Annotation of mandoc/mandoc.h, Revision 1.127
1.127 ! schwarze 1: /* $Id: mandoc.h,v 1.126 2014/07/02 03:48:07 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 */
1.54 kristaps 77: MANDOCERR_IGNNS, /* skipping no-space macro */
1.14 schwarze 78: MANDOCERR_SCOPENEST, /* blocks badly nested */
1.32 kristaps 79: MANDOCERR_CHILD, /* child violates parent syntax */
80: MANDOCERR_NESTEDDISP, /* nested displays are not portable */
81: MANDOCERR_SCOPEREP, /* already in literal mode */
1.68 kristaps 82: MANDOCERR_LINESCOPE, /* line scope broken */
1.32 kristaps 83:
84: /* related to missing macro arguments */
85: MANDOCERR_MACROEMPTY, /* skipping empty macro */
1.46 schwarze 86: MANDOCERR_ARGCWARN, /* argument count wrong */
1.32 kristaps 87: MANDOCERR_DISPTYPE, /* missing display type */
88: MANDOCERR_LISTFIRST, /* list type must come first */
89: MANDOCERR_NOWIDTHARG, /* tag lists require a width argument */
90: MANDOCERR_FONTTYPE, /* missing font type */
1.53 kristaps 91: MANDOCERR_WNOSCOPE, /* skipping end of block that is not open */
1.13 schwarze 92:
1.32 kristaps 93: /* related to bad macro arguments */
94: MANDOCERR_IGNARGV, /* skipping argument */
95: MANDOCERR_ARGVREP, /* duplicate argument */
96: MANDOCERR_DISPREP, /* duplicate display type */
97: MANDOCERR_LISTREP, /* duplicate list type */
98: MANDOCERR_BADATT, /* unknown AT&T UNIX version */
1.5 kristaps 99: MANDOCERR_BADBOOL, /* bad Boolean value */
1.31 kristaps 100: MANDOCERR_BADFONT, /* unknown font */
1.32 kristaps 101: MANDOCERR_BADSTANDARD, /* unknown standard specifier */
102: MANDOCERR_BADWIDTH, /* bad width argument */
103:
104: /* related to plain text */
105: MANDOCERR_NOBLANKLN, /* blank line in non-literal context */
106: MANDOCERR_BADTAB, /* tab in non-literal context */
107: MANDOCERR_EOLNSPACE, /* end of line whitespace */
1.5 kristaps 108: MANDOCERR_BADCOMMENT, /* bad comment style */
1.124 schwarze 109: MANDOCERR_BADESCAPE, /* bad escape sequence */
1.32 kristaps 110: MANDOCERR_BADQUOTE, /* unterminated quoted string */
1.79 kristaps 111:
1.32 kristaps 112: MANDOCERR_ERROR, /* ===== start of errors ===== */
1.80 kristaps 113:
114: /* related to equations */
1.84 kristaps 115: MANDOCERR_EQNNSCOPE, /* unexpected equation scope closure*/
116: MANDOCERR_EQNSCOPE, /* equation scope open on exit */
1.89 kristaps 117: MANDOCERR_EQNBADSCOPE, /* overlapping equation scopes */
118: MANDOCERR_EQNEOF, /* unexpected end of equation */
119: MANDOCERR_EQNSYNT, /* equation syntax error */
1.32 kristaps 120:
1.42 kristaps 121: /* related to tables */
1.36 kristaps 122: MANDOCERR_TBL, /* bad table syntax */
123: MANDOCERR_TBLOPT, /* bad table option */
1.37 kristaps 124: MANDOCERR_TBLLAYOUT, /* bad table layout */
125: MANDOCERR_TBLNOLAYOUT, /* no table layout cells specified */
1.39 kristaps 126: MANDOCERR_TBLNODATA, /* no table data cells specified */
1.47 kristaps 127: MANDOCERR_TBLIGNDATA, /* ignore data in cell */
1.48 kristaps 128: MANDOCERR_TBLBLOCK, /* data block still open */
1.50 kristaps 129: MANDOCERR_TBLEXTRADAT, /* ignoring extra data cells */
1.42 kristaps 130:
1.33 schwarze 131: MANDOCERR_ROFFLOOP, /* input stack limit exceeded, infinite loop? */
