Annotation of mandoc/mandoc.h, Revision 1.104
1.104 ! schwarze 1: /* $Id: mandoc.h,v 1.103 2012/07/12 15:11:14 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 */
1.58 schwarze 53: MANDOCERR_NODATE, /* date missing, using today's date */
54: MANDOCERR_BADDATE, /* cannot parse date, using it verbatim */
1.32 kristaps 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 */
1.26 kristaps 61: MANDOCERR_SO, /* .so is fragile, better use ln(1) */
1.32 kristaps 62: MANDOCERR_NAMESECFIRST, /* NAME section must come first */
63: MANDOCERR_BADNAMESEC, /* bad NAME section contents */
1.9 kristaps 64: MANDOCERR_SECOOO, /* sections out of conventional order */
1.32 kristaps 65: MANDOCERR_SECREP, /* duplicate section name */
1.5 kristaps 66: MANDOCERR_SECMSEC, /* section not in conventional manual section */
1.32 kristaps 67:
68: /* related to macros and nesting */
69: MANDOCERR_MACROOBS, /* skipping obsolete macro */
1.35 kristaps 70: MANDOCERR_IGNPAR, /* skipping paragraph macro */
1.104 ! schwarze 71: MANDOCERR_MOVEPAR, /* moving paragraph macro out of list */
1.54 kristaps 72: MANDOCERR_IGNNS, /* skipping no-space macro */
1.14 schwarze 73: MANDOCERR_SCOPENEST, /* blocks badly nested */
1.32 kristaps 74: MANDOCERR_CHILD, /* child violates parent syntax */
75: MANDOCERR_NESTEDDISP, /* nested displays are not portable */
76: MANDOCERR_SCOPEREP, /* already in literal mode */
1.68 kristaps 77: MANDOCERR_LINESCOPE, /* line scope broken */
1.32 kristaps 78:
79: /* related to missing macro arguments */
80: MANDOCERR_MACROEMPTY, /* skipping empty macro */
1.46 schwarze 81: MANDOCERR_ARGCWARN, /* argument count wrong */
1.32 kristaps 82: MANDOCERR_DISPTYPE, /* missing display type */
83: MANDOCERR_LISTFIRST, /* list type must come first */
84: MANDOCERR_NOWIDTHARG, /* tag lists require a width argument */
85: MANDOCERR_FONTTYPE, /* missing font type */
1.53 kristaps 86: MANDOCERR_WNOSCOPE, /* skipping end of block that is not open */
1.13 schwarze 87:
1.32 kristaps 88: /* related to bad macro arguments */
89: MANDOCERR_IGNARGV, /* skipping argument */
90: MANDOCERR_ARGVREP, /* duplicate argument */
91: MANDOCERR_DISPREP, /* duplicate display type */
92: MANDOCERR_LISTREP, /* duplicate list type */
93: MANDOCERR_BADATT, /* unknown AT&T UNIX version */
1.5 kristaps 94: MANDOCERR_BADBOOL, /* bad Boolean value */
1.31 kristaps 95: MANDOCERR_BADFONT, /* unknown font */
1.32 kristaps 96: MANDOCERR_BADSTANDARD, /* unknown standard specifier */
97: MANDOCERR_BADWIDTH, /* bad width argument */
98:
99: /* related to plain text */
100: MANDOCERR_NOBLANKLN, /* blank line in non-literal context */
101: MANDOCERR_BADTAB, /* tab in non-literal context */
102: MANDOCERR_EOLNSPACE, /* end of line whitespace */
1.5 kristaps 103: MANDOCERR_BADCOMMENT, /* bad comment style */
1.32 kristaps 104: MANDOCERR_BADESCAPE, /* unknown escape sequence */
105: MANDOCERR_BADQUOTE, /* unterminated quoted string */
1.79 kristaps 106:
107: /* related to equations */
108: MANDOCERR_EQNQUOTE, /* unexpected literal in equation */
1.32 kristaps 109:
110: MANDOCERR_ERROR, /* ===== start of errors ===== */
1.80 kristaps 111:
112: /* related to equations */
1.84 kristaps 113: MANDOCERR_EQNNSCOPE, /* unexpected equation scope closure*/
114: MANDOCERR_EQNSCOPE, /* equation scope open on exit */
1.89 kristaps 115: MANDOCERR_EQNBADSCOPE, /* overlapping equation scopes */
116: MANDOCERR_EQNEOF, /* unexpected end of equation */
117: MANDOCERR_EQNSYNT, /* equation syntax error */
