]>
git.cameronkatri.com Git - mandoc.git/blob - mdoc.c
1 /* $Id: mdoc.c,v 1.121 2010/04/06 11:33:00 kristaps 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.
21 #include <sys/types.h>
32 #include "libmandoc.h"
34 const char *const __mdoc_merrnames
[MERRMAX
] = {
35 "trailing whitespace", /* ETAILWS */
36 "unexpected quoted parameter", /* EQUOTPARM */
37 "unterminated quoted parameter", /* EQUOTTERM */
38 "argument parameter suggested", /* EARGVAL */
39 "macro disallowed in prologue", /* EBODYPROL */
40 "macro disallowed in body", /* EPROLBODY */
41 "text disallowed in prologue", /* ETEXTPROL */
42 "blank line disallowed", /* ENOBLANK */
43 "text parameter too long", /* ETOOLONG */
44 "invalid escape sequence", /* EESCAPE */
45 "invalid character", /* EPRINT */
46 "document has no body", /* ENODAT */
47 "document has no prologue", /* ENOPROLOGUE */
48 "expected line arguments", /* ELINE */
49 "invalid AT&T argument", /* EATT */
50 "default name not yet set", /* ENAME */
51 "missing list type", /* ELISTTYPE */
52 "missing display type", /* EDISPTYPE */
53 "too many display types", /* EMULTIDISP */
54 "too many list types", /* EMULTILIST */
55 "NAME section must be first", /* ESECNAME */
56 "badly-formed NAME section", /* ENAMESECINC */
57 "argument repeated", /* EARGREP */
58 "expected boolean parameter", /* EBOOL */
59 "inconsistent column syntax", /* ECOLMIS */
60 "nested display invalid", /* ENESTDISP */
61 "width argument missing", /* EMISSWIDTH */
62 "invalid section for this manual section", /* EWRONGMSEC */
63 "section out of conventional order", /* ESECOOO */
64 "section repeated", /* ESECREP */
65 "invalid standard argument", /* EBADSTAND */
66 "multi-line arguments discouraged", /* ENOMULTILINE */
67 "multi-line arguments suggested", /* EMULTILINE */
68 "line arguments discouraged", /* ENOLINE */
69 "prologue macro out of conventional order", /* EPROLOOO */
70 "prologue macro repeated", /* EPROLREP */
71 "invalid manual section", /* EBADMSEC */
72 "invalid section", /* EBADSEC */
73 "invalid font mode", /* EFONT */
74 "invalid date syntax", /* EBADDATE */
75 "invalid number format", /* ENUMFMT */
76 "superfluous width argument", /* ENOWIDTH */
77 "system: utsname error", /* EUTSNAME */
78 "obsolete macro", /* EOBS */
79 "end-of-line scope violation", /* EIMPBRK */
80 "empty macro ignored", /* EIGNE */
81 "unclosed explicit scope", /* EOPEN */
82 "unterminated quoted phrase", /* EQUOTPHR */
83 "closure macro without prior context", /* ENOCTX */
84 "no description found for library", /* ELIB */
85 "bad child for parent context", /* EBADCHILD */
86 "list arguments preceding type", /* ENOTYPE */
89 const char *const __mdoc_macronames
[MDOC_MAX
] = {
90 "Ap", "Dd", "Dt", "Os",
91 "Sh", "Ss", "Pp", "D1",
92 "Dl", "Bd", "Ed", "Bl",
93 "El", "It", "Ad", "An",
94 "Ar", "Cd", "Cm", "Dv",
95 "Er", "Ev", "Ex", "Fa",
96 "Fd", "Fl", "Fn", "Ft",
97 "Ic", "In", "Li", "Nd",
98 "Nm", "Op", "Ot", "Pa",
99 "Rv", "St", "Va", "Vt",
101 "Xr", "%A", "%B", "%D",
