mandoc.h

   1 /*      $Id: mandoc.h,v 1.160 2014/10/10 15:26:29 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2010, 2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
   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
  20
  21 #define ASCII_NBRSP      31  /* non-breaking space */
  22 #define ASCII_HYPH       30  /* breakable hyphen */
  23 #define ASCII_BREAK      29  /* breakable zero-width space */
  24
  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  */
  31 enum    mandoclevel {
  32         MANDOCLEVEL_OK = 0,
  33         MANDOCLEVEL_RESERVED,
  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 */
  39         MANDOCLEVEL_MAX
  40 };
  41
  42 /*
  43  * All possible things that can go wrong within a parse, be it libroff,
  44  * libmdoc, or libman.
  45  */
  46 enum    mandocerr {
  47         MANDOCERR_OK,
  48
  49         MANDOCERR_WARNING, /* ===== start of warnings ===== */
  50
  51         /* related to the prologue */
  52         MANDOCERR_DT_NOTITLE, /* missing manual title, using UNTITLED: line */
  53         MANDOCERR_TH_NOTITLE, /* missing manual title, using "": [macro] */
  54         MANDOCERR_TITLE_CASE, /* lower case character in document title */
  55         MANDOCERR_MSEC_MISSING, /* missing manual section, using "": macro */
  56         MANDOCERR_MSEC_BAD, /* unknown manual section: Dt ... section */
  57         MANDOCERR_ARCH_BAD, /* unknown manual volume or arch: Dt ... volume */
  58         MANDOCERR_DATE_MISSING, /* missing date, using today's date */
  59         MANDOCERR_DATE_BAD, /* cannot parse date, using it verbatim: date */
  60         MANDOCERR_OS_MISSING, /* missing Os macro, using "" */
  61         MANDOCERR_PROLOG_REP, /* duplicate prologue macro: macro */
  62         MANDOCERR_PROLOG_LATE, /* late prologue macro: macro */
  63         MANDOCERR_DT_LATE, /* skipping late title macro: Dt args */
  64         MANDOCERR_PROLOG_ORDER, /* prologue macros out of order: macros */
  65
  66         /* related to document structure */
  67         MANDOCERR_SO, /* .so is fragile, better use ln(1): so path */
  68         MANDOCERR_DOC_EMPTY, /* no document body */
  69         MANDOCERR_SEC_BEFORE, /* content before first section header: macro */
  70         MANDOCERR_NAMESEC_FIRST, /* first section is not NAME: Sh title */
  71         MANDOCERR_NAMESEC_BAD, /* bad NAME section contents: macro */
  72         MANDOCERR_SEC_ORDER, /* sections out of conventional order: Sh title */
  73         MANDOCERR_SEC_REP, /* duplicate section title: Sh title */
  74         MANDOCERR_SEC_MSEC, /* unexpected section: Sh title for ... only */
  75         MANDOCERR_XR_ORDER, /* unusual Xr order: ... after ... */
  76         MANDOCERR_XR_PUNCT, /* unusual Xr punctuation: ... after ... */
  77         MANDOCERR_AN_MISSING, /* AUTHORS section without An macro */
  78
  79         /* related to macros and nesting */
  80         MANDOCERR_MACRO_OBS, /* obsolete macro: macro */
  81         MANDOCERR_PAR_SKIP, /* skipping paragraph macro: macro ... */
  82         MANDOCERR_PAR_MOVE, /* moving paragraph macro out of list: macro */
  83         MANDOCERR_NS_SKIP, /* skipping no-space macro */
  84         MANDOCERR_BLK_NEST, /* blocks badly nested: macro ... */
  85         MANDOCERR_BD_NEST, /* nested displays are not portable: macro ... */
  86         MANDOCERR_BL_MOVE, /* moving content out of list: macro */
  87         MANDOCERR_VT_CHILD, /* .Vt block has child macro: macro */
  88         MANDOCERR_FI_SKIP, /* fill mode already enabled, skipping: fi */
  89         MANDOCERR_NF_SKIP, /* fill mode already disabled, skipping: nf */
  90         MANDOCERR_BLK_LINE, /* line scope broken: macro breaks macro */
  91
  92         /* related to missing arguments */
  93         MANDOCERR_REQ_EMPTY, /* skipping empty request: request */
  94         MANDOCERR_COND_EMPTY, /* conditional request controls empty scope */
