1 /* $Id: mdoc.c,v 1.30 2009/10/21 19:13:50 schwarze Exp $ */
3 * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
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.
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.
17 #include <sys/types.h>
28 const char *const __mdoc_merrnames[MERRMAX] = {
29 "trailing whitespace", /* ETAILWS */
30 "unexpected quoted parameter", /* EQUOTPARM */
31 "unterminated quoted parameter", /* EQUOTTERM */
32 "system: malloc error", /* EMALLOC */
33 "argument parameter suggested", /* EARGVAL */
34 "macro disallowed in prologue", /* EBODYPROL */
35 "macro disallowed in body", /* EPROLBODY */
36 "text disallowed in prologue", /* ETEXTPROL */
37 "blank line disallowed", /* ENOBLANK */
38 "text parameter too long", /* ETOOLONG */
39 "invalid escape sequence", /* EESCAPE */
40 "invalid character", /* EPRINT */
41 "document has no body", /* ENODAT */
42 "document has no prologue", /* ENOPROLOGUE */
43 "expected line arguments", /* ELINE */
44 "invalid AT&T argument", /* EATT */
45 "default name not yet set", /* ENAME */
46 "missing list type", /* ELISTTYPE */
47 "missing display type", /* EDISPTYPE */
48 "too many display types", /* EMULTIDISP */
49 "too many list types", /* EMULTILIST */
50 "NAME section must be first", /* ESECNAME */
51 "badly-formed NAME section", /* ENAMESECINC */
52 "argument repeated", /* EARGREP */
53 "expected boolean parameter", /* EBOOL */
54 "inconsistent column syntax", /* ECOLMIS */
55 "nested display invalid", /* ENESTDISP */
56 "width argument missing", /* EMISSWIDTH */
57 "invalid section for this manual section", /* EWRONGMSEC */
58 "section out of conventional order", /* ESECOOO */
59 "section repeated", /* ESECREP */
60 "invalid standard argument", /* EBADSTAND */
61 "multi-line arguments discouraged", /* ENOMULTILINE */
62 "multi-line arguments suggested", /* EMULTILINE */
63 "line arguments discouraged", /* ENOLINE */
64 "prologue macro out of conventional order", /* EPROLOOO */
65 "prologue macro repeated", /* EPROLREP */
66 "invalid manual section", /* EBADMSEC */
67 "invalid section", /* EBADSEC */
68 "invalid font mode", /* EFONT */
69 "invalid date syntax", /* EBADDATE */
70 "invalid number format", /* ENUMFMT */
71 "superfluous width argument", /* ENOWIDTH */
72 "system: utsname error", /* EUTSNAME */
73 "obsolete macro", /* EOBS */
74 "end-of-line scope violation", /* EIMPBRK */
75 "empty macro ignored", /* EIGNE */
76 "unclosed explicit scope", /* EOPEN */
77 "unterminated quoted phrase", /* EQUOTPHR */
78 "closure macro without prior context", /* ENOCTX */
79 "no description found for library", /* ELIB */
80 "bad child for parent context", /* EBADCHILD */
81 "list arguments preceding type", /* ENOTYPE */
84 const char *const __mdoc_macronames[MDOC_MAX] = {
85 "Ap", "Dd", "Dt", "Os",
86 "Sh", "Ss", "Pp", "D1",
87 "Dl", "Bd", "Ed", "Bl",
88 "El", "It", "Ad", "An",
89 "Ar", "Cd", "Cm", "Dv",
90 "Er", "Ev", "Ex", "Fa",
91 "Fd", "Fl", "Fn", "Ft",
92 "Ic", "In", "Li", "Nd",
93 "Nm", "Op", "Ot", "Pa",
94 "Rv", "St", "Va", "Vt",
96 "Xr", "\%A", "\%B", "\%D",
98 "\%I", "\%J", "\%N", "\%O",
100 "\%P", "\%R", "\%T", "\%V",
101 "Ac", "Ao", "Aq", "At",
102 "Bc", "Bf", "Bo", "Bq",
103 "Bsx", "Bx", "Db", "Dc",
104 "Do", "Dq", "Ec", "Ef",
105 "Em", "Eo", "Fx", "Ms",
