xref: /openbsd-src/usr.bin/mandoc/mandoc.h (revision 4c1e55dc91edd6e69ccc60ce855900fbc12cf34f)
1 /*	$Id: mandoc.h,v 1.47 2012/07/12 15:09:50 schwarze Exp $ */
2 /*
3  * Copyright (c) 2010, 2011 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 #ifndef MANDOC_H
18 #define MANDOC_H
19 
20 #define ASCII_NBRSP	 31  /* non-breaking space */
21 #define	ASCII_HYPH	 30  /* breakable hyphen */
22 
23 /*
24  * Status level.  This refers to both internal status (i.e., whilst
25  * running, when warnings/errors are reported) and an indicator of a
26  * threshold of when to halt (when said internal state exceeds the
27  * threshold).
28  */
29 enum	mandoclevel {
30 	MANDOCLEVEL_OK = 0,
31 	MANDOCLEVEL_RESERVED,
32 	MANDOCLEVEL_WARNING, /* warnings: syntax, whitespace, etc. */
33 	MANDOCLEVEL_ERROR, /* input has been thrown away */
34 	MANDOCLEVEL_FATAL, /* input is borked */
35 	MANDOCLEVEL_BADARG, /* bad argument in invocation */
36 	MANDOCLEVEL_SYSERR, /* system error */
37 	MANDOCLEVEL_MAX
38 };
39 
40 /*
41  * All possible things that can go wrong within a parse, be it libroff,
42  * libmdoc, or libman.
43  */
44 enum	mandocerr {
45 	MANDOCERR_OK,
46 
47 	MANDOCERR_WARNING, /* ===== start of warnings ===== */
48 
49 	/* related to the prologue */
50 	MANDOCERR_NOTITLE, /* no title in document */
51 	MANDOCERR_UPPERCASE, /* document title should be all caps */
52 	MANDOCERR_BADMSEC, /* unknown manual section */
53 	MANDOCERR_NODATE, /* date missing, using today's date */
54 	MANDOCERR_BADDATE, /* cannot parse date, using it verbatim */
55 	MANDOCERR_PROLOGOOO, /* prologue macros out of order */
56 	MANDOCERR_PROLOGREP, /* duplicate prologue macro */
57 	MANDOCERR_BADPROLOG, /* macro not allowed in prologue */
58 	MANDOCERR_BADBODY, /* macro not allowed in body */
59 
60 	/* related to document structure */
61 	MANDOCERR_SO, /* .so is fragile, better use ln(1) */
62 	MANDOCERR_NAMESECFIRST, /* NAME section must come first */
63 	MANDOCERR_BADNAMESEC, /* bad NAME section contents */
64 	MANDOCERR_SECOOO, /* sections out of conventional order */
65 	MANDOCERR_SECREP, /* duplicate section name */
66 	MANDOCERR_SECMSEC, /* section not in conventional manual section */
67 
68 	/* related to macros and nesting */
69 	MANDOCERR_MACROOBS, /* skipping obsolete macro */
70 	MANDOCERR_IGNPAR, /* skipping paragraph macro */
71 	MANDOCERR_IGNNS, /* skipping no-space macro */
72 	MANDOCERR_SCOPENEST, /* blocks badly nested */
73 	MANDOCERR_CHILD, /* child violates parent syntax */
74 	MANDOCERR_NESTEDDISP, /* nested displays are not portable */
75 	MANDOCERR_SCOPEREP, /* already in literal mode */
76 	MANDOCERR_LINESCOPE, /* line scope broken */
77 
78 	/* related to missing macro arguments */
79 	MANDOCERR_MACROEMPTY, /* skipping empty macro */
80 	MANDOCERR_ARGCWARN, /* argument count wrong */
81 	MANDOCERR_DISPTYPE, /* missing display type */
