1 /* $Id: mdoc.c,v 1.233 2014/11/28 06:27:05 schwarze Exp $ */
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
6 * Permission to use, copy, modify, and distribute this software for any
7 * purpose with or without fee is hereby granted, provided that the above
8 * copyright notice and this permission notice appear in all copies.
10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
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.
20 #include <sys/types.h>
32 #include "mandoc_aux.h"
34 #include "libmandoc.h"
36 const char *const __mdoc_macronames[MDOC_MAX + 1] = {
37 "Ap", "Dd", "Dt", "Os",
38 "Sh", "Ss", "Pp", "D1",
39 "Dl", "Bd", "Ed", "Bl",
40 "El", "It", "Ad", "An",
41 "Ar", "Cd", "Cm", "Dv",
42 "Er", "Ev", "Ex", "Fa",
43 "Fd", "Fl", "Fn", "Ft",
44 "Ic", "In", "Li", "Nd",
45 "Nm", "Op", "Ot", "Pa",
46 "Rv", "St", "Va", "Vt",
47 "Xr", "%A", "%B", "%D",
48 "%I", "%J", "%N", "%O",
49 "%P", "%R", "%T", "%V",
50 "Ac", "Ao", "Aq", "At",
51 "Bc", "Bf", "Bo", "Bq",
52 "Bsx", "Bx", "Db", "Dc",
53 "Do", "Dq", "Ec", "Ef",
54 "Em", "Eo", "Fx", "Ms",
55 "No", "Ns", "Nx", "Ox",
56 "Pc", "Pf", "Po", "Pq",
57 "Qc", "Ql", "Qo", "Qq",
58 "Re", "Rs", "Sc", "So",
59 "Sq", "Sm", "Sx", "Sy",
60 "Tn", "Ux", "Xc", "Xo",
61 "Fo", "Fc", "Oo", "Oc",
62 "Bk", "Ek", "Bt", "Hf",
63 "Fr", "Ud", "Lb", "Lp",
64 "Lk", "Mt", "Brq", "Bro",
65 "Brc", "%C", "Es", "En",
66 "Dx", "%Q", "br", "sp",
67 "%U", "Ta", "ll", "text",
70 const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
71 "split", "nosplit", "ragged",
72 "unfilled", "literal", "file",
73 "offset", "bullet", "dash",
74 "hyphen", "item", "enum",
75 "tag", "diag", "hang",
76 "ohang", "inset", "column",
77 "width", "compact", "std",
78 "filled", "words", "emphasis",
79 "symbolic", "nested", "centered"
82 const char * const *mdoc_macronames = __mdoc_macronames;
83 const char * const *mdoc_argnames = __mdoc_argnames;
85 static void mdoc_node_free(struct mdoc_node *);
86 static void mdoc_node_unlink(struct mdoc *,
88 static void mdoc_free1(struct mdoc *);
89 static void mdoc_alloc1(struct mdoc *);
90 static struct mdoc_node *node_alloc(struct mdoc *, int, int,
91 enum mdoct, enum mdoc_type);
92 static void node_append(struct mdoc *, struct mdoc_node *);
93 static int mdoc_ptext(struct mdoc *, int, char *, int);
94 static int mdoc_pmacro(struct mdoc *, int, char *, int);
97 const struct mdoc_node *
98 mdoc_node(const struct mdoc *mdoc)
104 const struct mdoc_meta *
105 mdoc_meta(const struct mdoc *mdoc)
112 * Frees volatile resources (parse tree, meta-data, fields).
115 mdoc_free1(struct mdoc *mdoc)
119 mdoc_node_delete(mdoc, mdoc->first);
120 free(mdoc->meta.msec);
121 free(mdoc->meta.vol);
122 free(mdoc->meta.arch);
123 free(mdoc->meta.date);
124 free(mdoc->meta.title);
126 free(mdoc->meta.name);
130 * Allocate all volatile resources (parse tree, meta-data, fields).
