1 /* $Id: mdoc.c,v 1.205 2013/10/21 23:47:58 schwarze Exp $ */
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010, 2012, 2013 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.
22 #include <sys/types.h>
34 #include "libmandoc.h"
36 const char *const __mdoc_macronames[MDOC_MAX] = {
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",
48 "Xr", "%A", "%B", "%D",
50 "%I", "%J", "%N", "%O",
52 "%P", "%R", "%T", "%V",
53 "Ac", "Ao", "Aq", "At",
54 "Bc", "Bf", "Bo", "Bq",
55 "Bsx", "Bx", "Db", "Dc",
56 "Do", "Dq", "Ec", "Ef",
57 "Em", "Eo", "Fx", "Ms",
58 "No", "Ns", "Nx", "Ox",
59 "Pc", "Pf", "Po", "Pq",
60 "Qc", "Ql", "Qo", "Qq",
61 "Re", "Rs", "Sc", "So",
62 "Sq", "Sm", "Sx", "Sy",
63 "Tn", "Ux", "Xc", "Xo",
64 "Fo", "Fc", "Oo", "Oc",
65 "Bk", "Ek", "Bt", "Hf",
66 "Fr", "Ud", "Lb", "Lp",
67 "Lk", "Mt", "Brq", "Bro",
69 "Brc", "%C", "Es", "En",
71 "Dx", "%Q", "br", "sp",
76 const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
77 "split", "nosplit", "ragged",
78 "unfilled", "literal", "file",
79 "offset", "bullet", "dash",
80 "hyphen", "item", "enum",
81 "tag", "diag", "hang",
82 "ohang", "inset", "column",
83 "width", "compact", "std",
84 "filled", "words", "emphasis",
85 "symbolic", "nested", "centered"
88 const char * const *mdoc_macronames = __mdoc_macronames;
89 const char * const *mdoc_argnames = __mdoc_argnames;
91 static void mdoc_node_free(struct mdoc_node *);
92 static void mdoc_node_unlink(struct mdoc *,
94 static void mdoc_free1(struct mdoc *);
95 static void mdoc_alloc1(struct mdoc *);
96 static struct mdoc_node *node_alloc(struct mdoc *, int, int,
97 enum mdoct, enum mdoc_type);
98 static int node_append(struct mdoc *,
101 static int mdoc_preptext(struct mdoc *, int, char *, int);
103 static int mdoc_ptext(struct mdoc *, int, char *, int);
104 static int mdoc_pmacro(struct mdoc *, int, char *, int);
106 const struct mdoc_node *
107 mdoc_node(const struct mdoc *mdoc)
110 assert( ! (MDOC_HALT & mdoc->flags));
115 const struct mdoc_meta *
116 mdoc_meta(const struct mdoc *mdoc)
119 assert( ! (MDOC_HALT & mdoc->flags));
125 * Frees volatile resources (parse tree, meta-data, fields).
128 mdoc_free1(struct mdoc *mdoc)
132 mdoc_node_delete(mdoc, mdoc->first);
133 if (mdoc->meta.title)
134 free(mdoc->meta.title);
138 free(mdoc->meta.name);
140 free(mdoc->meta.arch);
142 free(mdoc->meta.vol);
144 free(mdoc->meta.msec);
146 free(mdoc->meta.date);
151 * Allocate all volatile resources (parse tree, meta-data, fields).
154 mdoc_alloc1(struct mdoc *mdoc)
157 memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
159 mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
160 mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
161 mdoc->first = mdoc->last;
162 mdoc->last->type = MDOC_ROOT;
163 mdoc->last->tok = MDOC_MAX;
164 mdoc->next = MDOC_NEXT_CHILD;
169 * Free up volatile resources (see mdoc_free1()) then re-initialises the
170 * data with mdoc_alloc1(). After invocation, parse data has been reset
171 * and the parser is ready for re-invocation on a new tree; however,
172 * cross-parse non-volatile data is kept intact.
175 mdoc_reset(struct mdoc *mdoc)
184 * Completely free up all volatile and non-volatile parse resources.
185 * After invocation, the pointer is no longer usable.
188 mdoc_free(struct mdoc *mdoc)
197 * Allocate volatile and non-volatile parse resources.
200 mdoc_alloc(struct roff *roff, struct mparse *parse, char *defos)
204 p = mandoc_calloc(1, sizeof(struct mdoc));
217 * Climb back up the parse tree, validating open scopes. Mostly calls
218 * through to macro_end() in macro.c.