82 	MANDOCERR_LISTFIRST, /* list type must come first */
83 	MANDOCERR_NOWIDTHARG, /* tag lists require a width argument */
84 	MANDOCERR_FONTTYPE, /* missing font type */
85 	MANDOCERR_WNOSCOPE, /* skipping end of block that is not open */
86 
87 	/* related to bad macro arguments */
88 	MANDOCERR_IGNARGV, /* skipping argument */
89 	MANDOCERR_ARGVREP, /* duplicate argument */
90 	MANDOCERR_DISPREP, /* duplicate display type */
91 	MANDOCERR_LISTREP, /* duplicate list type */
92 	MANDOCERR_BADATT, /* unknown AT&T UNIX version */
93 	MANDOCERR_BADBOOL, /* bad Boolean value */
94 	MANDOCERR_BADFONT, /* unknown font */
95 	MANDOCERR_BADSTANDARD, /* unknown standard specifier */
96 	MANDOCERR_BADWIDTH, /* bad width argument */
97 
98 	/* related to plain text */
99 	MANDOCERR_NOBLANKLN, /* blank line in non-literal context */
100 	MANDOCERR_BADTAB, /* tab in non-literal context */
101 	MANDOCERR_EOLNSPACE, /* end of line whitespace */
102 	MANDOCERR_BADCOMMENT, /* bad comment style */
103 	MANDOCERR_BADESCAPE, /* unknown escape sequence */
104 	MANDOCERR_BADQUOTE, /* unterminated quoted string */
105 
106 	/* related to equations */
107 	MANDOCERR_EQNQUOTE, /* unexpected literal in equation */
108 
109 	MANDOCERR_ERROR, /* ===== start of errors ===== */
110 
111 	/* related to equations */
112 	MANDOCERR_EQNNSCOPE, /* unexpected equation scope closure*/
113 	MANDOCERR_EQNSCOPE, /* equation scope open on exit */
114 	MANDOCERR_EQNBADSCOPE, /* overlapping equation scopes */
115 	MANDOCERR_EQNEOF, /* unexpected end of equation */
116 	MANDOCERR_EQNSYNT, /* equation syntax error */
117 
118 	/* related to tables */
119 	MANDOCERR_TBL, /* bad table syntax */
120 	MANDOCERR_TBLOPT, /* bad table option */
121 	MANDOCERR_TBLLAYOUT, /* bad table layout */
122 	MANDOCERR_TBLNOLAYOUT, /* no table layout cells specified */
123 	MANDOCERR_TBLNODATA, /* no table data cells specified */
124 	MANDOCERR_TBLIGNDATA, /* ignore data in cell */
125 	MANDOCERR_TBLBLOCK, /* data block still open */
126 	MANDOCERR_TBLEXTRADAT, /* ignoring extra data cells */
127 
128 	MANDOCERR_ROFFLOOP, /* input stack limit exceeded, infinite loop? */
129 	MANDOCERR_BADCHAR, /* skipping bad character */
130 	MANDOCERR_NAMESC, /* escaped character not allowed in a name */
131 	MANDOCERR_NONAME, /* manual name not yet set */
132 	MANDOCERR_NOTEXT, /* skipping text before the first section header */
133 	MANDOCERR_MACRO, /* skipping unknown macro */
134 	MANDOCERR_REQUEST, /* NOT IMPLEMENTED: skipping request */
135 	MANDOCERR_ARGCOUNT, /* argument count wrong */
136 	MANDOCERR_NOSCOPE, /* skipping end of block that is not open */
137 	MANDOCERR_SCOPEBROKEN, /* missing end of block */
138 	MANDOCERR_SCOPEEXIT, /* scope open on exit */
139 	MANDOCERR_UNAME, /* uname(3) system call failed */
140 	/* FIXME: merge following with MANDOCERR_ARGCOUNT */
