Annotation of mandoc/mandoc.h, Revision 1.105
1.105 ! schwarze 1: /* $Id: mandoc.h,v 1.104 2012/07/18 11:11:12 schwarze Exp $ */
1.1 kristaps 2: /*
1.46 schwarze 3: * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.1 kristaps 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
1.7 kristaps 19:
20: #define ASCII_NBRSP 31 /* non-breaking space */
21: #define ASCII_HYPH 30 /* breakable hyphen */
22:
1.22 kristaps 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: */
1.18 schwarze 29: enum mandoclevel {
30: MANDOCLEVEL_OK = 0,
31: MANDOCLEVEL_RESERVED,
1.22 kristaps 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 */
1.18 schwarze 37: MANDOCLEVEL_MAX
38: };
39:
1.22 kristaps 40: /*
41: * All possible things that can go wrong within a parse, be it libroff,
42: * libmdoc, or libman.
43: */
1.1 kristaps 44: enum mandocerr {
45: MANDOCERR_OK,
1.13 schwarze 46:
1.21 kristaps 47: MANDOCERR_WARNING, /* ===== start of warnings ===== */
1.32 kristaps 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 */
1.105 ! schwarze 53: MANDOCERR_BADVOLARCH, /* unknown manual volume or arch */
1.58 schwarze 54: MANDOCERR_NODATE, /* date missing, using today's date */
55: MANDOCERR_BADDATE, /* cannot parse date, using it verbatim */
1.32 kristaps 56: MANDOCERR_PROLOGOOO, /* prologue macros out of order */
57: MANDOCERR_PROLOGREP, /* duplicate prologue macro */
58: MANDOCERR_BADPROLOG, /* macro not allowed in prologue */
59: MANDOCERR_BADBODY, /* macro not allowed in body */
60:
61: /* related to document structure */
1.26 kristaps 62: MANDOCERR_SO, /* .so is fragile, better use ln(1) */
1.32 kristaps 63: MANDOCERR_NAMESECFIRST, /* NAME section must come first */
64: MANDOCERR_BADNAMESEC, /* bad NAME section contents */
1.9 kristaps 65: MANDOCERR_SECOOO, /* sections out of conventional order */
1.32 kristaps 66: MANDOCERR_SECREP, /* duplicate section name */
1.5 kristaps 67: MANDOCERR_SECMSEC, /* section not in conventional manual section */
1.32 kristaps 68:
69: /* related to macros and nesting */
70: MANDOCERR_MACROOBS, /* skipping obsolete macro */
1.35 kristaps 71: MANDOCERR_IGNPAR, /* skipping paragraph macro */
1.104 schwarze 72: MANDOCERR_MOVEPAR, /* moving paragraph macro out of list */
1.54 kristaps 73: MANDOCERR_IGNNS, /* skipping no-space macro */
1.14 schwarze 74: MANDOCERR_SCOPENEST, /* blocks badly nested */
1.32 kristaps 75: MANDOCERR_CHILD, /* child violates parent syntax */
76: MANDOCERR_NESTEDDISP, /* nested displays are not portable */
77: MANDOCERR_SCOPEREP, /* already in literal mode */
1.68 kristaps 78: MANDOCERR_LINESCOPE, /* line scope broken */
1.32 kristaps 79:
80: /* related to missing macro arguments */
81: MANDOCERR_MACROEMPTY, /* skipping empty macro */
1.46 schwarze 82: MANDOCERR_ARGCWARN, /* argument count wrong */
1.32 kristaps 83: MANDOCERR_DISPTYPE, /* missing display type */
84: MANDOCERR_LISTFIRST, /* list type must come first */
85: MANDOCERR_NOWIDTHARG, /* tag lists require a width argument */
86: MANDOCERR_FONTTYPE, /* missing font type */
1.53 kristaps 87: MANDOCERR_WNOSCOPE, /* skipping end of block that is not open */
1.13 schwarze 88:
1.32 kristaps 89: /* related to bad macro arguments */
90: MANDOCERR_IGNARGV, /* skipping argument */
91: MANDOCERR_ARGVREP, /* duplicate argument */
92: MANDOCERR_DISPREP, /* duplicate display type */
93: MANDOCERR_LISTREP, /* duplicate list type */