1.32 kristaps 118:
1.42 kristaps 119: /* related to tables */
1.36 kristaps 120: MANDOCERR_TBL, /* bad table syntax */
121: MANDOCERR_TBLOPT, /* bad table option */
1.37 kristaps 122: MANDOCERR_TBLLAYOUT, /* bad table layout */
123: MANDOCERR_TBLNOLAYOUT, /* no table layout cells specified */
1.39 kristaps 124: MANDOCERR_TBLNODATA, /* no table data cells specified */
1.47 kristaps 125: MANDOCERR_TBLIGNDATA, /* ignore data in cell */
1.48 kristaps 126: MANDOCERR_TBLBLOCK, /* data block still open */
1.50 kristaps 127: MANDOCERR_TBLEXTRADAT, /* ignoring extra data cells */
1.42 kristaps 128:
1.33 schwarze 129: MANDOCERR_ROFFLOOP, /* input stack limit exceeded, infinite loop? */
1.32 kristaps 130: MANDOCERR_BADCHAR, /* skipping bad character */
1.52 schwarze 131: MANDOCERR_NAMESC, /* escaped character not allowed in a name */
1.103 schwarze 132: MANDOCERR_NONAME, /* manual name not yet set */
1.32 kristaps 133: MANDOCERR_NOTEXT, /* skipping text before the first section header */
134: MANDOCERR_MACRO, /* skipping unknown macro */
1.25 kristaps 135: MANDOCERR_REQUEST, /* NOT IMPLEMENTED: skipping request */
1.5 kristaps 136: MANDOCERR_ARGCOUNT, /* argument count wrong */
1.32 kristaps 137: MANDOCERR_NOSCOPE, /* skipping end of block that is not open */
1.30 kristaps 138: MANDOCERR_SCOPEBROKEN, /* missing end of block */
1.19 schwarze 139: MANDOCERR_SCOPEEXIT, /* scope open on exit */
1.28 kristaps 140: MANDOCERR_UNAME, /* uname(3) system call failed */
1.5 kristaps 141: /* FIXME: merge following with MANDOCERR_ARGCOUNT */
142: MANDOCERR_NOARGS, /* macro requires line argument(s) */
143: MANDOCERR_NOBODY, /* macro requires body argument(s) */
144: MANDOCERR_NOARGV, /* macro requires argument(s) */
1.10 kristaps 145: MANDOCERR_LISTTYPE, /* missing list type */
1.5 kristaps 146: MANDOCERR_ARGSLOST, /* line argument(s) will be lost */
147: MANDOCERR_BODYLOST, /* body argument(s) will be lost */
1.13 schwarze 148:
1.21 kristaps 149: MANDOCERR_FATAL, /* ===== start of fatal errors ===== */
1.32 kristaps 150:
1.66 kristaps 151: MANDOCERR_NOTMANUAL, /* manual isn't really a manual */
1.8 kristaps 152: MANDOCERR_COLUMNS, /* column syntax is inconsistent */
1.32 kristaps 153: MANDOCERR_BADDISP, /* NOT IMPLEMENTED: .Bd -file */
1.5 kristaps 154: MANDOCERR_SYNTARGVCOUNT, /* argument count wrong, violates syntax */
155: MANDOCERR_SYNTCHILD, /* child violates parent syntax */
156: MANDOCERR_SYNTARGCOUNT, /* argument count wrong, violates syntax */
1.26 kristaps 157: MANDOCERR_SOPATH, /* NOT IMPLEMENTED: .so with absolute path or ".." */
1.5 kristaps 158: MANDOCERR_NODOCBODY, /* no document body */
159: MANDOCERR_NODOCPROLOG, /* no document prologue */
1.18 schwarze 160: MANDOCERR_MEM, /* static buffer exhausted */
1.1 kristaps 161: MANDOCERR_MAX
1.43 kristaps 162: };
163:
164: struct tbl {
165: char tab; /* cell-separator */
166: char decimal; /* decimal point */
167: int linesize;
168: int opts;
169: #define TBL_OPT_CENTRE (1 << 0)
170: #define TBL_OPT_EXPAND (1 << 1)
171: #define TBL_OPT_BOX (1 << 2)
172: #define TBL_OPT_DBOX (1 << 3)
173: #define TBL_OPT_ALLBOX (1 << 4)
174: #define TBL_OPT_NOKEEP (1 << 5)
175: #define TBL_OPT_NOSPACE (1 << 6)
1.45 kristaps 176: int cols; /* number of columns */
1.38 kristaps 177: };
178:
1.41 kristaps 179: /*
180: * The head of a table specifies all of its columns. When formatting a
181: * tbl_span, iterate over these and plug in data from the tbl_span when
182: * appropriate, using tbl_cell as a guide to placement.
