1 /* $Id: mandoc.h,v 1.253 2018/08/25 16:53:39 schwarze Exp $ */
3 * Copyright (c) 2010, 2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010-2018 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 AUTHORS DISCLAIM ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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.
19 #define ASCII_NBRSP 31 /* non-breaking space */
20 #define ASCII_HYPH 30 /* breakable hyphen */
21 #define ASCII_BREAK 29 /* breakable zero-width space */
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
31 MANDOCLEVEL_STYLE
, /* style suggestions */
32 MANDOCLEVEL_WARNING
, /* warnings: syntax, whitespace, etc. */
33 MANDOCLEVEL_ERROR
, /* input has been thrown away */
34 MANDOCLEVEL_UNSUPP
, /* input needs unimplemented features */
35 MANDOCLEVEL_BADARG
, /* bad argument in invocation */
36 MANDOCLEVEL_SYSERR
, /* system error */
41 * All possible things that can go wrong within a parse, be it libroff,
47 MANDOCERR_BASE
, /* ===== start of base system conventions ===== */
49 MANDOCERR_MDOCDATE
, /* Mdocdate found: Dd ... */
50 MANDOCERR_MDOCDATE_MISSING
, /* Mdocdate missing: Dd ... */
51 MANDOCERR_ARCH_BAD
, /* unknown architecture: Dt ... arch */
52 MANDOCERR_OS_ARG
, /* operating system explicitly specified: Os ... */
53 MANDOCERR_RCS_MISSING
, /* RCS id missing */
54 MANDOCERR_XR_BAD
, /* referenced manual not found: Xr name sec */
56 MANDOCERR_STYLE
, /* ===== start of style suggestions ===== */
58 MANDOCERR_DATE_LEGACY
, /* legacy man(7) date format: Dd ... */
59 MANDOCERR_DATE_NORM
, /* normalizing date format to: ... */
60 MANDOCERR_TITLE_CASE
, /* lower case character in document title */
61 MANDOCERR_RCS_REP
, /* duplicate RCS id: ... */
62 MANDOCERR_SEC_TYPO
, /* possible typo in section name: Sh ... */
63 MANDOCERR_ARG_QUOTE
, /* unterminated quoted argument */
64 MANDOCERR_MACRO_USELESS
, /* useless macro: macro */
65 MANDOCERR_BX
, /* consider using OS macro: macro */
66 MANDOCERR_ER_ORDER
, /* errnos out of order: Er ... */
67 MANDOCERR_ER_REP
, /* duplicate errno: Er ... */
68 MANDOCERR_DELIM
, /* trailing delimiter: macro ... */
69 MANDOCERR_DELIM_NB
, /* no blank before trailing delimiter: macro ... */
70 MANDOCERR_FI_SKIP
, /* fill mode already enabled, skipping: fi */
71 MANDOCERR_NF_SKIP
, /* fill mode already disabled, skipping: nf */
72 MANDOCERR_DASHDASH
, /* verbatim "--", maybe consider using \(em */
73 MANDOCERR_FUNC
, /* function name without markup: name() */
74 MANDOCERR_SPACE_EOL
, /* whitespace at end of input line */
75 MANDOCERR_COMMENT_BAD
, /* bad comment style */
77 MANDOCERR_WARNING
, /* ===== start of warnings ===== */
79 /* related to the prologue */
80 MANDOCERR_DT_NOTITLE
, /* missing manual title, using UNTITLED: line */
81 MANDOCERR_TH_NOTITLE
, /* missing manual title, using "": [macro] */
82 MANDOCERR_MSEC_MISSING
, /* missing manual section, using "": macro */
83 MANDOCERR_MSEC_BAD