133 mdoc_alloc1(struct mdoc *mdoc)
136 memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
138 mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
139 mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
140 mdoc->first = mdoc->last;
141 mdoc->last->type = MDOC_ROOT;
142 mdoc->last->tok = MDOC_MAX;
143 mdoc->next = MDOC_NEXT_CHILD;
147 * Free up volatile resources (see mdoc_free1()) then re-initialises the
148 * data with mdoc_alloc1(). After invocation, parse data has been reset
149 * and the parser is ready for re-invocation on a new tree; however,
150 * cross-parse non-volatile data is kept intact.
153 mdoc_reset(struct mdoc *mdoc)
161 * Completely free up all volatile and non-volatile parse resources.
162 * After invocation, the pointer is no longer usable.
165 mdoc_free(struct mdoc *mdoc)
173 * Allocate volatile and non-volatile parse resources.
176 mdoc_alloc(struct roff *roff, struct mparse *parse,
177 const char *defos, int quick)
181 p = mandoc_calloc(1, sizeof(struct mdoc));
194 mdoc_endparse(struct mdoc *mdoc)
202 mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
206 n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
208 if (ep->ln > mdoc->last->line)
209 n->flags |= MDOC_LINE;
210 node_append(mdoc, n);
211 mdoc->next = MDOC_NEXT_SIBLING;
215 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
219 n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
221 node_append(mdoc, n);
222 mdoc->next = MDOC_NEXT_SIBLING;
226 * Main parse routine. Parses a single line -- really just hands off to
227 * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
230 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
233 if (mdoc->last->type != MDOC_EQN || ln > mdoc->last->line)
234 mdoc->flags |= MDOC_NEWLINE;
237 * Let the roff nS register switch SYNOPSIS mode early,
238 * such that the parser knows at all times
239 * whether this mode is on or off.
240 * Note that this mode is also switched by the Sh macro.
242 if (roff_getreg(mdoc->roff, "nS"))
243 mdoc->flags |= MDOC_SYNOPSIS;
245 mdoc->flags &= ~MDOC_SYNOPSIS;
247 return(roff_getcontrol(mdoc->roff, buf, &offs) ?
248 mdoc_pmacro(mdoc, ln, buf, offs) :
249 mdoc_ptext(mdoc, ln, buf, offs));
253 mdoc_macro(MACRO_PROT_ARGS)
255 assert(tok < MDOC_MAX);
257 if (mdoc->flags & MDOC_PBODY) {
258 if (tok == MDOC_Dt) {
259 mandoc_vmsg(MANDOCERR_DT_LATE,
260 mdoc->parse, line, ppos,
261 "Dt %s", buf + *pos);
264 } else if ( ! (mdoc_macros[tok].flags & MDOC_PROLOGUE)) {
265 if (mdoc->meta.title == NULL) {
266 mandoc_vmsg(MANDOCERR_DT_NOTITLE,
267 mdoc->parse, line, ppos, "%s %s",
268 mdoc_macronames[tok], buf + *pos);
269 mdoc->meta.title = mandoc_strdup("UNTITLED");
271 if (NULL == mdoc->meta.vol)
272 mdoc->meta.vol = mandoc_strdup("LOCAL");
273 mdoc->flags |= MDOC_PBODY;
275 (*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf);
280 node_append(struct mdoc *mdoc, struct mdoc_node *p)
285 assert(MDOC_ROOT != p->type);
287 switch (mdoc->next) {
288 case MDOC_NEXT_SIBLING:
289 mdoc->last->next = p;
290 p->prev = mdoc->last;
291 p->parent = mdoc->last->parent;
293 case MDOC_NEXT_CHILD:
294 mdoc->last->child = p;
295 p->parent = mdoc->last;
305 * Copy over the normalised-data pointer of our parent. Not
306 * everybody has one, but copying a null pointer is fine.