103 "%I", "%J", "%N", "%O",
105 "%P", "%R", "%T", "%V",
106 "Ac", "Ao", "Aq", "At",
107 "Bc", "Bf", "Bo", "Bq",
108 "Bsx", "Bx", "Db", "Dc",
109 "Do", "Dq", "Ec", "Ef",
110 "Em", "Eo", "Fx", "Ms",
111 "No", "Ns", "Nx", "Ox",
112 "Pc", "Pf", "Po", "Pq",
113 "Qc", "Ql", "Qo", "Qq",
114 "Re", "Rs", "Sc", "So",
115 "Sq", "Sm", "Sx", "Sy",
116 "Tn", "Ux", "Xc", "Xo",
117 "Fo", "Fc", "Oo", "Oc",
118 "Bk", "Ek", "Bt", "Hf",
119 "Fr", "Ud", "Lb", "Lp",
120 "Lk", "Mt", "Brq", "Bro",
122 "Brc", "%C", "Es", "En",
124 "Dx", "%Q", "br", "sp",
129 const char *const __mdoc_argnames
[MDOC_ARG_MAX
] = {
130 "split", "nosplit", "ragged",
131 "unfilled", "literal", "file",
132 "offset", "bullet", "dash",
133 "hyphen", "item", "enum",
134 "tag", "diag", "hang",
135 "ohang", "inset", "column",
136 "width", "compact", "std",
137 "filled", "words", "emphasis",
138 "symbolic", "nested", "centered"
141 const char * const *mdoc_macronames
= __mdoc_macronames
;
142 const char * const *mdoc_argnames
= __mdoc_argnames
;
144 static void mdoc_node_free(struct mdoc_node
*);
145 static void mdoc_node_unlink(struct mdoc
*,
147 static void mdoc_free1(struct mdoc
*);
148 static void mdoc_alloc1(struct mdoc
*);
149 static struct mdoc_node
*node_alloc(struct mdoc
*, int, int,
150 enum mdoct
, enum mdoc_type
);
151 static int node_append(struct mdoc
*,
153 static int parsetext(struct mdoc
*, int, char *);
154 static int parsemacro(struct mdoc
*, int, char *);
155 static int macrowarn(struct mdoc
*, int, const char *);
156 static int pstring(struct mdoc
*, int, int,
157 const char *, size_t);
159 const struct mdoc_node
*
160 mdoc_node(const struct mdoc
*m
)
163 return(MDOC_HALT
& m
->flags
? NULL
: m
->first
);
167 const struct mdoc_meta
*
168 mdoc_meta(const struct mdoc
*m
)
171 return(MDOC_HALT
& m
->flags
? NULL
: &m
->meta
);
176 * Frees volatile resources (parse tree, meta-data, fields).
179 mdoc_free1(struct mdoc
*mdoc
)
183 mdoc_node_delete(mdoc
, mdoc
->first
);
184 if (mdoc
->meta
.title
)
185 free(mdoc
->meta
.title
);
189 free(mdoc
->meta
.name
);
191 free(mdoc
->meta
.arch
);
193 free(mdoc
->meta
.vol
);
198 * Allocate all volatile resources (parse tree, meta-data, fields).
201 mdoc_alloc1(struct mdoc
*mdoc
)
204 memset(&mdoc
->meta
, 0, sizeof(struct mdoc_meta
));
206 mdoc
->lastnamed
= mdoc
->lastsec
= SEC_NONE
;
207 mdoc
->last
= mandoc_calloc(1, sizeof(struct mdoc_node
));
208 mdoc
->first
= mdoc
->last
;
209 mdoc
->last
->type
= MDOC_ROOT
;
210 mdoc
->next
= MDOC_NEXT_CHILD
;
215 * Free up volatile resources (see mdoc_free1()) then re-initialises the
216 * data with mdoc_alloc1(). After invocation, parse data has been reset
217 * and the parser is ready for re-invocation on a new tree; however,
218 * cross-parse non-volatile data is kept intact.
221 mdoc_reset(struct mdoc
*mdoc
)
230 * Completely free up all volatile and non-volatile parse resources.
231 * After invocation, the pointer is no longer usable.
234 mdoc_free(struct mdoc
*mdoc
)
243 * Allocate volatile and non-volatile parse resources.