  95         MANDOCERR_MACRO_EMPTY, /* skipping empty macro: macro */
  96         MANDOCERR_ARG_EMPTY, /* empty argument, using 0n: macro arg */
  97         MANDOCERR_ARGCWARN, /* argument count wrong */
  98         MANDOCERR_BD_NOTYPE, /* missing display type, using -ragged: Bd */
  99         MANDOCERR_BL_LATETYPE, /* list type is not the first argument: Bl arg */
 100         MANDOCERR_BL_NOWIDTH, /* missing -width in -tag list, using 8n */
 101         MANDOCERR_EX_NONAME, /* missing utility name, using "": Ex */
 102         MANDOCERR_IT_NOHEAD, /* empty head in list item: Bl -type It */
 103         MANDOCERR_IT_NOBODY, /* empty list item: Bl -type It */
 104         MANDOCERR_BF_NOFONT, /* missing font type, using \fR: Bf */
 105         MANDOCERR_BF_BADFONT, /* unknown font type, using \fR: Bf font */
 106         MANDOCERR_ARG_STD, /* missing -std argument, adding it: macro */
 107
 108         /* related to bad arguments */
 109         MANDOCERR_ARG_QUOTE, /* unterminated quoted argument */
 110         MANDOCERR_ARG_REP, /* duplicate argument: macro arg */
 111         MANDOCERR_AN_REP, /* skipping duplicate argument: An -arg */
 112         MANDOCERR_BD_REP, /* skipping duplicate display type: Bd -type */
 113         MANDOCERR_BL_REP, /* skipping duplicate list type: Bl -type */
 114         MANDOCERR_BL_SKIPW, /* skipping -width argument: Bl -type */
 115         MANDOCERR_AT_BAD, /* unknown AT&T UNIX version: At version */
 116         MANDOCERR_FA_COMMA, /* comma in function argument: arg */
 117         MANDOCERR_RS_BAD, /* invalid content in Rs block: macro */
 118         MANDOCERR_SM_BAD, /* invalid Boolean argument: macro arg */
 119         MANDOCERR_FT_BAD, /* unknown font, skipping request: ft font */
 120
 121         /* related to plain text */
 122         MANDOCERR_FI_BLANK, /* blank line in fill mode, using .sp */
 123         MANDOCERR_FI_TAB, /* tab in filled text */
 124         MANDOCERR_SPACE_EOL, /* whitespace at end of input line */
 125         MANDOCERR_COMMENT_BAD, /* bad comment style */
 126         MANDOCERR_ESC_BAD, /* invalid escape sequence: esc */
 127         MANDOCERR_STR_UNDEF, /* undefined string, using "": name */
 128
 129         MANDOCERR_ERROR, /* ===== start of errors ===== */
 130
 131         /* related to equations */
 132         MANDOCERR_EQNNSCOPE, /* unexpected equation scope closure*/
 133         MANDOCERR_EQNSCOPE, /* equation scope open on exit */
 134         MANDOCERR_EQNBADSCOPE, /* overlapping equation scopes */
 135         MANDOCERR_EQNEOF, /* unexpected end of equation */
 136         MANDOCERR_EQNSYNT, /* equation syntax error */
 137
 138         /* related to tables */
 139         MANDOCERR_TBL, /* bad table syntax */
 140         MANDOCERR_TBLOPT, /* bad table option */
 141         MANDOCERR_TBLLAYOUT, /* bad table layout */
 142         MANDOCERR_TBLNOLAYOUT, /* no table layout cells specified */
 143         MANDOCERR_TBLNODATA, /* no table data cells specified */
 144         MANDOCERR_TBLIGNDATA, /* ignore data in cell */
 145         MANDOCERR_TBLBLOCK, /* data block still open */
 146         MANDOCERR_TBLEXTRADAT, /* ignoring extra data cells */
 147
 148         /* related to document structure and macros */
 149         MANDOCERR_ROFFLOOP, /* input stack limit exceeded, infinite loop? */
 150         MANDOCERR_BADCHAR, /* skipping bad character: number */
 151         MANDOCERR_MACRO, /* skipping unknown macro: macro */
 152         MANDOCERR_IT_STRAY, /* skipping item outside list: It ... */
 153         MANDOCERR_TA_STRAY, /* skipping column outside column list: Ta */
 154         MANDOCERR_BLK_NOTOPEN, /* skipping end of block that is not open */
 155         MANDOCERR_BLK_BROKEN, /* inserting missing end of block: macro ... */
 156         MANDOCERR_BLK_NOEND, /* appending missing end of block: macro */
 157