141 	MANDOCERR_NOARGS, /* macro requires line argument(s) */
142 	MANDOCERR_NOBODY, /* macro requires body argument(s) */
143 	MANDOCERR_NOARGV, /* macro requires argument(s) */
144 	MANDOCERR_LISTTYPE, /* missing list type */
145 	MANDOCERR_ARGSLOST, /* line argument(s) will be lost */
146 	MANDOCERR_BODYLOST, /* body argument(s) will be lost */
147 
148 	MANDOCERR_FATAL, /* ===== start of fatal errors ===== */
149 
150 	MANDOCERR_NOTMANUAL, /* manual isn't really a manual */
151 	MANDOCERR_COLUMNS, /* column syntax is inconsistent */
152 	MANDOCERR_BADDISP, /* NOT IMPLEMENTED: .Bd -file */
153 	MANDOCERR_SYNTARGVCOUNT, /* argument count wrong, violates syntax */
154 	MANDOCERR_SYNTCHILD, /* child violates parent syntax */
155 	MANDOCERR_SYNTARGCOUNT, /* argument count wrong, violates syntax */
156 	MANDOCERR_SOPATH, /* NOT IMPLEMENTED: .so with absolute path or ".." */
157 	MANDOCERR_NODOCBODY, /* no document body */
158 	MANDOCERR_NODOCPROLOG, /* no document prologue */
159 	MANDOCERR_MEM, /* static buffer exhausted */
160 	MANDOCERR_MAX
161 };
162 
163 struct	tbl {
164 	char		  tab; /* cell-separator */
165 	char		  decimal; /* decimal point */
166 	int		  linesize;
167 	int		  opts;
168 #define	TBL_OPT_CENTRE	 (1 << 0)
169 #define	TBL_OPT_EXPAND	 (1 << 1)
170 #define	TBL_OPT_BOX	 (1 << 2)
171 #define	TBL_OPT_DBOX	 (1 << 3)
172 #define	TBL_OPT_ALLBOX	 (1 << 4)
173 #define	TBL_OPT_NOKEEP	 (1 << 5)
174 #define	TBL_OPT_NOSPACE	 (1 << 6)
175 	int		  cols; /* number of columns */
176 };
177 
178 /*
179  * The head of a table specifies all of its columns.  When formatting a
180  * tbl_span, iterate over these and plug in data from the tbl_span when
181  * appropriate, using tbl_cell as a guide to placement.
182  */
183 struct	tbl_head {
184 	int		  ident; /* 0 <= unique id < cols */
185 	int		  vert; /* width of preceding vertical line */
186 	struct tbl_head	 *next;
187 	struct tbl_head	 *prev;
188 };
189 
190 enum	tbl_cellt {
191 	TBL_CELL_CENTRE, /* c, C */
192 	TBL_CELL_RIGHT, /* r, R */
193 	TBL_CELL_LEFT, /* l, L */
194 	TBL_CELL_NUMBER, /* n, N */
195 	TBL_CELL_SPAN, /* s, S */
196 	TBL_CELL_LONG, /* a, A */
197 	TBL_CELL_DOWN, /* ^ */
198 	TBL_CELL_HORIZ, /* _, - */
199 	TBL_CELL_DHORIZ, /* = */
200 	TBL_CELL_MAX
201 };
202 
203 /*
204  * A cell in a layout row.
205  */
206 struct	tbl_cell {
207 	struct tbl_cell	 *next;
208 	int		  vert; /* width of preceding vertical line */
209 	enum tbl_cellt	  pos;
210 	size_t		  spacing;
211 	int		  flags;
212 #define	TBL_CELL_TALIGN	 (1 << 0) /* t, T */
213 #define	TBL_CELL_BALIGN	 (1 << 1) /* d, D */
214 #define	TBL_CELL_BOLD	 (1 << 2) /* fB, B, b */
215 #define	TBL_CELL_ITALIC	 (1 << 3) /* fI, I, i */
216 #define	TBL_CELL_EQUAL	 (1 << 4) /* e, E */
217 #define	TBL_CELL_UP	 (1 << 5) /* u, U */
218 #define	TBL_CELL_WIGN	 (1 << 6) /* z, Z */
219 	struct tbl_head	 *head;
220 };
221 
222 /*
223  * A layout row.