1.32 kristaps 132: MANDOCERR_BADCHAR, /* skipping bad character */
1.52 schwarze 133: MANDOCERR_NAMESC, /* escaped character not allowed in a name */
1.103 schwarze 134: MANDOCERR_NONAME, /* manual name not yet set */
1.125 schwarze 135: MANDOCERR_NOTEXT, /* skipping text before first section header */
1.32 kristaps 136: MANDOCERR_MACRO, /* skipping unknown macro */
1.25 kristaps 137: MANDOCERR_REQUEST, /* NOT IMPLEMENTED: skipping request */
1.5 kristaps 138: MANDOCERR_ARGCOUNT, /* argument count wrong */
1.106 schwarze 139: MANDOCERR_STRAYTA, /* skipping column outside column list */
1.32 kristaps 140: MANDOCERR_NOSCOPE, /* skipping end of block that is not open */
1.30 kristaps 141: MANDOCERR_SCOPEBROKEN, /* missing end of block */
1.19 schwarze 142: MANDOCERR_SCOPEEXIT, /* scope open on exit */
1.28 kristaps 143: MANDOCERR_UNAME, /* uname(3) system call failed */
1.5 kristaps 144: /* FIXME: merge following with MANDOCERR_ARGCOUNT */
145: MANDOCERR_NOARGS, /* macro requires line argument(s) */
146: MANDOCERR_NOBODY, /* macro requires body argument(s) */
147: MANDOCERR_NOARGV, /* macro requires argument(s) */
1.108 schwarze 148: MANDOCERR_NUMERIC, /* request requires a numeric argument */
1.10 kristaps 149: MANDOCERR_LISTTYPE, /* missing list type */
1.5 kristaps 150: MANDOCERR_ARGSLOST, /* line argument(s) will be lost */
1.13 schwarze 151:
1.21 kristaps 152: MANDOCERR_FATAL, /* ===== start of fatal errors ===== */
1.32 kristaps 153:
1.113 schwarze 154: MANDOCERR_TOOLARGE, /* input too large */
1.124 schwarze 155: MANDOCERR_NOTMANUAL, /* not a manual */
1.8 kristaps 156: MANDOCERR_COLUMNS, /* column syntax is inconsistent */
1.32 kristaps 157: MANDOCERR_BADDISP, /* NOT IMPLEMENTED: .Bd -file */
1.5 kristaps 158: MANDOCERR_SYNTARGVCOUNT, /* argument count wrong, violates syntax */
159: MANDOCERR_SYNTCHILD, /* child violates parent syntax */
160: MANDOCERR_SYNTARGCOUNT, /* argument count wrong, violates syntax */
1.123 schwarze 161: MANDOCERR_SO_PATH, /* NOT IMPLEMENTED: .so with absolute path or ".." */
162: MANDOCERR_SO_FAIL, /* .so request failed */
1.5 kristaps 163: MANDOCERR_NODOCPROLOG, /* no document prologue */
1.18 schwarze 164: MANDOCERR_MEM, /* static buffer exhausted */
1.113 schwarze 165:
166: /* ===== system errors ===== */
167:
168: MANDOCERR_SYSOPEN, /* cannot open file */
169: MANDOCERR_SYSSTAT, /* cannot stat file */
170: MANDOCERR_SYSREAD, /* cannot read file */
171:
1.1 kristaps 172: MANDOCERR_MAX
1.43 kristaps 173: };
174:
1.107 schwarze 175: struct tbl_opts {
1.43 kristaps 176: char tab; /* cell-separator */
177: char decimal; /* decimal point */
178: int linesize;
179: int opts;
180: #define TBL_OPT_CENTRE (1 << 0)
181: #define TBL_OPT_EXPAND (1 << 1)
182: #define TBL_OPT_BOX (1 << 2)
183: #define TBL_OPT_DBOX (1 << 3)
184: #define TBL_OPT_ALLBOX (1 << 4)
185: #define TBL_OPT_NOKEEP (1 << 5)
186: #define TBL_OPT_NOSPACE (1 << 6)
1.45 kristaps 187: int cols; /* number of columns */
1.38 kristaps 188: };
189:
1.41 kristaps 190: /*
191: * The head of a table specifies all of its columns. When formatting a
192: * tbl_span, iterate over these and plug in data from the tbl_span when
193: * appropriate, using tbl_cell as a guide to placement.
