp(p + 1, "fR|\\fP"))
447 447 return DELIM_MIDDLE;
448 448
449 449 return DELIM_NONE;
450 450 }
451 451
452 452 void
453 453 mdoc_validate(struct roff_man *mdoc)
454 454 {
455 455
456 456 mdoc->last = mdoc->first;
457 457 mdoc_node_validate(mdoc);
458 458 mdoc_state_reset(mdoc);
459 459 }
|
↓ open down ↓ |
148 lines elided |
↑ open up ↑ |
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX