Annotation of mandoc/mandoc.h, Revision 1.130
1.130 ! schwarze 1: /* $Id: mandoc.h,v 1.129 2014/07/02 13:10:45 schwarze Exp $ */
1.1 kristaps 2: /*
1.46 schwarze 3: * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.118 schwarze 4: * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
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.
9: *
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.
17: */
18: #ifndef MANDOC_H
19: #define MANDOC_H
1.7 kristaps 20:
21: #define ASCII_NBRSP 31 /* non-breaking space */
22: #define ASCII_HYPH 30 /* breakable hyphen */
1.115 schwarze 23: #define ASCII_BREAK 29 /* breakable zero-width space */
1.7 kristaps 24:
1.22 kristaps 25: /*
26: * Status level. This refers to both internal status (i.e., whilst
27: * running, when warnings/errors are reported) and an indicator of a
28: * threshold of when to halt (when said internal state exceeds the
29: * threshold).
30: */
1.18 schwarze 31: enum mandoclevel {
32: MANDOCLEVEL_OK = 0,
33: MANDOCLEVEL_RESERVED,
1.22 kristaps 34: MANDOCLEVEL_WARNING, /* warnings: syntax, whitespace, etc. */
35: MANDOCLEVEL_ERROR, /* input has been thrown away */
36: MANDOCLEVEL_FATAL, /* input is borked */
37: MANDOCLEVEL_BADARG, /* bad argument in invocation */
38: MANDOCLEVEL_SYSERR, /* system error */
1.18 schwarze 39: MANDOCLEVEL_MAX
40: };
41:
1.22 kristaps 42: /*
43: * All possible things that can go wrong within a parse, be it libroff,
44: * libmdoc, or libman.
45: */
1.1 kristaps 46: enum mandocerr {
47: MANDOCERR_OK,
1.13 schwarze 48:
1.21 kristaps 49: MANDOCERR_WARNING, /* ===== start of warnings ===== */
1.32 kristaps 50:
51: /* related to the prologue */
1.125 schwarze 52: MANDOCERR_TH_MISSING, /* missing .TH macro, using "unknown 1" */
53: MANDOCERR_TITLE_CASE, /* lower case character in document title */
54: MANDOCERR_MSEC_BAD, /* unknown manual section: section */
55: MANDOCERR_ARCH_BAD, /* unknown manual volume or arch: volume */
56: MANDOCERR_DATE_MISSING, /* missing date, using today's date */
57: MANDOCERR_DATE_BAD, /* cannot parse date, using it verbatim: date */
58: MANDOCERR_PROLOG_ORDER, /* prologue macros out of order: macro */
59: MANDOCERR_PROLOG_REP, /* duplicate prologue macro: macro */
60: MANDOCERR_PROLOG_BAD, /* incomplete prologue, terminated by: macro */
61: MANDOCERR_PROLOG_ONLY, /* skipping prologue macro in body: macro */
1.32 kristaps 62:
63: /* related to document structure */
1.124 schwarze 64: MANDOCERR_SO, /* .so is fragile, better use ln(1): .so path */
1.122 schwarze 65: MANDOCERR_DOC_EMPTY, /* no document body */
1.125 schwarze 66: MANDOCERR_SEC_BEFORE, /* content before first section header: macro */
67: MANDOCERR_NAMESEC_FIRST, /* first section is not "NAME": title */
68: MANDOCERR_NAMESEC_BAD, /* bad NAME section contents: macro */
69: MANDOCERR_SEC_ORDER, /* sections out of conventional order: title */
70: MANDOCERR_SEC_REP, /* duplicate section title: title */
71: MANDOCERR_SEC_MSEC, /* unexpected section: title for ... only */
1.32 kristaps 72:
73: /* related to macros and nesting */
1.126 schwarze 74: MANDOCERR_MACRO_OBS, /* obsolete macro: macro */
1.127 schwarze 75: MANDOCERR_PAR_SKIP, /* skipping paragraph macro: macro ... */
76: MANDOCERR_PAR_MOVE, /* moving paragraph macro out of list: macro */