194: */
195: struct tbl_head {
1.45 kristaps 196: int ident; /* 0 <= unique id < cols */
1.101 schwarze 197: int vert; /* width of preceding vertical line */
1.41 kristaps 198: struct tbl_head *next;
199: struct tbl_head *prev;
200: };
201:
1.38 kristaps 202: enum tbl_cellt {
203: TBL_CELL_CENTRE, /* c, C */
204: TBL_CELL_RIGHT, /* r, R */
205: TBL_CELL_LEFT, /* l, L */
206: TBL_CELL_NUMBER, /* n, N */
207: TBL_CELL_SPAN, /* s, S */
208: TBL_CELL_LONG, /* a, A */
209: TBL_CELL_DOWN, /* ^ */
210: TBL_CELL_HORIZ, /* _, - */
211: TBL_CELL_DHORIZ, /* = */
212: TBL_CELL_MAX
213: };
214:
215: /*
216: * A cell in a layout row.
217: */
218: struct tbl_cell {
219: struct tbl_cell *next;
1.101 schwarze 220: int vert; /* width of preceding vertical line */
1.38 kristaps 221: enum tbl_cellt pos;
1.51 kristaps 222: size_t spacing;
1.38 kristaps 223: int flags;
224: #define TBL_CELL_TALIGN (1 << 0) /* t, T */
225: #define TBL_CELL_BALIGN (1 << 1) /* d, D */
226: #define TBL_CELL_BOLD (1 << 2) /* fB, B, b */
227: #define TBL_CELL_ITALIC (1 << 3) /* fI, I, i */
228: #define TBL_CELL_EQUAL (1 << 4) /* e, E */
229: #define TBL_CELL_UP (1 << 5) /* u, U */
230: #define TBL_CELL_WIGN (1 << 6) /* z, Z */
1.41 kristaps 231: struct tbl_head *head;
1.38 kristaps 232: };
233:
234: /*
235: * A layout row.
236: */
237: struct tbl_row {
238: struct tbl_row *next;
239: struct tbl_cell *first;
240: struct tbl_cell *last;
1.119 schwarze 241: int vert; /* trailing vertical line */
1.38 kristaps 242: };
243:
1.40 kristaps 244: enum tbl_datt {
1.51 kristaps 245: TBL_DATA_NONE, /* has no data */
246: TBL_DATA_DATA, /* consists of data/string */
247: TBL_DATA_HORIZ, /* horizontal line */
248: TBL_DATA_DHORIZ, /* double-horizontal line */
249: TBL_DATA_NHORIZ, /* squeezed horizontal line */
250: TBL_DATA_NDHORIZ /* squeezed double-horizontal line */
1.40 kristaps 251: };
252:
1.38 kristaps 253: /*
254: * A cell within a row of data. The "string" field contains the actual
255: * string value that's in the cell. The rest is layout.
256: */
257: struct tbl_dat {
1.51 kristaps 258: struct tbl_cell *layout; /* layout cell */
1.50 kristaps 259: int spans; /* how many spans follow */
1.38 kristaps 260: struct tbl_dat *next;
1.51 kristaps 261: char *string; /* data (NULL if not TBL_DATA_DATA) */
1.40 kristaps 262: enum tbl_datt pos;
263: };
264:
265: enum tbl_spant {
266: TBL_SPAN_DATA, /* span consists of data */
267: TBL_SPAN_HORIZ, /* span is horizontal line */
268: TBL_SPAN_DHORIZ /* span is double horizontal line */
1.38 kristaps 269: };
270:
271: /*
272: * A row of data in a table.