221 mdoc_endparse(struct mdoc *mdoc)
224 assert( ! (MDOC_HALT & mdoc->flags));
225 if (mdoc_macroend(mdoc))
227 mdoc->flags |= MDOC_HALT;
232 mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
236 assert( ! (MDOC_HALT & mdoc->flags));
238 /* No text before an initial macro. */
240 if (SEC_NONE == mdoc->lastnamed) {
241 mdoc_pmsg(mdoc, ep->ln, ep->pos, MANDOCERR_NOTEXT);
245 n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
248 if ( ! node_append(mdoc, n))
251 mdoc->next = MDOC_NEXT_SIBLING;
256 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
260 assert( ! (MDOC_HALT & mdoc->flags));
262 /* No text before an initial macro. */
264 if (SEC_NONE == mdoc->lastnamed) {
265 mdoc_pmsg(mdoc, sp->line, 0, MANDOCERR_NOTEXT);
269 n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
272 if ( ! node_append(mdoc, n))
275 mdoc->next = MDOC_NEXT_SIBLING;
281 * Main parse routine. Parses a single line -- really just hands off to
282 * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
285 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
288 assert( ! (MDOC_HALT & mdoc->flags));
290 mdoc->flags |= MDOC_NEWLINE;
293 * Let the roff nS register switch SYNOPSIS mode early,
294 * such that the parser knows at all times
295 * whether this mode is on or off.
296 * Note that this mode is also switched by the Sh macro.
298 if (roff_getreg(mdoc->roff, "nS"))
299 mdoc->flags |= MDOC_SYNOPSIS;
301 mdoc->flags &= ~MDOC_SYNOPSIS;
303 return(roff_getcontrol(mdoc->roff, buf, &offs) ?
304 mdoc_pmacro(mdoc, ln, buf, offs) :
305 mdoc_ptext(mdoc, ln, buf, offs));
309 mdoc_macro(MACRO_PROT_ARGS)
311 assert(tok < MDOC_MAX);
313 /* If we're in the body, deny prologue calls. */
315 if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
316 MDOC_PBODY & mdoc->flags) {
317 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADBODY);
321 /* If we're in the prologue, deny "body" macros. */
323 if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
324 ! (MDOC_PBODY & mdoc->flags)) {
325 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADPROLOG);
326 if (NULL == mdoc->meta.msec)
327 mdoc->meta.msec = mandoc_strdup("1");
328 if (NULL == mdoc->meta.title)
329 mdoc->meta.title = mandoc_strdup("UNKNOWN");
330 if (NULL == mdoc->meta.vol)
331 mdoc->meta.vol = mandoc_strdup("LOCAL");
332 if (NULL == mdoc->meta.os)
333 mdoc->meta.os = mandoc_strdup("LOCAL");
334 if (NULL == mdoc->meta.date)
335 mdoc->meta.date = mandoc_normdate
336 (mdoc->parse, NULL, line, ppos);
337 mdoc->flags |= MDOC_PBODY;
340 return((*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf));
345 node_append(struct mdoc *mdoc, struct mdoc_node *p)
350 assert(MDOC_ROOT != p->type);
352 switch (mdoc->next) {
353 case (MDOC_NEXT_SIBLING):
354 mdoc->last->next = p;
355 p->prev = mdoc->last;
356 p->parent = mdoc->last->parent;
358 case (MDOC_NEXT_CHILD):
359 mdoc->last->child = p;
360 p->parent = mdoc->last;
370 * Copy over the normalised-data pointer of our parent. Not
371 * everybody has one, but copying a null pointer is fine.