94: MANDOCERR_BADATT, /* unknown AT&T UNIX version */
1.5 kristaps 95: MANDOCERR_BADBOOL, /* bad Boolean value */
1.31 kristaps 96: MANDOCERR_BADFONT, /* unknown font */
1.32 kristaps 97: MANDOCERR_BADSTANDARD, /* unknown standard specifier */
98: MANDOCERR_BADWIDTH, /* bad width argument */
99:
100: /* related to plain text */
101: MANDOCERR_NOBLANKLN, /* blank line in non-literal context */
102: MANDOCERR_BADTAB, /* tab in non-literal context */
103: MANDOCERR_EOLNSPACE, /* end of line whitespace */
1.5 kristaps 104: MANDOCERR_BADCOMMENT, /* bad comment style */
1.32 kristaps 105: MANDOCERR_BADESCAPE, /* unknown escape sequence */
106: MANDOCERR_BADQUOTE, /* unterminated quoted string */
1.79 kristaps 107:
108: /* related to equations */
109: MANDOCERR_EQNQUOTE, /* unexpected literal in equation */
1.32 kristaps 110:
111: MANDOCERR_ERROR, /* ===== start of errors ===== */
1.80 kristaps 112:
113: /* related to equations */
1.84 kristaps 114: MANDOCERR_EQNNSCOPE, /* unexpected equation scope closure*/
115: MANDOCERR_EQNSCOPE, /* equation scope open on exit */
1.89 kristaps 116: MANDOCERR_EQNBADSCOPE, /* overlapping equation scopes */
117: MANDOCERR_EQNEOF, /* unexpected end of equation */
118: MANDOCERR_EQNSYNT, /* equation syntax error */
1.32 kristaps 119:
1.42 kristaps 120: /* related to tables */
1.36 kristaps 121: MANDOCERR_TBL, /* bad table syntax */
122: MANDOCERR_TBLOPT, /* bad table option */
1.37 kristaps 123: MANDOCERR_TBLLAYOUT, /* bad table layout */
124: MANDOCERR_TBLNOLAYOUT, /* no table layout cells specified */
1.39 kristaps 125: MANDOCERR_TBLNODATA, /* no table data cells specified */
1.47 kristaps 126: MANDOCERR_TBLIGNDATA, /* ignore data in cell */
1.48 kristaps 127: MANDOCERR_TBLBLOCK, /* data block still open */
1.50 kristaps 128: MANDOCERR_TBLEXTRADAT, /* ignoring extra data cells */
1.42 kristaps 129:
1.33 schwarze 130: MANDOCERR_ROFFLOOP, /* input stack limit exceeded, infinite loop? */
1.32 kristaps 131: MANDOCERR_BADCHAR, /* skipping bad character */
1.52 schwarze 132: MANDOCERR_NAMESC, /* escaped character not allowed in a name */
1.103 schwarze 133: MANDOCERR_NONAME, /* manual name not yet set */
1.32 kristaps 134: MANDOCERR_NOTEXT, /* skipping text before the first section header */
135: MANDOCERR_MACRO, /* skipping unknown macro */
1.25 kristaps 136: MANDOCERR_REQUEST, /* NOT IMPLEMENTED: skipping request */
1.5 kristaps 137: MANDOCERR_ARGCOUNT, /* argument count wrong */
1.32 kristaps 138: MANDOCERR_NOSCOPE, /* skipping end of block that is not open */
1.30 kristaps 139: MANDOCERR_SCOPEBROKEN, /* missing end of block */
1.19 schwarze 140: MANDOCERR_SCOPEEXIT, /* scope open on exit */
1.28 kristaps 141: MANDOCERR_UNAME, /* uname(3) system call failed */
1.5 kristaps 142: /* FIXME: merge following with MANDOCERR_ARGCOUNT */
143: MANDOCERR_NOARGS, /* macro requires line argument(s) */
144: MANDOCERR_NOBODY, /* macro requires body argument(s) */
145: MANDOCERR_NOARGV, /* macro requires argument(s) */
1.10 kristaps 146: MANDOCERR_LISTTYPE, /* missing list type */
1.5 kristaps 147: MANDOCERR_ARGSLOST, /* line argument(s) will be lost */
148: MANDOCERR_BODYLOST, /* body argument(s) will be lost */
1.13 schwarze 149:
1.21 kristaps 150: MANDOCERR_FATAL, /* ===== start of fatal errors ===== */
1.32 kristaps 151:
1.66 kristaps 152: MANDOCERR_NOTMANUAL, /* manual isn't really a manual */