273: */
274: struct tbl_span {
1.107 schwarze 275: struct tbl_opts *opts;
1.44 kristaps 276: struct tbl_head *head;
1.51 kristaps 277: struct tbl_row *layout; /* layout row */
1.38 kristaps 278: struct tbl_dat *first;
279: struct tbl_dat *last;
1.56 kristaps 280: int line; /* parse line */
1.44 kristaps 281: int flags;
282: #define TBL_SPAN_FIRST (1 << 0)
283: #define TBL_SPAN_LAST (1 << 1)
1.40 kristaps 284: enum tbl_spant pos;
1.38 kristaps 285: struct tbl_span *next;
1.55 kristaps 286: };
287:
1.83 kristaps 288: enum eqn_boxt {
1.84 kristaps 289: EQN_ROOT, /* root of parse tree */
290: EQN_TEXT, /* text (number, variable, whatever) */
1.90 kristaps 291: EQN_SUBEXPR, /* nested `eqn' subexpression */
1.92 kristaps 292: EQN_LIST, /* subexpressions list */
293: EQN_MATRIX /* matrix subexpression */
1.83 kristaps 294: };
295:
1.85 kristaps 296: enum eqn_markt {
297: EQNMARK_NONE = 0,
298: EQNMARK_DOT,
299: EQNMARK_DOTDOT,
300: EQNMARK_HAT,
301: EQNMARK_TILDE,
302: EQNMARK_VEC,
303: EQNMARK_DYAD,
304: EQNMARK_BAR,
305: EQNMARK_UNDER,
306: EQNMARK__MAX
307: };
308:
1.86 kristaps 309: enum eqn_fontt {
310: EQNFONT_NONE = 0,
311: EQNFONT_ROMAN,
312: EQNFONT_BOLD,
1.93 kristaps 313: EQNFONT_FAT,
1.86 kristaps 314: EQNFONT_ITALIC,
315: EQNFONT__MAX
316: };
317:
1.87 kristaps 318: enum eqn_post {
319: EQNPOS_NONE = 0,
320: EQNPOS_OVER,
321: EQNPOS_SUP,
322: EQNPOS_SUB,
323: EQNPOS_TO,
324: EQNPOS_FROM,
325: EQNPOS__MAX
326: };
327:
1.89 kristaps 328: enum eqn_pilet {
329: EQNPILE_NONE = 0,
1.95 kristaps 330: EQNPILE_PILE,
1.89 kristaps 331: EQNPILE_CPILE,
332: EQNPILE_RPILE,
333: EQNPILE_LPILE,
1.95 kristaps 334: EQNPILE_COL,
1.92 kristaps 335: EQNPILE_CCOL,
336: EQNPILE_RCOL,
337: EQNPILE_LCOL,
1.89 kristaps 338: EQNPILE__MAX
339: };
340:
1.86 kristaps 341: /*
1.84 kristaps 342: * A "box" is a parsed mathematical expression as defined by the eqn.7
343: * grammar.
344: */
1.83 kristaps 345: struct eqn_box {
1.88 kristaps 346: int size; /* font size of expression */
347: #define EQN_DEFSIZE INT_MIN
1.84 kristaps 348: enum eqn_boxt type; /* type of node */
1.89 kristaps 349: struct eqn_box *first; /* first child node */
350: struct eqn_box *last; /* last child node */
351: struct eqn_box *next; /* node sibling */
352: struct eqn_box *parent; /* node sibling */
353: char *text; /* text (or NULL) */
354: char *left;
355: char *right;
1.87 kristaps 356: enum eqn_post pos; /* position of next box */
1.86 kristaps 357: enum eqn_markt mark; /* a mark about the box */
358: enum eqn_fontt font; /* font of box */
1.89 kristaps 359: enum eqn_pilet pile; /* equation piling */
1.83 kristaps 360: };
361:
1.87 kristaps 362: /*
363: * An equation consists of a tree of expressions starting at a given
1.120 schwarze 364: * line and position.