311 if (ENDBODY_NOT != p->end)
317 p->norm = p->parent->norm;
323 mdoc_valid_pre(mdoc, p);
327 assert(MDOC_BLOCK == p->parent->type);
331 assert(MDOC_BLOCK == p->parent->type);
337 assert(MDOC_BLOCK == p->parent->type);
350 mdoc_valid_post(mdoc);
357 static struct mdoc_node *
358 node_alloc(struct mdoc *mdoc, int line, int pos,
359 enum mdoct tok, enum mdoc_type type)
363 p = mandoc_calloc(1, sizeof(struct mdoc_node));
364 p->sec = mdoc->lastsec;
373 if (MDOC_SYNOPSIS & mdoc->flags)
374 p->flags |= MDOC_SYNPRETTY;
376 p->flags &= ~MDOC_SYNPRETTY;
377 if (MDOC_NEWLINE & mdoc->flags)
378 p->flags |= MDOC_LINE;
379 mdoc->flags &= ~MDOC_NEWLINE;
385 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
389 p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
390 node_append(mdoc, p);
391 mdoc->next = MDOC_NEXT_CHILD;
395 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
401 p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
402 node_append(mdoc, p);
403 mdoc->next = MDOC_NEXT_CHILD;
408 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
412 p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
413 node_append(mdoc, p);
414 mdoc->next = MDOC_NEXT_CHILD;
419 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
420 struct mdoc_node *body, enum mdoc_endbody end)
424 p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
426 p->norm = body->norm;
428 node_append(mdoc, p);
429 mdoc->next = MDOC_NEXT_SIBLING;
433 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
434 enum mdoct tok, struct mdoc_arg *args)
438 p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
453 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
458 node_append(mdoc, p);
459 mdoc->next = MDOC_NEXT_CHILD;
464 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
465 enum mdoct tok, struct mdoc_arg *args)
469 p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
476 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
481 node_append(mdoc, p);
482 mdoc->next = MDOC_NEXT_CHILD;
486 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
490 n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
491 n->string = roff_strdup(mdoc->roff, p);
492 node_append(mdoc, n);
493 mdoc->next = MDOC_NEXT_SIBLING;
497 mdoc_word_append(struct mdoc *mdoc, const char *p)
500 char *addstr, *newstr;
503 addstr = roff_strdup(mdoc->roff, p);
504 mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
508 mdoc->next = MDOC_NEXT_SIBLING;
512 mdoc_node_free(struct mdoc_node *p)
515 if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
520 mdoc_argv_free(p->args);
525 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
528 /* Adjust siblings. */
531 n->prev->next = n->next;
533 n->next->prev = n->prev;
539 if (n->parent->child == n)
540 n->parent->child = n->prev ? n->prev : n->next;
541 if (n->parent->last == n)
542 n->parent->last = n->prev ? n->prev : NULL;
545 /* Adjust parse point, if applicable. */
547 if (mdoc && mdoc->last == n) {
549 mdoc->last = n->prev;
550 mdoc->next = MDOC_NEXT_SIBLING;
552 mdoc->last = n->parent;
553 mdoc->next = MDOC_NEXT_CHILD;
557 if (mdoc && mdoc->first == n)
562 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
567 mdoc_node_delete(mdoc, p->child);
569 assert(0 == p->nchild);
571 mdoc_node_unlink(mdoc, p);
576 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
579 mdoc_node_unlink(mdoc, p);
580 node_append(mdoc, p);
584 * Parse free-form text, that is, a line that does not begin with the
588 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
597 * Divert directly to list processing if we're encountering a
598 * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
599 * (a MDOC_BODY means it's already open, in which case we should
600 * process within its context in the normal way).
603 if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
604 LIST_column == n->norm->Bl.type) {
605 /* `Bl' is open without any children. */
606 mdoc->flags |= MDOC_FREECOL;
607 mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
611 if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
613 MDOC_Bl == n->parent->tok &&
614 LIST_column == n->parent->norm->Bl.type) {
615 /* `Bl' has block-level `It' children. */
616 mdoc->flags |= MDOC_FREECOL;
617 mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
622 * Search for the beginning of unescaped trailing whitespace (ws)
623 * and for the first character not to be output (end).
