Import mdocml-1.11.3
[dragonfly.git] / contrib / mdocml / man_macro.c
CommitLineData
a4c7eb57 1/* $Id: man_macro.c,v 1.62 2011/04/19 16:38:48 kristaps Exp $ */
80387638
SW
2/*
3 * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
4 *
5 * Permission to use, copy, modify, and distribute this software for any
6 * purpose with or without fee is hereby granted, provided that the above
7 * copyright notice and this permission notice appear in all copies.
8 *
9 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16 */
17#ifdef HAVE_CONFIG_H
18#include "config.h"
19#endif
20
21#include <assert.h>
22#include <ctype.h>
23#include <stdlib.h>
24#include <string.h>
25
60e1e752 26#include "man.h"
80387638 27#include "mandoc.h"
60e1e752 28#include "libmandoc.h"
80387638
SW
29#include "libman.h"
30
31enum rew {
32 REW_REWIND,
33 REW_NOHALT,
34 REW_HALT
35};
36
37static int blk_close(MACRO_PROT_ARGS);
38static int blk_exp(MACRO_PROT_ARGS);
39static int blk_imp(MACRO_PROT_ARGS);
40static int in_line_eoln(MACRO_PROT_ARGS);
60e1e752
SW
41static int man_args(struct man *, int,
42 int *, char *, char **);
80387638
SW
43
44static int rew_scope(enum man_type,
45 struct man *, enum mant);
46static enum rew rew_dohalt(enum mant, enum man_type,
47 const struct man_node *);
48static enum rew rew_block(enum mant, enum man_type,
49 const struct man_node *);
60e1e752 50static void rew_warn(struct man *,
80387638
SW
51 struct man_node *, enum mandocerr);
52
53const struct man_macro __man_macros[MAN_MAX] = {
54 { in_line_eoln, MAN_NSCOPED }, /* br */
55 { in_line_eoln, 0 }, /* TH */
56 { blk_imp, MAN_SCOPED }, /* SH */
57 { blk_imp, MAN_SCOPED }, /* SS */
58 { blk_imp, MAN_SCOPED | MAN_FSCOPED }, /* TP */
59 { blk_imp, 0 }, /* LP */
60 { blk_imp, 0 }, /* PP */
61 { blk_imp, 0 }, /* P */
62 { blk_imp, 0 }, /* IP */
63 { blk_imp, 0 }, /* HP */
64 { in_line_eoln, MAN_SCOPED }, /* SM */
65 { in_line_eoln, MAN_SCOPED }, /* SB */
66 { in_line_eoln, 0 }, /* BI */
67 { in_line_eoln, 0 }, /* IB */
68 { in_line_eoln, 0 }, /* BR */
69 { in_line_eoln, 0 }, /* RB */
70 { in_line_eoln, MAN_SCOPED }, /* R */
71 { in_line_eoln, MAN_SCOPED }, /* B */
72 { in_line_eoln, MAN_SCOPED }, /* I */
73 { in_line_eoln, 0 }, /* IR */
74 { in_line_eoln, 0 }, /* RI */
75 { in_line_eoln, MAN_NSCOPED }, /* na */
76 { in_line_eoln, MAN_NSCOPED }, /* sp */
77 { in_line_eoln, 0 }, /* nf */
78 { in_line_eoln, 0 }, /* fi */
79 { blk_close, 0 }, /* RE */
80 { blk_exp, MAN_EXPLICIT }, /* RS */
81 { in_line_eoln, 0 }, /* DT */
82 { in_line_eoln, 0 }, /* UC */
83 { in_line_eoln, 0 }, /* PD */
84 { in_line_eoln, 0 }, /* AT */
85 { in_line_eoln, 0 }, /* in */
86 { in_line_eoln, 0 }, /* ft */
87};
88
89const struct man_macro * const man_macros = __man_macros;
90
91
92/*
93 * Warn when "n" is an explicit non-roff macro.