1.87 kristaps 365: */
1.55 kristaps 366: struct eqn {
1.94 kristaps 367: char *name; /* identifier (or NULL) */
1.84 kristaps 368: struct eqn_box *root; /* root mathematical expression */
1.82 kristaps 369: int ln; /* invocation line */
1.57 kristaps 370: int pos; /* invocation position */
1.16 kristaps 371: };
1.22 kristaps 372:
1.60 kristaps 373: /*
1.116 schwarze 374: * Parse options.
375: */
376: #define MPARSE_MDOC 1 /* assume -mdoc */
377: #define MPARSE_MAN 2 /* assume -man */
378: #define MPARSE_SO 4 /* honour .so requests */
379: #define MPARSE_QUICK 8 /* abort the parse early */
1.62 kristaps 380:
1.70 kristaps 381: enum mandoc_esc {
382: ESCAPE_ERROR = 0, /* bail! unparsable escape */
383: ESCAPE_IGNORE, /* escape to be ignored */
384: ESCAPE_SPECIAL, /* a regular special character */
385: ESCAPE_FONT, /* a generic font mode */
386: ESCAPE_FONTBOLD, /* bold font mode */
387: ESCAPE_FONTITALIC, /* italic font mode */
1.109 schwarze 388: ESCAPE_FONTBI, /* bold italic font mode */
1.70 kristaps 389: ESCAPE_FONTROMAN, /* roman font mode */
390: ESCAPE_FONTPREV, /* previous font mode */
391: ESCAPE_NUMBERED, /* a numbered glyph */
1.75 kristaps 392: ESCAPE_UNICODE, /* a unicode codepoint */
1.102 schwarze 393: ESCAPE_NOSPACE, /* suppress space if the last on a line */
394: ESCAPE_SKIPCHAR /* skip the next character */
1.70 kristaps 395: };
396:
1.63 kristaps 397: typedef void (*mandocmsg)(enum mandocerr, enum mandoclevel,
398: const char *, int, int, const char *);
1.62 kristaps 399:
400: struct mparse;
1.71 kristaps 401: struct mchars;
1.62 kristaps 402: struct mdoc;
403: struct man;
1.59 kristaps 404:
1.22 kristaps 405: __BEGIN_DECLS
1.62 kristaps 406:
1.112 schwarze 407: enum mandoc_esc mandoc_escape(const char **, const char **, int *);
1.73 kristaps 408: struct mchars *mchars_alloc(void);
1.96 kristaps 409: void mchars_free(struct mchars *);
1.120 schwarze 410: char mchars_num2char(const char *, size_t);
1.76 kristaps 411: int mchars_num2uc(const char *, size_t);
1.120 schwarze 412: int mchars_spec2cp(const struct mchars *,
1.96 kristaps 413: const char *, size_t);
1.120 schwarze 414: const char *mchars_spec2str(const struct mchars *,
1.96 kristaps 415: const char *, size_t, size_t *);
1.116 schwarze 416: struct mparse *mparse_alloc(int, enum mandoclevel, mandocmsg, char *);
1.96 kristaps 417: void mparse_free(struct mparse *);
418: void mparse_keep(struct mparse *);
419: enum mandoclevel mparse_readfd(struct mparse *, int, const char *);
1.99 joerg 420: enum mandoclevel mparse_readmem(struct mparse *, const void *, size_t,
421: const char *);
1.96 kristaps 422: void mparse_reset(struct mparse *);
1.120 schwarze 423: void mparse_result(struct mparse *,
1.117 schwarze 424: struct mdoc **, struct man **, char **);
1.96 kristaps 425: const char *mparse_getkeep(const struct mparse *);
426: const char *mparse_strerror(enum mandocerr);
427: const char *mparse_strlevel(enum mandoclevel);
1.1 kristaps 428:
429: __END_DECLS
430:
431: #endif /*!MANDOC_H*/
CVSweb