626 /* FIXME: replace with strcspn(). */
628 for (c = end = buf + offs; *c; c++) {
636 * Always warn about trailing tabs,
637 * even outside literal context,
638 * where they should be put on the next line.
643 * Strip trailing tabs in literal context only;
644 * outside, they affect the next line.
646 if (MDOC_LITERAL & mdoc->flags)
650 /* Skip the escaped character, too, if any. */
663 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
664 line, (int)(ws-buf), NULL);
666 if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
667 mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
668 line, (int)(c - buf), NULL);
671 * Insert a `sp' in the case of a blank line. Technically,
672 * blank lines aren't allowed, but enough manuals assume this
673 * behaviour that we want to work around it.
675 mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL);
676 mdoc->next = MDOC_NEXT_SIBLING;
677 mdoc_valid_post(mdoc);
681 mdoc_word_alloc(mdoc, line, offs, buf+offs);
683 if (mdoc->flags & MDOC_LITERAL)
687 * End-of-sentence check. If the last character is an unescaped
688 * EOS character, then flag the node as being the end of a
689 * sentence. The front-end will know how to interpret this.
694 if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
695 mdoc->last->flags |= MDOC_EOS;
700 * Parse a macro line, that is, a line beginning with the control
704 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
715 * Copy the first word into a nil-terminated buffer.
716 * Stop when a space, tab, escape, or eoln is encountered.
720 while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
721 mac[i++] = buf[offs++];
725 tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
727 if (tok == MDOC_MAX) {
728 mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
729 ln, sv, buf + sv - 1);
733 /* Skip a leading escape sequence or tab. */
738 mandoc_escape(&cp, NULL, NULL);
748 /* Jump to the next non-whitespace word. */
750 while (buf[offs] && ' ' == buf[offs])
754 * Trailing whitespace. Note that tabs are allowed to be passed
755 * into the parser as "text", so we only warn about spaces here.
758 if ('\0' == buf[offs] && ' ' == buf[offs - 1])
759 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
763 * If an initial macro or a list invocation, divert directly
764 * into macro processing.
767 if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
768 mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
776 * If the first macro of a `Bl -column', open an `It' block
777 * context around the parsed macro.
780 if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
781 LIST_column == n->norm->Bl.type) {
782 mdoc->flags |= MDOC_FREECOL;
783 mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
788 * If we're following a block-level `It' within a `Bl -column'
789 * context (perhaps opened in the above block or in ptext()),
790 * then open an `It' block context around the parsed macro.
793 if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
795 MDOC_Bl == n->parent->tok &&
796 LIST_column == n->parent->norm->Bl.type) {
797 mdoc->flags |= MDOC_FREECOL;
798 mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
802 /* Normal processing of a macro. */
804 mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
806 /* In quick mode (for mandocdb), abort after the NAME section. */
808 if (mdoc->quick && MDOC_Sh == tok &&
809 SEC_NAME != mdoc->last->sec)
816 mdoc_isdelim(const char *p)
829 return(DELIM_MIDDLE);
853 if (0 == strcmp(p + 1, "."))
855 if (0 == strcmp(p + 1, "fR|\\fP"))
856 return(DELIM_MIDDLE);
862 mdoc_deroff(char **dest, const struct mdoc_node *n)
867 if (MDOC_TEXT != n->type) {
868 for (n = n->child; n; n = n->next)
869 mdoc_deroff(dest, n);
873 /* Skip leading whitespace. */
875 for (cp = n->string; '\0' != *cp; cp++)
876 if (0 == isspace((unsigned char)*cp))
879 /* Skip trailing whitespace. */
881 for (sz = strlen(cp); sz; sz--)
882 if (0 == isspace((unsigned char)cp[sz-1]))
885 /* Skip empty strings. */
891 *dest = mandoc_strndup(cp, sz);
895 mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);