1.128 schwarze 77: MANDOCERR_NS_SKIP, /* skipping no-space macro */
78: MANDOCERR_BLOCK_NEST, /* blocks badly nested: macro ... */
79: MANDOCERR_BD_NEST, /* nested displays are not portable: macro ... */
1.129 schwarze 80: MANDOCERR_BL_MOVE, /* moving content out of list: macro */
81: MANDOCERR_VT_CHILD, /* .Vt block has child macro: macro */
1.128 schwarze 82: MANDOCERR_FI_SKIP, /* fill mode already enabled, skipping .fi */
83: MANDOCERR_NF_SKIP, /* fill mode already disabled, skipping .nf */
84: MANDOCERR_LINESCOPE, /* line scope broken: macro breaks macro */
1.32 kristaps 85:
86: /* related to missing macro arguments */
1.130 ! schwarze 87: MANDOCERR_MACRO_EMPTY, /* skipping empty macro: macro */
1.46 schwarze 88: MANDOCERR_ARGCWARN, /* argument count wrong */
1.32 kristaps 89: MANDOCERR_DISPTYPE, /* missing display type */
90: MANDOCERR_LISTFIRST, /* list type must come first */
91: MANDOCERR_NOWIDTHARG, /* tag lists require a width argument */
92: MANDOCERR_FONTTYPE, /* missing font type */
1.13 schwarze 93:
1.32 kristaps 94: /* related to bad macro arguments */
95: MANDOCERR_IGNARGV, /* skipping argument */
96: MANDOCERR_ARGVREP, /* duplicate argument */
97: MANDOCERR_DISPREP, /* duplicate display type */
98: MANDOCERR_LISTREP, /* duplicate list type */
99: MANDOCERR_BADATT, /* unknown AT&T UNIX version */
1.5 kristaps 100: MANDOCERR_BADBOOL, /* bad Boolean value */
1.31 kristaps 101: MANDOCERR_BADFONT, /* unknown font */
1.32 kristaps 102: MANDOCERR_BADSTANDARD, /* unknown standard specifier */
103: MANDOCERR_BADWIDTH, /* bad width argument */
104:
105: /* related to plain text */
106: MANDOCERR_NOBLANKLN, /* blank line in non-literal context */
107: MANDOCERR_BADTAB, /* tab in non-literal context */
108: MANDOCERR_EOLNSPACE, /* end of line whitespace */
1.5 kristaps 109: MANDOCERR_BADCOMMENT, /* bad comment style */
1.124 schwarze 110: MANDOCERR_BADESCAPE, /* bad escape sequence */
1.32 kristaps 111: MANDOCERR_BADQUOTE, /* unterminated quoted string */
1.79 kristaps 112:
1.32 kristaps 113: MANDOCERR_ERROR, /* ===== start of errors ===== */
1.80 kristaps 114:
115: /* related to equations */
1.84 kristaps 116: MANDOCERR_EQNNSCOPE, /* unexpected equation scope closure*/
117: MANDOCERR_EQNSCOPE, /* equation scope open on exit */
1.89 kristaps 118: MANDOCERR_EQNBADSCOPE, /* overlapping equation scopes */
119: MANDOCERR_EQNEOF, /* unexpected end of equation */
120: MANDOCERR_EQNSYNT, /* equation syntax error */
1.32 kristaps 121:
1.42 kristaps 122: /* related to tables */
1.36 kristaps 123: MANDOCERR_TBL, /* bad table syntax */
124: MANDOCERR_TBLOPT, /* bad table option */
1.37 kristaps 125: MANDOCERR_TBLLAYOUT, /* bad table layout */
126: MANDOCERR_TBLNOLAYOUT, /* no table layout cells specified */
1.39 kristaps 127: MANDOCERR_TBLNODATA, /* no table data cells specified */
1.47 kristaps 128: MANDOCERR_TBLIGNDATA, /* ignore data in cell */
1.48 kristaps 129: MANDOCERR_TBLBLOCK, /* data block still open */
1.50 kristaps 130: MANDOCERR_TBLEXTRADAT, /* ignoring extra data cells */
1.42 kristaps 131:
1.33 schwarze 132: MANDOCERR_ROFFLOOP, /* input stack limit exceeded, infinite loop? */
1.32 kristaps 133: MANDOCERR_BADCHAR, /* skipping bad character */