106 "No", "Ns", "Nx", "Ox",
107 "Pc", "Pf", "Po", "Pq",
108 "Qc", "Ql", "Qo", "Qq",
109 "Re", "Rs", "Sc", "So",
110 "Sq", "Sm", "Sx", "Sy",
111 "Tn", "Ux", "Xc", "Xo",
112 "Fo", "Fc", "Oo", "Oc",
113 "Bk", "Ek", "Bt", "Hf",
114 "Fr", "Ud", "Lb", "Lp",
115 "Lk", "Mt", "Brq", "Bro",
117 "Brc", "\%C", "Es", "En",
119 "Dx", "\%Q", "br", "sp"
122 const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
123 "split", "nosplit", "ragged",
124 "unfilled", "literal", "file",
125 "offset", "bullet", "dash",
126 "hyphen", "item", "enum",
127 "tag", "diag", "hang",
128 "ohang", "inset", "column",
129 "width", "compact", "std",
130 "filled", "words", "emphasis",
131 "symbolic", "nested", "centered"
134 const char * const *mdoc_macronames = __mdoc_macronames;
135 const char * const *mdoc_argnames = __mdoc_argnames;
137 static void mdoc_free1(struct mdoc *);
138 static int mdoc_alloc1(struct mdoc *);
139 static struct mdoc_node *node_alloc(struct mdoc *, int, int,
140 int, enum mdoc_type);
141 static int node_append(struct mdoc *,
143 static int parsetext(struct mdoc *, int, char *);
144 static int parsemacro(struct mdoc *, int, char *);
145 static int macrowarn(struct mdoc *, int, const char *);
146 static int pstring(struct mdoc *, int, int,
147 const char *, size_t);
150 const struct mdoc_node *
151 mdoc_node(const struct mdoc *m)
154 return(MDOC_HALT & m->flags ? NULL : m->first);
158 const struct mdoc_meta *
159 mdoc_meta(const struct mdoc *m)
162 return(MDOC_HALT & m->flags ? NULL : &m->meta);
167 * Frees volatile resources (parse tree, meta-data, fields).
170 mdoc_free1(struct mdoc *mdoc)
174 mdoc_node_freelist(mdoc->first);
175 if (mdoc->meta.title)
176 free(mdoc->meta.title);
180 free(mdoc->meta.name);
182 free(mdoc->meta.arch);
184 free(mdoc->meta.vol);
189 * Allocate all volatile resources (parse tree, meta-data, fields).
192 mdoc_alloc1(struct mdoc *mdoc)
195 bzero(&mdoc->meta, sizeof(struct mdoc_meta));
197 mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
198 mdoc->last = calloc(1, sizeof(struct mdoc_node));
199 if (NULL == mdoc->last)
202 mdoc->first = mdoc->last;
203 mdoc->last->type = MDOC_ROOT;
204 mdoc->next = MDOC_NEXT_CHILD;
210 * Free up volatile resources (see mdoc_free1()) then re-initialises the
211 * data with mdoc_alloc1(). After invocation, parse data has been reset
212 * and the parser is ready for re-invocation on a new tree; however,
213 * cross-parse non-volatile data is kept intact.
216 mdoc_reset(struct mdoc *mdoc)
220 return(mdoc_alloc1(mdoc));
225 * Completely free up all volatile and non-volatile parse resources.
226 * After invocation, the pointer is no longer usable.
229 mdoc_free(struct mdoc *mdoc)
238 * Allocate volatile and non-volatile parse resources.
241 mdoc_alloc(void *data, int pflags, const struct mdoc_cb *cb)
245 if (NULL == (p = calloc(1, sizeof(struct mdoc))))
248 (void)memcpy(&p->cb, cb, sizeof(struct mdoc_cb));
264 * Climb back up the parse tree, validating open scopes. Mostly calls
265 * through to macro_end() in macro.c.
268 mdoc_endparse(struct mdoc *m)
271 if (MDOC_HALT & m->flags)
273 else if (mdoc_macroend(m))
275 m->flags |= MDOC_HALT;
281 * Main parse routine. Parses a single line -- really just hands off to
282 * the macro (parsemacro()) or text parser (parsetext()).