1.8 kristaps 153: MANDOCERR_COLUMNS, /* column syntax is inconsistent */
1.32 kristaps 154: MANDOCERR_BADDISP, /* NOT IMPLEMENTED: .Bd -file */
1.5 kristaps 155: MANDOCERR_SYNTARGVCOUNT, /* argument count wrong, violates syntax */
156: MANDOCERR_SYNTCHILD, /* child violates parent syntax */
157: MANDOCERR_SYNTARGCOUNT, /* argument count wrong, violates syntax */
1.26 kristaps 158: MANDOCERR_SOPATH, /* NOT IMPLEMENTED: .so with absolute path or ".." */
1.5 kristaps 159: MANDOCERR_NODOCBODY, /* no document body */
160: MANDOCERR_NODOCPROLOG, /* no document prologue */
1.18 schwarze 161: MANDOCERR_MEM, /* static buffer exhausted */
1.1 kristaps 162: MANDOCERR_MAX
1.43 kristaps 163: };
164:
165: struct tbl {
166: char tab; /* cell-separator */
167: char decimal; /* decimal point */
168: int linesize;
169: int opts;
170: #define TBL_OPT_CENTRE (1 << 0)
171: #define TBL_OPT_EXPAND (1 << 1)
172: #define TBL_OPT_BOX (1 << 2)
173: #define TBL_OPT_DBOX (1 << 3)
174: #define TBL_OPT_ALLBOX (1 << 4)
175: #define TBL_OPT_NOKEEP (1 << 5)
176: #define TBL_OPT_NOSPACE (1 << 6)
1.45 kristaps 177: int cols; /* number of columns */
1.38 kristaps 178: };
179:
1.41 kristaps 180: /*
181: * The head of a table specifies all of its columns. When formatting a
182: * tbl_span, iterate over these and plug in data from the tbl_span when
183: * appropriate, using tbl_cell as a guide to placement.
184: */
185: struct tbl_head {
1.45 kristaps 186: int ident; /* 0 <= unique id < cols */
1.101 schwarze 187: int vert; /* width of preceding vertical line */
1.41 kristaps 188: struct tbl_head *next;
189: struct tbl_head *prev;
190: };
191:
1.38 kristaps 192: enum tbl_cellt {
193: TBL_CELL_CENTRE, /* c, C */
194: TBL_CELL_RIGHT, /* r, R */
195: TBL_CELL_LEFT, /* l, L */
196: TBL_CELL_NUMBER, /* n, N */
197: TBL_CELL_SPAN, /* s, S */
198: TBL_CELL_LONG, /* a, A */
199: TBL_CELL_DOWN, /* ^ */
200: TBL_CELL_HORIZ, /* _, - */
201: TBL_CELL_DHORIZ, /* = */
202: TBL_CELL_MAX
203: };
204:
205: /*
206: * A cell in a layout row.
207: */
208: struct tbl_cell {
209: struct tbl_cell *next;
1.101 schwarze 210: int vert; /* width of preceding vertical line */
1.38 kristaps 211: enum tbl_cellt pos;
1.51 kristaps 212: size_t spacing;
1.38 kristaps 213: int flags;
214: #define TBL_CELL_TALIGN (1 << 0) /* t, T */
215: #define TBL_CELL_BALIGN (1 << 1) /* d, D */
216: #define TBL_CELL_BOLD (1 << 2) /* fB, B, b */
217: #define TBL_CELL_ITALIC (1 << 3) /* fI, I, i */
218: #define TBL_CELL_EQUAL (1 << 4) /* e, E */
219: #define TBL_CELL_UP (1 << 5) /* u, U */
220: #define TBL_CELL_WIGN (1 << 6) /* z, Z */
1.41 kristaps 221: struct tbl_head *head;
1.38 kristaps 222: };
223:
224: /*
225: * A layout row.
226: */
227: struct tbl_row {
228: struct tbl_row *next;
229: struct tbl_cell *first;
230: struct tbl_cell *last;
231: };
232:
1.40 kristaps 233: enum tbl_datt {
1.51 kristaps 234: TBL_DATA_NONE, /* has no data */
235: TBL_DATA_DATA, /* consists of data/string */
236: TBL_DATA_HORIZ, /* horizontal line */
237: TBL_DATA_DHORIZ, /* double-horizontal line */
238: TBL_DATA_NHORIZ, /* squeezed horizontal line */
239: TBL_DATA_NDHORIZ /* squeezed double-horizontal line */
1.40 kristaps 240: };
241:
1.38 kristaps 242: /*
243: * A cell within a row of data. The "string" field contains the actual
244: * string value that's in the cell. The rest is layout.