246 mdoc_alloc(void *data
, int pflags
, const struct mdoc_cb
*cb
)
250 p
= mandoc_calloc(1, sizeof(struct mdoc
));
253 memcpy(&p
->cb
, cb
, sizeof(struct mdoc_cb
));
265 * Climb back up the parse tree, validating open scopes. Mostly calls
266 * through to macro_end() in macro.c.
269 mdoc_endparse(struct mdoc
*m
)
272 if (MDOC_HALT
& m
->flags
)
274 else if (mdoc_macroend(m
))
276 m
->flags
|= MDOC_HALT
;
282 * Main parse routine. Parses a single line -- really just hands off to
283 * the macro (parsemacro()) or text parser (parsetext()).
286 mdoc_parseln(struct mdoc
*m
, int ln
, char *buf
)
289 if (MDOC_HALT
& m
->flags
)
292 return('.' == *buf
? parsemacro(m
, ln
, buf
) :
293 parsetext(m
, ln
, buf
));
298 mdoc_verr(struct mdoc
*mdoc
, int ln
, int pos
,
299 const char *fmt
, ...)
304 if (NULL
== mdoc
->cb
.mdoc_err
)
308 (void)vsnprintf(buf
, sizeof(buf
) - 1, fmt
, ap
);
311 return((*mdoc
->cb
.mdoc_err
)(mdoc
->data
, ln
, pos
, buf
));
316 mdoc_vwarn(struct mdoc
*mdoc
, int ln
, int pos
, const char *fmt
, ...)
321 if (NULL
== mdoc
->cb
.mdoc_warn
)
325 (void)vsnprintf(buf
, sizeof(buf
) - 1, fmt
, ap
);
328 return((*mdoc
->cb
.mdoc_warn
)(mdoc
->data
, ln
, pos
, buf
));
333 mdoc_err(struct mdoc
*m
, int line
, int pos
, int iserr
, enum merr type
)
337 p
= __mdoc_merrnames
[(int)type
];
341 return(mdoc_verr(m
, line
, pos
, p
));
343 return(mdoc_vwarn(m
, line
, pos
, p
));
348 mdoc_macro(struct mdoc
*m
, enum mdoct tok
,
349 int ln
, int pp
, int *pos
, char *buf
)
352 assert(tok
< MDOC_MAX
);
354 * If we're in the prologue, deny "body" macros. Similarly, if
355 * we're in the body, deny prologue calls.
357 if (MDOC_PROLOGUE
& mdoc_macros
[tok
].flags
&&
358 MDOC_PBODY
& m
->flags
) {
359 if ( ! mdoc_pwarn(m
, ln
, pp
, EBODYPROL
))
362 * FIXME: do this in mdoc_action.c.
364 if (NULL
== m
->meta
.title
)
365 m
->meta
.title
= mandoc_strdup("unknown");
366 if (NULL
== m
->meta
.vol
)
367 m
->meta
.vol
= mandoc_strdup("local");
368 if (NULL
== m
->meta
.os
)
369 m
->meta
.os
= mandoc_strdup("local");
370 if (0 == m
->meta
.date
)
371 m
->meta
.date
= time(NULL
);
372 m
->flags
|= MDOC_PBODY
;
374 if ( ! (MDOC_PROLOGUE
& mdoc_macros
[tok
].flags
) &&
375 ! (MDOC_PBODY
& m
->flags
))
376 return(mdoc_perr(m
, ln
, pp
, EBODYPROL
));
378 return((*mdoc_macros
[tok
].fp
)(m
, tok
, ln
, pp
, pos
, buf
));
383 node_append(struct mdoc
*mdoc
, struct mdoc_node
*p
)
388 assert(MDOC_ROOT
!= p
->type
);
390 switch (mdoc
->next
) {
391 case (MDOC_NEXT_SIBLING
):
392 mdoc
->last
->next
= p
;
393 p
->prev
= mdoc
->last
;
394 p
->parent
= mdoc
->last
->parent
;
396 case (MDOC_NEXT_CHILD
):
397 mdoc
->last
->child
= p
;
398 p
->parent
= mdoc
->last
;
407 if ( ! mdoc_valid_pre(mdoc
, p
))
409 if ( ! mdoc_action_pre(mdoc
, p
))
414 assert(MDOC_BLOCK
== p
->parent
->type
);