376 if (ENDBODY_NOT != p->end)
382 p->norm = p->parent->norm;
388 if ( ! mdoc_valid_pre(mdoc, p))
393 assert(MDOC_BLOCK == p->parent->type);
397 assert(MDOC_BLOCK == p->parent->type);
403 assert(MDOC_BLOCK == p->parent->type);
416 if ( ! mdoc_valid_post(mdoc))
427 static struct mdoc_node *
428 node_alloc(struct mdoc *mdoc, int line, int pos,
429 enum mdoct tok, enum mdoc_type type)
433 p = mandoc_calloc(1, sizeof(struct mdoc_node));
434 p->sec = mdoc->lastsec;
442 if (MDOC_SYNOPSIS & mdoc->flags)
443 p->flags |= MDOC_SYNPRETTY;
445 p->flags &= ~MDOC_SYNPRETTY;
446 if (MDOC_NEWLINE & mdoc->flags)
447 p->flags |= MDOC_LINE;
448 mdoc->flags &= ~MDOC_NEWLINE;
455 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
459 p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
460 if ( ! node_append(mdoc, p))
462 mdoc->next = MDOC_NEXT_CHILD;
468 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
475 p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
476 if ( ! node_append(mdoc, p))
478 mdoc->next = MDOC_NEXT_CHILD;
484 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
488 p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
489 if ( ! node_append(mdoc, p))
491 mdoc->next = MDOC_NEXT_CHILD;
497 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
498 struct mdoc_node *body, enum mdoc_endbody end)
502 p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
504 p->norm = body->norm;
506 if ( ! node_append(mdoc, p))
508 mdoc->next = MDOC_NEXT_SIBLING;
514 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
515 enum mdoct tok, struct mdoc_arg *args)
519 p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
532 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
538 if ( ! node_append(mdoc, p))
540 mdoc->next = MDOC_NEXT_CHILD;
546 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
547 enum mdoct tok, struct mdoc_arg *args)
551 p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
558 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
564 if ( ! node_append(mdoc, p))
566 mdoc->next = MDOC_NEXT_CHILD;
571 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
575 n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
576 n->string = roff_strdup(mdoc->roff, p);
578 if ( ! node_append(mdoc, n))
581 mdoc->next = MDOC_NEXT_SIBLING;
586 mdoc_word_append(struct mdoc *mdoc, const char *p)
589 char *addstr, *newstr;
592 addstr = roff_strdup(mdoc->roff, p);
593 if (-1 == asprintf(&newstr, "%s %s", n->string, addstr)) {
595 exit((int)MANDOCLEVEL_SYSERR);
600 mdoc->next = MDOC_NEXT_SIBLING;
604 mdoc_node_free(struct mdoc_node *p)
607 if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
612 mdoc_argv_free(p->args);
618 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
621 /* Adjust siblings. */
624 n->prev->next = n->next;
626 n->next->prev = n->prev;
632 if (n->parent->child == n)
633 n->parent->child = n->prev ? n->prev : n->next;
634 if (n->parent->last == n)
635 n->parent->last = n->prev ? n->prev : NULL;
638 /* Adjust parse point, if applicable. */
640 if (mdoc && mdoc->last == n) {
642 mdoc->last = n->prev;
643 mdoc->next = MDOC_NEXT_SIBLING;
645 mdoc->last = n->parent;
646 mdoc->next = MDOC_NEXT_CHILD;
650 if (mdoc && mdoc->first == n)
656 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
661 mdoc_node_delete(mdoc, p->child);
663 assert(0 == p->nchild);
665 mdoc_node_unlink(mdoc, p);
670 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
673 mdoc_node_unlink(mdoc, p);
674 return(node_append(mdoc, p));
679 * Pre-treat a text line.
680 * Text lines can consist of equations, which must be handled apart from
682 * Thus, use this function to step through a line checking if it has any
683 * equations embedded in it.
684 * This must handle multiple equations AND equations that do not end at
685 * the end-of-line, i.e., will re-enter in the next roff parse.
688 mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
693 while ('\0' != buf[offs]) {
694 /* Mark starting position if eqn is set. */
696 if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
697 if (NULL != (start = strchr(buf + offs, delim)))
700 /* Parse text as normal. */
701 if ( ! mdoc_ptext(mdoc, line, buf, offs))
704 /* Continue only if an equation exists. */
708 /* Read past the end of the equation. */
709 offs += start - (buf + offs);
710 assert(start == &buf[offs]);
711 if (NULL != (end = strchr(buf + offs, delim))) {
717 /* Parse the equation itself. */
718 roff_openeqn(mdoc->roff, NULL, line, offs, buf);
720 /* Process a finished equation? */
721 if (roff_closeeqn(mdoc->roff))
722 if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
724 offs += (end - (buf + offs));
732 * Parse free-form text, that is, a line that does not begin with the
736 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
741 /* No text before an initial macro. */
743 if (SEC_NONE == mdoc->lastnamed) {
744 mdoc_pmsg(mdoc, line, offs, MANDOCERR_NOTEXT);
752 * Divert directly to list processing if we're encountering a
753 * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
754 * (a MDOC_BODY means it's already open, in which case we should
755 * process within its context in the normal way).