285 mdoc_parseln(struct mdoc *m, int ln, char *buf)
288 if (MDOC_HALT & m->flags)
291 return('.' == *buf ? parsemacro(m, ln, buf) :
292 parsetext(m, ln, buf));
297 mdoc_verr(struct mdoc *mdoc, int ln, int pos,
298 const char *fmt, ...)
303 if (NULL == mdoc->cb.mdoc_err)
307 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
310 return((*mdoc->cb.mdoc_err)(mdoc->data, ln, pos, buf));
315 mdoc_vwarn(struct mdoc *mdoc, int ln, int pos, const char *fmt, ...)
320 if (NULL == mdoc->cb.mdoc_warn)
324 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
327 return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, buf));
332 mdoc_err(struct mdoc *m, int line, int pos, int iserr, enum merr type)
336 p = __mdoc_merrnames[(int)type];
340 return(mdoc_verr(m, line, pos, p));
342 return(mdoc_vwarn(m, line, pos, p));
347 mdoc_macro(struct mdoc *m, int tok,
348 int ln, int pp, int *pos, char *buf)
351 * If we're in the prologue, deny "body" macros. Similarly, if
352 * we're in the body, deny prologue calls.
354 if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
355 MDOC_PBODY & m->flags)
356 return(mdoc_perr(m, ln, pp, EPROLBODY));
357 if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
358 ! (MDOC_PBODY & m->flags))
359 return(mdoc_perr(m, ln, pp, EBODYPROL));
361 return((*mdoc_macros[tok].fp)(m, tok, ln, pp, pos, buf));
366 node_append(struct mdoc *mdoc, struct mdoc_node *p)
371 assert(MDOC_ROOT != p->type);
373 switch (mdoc->next) {
374 case (MDOC_NEXT_SIBLING):
375 mdoc->last->next = p;
376 p->prev = mdoc->last;
377 p->parent = mdoc->last->parent;
379 case (MDOC_NEXT_CHILD):
380 mdoc->last->child = p;
381 p->parent = mdoc->last;
390 if ( ! mdoc_valid_pre(mdoc, p))
392 if ( ! mdoc_action_pre(mdoc, p))
397 assert(MDOC_BLOCK == p->parent->type);
401 assert(MDOC_BLOCK == p->parent->type);
405 assert(MDOC_BLOCK == p->parent->type);
416 if ( ! mdoc_valid_post(mdoc))
418 if ( ! mdoc_action_post(mdoc))
429 static struct mdoc_node *
430 node_alloc(struct mdoc *m, int line,
431 int pos, int tok, enum mdoc_type type)
435 if (NULL == (p = calloc(1, sizeof(struct mdoc_node)))) {
436 (void)mdoc_nerr(m, m->last, EMALLOC);
444 if (MDOC_TEXT != (p->type = type))
452 mdoc_tail_alloc(struct mdoc *m, int line, int pos, int tok)
456 p = node_alloc(m, line, pos, tok, MDOC_TAIL);
459 if ( ! node_append(m, p))
461 m->next = MDOC_NEXT_CHILD;
467 mdoc_head_alloc(struct mdoc *m, int line, int pos, int tok)
474 p = node_alloc(m, line, pos, tok, MDOC_HEAD);
477 if ( ! node_append(m, p))
479 m->next = MDOC_NEXT_CHILD;
485 mdoc_body_alloc(struct mdoc *m, int line, int pos, int tok)
489 p = node_alloc(m, line, pos, tok, MDOC_BODY);
492 if ( ! node_append(m, p))
494 m->next = MDOC_NEXT_CHILD;
500 mdoc_block_alloc(struct mdoc *m, int line, int pos,
501 int tok, struct mdoc_arg *args)
505 p = node_alloc(m, line, pos, tok, MDOC_BLOCK);
511 if ( ! node_append(m, p))
513 m->next = MDOC_NEXT_CHILD;
519 mdoc_elem_alloc(struct mdoc *m, int line, int pos,
520 int tok, struct mdoc_arg *args)
524 p = node_alloc(m, line, pos, tok, MDOC_ELEM);
530 if ( ! node_append(m, p))