245: */
246: struct tbl_dat {
1.51 kristaps 247: struct tbl_cell *layout; /* layout cell */
1.50 kristaps 248: int spans; /* how many spans follow */
1.38 kristaps 249: struct tbl_dat *next;
1.51 kristaps 250: char *string; /* data (NULL if not TBL_DATA_DATA) */
1.40 kristaps 251: enum tbl_datt pos;
252: };
253:
254: enum tbl_spant {
255: TBL_SPAN_DATA, /* span consists of data */
256: TBL_SPAN_HORIZ, /* span is horizontal line */
257: TBL_SPAN_DHORIZ /* span is double horizontal line */
1.38 kristaps 258: };
259:
260: /*
261: * A row of data in a table.
262: */
263: struct tbl_span {
1.44 kristaps 264: struct tbl *tbl;
265: struct tbl_head *head;
1.51 kristaps 266: struct tbl_row *layout; /* layout row */
1.38 kristaps 267: struct tbl_dat *first;
268: struct tbl_dat *last;
1.56 kristaps 269: int line; /* parse line */
1.44 kristaps 270: int flags;
271: #define TBL_SPAN_FIRST (1 << 0)
272: #define TBL_SPAN_LAST (1 << 1)
1.40 kristaps 273: enum tbl_spant pos;
1.38 kristaps 274: struct tbl_span *next;
1.55 kristaps 275: };
276:
1.83 kristaps 277: enum eqn_boxt {
1.84 kristaps 278: EQN_ROOT, /* root of parse tree */
279: EQN_TEXT, /* text (number, variable, whatever) */
1.90 kristaps 280: EQN_SUBEXPR, /* nested `eqn' subexpression */
1.92 kristaps 281: EQN_LIST, /* subexpressions list */
282: EQN_MATRIX /* matrix subexpression */
1.83 kristaps 283: };
284:
1.85 kristaps 285: enum eqn_markt {
286: EQNMARK_NONE = 0,
287: EQNMARK_DOT,
288: EQNMARK_DOTDOT,
289: EQNMARK_HAT,
290: EQNMARK_TILDE,
291: EQNMARK_VEC,
292: EQNMARK_DYAD,
293: EQNMARK_BAR,
294: EQNMARK_UNDER,
295: EQNMARK__MAX
296: };
297:
1.86 kristaps 298: enum eqn_fontt {
299: EQNFONT_NONE = 0,
300: EQNFONT_ROMAN,
301: EQNFONT_BOLD,
1.93 kristaps 302: EQNFONT_FAT,
1.86 kristaps 303: EQNFONT_ITALIC,
304: EQNFONT__MAX
305: };
306:
1.87 kristaps 307: enum eqn_post {
308: EQNPOS_NONE = 0,
309: EQNPOS_OVER,
310: EQNPOS_SUP,
311: EQNPOS_SUB,
312: EQNPOS_TO,
313: EQNPOS_FROM,
314: EQNPOS__MAX
315: };
316:
1.89 kristaps 317: enum eqn_pilet {
318: EQNPILE_NONE = 0,
1.95 kristaps 319: EQNPILE_PILE,
1.89 kristaps 320: EQNPILE_CPILE,
321: EQNPILE_RPILE,
322: EQNPILE_LPILE,
1.95 kristaps 323: EQNPILE_COL,
1.92 kristaps 324: EQNPILE_CCOL,
325: EQNPILE_RCOL,
326: EQNPILE_LCOL,
1.89 kristaps 327: EQNPILE__MAX
328: };
329:
1.86 kristaps 330: /*
1.84 kristaps 331: * A "box" is a parsed mathematical expression as defined by the eqn.7
332: * grammar.
333: */
1.83 kristaps 334: struct eqn_box {
1.88 kristaps 335: int size; /* font size of expression */
336: #define EQN_DEFSIZE INT_MIN
1.84 kristaps 337: enum eqn_boxt type; /* type of node */
1.89 kristaps 338: struct eqn_box *first; /* first child node */
339: struct eqn_box *last; /* last child node */
340: struct eqn_box *next; /* node sibling */
341: struct eqn_box *parent; /* node sibling */
342: char *text; /* text (or NULL) */
343: char *left;
344: char *right;
1.87 kristaps 345: enum eqn_post pos; /* position of next box */
1.86 kristaps 346: enum eqn_markt mark; /* a mark about the box */
347: enum eqn_fontt font; /* font of box */
1.89 kristaps 348: enum eqn_pilet pile; /* equation piling */
1.83 kristaps 349: };
350:
1.87 kristaps 351: /*
352: * An equation consists of a tree of expressions starting at a given
353: * line and position.