224  */
225 struct	tbl_row {
226 	struct tbl_row	 *next;
227 	struct tbl_cell	 *first;
228 	struct tbl_cell	 *last;
229 };
230 
231 enum	tbl_datt {
232 	TBL_DATA_NONE, /* has no data */
233 	TBL_DATA_DATA, /* consists of data/string */
234 	TBL_DATA_HORIZ, /* horizontal line */
235 	TBL_DATA_DHORIZ, /* double-horizontal line */
236 	TBL_DATA_NHORIZ, /* squeezed horizontal line */
237 	TBL_DATA_NDHORIZ /* squeezed double-horizontal line */
238 };
239 
240 /*
241  * A cell within a row of data.  The "string" field contains the actual
242  * string value that's in the cell.  The rest is layout.
243  */
244 struct	tbl_dat {
245 	struct tbl_cell	 *layout; /* layout cell */
246 	int		  spans; /* how many spans follow */
247 	struct tbl_dat	 *next;
248 	char		 *string; /* data (NULL if not TBL_DATA_DATA) */
249 	enum tbl_datt	  pos;
250 };
251 
252 enum	tbl_spant {
253 	TBL_SPAN_DATA, /* span consists of data */
254 	TBL_SPAN_HORIZ, /* span is horizontal line */
255 	TBL_SPAN_DHORIZ /* span is double horizontal line */
256 };
257 
258 /*
259  * A row of data in a table.
260  */
261 struct	tbl_span {
262 	struct tbl	 *tbl;
263 	struct tbl_head	 *head;
264 	struct tbl_row	 *layout; /* layout row */
265 	struct tbl_dat	 *first;
266 	struct tbl_dat	 *last;
267 	int		  line; /* parse line */
268 	int		  flags;
269 #define	TBL_SPAN_FIRST	 (1 << 0)
270 #define	TBL_SPAN_LAST	 (1 << 1)
271 	enum tbl_spant	  pos;
272 	struct tbl_span	 *next;
273 };
274 
275 enum	eqn_boxt {
276 	EQN_ROOT, /* root of parse tree */
277 	EQN_TEXT, /* text (number, variable, whatever) */
278 	EQN_SUBEXPR, /* nested `eqn' subexpression */
279 	EQN_LIST, /* subexpressions list */
280 	EQN_MATRIX /* matrix subexpression */
281 };
282 
283 enum	eqn_markt {
284 	EQNMARK_NONE = 0,
285 	EQNMARK_DOT,
286 	EQNMARK_DOTDOT,
287 	EQNMARK_HAT,
288 	EQNMARK_TILDE,
289 	EQNMARK_VEC,
290 	EQNMARK_DYAD,
291 	EQNMARK_BAR,
292 	EQNMARK_UNDER,
293 	EQNMARK__MAX
294 };
295 
296 enum	eqn_fontt {
297 	EQNFONT_NONE = 0,
298 	EQNFONT_ROMAN,
299 	EQNFONT_BOLD,
300 	EQNFONT_FAT,
301 	EQNFONT_ITALIC,
302 	EQNFONT__MAX
303 };
304 
305 enum	eqn_post {
306 	EQNPOS_NONE = 0,
307 	EQNPOS_OVER,
308 	EQNPOS_SUP,
309 	EQNPOS_SUB,
310 	EQNPOS_TO,
311 	EQNPOS_FROM,
312 	EQNPOS__MAX
313 };
314 
315 enum	eqn_pilet {
316 	EQNPILE_NONE = 0,
317 	EQNPILE_PILE,
318 	EQNPILE_CPILE,
319 	EQNPILE_RPILE,
320 	EQNPILE_LPILE,
321 	EQNPILE_COL,
322 	EQNPILE_CCOL,
323 	EQNPILE_RCOL,
324 	EQNPILE_LCOL,
325 	EQNPILE__MAX
326 };
327 
328  /*
329  * A "box" is a parsed mathematical expression as defined by the eqn.7
330  * grammar.