, /* unknown manual section: Dt ... section */
84 MANDOCERR_DATE_MISSING
, /* missing date, using today's date */
85 MANDOCERR_DATE_BAD
, /* cannot parse date, using it verbatim: date */
86 MANDOCERR_DATE_FUTURE
, /* date in the future, using it anyway: date */
87 MANDOCERR_OS_MISSING
, /* missing Os macro, using "" */
88 MANDOCERR_PROLOG_LATE
, /* late prologue macro: macro */
89 MANDOCERR_PROLOG_ORDER
, /* prologue macros out of order: macros */
91 /* related to document structure */
92 MANDOCERR_SO
, /* .so is fragile, better use ln(1): so path */
93 MANDOCERR_DOC_EMPTY
, /* no document body */
94 MANDOCERR_SEC_BEFORE
, /* content before first section header: macro */
95 MANDOCERR_NAMESEC_FIRST
, /* first section is not NAME: Sh title */
96 MANDOCERR_NAMESEC_NONM
, /* NAME section without Nm before Nd */
97 MANDOCERR_NAMESEC_NOND
, /* NAME section without description */
98 MANDOCERR_NAMESEC_ND
, /* description not at the end of NAME */
99 MANDOCERR_NAMESEC_BAD
, /* bad NAME section content: macro */
100 MANDOCERR_NAMESEC_PUNCT
, /* missing comma before name: Nm name */
101 MANDOCERR_ND_EMPTY
, /* missing description line, using "" */
102 MANDOCERR_ND_LATE
, /* description line outside NAME section */
103 MANDOCERR_SEC_ORDER
, /* sections out of conventional order: Sh title */
104 MANDOCERR_SEC_REP
, /* duplicate section title: Sh title */
105 MANDOCERR_SEC_MSEC
, /* unexpected section: Sh title for ... only */
106 MANDOCERR_XR_SELF
, /* cross reference to self: Xr name sec */
107 MANDOCERR_XR_ORDER
, /* unusual Xr order: ... after ... */
108 MANDOCERR_XR_PUNCT
, /* unusual Xr punctuation: ... after ... */
109 MANDOCERR_AN_MISSING
, /* AUTHORS section without An macro */
111 /* related to macros and nesting */
112 MANDOCERR_MACRO_OBS
, /* obsolete macro: macro */
113 MANDOCERR_MACRO_CALL
, /* macro neither callable nor escaped: macro */
114 MANDOCERR_PAR_SKIP
, /* skipping paragraph macro: macro ... */
115 MANDOCERR_PAR_MOVE
, /* moving paragraph macro out of list: macro */
116 MANDOCERR_NS_SKIP
, /* skipping no-space macro */
117 MANDOCERR_BLK_NEST
, /* blocks badly nested: macro ... */
118 MANDOCERR_BD_NEST
, /* nested displays are not portable: macro ... */
119 MANDOCERR_BL_MOVE
, /* moving content out of list: macro */
120 MANDOCERR_TA_LINE
, /* first macro on line: Ta */
121 MANDOCERR_BLK_LINE
, /* line scope broken: macro breaks macro */
122 MANDOCERR_BLK_BLANK
, /* skipping blank line in line scope */
124 /* related to missing arguments */
125 MANDOCERR_REQ_EMPTY
, /* skipping empty request: request */
126 MANDOCERR_COND_EMPTY
, /* conditional request controls empty scope */
127 MANDOCERR_MACRO_EMPTY
, /* skipping empty macro: macro */
128 MANDOCERR_BLK_EMPTY
, /* empty block: macro */
129 MANDOCERR_ARG_EMPTY
, /* empty argument, using 0n: macro arg */
130 MANDOCERR_BD_NOTYPE
, /* missing display type, using -ragged: Bd */
131 MANDOCERR_BL_LATETYPE
, /* list type is not the first argument: Bl arg */