418 assert(MDOC_BLOCK
== p
->parent
->type
);
422 assert(MDOC_BLOCK
== p
->parent
->type
);
433 if ( ! mdoc_valid_post(mdoc
))
435 if ( ! mdoc_action_post(mdoc
))
446 static struct mdoc_node
*
447 node_alloc(struct mdoc
*m
, int line
, int pos
,
448 enum mdoct tok
, enum mdoc_type type
)
452 p
= mandoc_calloc(1, sizeof(struct mdoc_node
));
464 mdoc_tail_alloc(struct mdoc
*m
, int line
, int pos
, enum mdoct tok
)
468 p
= node_alloc(m
, line
, pos
, tok
, MDOC_TAIL
);
469 if ( ! node_append(m
, p
))
471 m
->next
= MDOC_NEXT_CHILD
;
477 mdoc_head_alloc(struct mdoc
*m
, int line
, int pos
, enum mdoct tok
)
484 p
= node_alloc(m
, line
, pos
, tok
, MDOC_HEAD
);
485 if ( ! node_append(m
, p
))
487 m
->next
= MDOC_NEXT_CHILD
;
493 mdoc_body_alloc(struct mdoc
*m
, int line
, int pos
, enum mdoct tok
)
497 p
= node_alloc(m
, line
, pos
, tok
, MDOC_BODY
);
498 if ( ! node_append(m
, p
))
500 m
->next
= MDOC_NEXT_CHILD
;
506 mdoc_block_alloc(struct mdoc
*m
, int line
, int pos
,
507 enum mdoct tok
, struct mdoc_arg
*args
)
511 p
= node_alloc(m
, line
, pos
, tok
, MDOC_BLOCK
);
515 if ( ! node_append(m
, p
))
517 m
->next
= MDOC_NEXT_CHILD
;
523 mdoc_elem_alloc(struct mdoc
*m
, int line
, int pos
,
524 enum mdoct tok
, struct mdoc_arg
*args
)
528 p
= node_alloc(m
, line
, pos
, tok
, MDOC_ELEM
);
532 if ( ! node_append(m
, p
))
534 m
->next
= MDOC_NEXT_CHILD
;
540 pstring(struct mdoc
*m
, int line
, int pos
, const char *p
, size_t len
)
545 n
= node_alloc(m
, line
, pos
, -1, MDOC_TEXT
);
546 n
->string
= mandoc_malloc(len
+ 1);
547 sv
= strlcpy(n
->string
, p
, len
+ 1);
549 /* Prohibit truncation. */
550 assert(sv
< len
+ 1);
552 if ( ! node_append(m
, n
))
554 m
->next
= MDOC_NEXT_SIBLING
;
560 mdoc_word_alloc(struct mdoc
*m
, int line
, int pos
, const char *p
)
563 return(pstring(m
, line
, pos
, p
, strlen(p
)));
568 mdoc_node_free(struct mdoc_node
*p
)
574 mdoc_argv_free(p
->args
);
580 mdoc_node_unlink(struct mdoc
*m
, struct mdoc_node
*n
)
583 /* Adjust siblings. */
586 n
->prev
->next
= n
->next
;
588 n
->next
->prev
= n
->prev
;
594 if (n
->parent
->child
== n
)
595 n
->parent
->child
= n
->prev
? n
->prev
: n
->next
;
598 /* Adjust parse point, if applicable. */
600 if (m
&& m
->last
== n
) {
603 m
->next
= MDOC_NEXT_SIBLING
;
606 m
->next
= MDOC_NEXT_CHILD
;
610 if (m
&& m
->first
== n
)
616 mdoc_node_delete(struct mdoc
*m
, struct mdoc_node
*p
)
621 mdoc_node_delete(m
, p
->child
);
623 assert(0 == p
->nchild
);
625 mdoc_node_unlink(m
, p
);
631 * Parse free-form text, that is, a line that does not begin with the
635 parsetext(struct mdoc
*m
, int line
, char *buf
)
640 if (SEC_NONE
== m
->lastnamed
)
641 return(mdoc_perr(m
, line
, 0, ETEXTPROL
));
644 * If in literal mode, then pass the buffer directly to the
645 * back-end, as it should be preserved as a single term.