94 */
60e1e752 95static void
80387638
SW
96rew_warn(struct man *m, struct man_node *n, enum mandocerr er)
97{
98
99 if (er == MANDOCERR_MAX || MAN_BLOCK != n->type)
60e1e752 100 return;
80387638 101 if (MAN_VALID & n->flags)
60e1e752 102 return;
80387638 103 if ( ! (MAN_EXPLICIT & man_macros[n->tok].flags))
60e1e752
SW
104 return;
105
106 assert(er < MANDOCERR_FATAL);
107 man_nmsg(m, n, er);
80387638
SW
108}
109
110
111/*
112 * Rewind scope. If a code "er" != MANDOCERR_MAX has been provided, it
113 * will be used if an explicit block scope is being closed out.
114 */
115int
60e1e752 116man_unscope(struct man *m, const struct man_node *to,
80387638
SW
117 enum mandocerr er)
118{
60e1e752 119 struct man_node *n;
80387638 120
60e1e752 121 assert(to);
80387638
SW
122
123 /* LINTED */
60e1e752
SW
124 while (m->last != to) {
125 /*
126 * Save the parent here, because we may delete the
127 * m->last node in the post-validation phase and reset
128 * it to m->last->parent, causing a step in the closing
129 * out to be lost.
130 */
131 n = m->last->parent;
132 rew_warn(m, m->last, er);
80387638
SW
133 if ( ! man_valid_post(m))
134 return(0);
60e1e752 135 m->last = n;
80387638
SW
136 assert(m->last);
137 }
138
60e1e752 139 rew_warn(m, m->last, er);
80387638
SW
140 if ( ! man_valid_post(m))
141 return(0);
142
143 m->next = MAN_ROOT == m->last->type ?
144 MAN_NEXT_CHILD : MAN_NEXT_SIBLING;
145
146 return(1);
147}
148
149
150static enum rew
151rew_block(enum mant ntok, enum man_type type, const struct man_node *n)
152{
153
154 if (MAN_BLOCK == type && ntok == n->parent->tok &&
155 MAN_BODY == n->parent->type)
156 return(REW_REWIND);
157 return(ntok == n->tok ? REW_HALT : REW_NOHALT);
158}
159
160
161/*
162 * There are three scope levels: scoped to the root (all), scoped to the
163 * section (all less sections), and scoped to subsections (all less
164 * sections and subsections).
165 */
166static enum rew
167rew_dohalt(enum mant tok, enum man_type type, const struct man_node *n)
168{
169 enum rew c;
170
171 /* We cannot progress beyond the root ever. */
172 if (MAN_ROOT == n->type)
173 return(REW_HALT);
174
175 assert(n->parent);
176
177 /* Normal nodes shouldn't go to the level of the root. */
178 if (MAN_ROOT == n->parent->type)
179 return(REW_REWIND);
180
181 /* Already-validated nodes should be closed out. */
182 if (MAN_VALID & n->flags)
183 return(REW_NOHALT);
184
185 /* First: rewind to ourselves. */
186 if (type == n->type && tok == n->tok)
187 return(REW_REWIND);
188
189 /*
190 * Next follow the implicit scope-smashings as defined by man.7:
191 * section, sub-section, etc.
192 */
193
194 switch (tok) {
195 case (MAN_SH):
196 break;
197 case (MAN_SS):
198 /* Rewind to a section, if a block. */
199 if (REW_NOHALT != (c = rew_block(MAN_SH, type, n)))
200 return(c);
201 break;
202 case (MAN_RS):
203 /* Rewind to a subsection, if a block. */
204 if (REW_NOHALT != (c = rew_block(MAN_SS, type, n)))
205 return(c);
206 /* Rewind to a section, if a block. */
207 if (REW_NOHALT != (c = rew_block(MAN_SH, type, n)))
208 return(c);
209 break;
210 default:
211 /* Rewind to an offsetter, if a block. */
212 if (REW_NOHALT != (c = rew_block(MAN_RS, type, n)))
213 return(c);
214 /* Rewind to a subsection, if a block. */
215 if (REW_NOHALT != (c = rew_block(MAN_SS, type, n)))
216 return(c);
217 /* Rewind to a section, if a block. */
218 if (REW_NOHALT != (c = rew_block(MAN_SH, type, n)))
219 return(c);
220 break;
221 }
222
223 return(REW_NOHALT);
224}
225
226
227/*
228 * Rewinding entails ascending the parse tree until a coherent point,
229 * for example, the `SH' macro will close out any intervening `SS'
230 * scopes. When a scope is closed, it must be validated and actioned.