132 MANDOCERR_BL_NOWIDTH
, /* missing -width in -tag list, using 6n */
133 MANDOCERR_EX_NONAME
, /* missing utility name, using "": Ex */
134 MANDOCERR_FO_NOHEAD
, /* missing function name, using "": Fo */
135 MANDOCERR_IT_NOHEAD
, /* empty head in list item: Bl -type It */
136 MANDOCERR_IT_NOBODY
, /* empty list item: Bl -type It */
137 MANDOCERR_IT_NOARG
, /* missing argument, using next line: Bl -c It */
138 MANDOCERR_BF_NOFONT
, /* missing font type, using \fR: Bf */
139 MANDOCERR_BF_BADFONT
, /* unknown font type, using \fR: Bf font */
140 MANDOCERR_PF_SKIP
, /* nothing follows prefix: Pf arg */
141 MANDOCERR_RS_EMPTY
, /* empty reference block: Rs */
142 MANDOCERR_XR_NOSEC
, /* missing section argument: Xr arg */
143 MANDOCERR_ARG_STD
, /* missing -std argument, adding it: macro */
144 MANDOCERR_OP_EMPTY
, /* missing option string, using "": OP */
145 MANDOCERR_UR_NOHEAD
, /* missing resource identifier, using "": UR */
146 MANDOCERR_EQN_NOBOX
, /* missing eqn box, using "": op */
148 /* related to bad arguments */
149 MANDOCERR_ARG_REP
, /* duplicate argument: macro arg */
150 MANDOCERR_AN_REP
, /* skipping duplicate argument: An -arg */
151 MANDOCERR_BD_REP
, /* skipping duplicate display type: Bd -type */
152 MANDOCERR_BL_REP
, /* skipping duplicate list type: Bl -type */
153 MANDOCERR_BL_SKIPW
, /* skipping -width argument: Bl -type */
154 MANDOCERR_BL_COL
, /* wrong number of cells */
155 MANDOCERR_AT_BAD
, /* unknown AT&T UNIX version: At version */
156 MANDOCERR_FA_COMMA
, /* comma in function argument: arg */
157 MANDOCERR_FN_PAREN
, /* parenthesis in function name: arg */
158 MANDOCERR_LB_BAD
, /* unknown library name: Lb ... */
159 MANDOCERR_RS_BAD
, /* invalid content in Rs block: macro */
160 MANDOCERR_SM_BAD
, /* invalid Boolean argument: macro arg */
161 MANDOCERR_CHAR_FONT
, /* argument contains two font escapes */
162 MANDOCERR_FT_BAD
, /* unknown font, skipping request: ft font */
163 MANDOCERR_TR_ODD
, /* odd number of characters in request: tr char */
165 /* related to plain text */
166 MANDOCERR_FI_BLANK
, /* blank line in fill mode, using .sp */
167 MANDOCERR_FI_TAB
, /* tab in filled text */
168 MANDOCERR_EOS
, /* new sentence, new line */
169 MANDOCERR_ESC_BAD
, /* invalid escape sequence: esc */
170 MANDOCERR_STR_UNDEF
, /* undefined string, using "": name */
172 /* related to tables */
173 MANDOCERR_TBLLAYOUT_SPAN
, /* tbl line starts with span */
174 MANDOCERR_TBLLAYOUT_DOWN
, /* tbl column starts with span */
175 MANDOCERR_TBLLAYOUT_VERT
, /* skipping vertical bar in tbl layout */
177 MANDOCERR_ERROR
, /* ===== start of errors ===== */
179 /* related to tables */
180 MANDOCERR_TBLOPT_ALPHA
, /* non-alphabetic character in tbl options */
181 MANDOCERR_TBLOPT_BAD
, /* skipping unknown tbl option: option */
182 MANDOCERR_TBLOPT_NOARG
, /* missing tbl option argument: option */
183 MANDOCERR_TBLOPT_ARGSZ
, /* wrong tbl option argument size: option */