183: */
184: struct tbl_head {
1.45 kristaps 185: int ident; /* 0 <= unique id < cols */
1.101 schwarze 186: int vert; /* width of preceding vertical line */
1.41 kristaps 187: struct tbl_head *next;
188: struct tbl_head *prev;
189: };
190:
1.38 kristaps 191: enum tbl_cellt {
192: TBL_CELL_CENTRE, /* c, C */
193: TBL_CELL_RIGHT, /* r, R */
194: TBL_CELL_LEFT, /* l, L */
195: TBL_CELL_NUMBER, /* n, N */
196: TBL_CELL_SPAN, /* s, S */
197: TBL_CELL_LONG, /* a, A */
198: TBL_CELL_DOWN, /* ^ */
199: TBL_CELL_HORIZ, /* _, - */
200: TBL_CELL_DHORIZ, /* = */
201: TBL_CELL_MAX
202: };
203:
204: /*
205: * A cell in a layout row.
206: */
207: struct tbl_cell {
208: struct tbl_cell *next;
1.101 schwarze 209: int vert; /* width of preceding vertical line */
1.38 kristaps 210: enum tbl_cellt pos;
1.51 kristaps 211: size_t spacing;
1.38 kristaps 212: int flags;
213: #define TBL_CELL_TALIGN (1 << 0) /* t, T */
214: #define TBL_CELL_BALIGN (1 << 1) /* d, D */
215: #define TBL_CELL_BOLD (1 << 2) /* fB, B, b */
216: #define TBL_CELL_ITALIC (1 << 3) /* fI, I, i */
217: #define TBL_CELL_EQUAL (1 << 4) /* e, E */
218: #define TBL_CELL_UP (1 << 5) /* u, U */
219: #define TBL_CELL_WIGN (1 << 6) /* z, Z */
1.41 kristaps 220: struct tbl_head *head;
1.38 kristaps 221: };
222:
223: /*
224: * A layout row.
225: */
226: struct tbl_row {
227: struct tbl_row *next;
228: struct tbl_cell *first;
229: struct tbl_cell *last;
230: };
231:
1.40 kristaps 232: enum tbl_datt {
1.51 kristaps 233: TBL_DATA_NONE, /* has no data */
234: TBL_DATA_DATA, /* consists of data/string */
235: TBL_DATA_HORIZ, /* horizontal line */
236: TBL_DATA_DHORIZ, /* double-horizontal line */
237: TBL_DATA_NHORIZ, /* squeezed horizontal line */
238: TBL_DATA_NDHORIZ /* squeezed double-horizontal line */
1.40 kristaps 239: };
240:
1.38 kristaps 241: /*
242: * A cell within a row of data. The "string" field contains the actual
243: * string value that's in the cell. The rest is layout.
244: */
245: struct tbl_dat {
1.51 kristaps 246: struct tbl_cell *layout; /* layout cell */
1.50 kristaps 247: int spans; /* how many spans follow */
1.38 kristaps 248: struct tbl_dat *next;
1.51 kristaps 249: char *string; /* data (NULL if not TBL_DATA_DATA) */
1.40 kristaps 250: enum tbl_datt pos;
251: };
252:
253: enum tbl_spant {
254: TBL_SPAN_DATA, /* span consists of data */
255: TBL_SPAN_HORIZ, /* span is horizontal line */
256: TBL_SPAN_DHORIZ /* span is double horizontal line */
1.38 kristaps 257: };
258:
259: /*
260: * A row of data in a table.