354: */
1.55 kristaps 355: struct eqn {
1.94 kristaps 356: char *name; /* identifier (or NULL) */
1.84 kristaps 357: struct eqn_box *root; /* root mathematical expression */
1.82 kristaps 358: int ln; /* invocation line */
1.57 kristaps 359: int pos; /* invocation position */
1.16 kristaps 360: };
1.22 kristaps 361:
1.60 kristaps 362: /*
1.62 kristaps 363: * The type of parse sequence. This value is usually passed via the
364: * mandoc(1) command line of -man and -mdoc. It's almost exclusively
365: * -mandoc but the others have been retained for compatibility.
366: */
367: enum mparset {
368: MPARSE_AUTO, /* magically determine the document type */
369: MPARSE_MDOC, /* assume -mdoc */
370: MPARSE_MAN /* assume -man */
371: };
372:
1.70 kristaps 373: enum mandoc_esc {
374: ESCAPE_ERROR = 0, /* bail! unparsable escape */
375: ESCAPE_IGNORE, /* escape to be ignored */
376: ESCAPE_SPECIAL, /* a regular special character */
377: ESCAPE_FONT, /* a generic font mode */
378: ESCAPE_FONTBOLD, /* bold font mode */
379: ESCAPE_FONTITALIC, /* italic font mode */
380: ESCAPE_FONTROMAN, /* roman font mode */
381: ESCAPE_FONTPREV, /* previous font mode */
382: ESCAPE_NUMBERED, /* a numbered glyph */
1.75 kristaps 383: ESCAPE_UNICODE, /* a unicode codepoint */
1.102 schwarze 384: ESCAPE_NOSPACE, /* suppress space if the last on a line */
385: ESCAPE_SKIPCHAR /* skip the next character */
1.70 kristaps 386: };
387:
1.63 kristaps 388: typedef void (*mandocmsg)(enum mandocerr, enum mandoclevel,
389: const char *, int, int, const char *);
1.62 kristaps 390:
391: struct mparse;
1.71 kristaps 392: struct mchars;
1.62 kristaps 393: struct mdoc;
394: struct man;
1.59 kristaps 395:
1.22 kristaps 396: __BEGIN_DECLS
1.62 kristaps 397:
1.59 kristaps 398: void *mandoc_calloc(size_t, size_t);
1.96 kristaps 399: enum mandoc_esc mandoc_escape(const char **, const char **, int *);
1.59 kristaps 400: void *mandoc_malloc(size_t);
401: void *mandoc_realloc(void *, size_t);
1.78 kristaps 402: char *mandoc_strdup(const char *);
1.89 kristaps 403: char *mandoc_strndup(const char *, size_t);
1.73 kristaps 404: struct mchars *mchars_alloc(void);
1.96 kristaps 405: void mchars_free(struct mchars *);
1.74 kristaps 406: char mchars_num2char(const char *, size_t);
1.76 kristaps 407: int mchars_num2uc(const char *, size_t);
1.98 kristaps 408: int mchars_spec2cp(const struct mchars *,
1.96 kristaps 409: const char *, size_t);
1.98 kristaps 410: const char *mchars_spec2str(const struct mchars *,
1.96 kristaps 411: const char *, size_t, size_t *);
1.100 schwarze 412: struct mparse *mparse_alloc(enum mparset, enum mandoclevel,
413: mandocmsg, void *, char *);
1.96 kristaps 414: void mparse_free(struct mparse *);
415: void mparse_keep(struct mparse *);
416: enum mandoclevel mparse_readfd(struct mparse *, int, const char *);
1.99 joerg 417: enum mandoclevel mparse_readmem(struct mparse *, const void *, size_t,
418: const char *);
1.96 kristaps 419: void mparse_reset(struct mparse *);
420: void mparse_result(struct mparse *,
421: struct mdoc **, struct man **);
422: const char *mparse_getkeep(const struct mparse *);
423: const char *mparse_strerror(enum mandocerr);
424: const char *mparse_strlevel(enum mandoclevel);
1.1 kristaps 425:
426: __END_DECLS
427:
428: #endif /*!MANDOC_H*/
CVSweb