184 MANDOCERR_TBLLAYOUT_NONE
, /* empty tbl layout */
185 MANDOCERR_TBLLAYOUT_CHAR
, /* invalid character in tbl layout: char */
186 MANDOCERR_TBLLAYOUT_PAR
, /* unmatched parenthesis in tbl layout */
187 MANDOCERR_TBLDATA_NONE
, /* tbl without any data cells */
188 MANDOCERR_TBLDATA_SPAN
, /* ignoring data in spanned tbl cell: data */
189 MANDOCERR_TBLDATA_EXTRA
, /* ignoring extra tbl data cells: data */
190 MANDOCERR_TBLDATA_BLK
, /* data block open at end of tbl: macro */
192 /* related to document structure and macros */
193 MANDOCERR_FILE
, /* cannot open file */
194 MANDOCERR_PROLOG_REP
, /* duplicate prologue macro: macro */
195 MANDOCERR_DT_LATE
, /* skipping late title macro: Dt args */
196 MANDOCERR_ROFFLOOP
, /* input stack limit exceeded, infinite loop? */
197 MANDOCERR_CHAR_BAD
, /* skipping bad character: number */
198 MANDOCERR_MACRO
, /* skipping unknown macro: macro */
199 MANDOCERR_REQ_NOMAC
, /* skipping request outside macro: ... */
200 MANDOCERR_REQ_INSEC
, /* skipping insecure request: request */
201 MANDOCERR_IT_STRAY
, /* skipping item outside list: It ... */
202 MANDOCERR_TA_STRAY
, /* skipping column outside column list: Ta */
203 MANDOCERR_BLK_NOTOPEN
, /* skipping end of block that is not open */
204 MANDOCERR_RE_NOTOPEN
, /* fewer RS blocks open, skipping: RE arg */
205 MANDOCERR_BLK_BROKEN
, /* inserting missing end of block: macro ... */
206 MANDOCERR_BLK_NOEND
, /* appending missing end of block: macro */
208 /* related to request and macro arguments */
209 MANDOCERR_NAMESC
, /* escaped character not allowed in a name: name */
210 MANDOCERR_ARG_UNDEF
, /* using macro argument outside macro */
211 MANDOCERR_ARG_NONUM
, /* argument number is not numeric */
212 MANDOCERR_BD_FILE
, /* NOT IMPLEMENTED: Bd -file */
213 MANDOCERR_BD_NOARG
, /* skipping display without arguments: Bd */
214 MANDOCERR_BL_NOTYPE
, /* missing list type, using -item: Bl */
215 MANDOCERR_CE_NONUM
, /* argument is not numeric, using 1: ce ... */
216 MANDOCERR_CHAR_ARG
, /* argument is not a character: char ... */
217 MANDOCERR_NM_NONAME
, /* missing manual name, using "": Nm */
218 MANDOCERR_OS_UNAME
, /* uname(3) system call failed, using UNKNOWN */
219 MANDOCERR_ST_BAD
, /* unknown standard specifier: St standard */
220 MANDOCERR_IT_NONUM
, /* skipping request without numeric argument */
221 MANDOCERR_SHIFT
, /* excessive shift: ..., but max is ... */
222 MANDOCERR_SO_PATH
, /* NOT IMPLEMENTED: .so with absolute path or ".." */
223 MANDOCERR_SO_FAIL
, /* .so request failed */
224 MANDOCERR_ARG_SKIP
, /* skipping all arguments: macro args */
225 MANDOCERR_ARG_EXCESS
, /* skipping excess arguments: macro ... args */
226 MANDOCERR_DIVZERO
, /* divide by zero */
228 MANDOCERR_UNSUPP
, /* ===== start of unsupported features ===== */
230 MANDOCERR_TOOLARGE
, /* input too large */
231 MANDOCERR_CHAR_UNSUPP
, /* unsupported control character: number */
232 MANDOCERR_REQ_UNSUPP