261: */
262: struct tbl_span {
1.44 kristaps 263: struct tbl *tbl;
264: struct tbl_head *head;
1.51 kristaps 265: struct tbl_row *layout; /* layout row */
1.38 kristaps 266: struct tbl_dat *first;
267: struct tbl_dat *last;
1.56 kristaps 268: int line; /* parse line */
1.44 kristaps 269: int flags;
270: #define TBL_SPAN_FIRST (1 << 0)
271: #define TBL_SPAN_LAST (1 << 1)
1.40 kristaps 272: enum tbl_spant pos;
1.38 kristaps 273: struct tbl_span *next;
1.55 kristaps 274: };
275:
1.83 kristaps 276: enum eqn_boxt {
1.84 kristaps 277: EQN_ROOT, /* root of parse tree */
278: EQN_TEXT, /* text (number, variable, whatever) */
1.90 kristaps 279: EQN_SUBEXPR, /* nested `eqn' subexpression */
1.92 kristaps 280: EQN_LIST, /* subexpressions list */
281: EQN_MATRIX /* matrix subexpression */
1.83 kristaps 282: };
283:
1.85 kristaps 284: enum eqn_markt {
285: EQNMARK_NONE = 0,
286: EQNMARK_DOT,
287: EQNMARK_DOTDOT,
288: EQNMARK_HAT,
289: EQNMARK_TILDE,
290: EQNMARK_VEC,
291: EQNMARK_DYAD,
292: EQNMARK_BAR,
293: EQNMARK_UNDER,
294: EQNMARK__MAX
295: };
296:
1.86 kristaps 297: enum eqn_fontt {
298: EQNFONT_NONE = 0,
299: EQNFONT_ROMAN,
300: EQNFONT_BOLD,
1.93 kristaps 301: EQNFONT_FAT,
1.86 kristaps 302: EQNFONT_ITALIC,
303: EQNFONT__MAX
304: };
305:
1.87 kristaps 306: enum eqn_post {
307: EQNPOS_NONE = 0,
308: EQNPOS_OVER,
309: EQNPOS_SUP,
310: EQNPOS_SUB,
311: EQNPOS_TO,
312: EQNPOS_FROM,
313: EQNPOS__MAX
314: };
315:
1.89 kristaps 316: enum eqn_pilet {
317: EQNPILE_NONE = 0,
1.95 kristaps 318: EQNPILE_PILE,
1.89 kristaps 319: EQNPILE_CPILE,
320: EQNPILE_RPILE,
321: EQNPILE_LPILE,
1.95 kristaps 322: EQNPILE_COL,
1.92 kristaps 323: EQNPILE_CCOL,
324: EQNPILE_RCOL,
325: EQNPILE_LCOL,
1.89 kristaps 326: EQNPILE__MAX
327: };
328:
1.86 kristaps 329: /*
1.84 kristaps 330: * A "box" is a parsed mathematical expression as defined by the eqn.7
331: * grammar.
332: */
1.83 kristaps 333: struct eqn_box {
1.88 kristaps 334: int size; /* font size of expression */
335: #define EQN_DEFSIZE INT_MIN
1.84 kristaps 336: enum eqn_boxt type; /* type of node */
1.89 kristaps 337: struct eqn_box *first; /* first child node */
338: struct eqn_box *last; /* last child node */
339: struct eqn_box *next; /* node sibling */
340: struct eqn_box *parent; /* node sibling */
341: char *text; /* text (or NULL) */
342: char *left;
343: char *right;
1.87 kristaps 344: enum eqn_post pos; /* position of next box */
1.86 kristaps 345: enum eqn_markt mark; /* a mark about the box */
346: enum eqn_fontt font; /* font of box */
1.89 kristaps 347: enum eqn_pilet pile; /* equation piling */
1.83 kristaps 348: };
349:
1.87 kristaps 350: /*
351: * An equation consists of a tree of expressions starting at a given
352: * line and position.