532 m->next = MDOC_NEXT_CHILD;
538 pstring(struct mdoc *m, int line, int pos, const char *p, size_t len)
543 n = node_alloc(m, line, pos, -1, MDOC_TEXT);
545 return(mdoc_nerr(m, m->last, EMALLOC));
547 n->string = malloc(len + 1);
548 if (NULL == n->string) {
550 return(mdoc_nerr(m, m->last, EMALLOC));
553 sv = strlcpy(n->string, p, len + 1);
555 /* Prohibit truncation. */
556 assert(sv < len + 1);
558 if ( ! node_append(m, n))
560 m->next = MDOC_NEXT_SIBLING;
566 mdoc_word_alloc(struct mdoc *m, int line, int pos, const char *p)
569 return(pstring(m, line, pos, p, strlen(p)));
574 mdoc_node_free(struct mdoc_node *p)
582 mdoc_argv_free(p->args);
588 mdoc_node_freelist(struct mdoc_node *p)
592 mdoc_node_freelist(p->child);
594 mdoc_node_freelist(p->next);
596 assert(0 == p->nchild);
602 * Parse free-form text, that is, a line that does not begin with the
606 parsetext(struct mdoc *m, int line, char *buf)
610 if (SEC_NONE == m->lastnamed)
611 return(mdoc_perr(m, line, 0, ETEXTPROL));
614 * If in literal mode, then pass the buffer directly to the
615 * back-end, as it should be preserved as a single term.
618 if (MDOC_LITERAL & m->flags)
619 return(mdoc_word_alloc(m, line, 0, buf));
621 /* Disallow blank/white-space lines in non-literal mode. */
623 for (i = 0; ' ' == buf[i]; i++)
624 /* Skip leading whitespace. */ ;
626 return(mdoc_perr(m, line, 0, ENOBLANK));
629 * Break apart a free-form line into tokens. Spaces are
630 * stripped out of the input.
633 for (j = i; buf[i]; i++) {
637 /* Escaped whitespace. */
638 if (i && ' ' == buf[i] && '\\' == buf[i - 1])
642 if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
645 for ( ; ' ' == buf[i]; i++)
646 /* Skip trailing whitespace. */ ;
653 if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
656 m->next = MDOC_NEXT_SIBLING;
664 macrowarn(struct mdoc *m, int ln, const char *buf)
666 if ( ! (MDOC_IGN_MACRO & m->pflags))
667 return(mdoc_verr(m, ln, 0,
668 "unknown macro: %s%s",
669 buf, strlen(buf) > 3 ? "..." : ""));
670 return(mdoc_vwarn(m, ln, 0, "unknown macro: %s%s",
671 buf, strlen(buf) > 3 ? "..." : ""));
676 * Parse a macro line, that is, a line beginning with the control
680 parsemacro(struct mdoc *m, int ln, char *buf)
685 /* Empty lines are ignored. */
692 /* Accept whitespace after the initial control char. */
696 while (buf[i] && ' ' == buf[i])
702 /* Copy the first word into a nil-terminated buffer. */
704 for (j = 0; j < 4; j++, i++) {
705 if (0 == (mac[j] = buf[i]))
707 else if (' ' == buf[i])
710 /* Check for invalid characters. */
712 if (isgraph((u_char)buf[i]))
714 return(mdoc_perr(m, ln, i, EPRINT));
719 if (j == 4 || j < 2) {
720 if ( ! macrowarn(m, ln, mac))
725 if (MDOC_MAX == (c = mdoc_hash_find(mac))) {
726 if ( ! macrowarn(m, ln, mac))
731 /* The macro is sane. Jump to the next word. */
733 while (buf[i] && ' ' == buf[i])
737 * Begin recursive parse sequence. Since we're at the start of
738 * the line, we don't need to do callable/parseable checks.
740 if ( ! mdoc_macro(m, c, ln, 1, &i, buf))
745 err: /* Error out. */
747 m->flags |= MDOC_HALT;