 158         /* related to request and macro arguments */
 159         MANDOCERR_NAMESC, /* escaped character not allowed in a name: name */
 160         MANDOCERR_ARGCOUNT, /* argument count wrong */
 161         MANDOCERR_BL_NOTYPE, /* missing list type, using -item: Bl */
 162         MANDOCERR_NM_NONAME, /* missing manual name, using "": Nm */
 163         MANDOCERR_OS_UNAME, /* uname(3) system call failed, using UNKNOWN */
 164         MANDOCERR_ST_BAD, /* unknown standard specifier: St standard */
 165         MANDOCERR_IT_NONUM, /* skipping request without numeric argument */
 166         MANDOCERR_ARG_SKIP, /* skipping all arguments: macro args */
 167         MANDOCERR_ARG_EXCESS, /* skipping excess arguments: macro ... args */
 168
 169         MANDOCERR_FATAL, /* ===== start of fatal errors ===== */
 170
 171         MANDOCERR_TOOLARGE, /* input too large */
 172         MANDOCERR_BD_FILE, /* NOT IMPLEMENTED: Bd -file */
 173         MANDOCERR_SO_PATH, /* NOT IMPLEMENTED: .so with absolute path or ".." */
 174         MANDOCERR_SO_FAIL, /* .so request failed */
 175
 176         /* ===== system errors ===== */
 177
 178         MANDOCERR_SYSDUP, /* cannot dup file descriptor */
 179         MANDOCERR_SYSEXEC, /* cannot exec */
 180         MANDOCERR_SYSEXIT, /* gunzip failed with code */
 181         MANDOCERR_SYSFORK, /* cannot fork */
 182         MANDOCERR_SYSOPEN, /* cannot open file */
 183         MANDOCERR_SYSPIPE, /* cannot open pipe */
 184         MANDOCERR_SYSREAD, /* cannot read file */
 185         MANDOCERR_SYSSIG, /* gunzip died from signal */
 186         MANDOCERR_SYSSTAT, /* cannot stat file */
 187         MANDOCERR_SYSWAIT, /* wait failed */
 188
 189         MANDOCERR_MAX
 190 };
 191
 192 struct  tbl_opts {
 193         char              tab; /* cell-separator */
 194         char              decimal; /* decimal point */
 195         int               linesize;
 196         int               opts;
 197 #define TBL_OPT_CENTRE   (1 << 0)
 198 #define TBL_OPT_EXPAND   (1 << 1)
 199 #define TBL_OPT_BOX      (1 << 2)
 200 #define TBL_OPT_DBOX     (1 << 3)
 201 #define TBL_OPT_ALLBOX   (1 << 4)
 202 #define TBL_OPT_NOKEEP   (1 << 5)
 203 #define TBL_OPT_NOSPACE  (1 << 6)
 204         int               cols; /* number of columns */
 205 };
 206
 207 /*
 208  * The head of a table specifies all of its columns.  When formatting a
 209  * tbl_span, iterate over these and plug in data from the tbl_span when
 210  * appropriate, using tbl_cell as a guide to placement.
 211  */
 212 struct  tbl_head {
 213         int               ident; /* 0 <= unique id < cols */
 214         int               vert; /* width of preceding vertical line */
 215         struct tbl_head  *next;
 216         struct tbl_head  *prev;
 217 };
 218
 219 enum    tbl_cellt {
 220         TBL_CELL_CENTRE, /* c, C */
 221         TBL_CELL_RIGHT, /* r, R */
 222         TBL_CELL_LEFT, /* l, L */
 223         TBL_CELL_NUMBER, /* n, N */
 224         TBL_CELL_SPAN, /* s, S */
 225         TBL_CELL_LONG, /* a, A */
 226         TBL_CELL_DOWN, /* ^ */
 227         TBL_CELL_HORIZ, /* _, - */
 228         TBL_CELL_DHORIZ, /* = */
 229         TBL_CELL_MAX
 230 };
 231
 232 /*
 233  * A cell in a layout row.
 234  */
 235 struct  tbl_cell {
 236         struct tbl_cell  *next;
 237         int               vert; /* width of preceding vertical line */
 238         enum tbl_cellt    pos;
 239         size_t            spacing;
 240         int               flags;
 241 #define TBL_CELL_TALIGN  (1 << 0) /* t, T */
 242 #define TBL_CELL_BALIGN  (1 << 1) /* d, D */
 243 #define TBL_CELL_BOLD    (1 << 2) /* fB, B, b */
 244 #define TBL_CELL_ITALIC  (1 << 3) /* fI, I, i */
 245 #define TBL_CELL_EQUAL   (1 << 4) /* e, E */
 246 #define TBL_CELL_UP      (1 << 5) /* u, U */
 247 #define TBL_CELL_WIGN    (1 << 6) /* z, Z */
 248         struct tbl_head  *head;
 249 };
 250
 251 /*
 252  * A layout row.