758 if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
759 LIST_column == n->norm->Bl.type) {
760 /* `Bl' is open without any children. */
761 mdoc->flags |= MDOC_FREECOL;
762 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
765 if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
767 MDOC_Bl == n->parent->tok &&
768 LIST_column == n->parent->norm->Bl.type) {
769 /* `Bl' has block-level `It' children. */
770 mdoc->flags |= MDOC_FREECOL;
771 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
775 * Search for the beginning of unescaped trailing whitespace (ws)
776 * and for the first character not to be output (end).
779 /* FIXME: replace with strcspn(). */
781 for (c = end = buf + offs; *c; c++) {
789 * Always warn about trailing tabs,
790 * even outside literal context,
791 * where they should be put on the next line.
796 * Strip trailing tabs in literal context only;
797 * outside, they affect the next line.
799 if (MDOC_LITERAL & mdoc->flags)
803 /* Skip the escaped character, too, if any. */
816 mdoc_pmsg(mdoc, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
818 if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
819 mdoc_pmsg(mdoc, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
822 * Insert a `sp' in the case of a blank line. Technically,
823 * blank lines aren't allowed, but enough manuals assume this
824 * behaviour that we want to work around it.
826 if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
829 mdoc->next = MDOC_NEXT_SIBLING;
831 return(mdoc_valid_post(mdoc));
834 if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
837 if (MDOC_LITERAL & mdoc->flags)
841 * End-of-sentence check. If the last character is an unescaped
842 * EOS character, then flag the node as being the end of a
843 * sentence. The front-end will know how to interpret this.
848 if (mandoc_eos(buf+offs, (size_t)(end-buf-offs), 0))
849 mdoc->last->flags |= MDOC_EOS;
856 * Parse a macro line, that is, a line beginning with the control
860 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
867 /* Empty post-control lines are ignored. */
869 if ('"' == buf[offs]) {
870 mdoc_pmsg(mdoc, ln, offs, MANDOCERR_BADCOMMENT);
872 } else if ('\0' == buf[offs])
878 * Copy the first word into a nil-terminated buffer.
879 * Stop copying when a tab, space, or eoln is encountered.
883 while (i < 4 && '\0' != buf[offs] &&
884 ' ' != buf[offs] && '\t' != buf[offs])
885 mac[i++] = buf[offs++];
889 tok = (i > 1 || i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
891 if (MDOC_MAX == tok) {
892 mandoc_vmsg(MANDOCERR_MACRO, mdoc->parse,
893 ln, sv, "%s", buf + sv - 1);
897 /* Disregard the first trailing tab, if applicable. */
899 if ('\t' == buf[offs])
902 /* Jump to the next non-whitespace word. */
904 while (buf[offs] && ' ' == buf[offs])
908 * Trailing whitespace. Note that tabs are allowed to be passed
909 * into the parser as "text", so we only warn about spaces here.
912 if ('\0' == buf[offs] && ' ' == buf[offs - 1])
913 mdoc_pmsg(mdoc, ln, offs - 1, MANDOCERR_EOLNSPACE);
916 * If an initial macro or a list invocation, divert directly
917 * into macro processing.
920 if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
921 if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
930 * If the first macro of a `Bl -column', open an `It' block
931 * context around the parsed macro.
934 if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
935 LIST_column == n->norm->Bl.type) {
936 mdoc->flags |= MDOC_FREECOL;
937 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
943 * If we're following a block-level `It' within a `Bl -column'
944 * context (perhaps opened in the above block or in ptext()),
945 * then open an `It' block context around the parsed macro.
948 if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
950 MDOC_Bl == n->parent->tok &&
951 LIST_column == n->parent->norm->Bl.type) {
952 mdoc->flags |= MDOC_FREECOL;
953 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
958 /* Normal processing of a macro. */
960 if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
965 err: /* Error out. */
967 mdoc->flags |= MDOC_HALT;
972 mdoc_isdelim(const char *p)
985 return(DELIM_MIDDLE);
1001 return(DELIM_CLOSE);
1009 if (0 == strcmp(p + 1, "."))
1010 return(DELIM_CLOSE);
1011 if (0 == strcmp(p + 1, "fR|\\fP"))
1012 return(DELIM_MIDDLE);