331  */
332 struct	eqn_box {
333 	int		  size; /* font size of expression */
334 #define	EQN_DEFSIZE	  INT_MIN
335 	enum eqn_boxt	  type; /* type of node */
336 	struct eqn_box	 *first; /* first child node */
337 	struct eqn_box	 *last; /* last child node */
338 	struct eqn_box	 *next; /* node sibling */
339 	struct eqn_box	 *parent; /* node sibling */
340 	char		 *text; /* text (or NULL) */
341 	char		 *left;
342 	char		 *right;
343 	enum eqn_post	  pos; /* position of next box */
344 	enum eqn_markt	  mark; /* a mark about the box */
345 	enum eqn_fontt	  font; /* font of box */
346 	enum eqn_pilet	  pile; /* equation piling */
347 };
348 
349 /*
350  * An equation consists of a tree of expressions starting at a given
351  * line and position.
352  */
353 struct	eqn {
354 	char		 *name; /* identifier (or NULL) */
355 	struct eqn_box	 *root; /* root mathematical expression */
356 	int		  ln; /* invocation line */
357 	int		  pos; /* invocation position */
358 };
359 
360 /*
361  * The type of parse sequence.  This value is usually passed via the
362  * mandoc(1) command line of -man and -mdoc.  It's almost exclusively
363  * -mandoc but the others have been retained for compatibility.
364  */
365 enum	mparset {
366 	MPARSE_AUTO, /* magically determine the document type */
367 	MPARSE_MDOC, /* assume -mdoc */
368 	MPARSE_MAN /* assume -man */
369 };
370 
371 enum	mandoc_esc {
372 	ESCAPE_ERROR = 0, /* bail! unparsable escape */
373 	ESCAPE_IGNORE, /* escape to be ignored */
374 	ESCAPE_SPECIAL, /* a regular special character */
375 	ESCAPE_FONT, /* a generic font mode */
376 	ESCAPE_FONTBOLD, /* bold font mode */
377 	ESCAPE_FONTITALIC, /* italic font mode */
378 	ESCAPE_FONTROMAN, /* roman font mode */
379 	ESCAPE_FONTPREV, /* previous font mode */
380 	ESCAPE_NUMBERED, /* a numbered glyph */
381 	ESCAPE_UNICODE, /* a unicode codepoint */
382 	ESCAPE_NOSPACE, /* suppress space if the last on a line */
383 	ESCAPE_SKIPCHAR /* skip the next character */
384 };
385 
386 typedef	void	(*mandocmsg)(enum mandocerr, enum mandoclevel,
387 			const char *, int, int, const char *);
388 
389 struct	mparse;
390 struct	mchars;
391 struct	mdoc;
392 struct	man;
393 
394 __BEGIN_DECLS
395 
396 void		 *mandoc_calloc(size_t, size_t);
397 enum mandoc_esc	  mandoc_escape(const char **, const char **, int *);
398 void		 *mandoc_malloc(size_t);
399 void		 *mandoc_realloc(void *, size_t);
400 char		 *mandoc_strdup(const char *);
401 char		 *mandoc_strndup(const char *, size_t);
402 struct mchars	 *mchars_alloc(void);
403 void		  mchars_free(struct mchars *);
404 char	 	  mchars_num2char(const char *, size_t);
405 int		  mchars_num2uc(const char *, size_t);
406 int		  mchars_spec2cp(const struct mchars *,
407 			const char *, size_t);
408 const char	 *mchars_spec2str(const struct mchars *,
409 			const char *, size_t, size_t *);
410 struct mparse	 *mparse_alloc(enum mparset, enum mandoclevel,
411 			mandocmsg, void *, char *);
412 void		  mparse_free(struct mparse *);
413 void		  mparse_keep(struct mparse *);
414 enum mandoclevel  mparse_readfd(struct mparse *, int, const char *);
415 void		  mparse_reset(struct mparse *);
416 void		  mparse_result(struct mparse *,
417 			struct mdoc **, struct man **);
418 const char	 *mparse_getkeep(const struct mparse *);
419 const char	 *mparse_strerror(enum mandocerr);
420 const char	 *mparse_strlevel(enum mandoclevel);
421 
422 __END_DECLS
423 
424 #endif /*!MANDOC_H*/
425