, /* unsupported roff request: request */
233 MANDOCERR_WHILE_NEST
, /* nested .while loops */
234 MANDOCERR_WHILE_OUTOF
, /* end of scope with open .while loop */
235 MANDOCERR_WHILE_INTO
, /* end of .while loop in inner scope */
236 MANDOCERR_WHILE_FAIL
, /* cannot continue this .while loop */
237 MANDOCERR_TBLOPT_EQN
, /* eqn delim option in tbl: arg */
238 MANDOCERR_TBLLAYOUT_MOD
, /* unsupported tbl layout modifier: m */
239 MANDOCERR_TBLMACRO
, /* ignoring macro in table: macro */
245 char tab
; /* cell-separator */
246 char decimal
; /* decimal point */
248 #define TBL_OPT_CENTRE (1 << 0)
249 #define TBL_OPT_EXPAND (1 << 1)
250 #define TBL_OPT_BOX (1 << 2)
251 #define TBL_OPT_DBOX (1 << 3)
252 #define TBL_OPT_ALLBOX (1 << 4)
253 #define TBL_OPT_NOKEEP (1 << 5)
254 #define TBL_OPT_NOSPACE (1 << 6)
255 #define TBL_OPT_NOWARN (1 << 7)
256 int cols
; /* number of columns */
257 int lvert
; /* width of left vertical line */
258 int rvert
; /* width of right vertical line */
262 TBL_CELL_CENTRE
, /* c, C */
263 TBL_CELL_RIGHT
, /* r, R */
264 TBL_CELL_LEFT
, /* l, L */
265 TBL_CELL_NUMBER
, /* n, N */
266 TBL_CELL_SPAN
, /* s, S */
267 TBL_CELL_LONG
, /* a, A */
268 TBL_CELL_DOWN
, /* ^ */
269 TBL_CELL_HORIZ
, /* _, - */
270 TBL_CELL_DHORIZ
, /* = */
275 * A cell in a layout row.
278 struct tbl_cell
*next
;
279 char *wstr
; /* min width represented as a string */
280 size_t width
; /* minimum column width */
281 size_t spacing
; /* to the right of the column */
282 int vert
; /* width of subsequent vertical line */
283 int col
; /* column number, starting from 0 */
285 #define TBL_CELL_TALIGN (1 << 0) /* t, T */
286 #define TBL_CELL_BALIGN (1 << 1) /* d, D */
287 #define TBL_CELL_BOLD (1 << 2) /* fB, B, b */
288 #define TBL_CELL_ITALIC (1 << 3) /* fI, I, i */
289 #define TBL_CELL_EQUAL (1 << 4) /* e, E */
290 #define TBL_CELL_UP (1 << 5) /* u, U */
291 #define TBL_CELL_WIGN (1 << 6) /* z, Z */
292 #define TBL_CELL_WMAX (1 << 7) /* x, X */
300 struct tbl_row
*next
;
301 struct tbl_cell
*first
;
302 struct tbl_cell
*last
;
303 int vert
; /* width of left vertical line */
307 TBL_DATA_NONE
, /* has no data */
308 TBL_DATA_DATA
, /* consists of data/string */
309 TBL_DATA_HORIZ
, /* horizontal line */
310 TBL_DATA_DHORIZ
, /* double-horizontal line */
311 TBL_DATA_NHORIZ
, /* squeezed horizontal line */
312 TBL_DATA_NDHORIZ
/* squeezed double-horizontal line */
316 * A cell within a row of data. The "string" field contains the actual
317 * string value that's in the cell. The rest is layout.
320 struct tbl_cell
*layout
; /* layout cell */
321 struct tbl_dat
*next
;
322 char *string
; /* data (NULL if not TBL_DATA_DATA) */
323 int spans
; /* how many spans follow */
324 int block
; /* T{ text block T} */
329 TBL_SPAN_DATA
, /* span consists of data */
330 TBL_SPAN_HORIZ
, /* span is horizontal line */
331 TBL_SPAN_DHORIZ
/* span is double horizontal line */
335 * A row of data in a table.