1.52 schwarze 134: MANDOCERR_NAMESC, /* escaped character not allowed in a name */
1.103 schwarze 135: MANDOCERR_NONAME, /* manual name not yet set */
1.125 schwarze 136: MANDOCERR_NOTEXT, /* skipping text before first section header */
1.32 kristaps 137: MANDOCERR_MACRO, /* skipping unknown macro */
1.25 kristaps 138: MANDOCERR_REQUEST, /* NOT IMPLEMENTED: skipping request */
1.5 kristaps 139: MANDOCERR_ARGCOUNT, /* argument count wrong */
1.129 schwarze 140: MANDOCERR_RS_SKIP, /* skipping invalid content in .Rs block: macro */
1.106 schwarze 141: MANDOCERR_STRAYTA, /* skipping column outside column list */
1.32 kristaps 142: MANDOCERR_NOSCOPE, /* skipping end of block that is not open */
1.30 kristaps 143: MANDOCERR_SCOPEBROKEN, /* missing end of block */
1.19 schwarze 144: MANDOCERR_SCOPEEXIT, /* scope open on exit */
1.28 kristaps 145: MANDOCERR_UNAME, /* uname(3) system call failed */
1.5 kristaps 146: /* FIXME: merge following with MANDOCERR_ARGCOUNT */
147: MANDOCERR_NOARGS, /* macro requires line argument(s) */
148: MANDOCERR_NOBODY, /* macro requires body argument(s) */
149: MANDOCERR_NOARGV, /* macro requires argument(s) */
1.108 schwarze 150: MANDOCERR_NUMERIC, /* request requires a numeric argument */
1.10 kristaps 151: MANDOCERR_LISTTYPE, /* missing list type */
1.5 kristaps 152: MANDOCERR_ARGSLOST, /* line argument(s) will be lost */
1.13 schwarze 153:
1.21 kristaps 154: MANDOCERR_FATAL, /* ===== start of fatal errors ===== */
1.32 kristaps 155:
1.113 schwarze 156: MANDOCERR_TOOLARGE, /* input too large */
1.124 schwarze 157: MANDOCERR_NOTMANUAL, /* not a manual */
1.8 kristaps 158: MANDOCERR_COLUMNS, /* column syntax is inconsistent */
1.32 kristaps 159: MANDOCERR_BADDISP, /* NOT IMPLEMENTED: .Bd -file */
1.5 kristaps 160: MANDOCERR_SYNTARGVCOUNT, /* argument count wrong, violates syntax */
161: MANDOCERR_SYNTCHILD, /* child violates parent syntax */
162: MANDOCERR_SYNTARGCOUNT, /* argument count wrong, violates syntax */
1.123 schwarze 163: MANDOCERR_SO_PATH, /* NOT IMPLEMENTED: .so with absolute path or ".." */
164: MANDOCERR_SO_FAIL, /* .so request failed */
1.5 kristaps 165: MANDOCERR_NODOCPROLOG, /* no document prologue */
1.18 schwarze 166: MANDOCERR_MEM, /* static buffer exhausted */
1.113 schwarze 167:
168: /* ===== system errors ===== */
169:
170: MANDOCERR_SYSOPEN, /* cannot open file */
171: MANDOCERR_SYSSTAT, /* cannot stat file */
172: MANDOCERR_SYSREAD, /* cannot read file */
173:
1.1 kristaps 174: MANDOCERR_MAX
1.43 kristaps 175: };
176:
1.107 schwarze 177: struct tbl_opts {
1.43 kristaps 178: char tab; /* cell-separator */
179: char decimal; /* decimal point */
180: int linesize;
181: int opts;
182: #define TBL_OPT_CENTRE (1 << 0)
183: #define TBL_OPT_EXPAND (1 << 1)
184: #define TBL_OPT_BOX (1 << 2)
185: #define TBL_OPT_DBOX (1 << 3)
186: #define TBL_OPT_ALLBOX (1 << 4)
187: #define TBL_OPT_NOKEEP (1 << 5)
188: #define TBL_OPT_NOSPACE (1 << 6)
1.45 kristaps 189: int cols; /* number of columns */
1.38 kristaps 190: };
191:
1.41 kristaps 192: /*
193: * The head of a table specifies all of its columns. When formatting a
194: * tbl_span, iterate over these and plug in data from the tbl_span when
195: * appropriate, using tbl_cell as a guide to placement.