648 if (MDOC_LITERAL
& m
->flags
)
649 return(mdoc_word_alloc(m
, line
, 0, buf
));
651 /* Disallow blank/white-space lines in non-literal mode. */
653 for (i
= 0; ' ' == buf
[i
]; i
++)
654 /* Skip leading whitespace. */ ;
656 if ('\0' == buf
[i
]) {
657 if ( ! mdoc_pwarn(m
, line
, 0, ENOBLANK
))
660 * Assume that a `Pp' should be inserted in the case of
661 * a blank line. Technically, blank lines aren't
662 * allowed, but enough manuals assume this behaviour
663 * that we want to work around it.
665 if ( ! mdoc_elem_alloc(m
, line
, 0, MDOC_Pp
, NULL
))
670 * Break apart a free-form line into tokens. Spaces are
671 * stripped out of the input.
674 for (j
= i
; buf
[i
]; i
++) {
678 /* Escaped whitespace. */
679 if (i
&& ' ' == buf
[i
] && '\\' == buf
[i
- 1])
685 if ( ! pstring(m
, line
, j
, &buf
[j
], (size_t)(i
- j
)))
688 /* Trailing whitespace? Check at overwritten byte. */
690 if (' ' == sv
&& '\0' == buf
[i
])
691 if ( ! mdoc_pwarn(m
, line
, i
- 1, ETAILWS
))
694 for ( ; ' ' == buf
[i
]; i
++)
695 /* Skip trailing whitespace. */ ;
699 /* Trailing whitespace? */
701 if (' ' == buf
[i
- 1] && '\0' == buf
[i
])
702 if ( ! mdoc_pwarn(m
, line
, i
- 1, ETAILWS
))
709 if (j
!= i
&& ! pstring(m
, line
, j
, &buf
[j
], (size_t)(i
- j
)))
712 m
->next
= MDOC_NEXT_SIBLING
;
719 macrowarn(struct mdoc
*m
, int ln
, const char *buf
)
721 if ( ! (MDOC_IGN_MACRO
& m
->pflags
))
722 return(mdoc_verr(m
, ln
, 0,
723 "unknown macro: %s%s",
724 buf
, strlen(buf
) > 3 ? "..." : ""));
725 return(mdoc_vwarn(m
, ln
, 0, "unknown macro: %s%s",
726 buf
, strlen(buf
) > 3 ? "..." : ""));
731 * Parse a macro line, that is, a line beginning with the control
735 parsemacro(struct mdoc
*m
, int ln
, char *buf
)
740 /* Empty lines are ignored. */
747 /* Accept whitespace after the initial control char. */
751 while (buf
[i
] && ' ' == buf
[i
])
757 /* Copy the first word into a nil-terminated buffer. */
759 for (j
= 0; j
< 4; j
++, i
++) {
760 if ('\0' == (mac
[j
] = buf
[i
]))
762 else if (' ' == buf
[i
])
765 /* Check for invalid characters. */
767 if (isgraph((u_char
)buf
[i
]))
769 return(mdoc_perr(m
, ln
, i
, EPRINT
));
774 if (j
== 4 || j
< 2) {
775 if ( ! macrowarn(m
, ln
, mac
))
780 if (MDOC_MAX
== (c
= mdoc_hash_find(mac
))) {
781 if ( ! macrowarn(m
, ln
, mac
))
786 /* The macro is sane. Jump to the next word. */
788 while (buf
[i
] && ' ' == buf
[i
])
791 /* Trailing whitespace? */
793 if ('\0' == buf
[i
] && ' ' == buf
[i
- 1])
794 if ( ! mdoc_pwarn(m
, ln
, i
- 1, ETAILWS
))
798 * Begin recursive parse sequence. Since we're at the start of
799 * the line, we don't need to do callable/parseable checks.
801 if ( ! mdoc_macro(m
, c
, ln
, 1, &i
, buf
))
806 err
: /* Error out. */
808 m
->flags
|= MDOC_HALT
;