338 struct tbl_opts
*opts
;
339 struct tbl_row
*layout
; /* layout row */
340 struct tbl_dat
*first
;
341 struct tbl_dat
*last
;
342 struct tbl_span
*prev
;
343 struct tbl_span
*next
;
344 int line
; /* parse line */
349 EQN_TEXT
, /* text (number, variable, whatever) */
350 EQN_SUBEXPR
, /* nested `eqn' subexpression */
351 EQN_LIST
, /* list (braces, etc.) */
352 EQN_PILE
, /* vertical pile */
353 EQN_MATRIX
/* pile of piles */
392 * A "box" is a parsed mathematical expression as defined by the eqn.7
396 int size
; /* font size of expression */
397 #define EQN_DEFSIZE INT_MIN
398 enum eqn_boxt type
; /* type of node */
399 struct eqn_box
*first
; /* first child node */
400 struct eqn_box
*last
; /* last child node */
401 struct eqn_box
*next
; /* node sibling */
402 struct eqn_box
*prev
; /* node sibling */
403 struct eqn_box
*parent
; /* node sibling */
404 char *text
; /* text (or NULL) */
405 char *left
; /* fence left-hand */
406 char *right
; /* fence right-hand */
407 char *top
; /* expression over-symbol */
408 char *bottom
; /* expression under-symbol */
409 size_t args
; /* arguments in parent */
410 size_t expectargs
; /* max arguments in parent */
411 enum eqn_post pos
; /* position of next box */
412 enum eqn_fontt font
; /* font of box */
413 enum eqn_pilet pile
; /* equation piling */
419 #define MPARSE_MDOC 1 /* assume -mdoc */
420 #define MPARSE_MAN 2 /* assume -man */
421 #define MPARSE_SO 4 /* honour .so requests */
422 #define MPARSE_QUICK 8 /* abort the parse early */
423 #define MPARSE_UTF8 16 /* accept UTF-8 input */
424 #define MPARSE_LATIN1 32 /* accept ISO-LATIN-1 input */
433 ESCAPE_ERROR
= 0, /* bail! unparsable escape */
434 ESCAPE_IGNORE
, /* escape to be ignored */
435 ESCAPE_SPECIAL
, /* a regular special character */
436 ESCAPE_FONT
, /* a generic font mode */
437 ESCAPE_FONTBOLD
, /* bold font mode */
438 ESCAPE_FONTITALIC
, /* italic font mode */
439 ESCAPE_FONTBI
, /* bold italic font mode */
440 ESCAPE_FONTROMAN
, /* roman font mode */
441 ESCAPE_FONTPREV
, /* previous font mode */
442 ESCAPE_NUMBERED
, /* a numbered glyph */
443 ESCAPE_UNICODE
, /* a unicode codepoint */
444 ESCAPE_DEVICE
, /* print the output device name */
445 ESCAPE_BREAK
, /* break the output line */
446 ESCAPE_NOSPACE
, /* suppress space if the last on a line */
447 ESCAPE_HORIZ
, /* horizontal movement */
448 ESCAPE_HLINE
, /* horizontal line drawing */
449 ESCAPE_SKIPCHAR
, /* skip the next character */
450 ESCAPE_OVERSTRIKE
/* overstrike all chars in the argument */
453 typedef void (*mandocmsg
)(enum mandocerr
, enum mandoclevel
,
454 const char *, int, int, const char *);
460 enum mandoc_esc
mandoc_escape(const char **, const char **, int *);
461 void mchars_alloc(void);
462 void mchars_free(void);
463 int mchars_num2char(const char *, size_t);
464 const char *mchars_uc2str(int);
465 int mchars_num2uc(const char *, size_t);
466 int mchars_spec2cp(const char *, size_t);
467 const char *mchars_spec2str(const char *, size_t, size_t *);
468 struct mparse
*mparse_alloc(int, enum mandocerr
, mandocmsg
,
469 enum mandoc_os
, const char *);
470 void mparse_free(struct mparse
*);
471 int mparse_open(struct mparse
*, const char *);
472 enum mandoclevel
mparse_readfd(struct mparse
*, int, const char *);
473 enum mandoclevel
mparse_readmem(struct mparse
*, void *, size_t,
475 void mparse_reset(struct mparse
*);
476 void mparse_result(struct mparse
*,
477 struct roff_man
**, char **);
478 void mparse_copy(const struct mparse
*);
479 const char *mparse_strerror(enum mandocerr
);
480 const char *mparse_strlevel(enum mandoclevel
);
481 void mparse_updaterc(struct mparse
*, enum mandoclevel
*);