196: */
197: struct tbl_head {
1.45 kristaps 198: int ident; /* 0 <= unique id < cols */
1.101 schwarze 199: int vert; /* width of preceding vertical line */
1.41 kristaps 200: struct tbl_head *next;
201: struct tbl_head *prev;
202: };
203:
1.38 kristaps 204: enum tbl_cellt {
205: TBL_CELL_CENTRE, /* c, C */
206: TBL_CELL_RIGHT, /* r, R */
207: TBL_CELL_LEFT, /* l, L */
208: TBL_CELL_NUMBER, /* n, N */
209: TBL_CELL_SPAN, /* s, S */
210: TBL_CELL_LONG, /* a, A */
211: TBL_CELL_DOWN, /* ^ */
212: TBL_CELL_HORIZ, /* _, - */
213: TBL_CELL_DHORIZ, /* = */
214: TBL_CELL_MAX
215: };
216:
217: /*
218: * A cell in a layout row.
219: */
220: struct tbl_cell {
221: struct tbl_cell *next;
1.101 schwarze 222: int vert; /* width of preceding vertical line */
1.38 kristaps 223: enum tbl_cellt pos;
1.51 kristaps 224: size_t spacing;
1.38 kristaps 225: int flags;
226: #define TBL_CELL_TALIGN (1 << 0) /* t, T */
227: #define TBL_CELL_BALIGN (1 << 1) /* d, D */
228: #define TBL_CELL_BOLD (1 << 2) /* fB, B, b */
229: #define TBL_CELL_ITALIC (1 << 3) /* fI, I, i */
230: #define TBL_CELL_EQUAL (1 << 4) /* e, E */
231: #define TBL_CELL_UP (1 << 5) /* u, U */
232: #define TBL_CELL_WIGN (1 << 6) /* z, Z */
1.41 kristaps 233: struct tbl_head *head;
1.38 kristaps 234: };
235:
236: /*
237: * A layout row.
238: */
239: struct tbl_row {
240: struct tbl_row *next;
241: struct tbl_cell *first;
242: struct tbl_cell *last;
1.119 schwarze 243: int vert; /* trailing vertical line */
1.38 kristaps 244: };
245:
1.40 kristaps 246: enum tbl_datt {
1.51 kristaps 247: TBL_DATA_NONE, /* has no data */
248: TBL_DATA_DATA, /* consists of data/string */
249: TBL_DATA_HORIZ, /* horizontal line */
250: TBL_DATA_DHORIZ, /* double-horizontal line */
251: TBL_DATA_NHORIZ, /* squeezed horizontal line */
252: TBL_DATA_NDHORIZ /* squeezed double-horizontal line */
1.40 kristaps 253: };
254:
1.38 kristaps 255: /*
256: * A cell within a row of data. The "string" field contains the actual
257: * string value that's in the cell. The rest is layout.
258: */
259: struct tbl_dat {
1.51 kristaps 260: struct tbl_cell *layout; /* layout cell */
1.50 kristaps 261: int spans; /* how many spans follow */
1.38 kristaps 262: struct tbl_dat *next;
1.51 kristaps 263: char *string; /* data (NULL if not TBL_DATA_DATA) */
1.40 kristaps 264: enum tbl_datt pos;
265: };
266:
267: enum tbl_spant {
268: TBL_SPAN_DATA, /* span consists of data */
269: TBL_SPAN_HORIZ, /* span is horizontal line */
270: TBL_SPAN_DHORIZ /* span is double horizontal line */
1.38 kristaps 271: };
272:
273: /*
274: * A row of data in a table.