231 */
232static int
233rew_scope(enum man_type type, struct man *m, enum mant tok)
234{
235 struct man_node *n;
236 enum rew c;
237
238 /* LINTED */
239 for (n = m->last; n; n = n->parent) {
240 /*
241 * Whether we should stop immediately (REW_HALT), stop
242 * and rewind until this point (REW_REWIND), or keep
243 * rewinding (REW_NOHALT).
244 */
245 c = rew_dohalt(tok, type, n);
246 if (REW_HALT == c)
247 return(1);
248 if (REW_REWIND == c)
249 break;
250 }
251
252 /*
253 * Rewind until the current point. Warn if we're a roff
254 * instruction that's mowing over explicit scopes.
255 */
256 assert(n);
257
258 return(man_unscope(m, n, MANDOCERR_MAX));
259}
260
261
262/*
263 * Close out a generic explicit macro.
264 */
265/* ARGSUSED */
266int
267blk_close(MACRO_PROT_ARGS)
268{
269 enum mant ntok;
270 const struct man_node *nn;
271
272 switch (tok) {
273 case (MAN_RE):
274 ntok = MAN_RS;
275 break;
276 default:
277 abort();
278 /* NOTREACHED */
279 }
280
281 for (nn = m->last->parent; nn; nn = nn->parent)
282 if (ntok == nn->tok)
283 break;
284
285 if (NULL == nn)
60e1e752 286 man_pmsg(m, line, ppos, MANDOCERR_NOSCOPE);
80387638
SW
287
288 if ( ! rew_scope(MAN_BODY, m, ntok))
289 return(0);
290 if ( ! rew_scope(MAN_BLOCK, m, ntok))
291 return(0);
292
293 return(1);
294}
295
296
297/* ARGSUSED */
298int
299blk_exp(MACRO_PROT_ARGS)
300{
60e1e752 301 int la;
80387638
SW
302 char *p;
303
304 /*
305 * Close out prior scopes. "Regular" explicit macros cannot be
306 * nested, but we allow roff macros to be placed just about
307 * anywhere.
308 */
309
310 if ( ! rew_scope(MAN_BODY, m, tok))
311 return(0);
312 if ( ! rew_scope(MAN_BLOCK, m, tok))
313 return(0);
314
315 if ( ! man_block_alloc(m, line, ppos, tok))
316 return(0);
317 if ( ! man_head_alloc(m, line, ppos, tok))
318 return(0);
319
320 for (;;) {
321 la = *pos;
60e1e752 322 if ( ! man_args(m, line, pos, buf, &p))
80387638 323 break;
80387638
SW
324 if ( ! man_word_alloc(m, line, la, p))
325 return(0);
326 }
327
328 assert(m);
329 assert(tok != MAN_MAX);
330
331 if ( ! rew_scope(MAN_HEAD, m, tok))
332 return(0);
333 return(man_body_alloc(m, line, ppos, tok));
334}
335
336
337
338/*
339 * Parse an implicit-block macro. These contain a MAN_HEAD and a
340 * MAN_BODY contained within a MAN_BLOCK. Rules for closing out other
341 * scopes, such as `SH' closing out an `SS', are defined in the rew
342 * routines.