353: */
1.55 kristaps 354: struct eqn {
1.94 kristaps 355: char *name; /* identifier (or NULL) */
1.84 kristaps 356: struct eqn_box *root; /* root mathematical expression */
1.82 kristaps 357: int ln; /* invocation line */
1.57 kristaps 358: int pos; /* invocation position */
1.16 kristaps 359: };
1.22 kristaps 360:
1.60 kristaps 361: /*
1.62 kristaps 362: * The type of parse sequence. This value is usually passed via the
363: * mandoc(1) command line of -man and -mdoc. It's almost exclusively
364: * -mandoc but the others have been retained for compatibility.
365: */
366: enum mparset {
367: MPARSE_AUTO, /* magically determine the document type */
368: MPARSE_MDOC, /* assume -mdoc */
369: MPARSE_MAN /* assume -man */
370: };
371:
1.70 kristaps 372: enum mandoc_esc {
373: ESCAPE_ERROR = 0, /* bail! unparsable escape */
374: ESCAPE_IGNORE, /* escape to be ignored */
375: ESCAPE_SPECIAL, /* a regular special character */
376: ESCAPE_FONT, /* a generic font mode */
377: ESCAPE_FONTBOLD, /* bold font mode */
378: ESCAPE_FONTITALIC, /* italic font mode */
379: ESCAPE_FONTROMAN, /* roman font mode */
380: ESCAPE_FONTPREV, /* previous font mode */
381: ESCAPE_NUMBERED, /* a numbered glyph */
1.75 kristaps 382: ESCAPE_UNICODE, /* a unicode codepoint */
1.102 schwarze 383: ESCAPE_NOSPACE, /* suppress space if the last on a line */
384: ESCAPE_SKIPCHAR /* skip the next character */
1.70 kristaps 385: };
386:
1.63 kristaps 387: typedef void (*mandocmsg)(enum mandocerr, enum mandoclevel,
388: const char *, int, int, const char *);
1.62 kristaps 389:
390: struct mparse;
1.71 kristaps 391: struct mchars;
1.62 kristaps 392: struct mdoc;
393: struct man;
1.59 kristaps 394:
1.22 kristaps 395: __BEGIN_DECLS
1.62 kristaps 396:
1.59 kristaps 397: void *mandoc_calloc(size_t, size_t);
1.96 kristaps 398: enum mandoc_esc mandoc_escape(const char **, const char **, int *);
1.59 kristaps 399: void *mandoc_malloc(size_t);
400: void *mandoc_realloc(void *, size_t);
1.78 kristaps 401: char *mandoc_strdup(const char *);
1.89 kristaps 402: char *mandoc_strndup(const char *, size_t);
1.73 kristaps 403: struct mchars *mchars_alloc(void);
1.96 kristaps 404: void mchars_free(struct mchars *);
1.74 kristaps 405: char mchars_num2char(const char *, size_t);
1.76 kristaps 406: int mchars_num2uc(const char *, size_t);
1.98 kristaps 407: int mchars_spec2cp(const struct mchars *,
1.96 kristaps 408: const char *, size_t);
1.98 kristaps 409: const char *mchars_spec2str(const struct mchars *,
1.96 kristaps 410: const char *, size_t, size_t *);
1.100 schwarze 411: struct mparse *mparse_alloc(enum mparset, enum mandoclevel,
412: mandocmsg, void *, char *);
1.96 kristaps 413: void mparse_free(struct mparse *);
414: void mparse_keep(struct mparse *);
415: enum mandoclevel mparse_readfd(struct mparse *, int, const char *);
1.99 joerg 416: enum mandoclevel mparse_readmem(struct mparse *, const void *, size_t,
417: const char *);
1.96 kristaps 418: void mparse_reset(struct mparse *);
419: void mparse_result(struct mparse *,
420: struct mdoc **, struct man **);
421: const char *mparse_getkeep(const struct mparse *);
422: const char *mparse_strerror(enum mandocerr);
423: const char *mparse_strlevel(enum mandoclevel);
1.1 kristaps 424:
425: __END_DECLS
426:
427: #endif /*!MANDOC_H*/
CVSweb