275: */
276: struct tbl_span {
1.107 schwarze 277: struct tbl_opts *opts;
1.44 kristaps 278: struct tbl_head *head;
1.51 kristaps 279: struct tbl_row *layout; /* layout row */
1.38 kristaps 280: struct tbl_dat *first;
281: struct tbl_dat *last;
1.56 kristaps 282: int line; /* parse line */
1.44 kristaps 283: int flags;
284: #define TBL_SPAN_FIRST (1 << 0)
285: #define TBL_SPAN_LAST (1 << 1)
1.40 kristaps 286: enum tbl_spant pos;
1.38 kristaps 287: struct tbl_span *next;
1.55 kristaps 288: };
289:
1.83 kristaps 290: enum eqn_boxt {
1.84 kristaps 291: EQN_ROOT, /* root of parse tree */
292: EQN_TEXT, /* text (number, variable, whatever) */
1.90 kristaps 293: EQN_SUBEXPR, /* nested `eqn' subexpression */
1.92 kristaps 294: EQN_LIST, /* subexpressions list */
295: EQN_MATRIX /* matrix subexpression */
1.83 kristaps 296: };
297:
1.85 kristaps 298: enum eqn_markt {
299: EQNMARK_NONE = 0,
300: EQNMARK_DOT,
301: EQNMARK_DOTDOT,
302: EQNMARK_HAT,
303: EQNMARK_TILDE,
304: EQNMARK_VEC,
305: EQNMARK_DYAD,
306: EQNMARK_BAR,
307: EQNMARK_UNDER,
308: EQNMARK__MAX
309: };
310:
1.86 kristaps 311: enum eqn_fontt {
312: EQNFONT_NONE = 0,
313: EQNFONT_ROMAN,
314: EQNFONT_BOLD,
1.93 kristaps 315: EQNFONT_FAT,
1.86 kristaps 316: EQNFONT_ITALIC,
317: EQNFONT__MAX
318: };
319:
1.87 kristaps 320: enum eqn_post {
321: EQNPOS_NONE = 0,
322: EQNPOS_OVER,
323: EQNPOS_SUP,
324: EQNPOS_SUB,
325: EQNPOS_TO,
326: EQNPOS_FROM,
327: EQNPOS__MAX
328: };
329:
1.89 kristaps 330: enum eqn_pilet {
331: EQNPILE_NONE = 0,
1.95 kristaps 332: EQNPILE_PILE,
1.89 kristaps 333: EQNPILE_CPILE,
334: EQNPILE_RPILE,
335: EQNPILE_LPILE,
1.95 kristaps 336: EQNPILE_COL,
1.92 kristaps 337: EQNPILE_CCOL,
338: EQNPILE_RCOL,
339: EQNPILE_LCOL,
1.89 kristaps 340: EQNPILE__MAX
341: };
342:
1.86 kristaps 343: /*
1.84 kristaps 344: * A "box" is a parsed mathematical expression as defined by the eqn.7
345: * grammar.
346: */
1.83 kristaps 347: struct eqn_box {
1.88 kristaps 348: int size; /* font size of expression */
349: #define EQN_DEFSIZE INT_MIN
1.84 kristaps 350: enum eqn_boxt type; /* type of node */
1.89 kristaps 351: struct eqn_box *first; /* first child node */
352: struct eqn_box *last; /* last child node */
353: struct eqn_box *next; /* node sibling */
354: struct eqn_box *parent; /* node sibling */
355: char *text; /* text (or NULL) */
356: char *left;
357: char *right;
1.87 kristaps 358: enum eqn_post pos; /* position of next box */
1.86 kristaps 359: enum eqn_markt mark; /* a mark about the box */
360: enum eqn_fontt font; /* font of box */
1.89 kristaps 361: enum eqn_pilet pile; /* equation piling */
1.83 kristaps 362: };
363:
1.87 kristaps 364: /*
365: * An equation consists of a tree of expressions starting at a given
1.120 schwarze 366: * line and position.