343 */
344/* ARGSUSED */
345int
346blk_imp(MACRO_PROT_ARGS)
347{
60e1e752 348 int la;
80387638
SW
349 char *p;
350 struct man_node *n;
351
352 /* Close out prior scopes. */
353
354 if ( ! rew_scope(MAN_BODY, m, tok))
355 return(0);
356 if ( ! rew_scope(MAN_BLOCK, m, tok))
357 return(0);
358
359 /* Allocate new block & head scope. */
360
361 if ( ! man_block_alloc(m, line, ppos, tok))
362 return(0);
363 if ( ! man_head_alloc(m, line, ppos, tok))
364 return(0);
365
366 n = m->last;
367
368 /* Add line arguments. */
369
370 for (;;) {
371 la = *pos;
60e1e752 372 if ( ! man_args(m, line, pos, buf, &p))
80387638 373 break;
80387638
SW
374 if ( ! man_word_alloc(m, line, la, p))
375 return(0);
376 }
377
378 /* Close out head and open body (unless MAN_SCOPE). */
379
380 if (MAN_SCOPED & man_macros[tok].flags) {
381 /* If we're forcing scope (`TP'), keep it open. */
382 if (MAN_FSCOPED & man_macros[tok].flags) {
383 m->flags |= MAN_BLINE;
384 return(1);
385 } else if (n == m->last) {
386 m->flags |= MAN_BLINE;
387 return(1);
388 }
389 }
390
391 if ( ! rew_scope(MAN_HEAD, m, tok))
392 return(0);
393 return(man_body_alloc(m, line, ppos, tok));
394}
395
396
397/* ARGSUSED */
398int
399in_line_eoln(MACRO_PROT_ARGS)
400{
60e1e752 401 int la;
80387638
SW
402 char *p;
403 struct man_node *n;
404
405 if ( ! man_elem_alloc(m, line, ppos, tok))
406 return(0);
407
408 n = m->last;
409
410 for (;;) {
411 la = *pos;
60e1e752 412 if ( ! man_args(m, line, pos, buf, &p))
80387638
SW
413 break;
414 if ( ! man_word_alloc(m, line, la, p))
415 return(0);
416 }
417
418 /*
419 * If no arguments are specified and this is MAN_SCOPED (i.e.,
420 * next-line scoped), then set our mode to indicate that we're
421 * waiting for terms to load into our context.
422 */
423
424 if (n == m->last && MAN_SCOPED & man_macros[tok].flags) {
425 assert( ! (MAN_NSCOPED & man_macros[tok].flags));
426 m->flags |= MAN_ELINE;
427 return(1);
428 }
429
430 /* Set ignorable context, if applicable. */
431
432 if (MAN_NSCOPED & man_macros[tok].flags) {
433 assert( ! (MAN_SCOPED & man_macros[tok].flags));
434 m->flags |= MAN_ILINE;
435 }
436
437 /*
438 * Rewind our element scope. Note that when TH is pruned, we'll
439 * be back at the root, so make sure that we don't clobber as
440 * its sibling.
441 */
442
443 for ( ; m->last; m->last = m->last->parent) {
444 if (m->last == n)
445 break;
446 if (m->last->type == MAN_ROOT)
447 break;
448 if ( ! man_valid_post(m))
449 return(0);
450 }
451
452 assert(m->last);
453
454 /*
455 * Same here regarding whether we're back at the root.
456 */
457
458 if (m->last->type != MAN_ROOT && ! man_valid_post(m))
459 return(0);
460
461 m->next = MAN_ROOT == m->last->type ?
462 MAN_NEXT_CHILD : MAN_NEXT_SIBLING;
463
464 return(1);
465}
466
467
468int
469man_macroend(struct man *m)
470{
471
472 return(man_unscope(m, m->first, MANDOCERR_SCOPEEXIT));
473}
474
60e1e752
SW
475static int
476man_args(struct man *m, int line, int *pos, char *buf, char **v)
477{
478 char *start;
479
480 assert(*pos);
481 *v = start = buf + *pos;
482 assert(' ' != *start);
483
484 if ('\0' == *start)
485 return(0);
486
487 *v = mandoc_getarg(m->parse, v, line, pos);
488 return(1);
489}