1.87 kristaps 367: */
1.55 kristaps 368: struct eqn {
1.94 kristaps 369: char *name; /* identifier (or NULL) */
1.84 kristaps 370: struct eqn_box *root; /* root mathematical expression */
1.82 kristaps 371: int ln; /* invocation line */
1.57 kristaps 372: int pos; /* invocation position */
1.16 kristaps 373: };
1.22 kristaps 374:
1.60 kristaps 375: /*
1.116 schwarze 376: * Parse options.
377: */
378: #define MPARSE_MDOC 1 /* assume -mdoc */
379: #define MPARSE_MAN 2 /* assume -man */
380: #define MPARSE_SO 4 /* honour .so requests */
381: #define MPARSE_QUICK 8 /* abort the parse early */
1.62 kristaps 382:
1.70 kristaps 383: enum mandoc_esc {
384: ESCAPE_ERROR = 0, /* bail! unparsable escape */
385: ESCAPE_IGNORE, /* escape to be ignored */
386: ESCAPE_SPECIAL, /* a regular special character */
387: ESCAPE_FONT, /* a generic font mode */
388: ESCAPE_FONTBOLD, /* bold font mode */
389: ESCAPE_FONTITALIC, /* italic font mode */
1.109 schwarze 390: ESCAPE_FONTBI, /* bold italic font mode */
1.70 kristaps 391: ESCAPE_FONTROMAN, /* roman font mode */
392: ESCAPE_FONTPREV, /* previous font mode */
393: ESCAPE_NUMBERED, /* a numbered glyph */
1.75 kristaps 394: ESCAPE_UNICODE, /* a unicode codepoint */
1.102 schwarze 395: ESCAPE_NOSPACE, /* suppress space if the last on a line */
396: ESCAPE_SKIPCHAR /* skip the next character */
1.70 kristaps 397: };
398:
1.63 kristaps 399: typedef void (*mandocmsg)(enum mandocerr, enum mandoclevel,
400: const char *, int, int, const char *);
1.62 kristaps 401:
402: struct mparse;
1.71 kristaps 403: struct mchars;
1.62 kristaps 404: struct mdoc;
405: struct man;
1.59 kristaps 406:
1.22 kristaps 407: __BEGIN_DECLS
1.62 kristaps 408:
1.112 schwarze 409: enum mandoc_esc mandoc_escape(const char **, const char **, int *);
1.73 kristaps 410: struct mchars *mchars_alloc(void);
1.96 kristaps 411: void mchars_free(struct mchars *);
1.120 schwarze 412: char mchars_num2char(const char *, size_t);
1.76 kristaps 413: int mchars_num2uc(const char *, size_t);
1.120 schwarze 414: int mchars_spec2cp(const struct mchars *,
1.96 kristaps 415: const char *, size_t);
1.120 schwarze 416: const char *mchars_spec2str(const struct mchars *,
1.96 kristaps 417: const char *, size_t, size_t *);
1.116 schwarze 418: struct mparse *mparse_alloc(int, enum mandoclevel, mandocmsg, char *);
1.96 kristaps 419: void mparse_free(struct mparse *);
420: void mparse_keep(struct mparse *);
421: enum mandoclevel mparse_readfd(struct mparse *, int, const char *);
1.99 joerg 422: enum mandoclevel mparse_readmem(struct mparse *, const void *, size_t,
423: const char *);
1.96 kristaps 424: void mparse_reset(struct mparse *);
1.120 schwarze 425: void mparse_result(struct mparse *,
1.117 schwarze 426: struct mdoc **, struct man **, char **);
1.96 kristaps 427: const char *mparse_getkeep(const struct mparse *);
428: const char *mparse_strerror(enum mandocerr);
429: const char *mparse_strlevel(enum mandoclevel);
1.1 kristaps 430:
431: __END_DECLS
432:
433: #endif /*!MANDOC_H*/
CVSweb