 253  */
 254 struct  tbl_row {
 255         struct tbl_row   *next;
 256         struct tbl_cell  *first;
 257         struct tbl_cell  *last;
 258         int               vert; /* trailing vertical line */
 259 };
 260
 261 enum    tbl_datt {
 262         TBL_DATA_NONE, /* has no data */
 263         TBL_DATA_DATA, /* consists of data/string */
 264         TBL_DATA_HORIZ, /* horizontal line */
 265         TBL_DATA_DHORIZ, /* double-horizontal line */
 266         TBL_DATA_NHORIZ, /* squeezed horizontal line */
 267         TBL_DATA_NDHORIZ /* squeezed double-horizontal line */
 268 };
 269
 270 /*
 271  * A cell within a row of data.  The "string" field contains the actual
 272  * string value that's in the cell.  The rest is layout.
 273  */
 274 struct  tbl_dat {
 275         struct tbl_cell  *layout; /* layout cell */
 276         int               spans; /* how many spans follow */
 277         struct tbl_dat   *next;
 278         char             *string; /* data (NULL if not TBL_DATA_DATA) */
 279         enum tbl_datt     pos;
 280 };
 281
 282 enum    tbl_spant {
 283         TBL_SPAN_DATA, /* span consists of data */
 284         TBL_SPAN_HORIZ, /* span is horizontal line */
 285         TBL_SPAN_DHORIZ /* span is double horizontal line */
 286 };
 287
 288 /*
 289  * A row of data in a table.
 290  */
 291 struct  tbl_span {
 292         struct tbl_opts  *opts;
 293         struct tbl_head  *head;
 294         struct tbl_row   *layout; /* layout row */
 295         struct tbl_dat   *first;
 296         struct tbl_dat   *last;
 297         int               line; /* parse line */
 298         int               flags;
 299 #define TBL_SPAN_FIRST   (1 << 0)
 300 #define TBL_SPAN_LAST    (1 << 1)
 301         enum tbl_spant    pos;
 302         struct tbl_span  *next;
 303 };
 304
 305 enum    eqn_boxt {
 306         EQN_ROOT, /* root of parse tree */
 307         EQN_TEXT, /* text (number, variable, whatever) */
 308         EQN_SUBEXPR, /* nested `eqn' subexpression */
 309         EQN_LIST, /* list (braces, etc.) */
 310         EQN_LISTONE, /* singleton list */
 311         EQN_PILE, /* vertical pile */
 312         EQN_MATRIX /* pile of piles */
 313 };
 314
 315 enum    eqn_fontt {
 316         EQNFONT_NONE = 0,
 317         EQNFONT_ROMAN,
 318         EQNFONT_BOLD,
 319         EQNFONT_FAT,
 320         EQNFONT_ITALIC,
 321         EQNFONT__MAX
 322 };
 323
 324 enum    eqn_post {
 325         EQNPOS_NONE = 0,
 326         EQNPOS_SUP,
 327         EQNPOS_SUBSUP,
 328         EQNPOS_SUB,
 329         EQNPOS_TO,
 330         EQNPOS_FROM,
 331         EQNPOS_FROMTO,
 332         EQNPOS_OVER,
 333         EQNPOS_SQRT,
 334         EQNPOS__MAX
 335 };
 336
 337 enum    eqn_pilet {
 338         EQNPILE_NONE = 0,
 339         EQNPILE_PILE,
 340         EQNPILE_CPILE,
 341         EQNPILE_RPILE,
 342         EQNPILE_LPILE,
 343         EQNPILE_COL,
 344         EQNPILE_CCOL,
 345         EQNPILE_RCOL,
 346         EQNPILE_LCOL,
 347         EQNPILE__MAX
 348 };
 349
 350  /*
 351  * A "box" is a parsed mathematical expression as defined by the eqn.7
 352  * grammar.
 353  */
 354 struct  eqn_box {
 355         int               size; /* font size of expression */
 356 #define EQN_DEFSIZE       INT_MIN
 357         enum eqn_boxt     type; /* type of node */
 358         struct eqn_box   *first; /* first child node */
 359         struct eqn_box   *last; /* last child node */
 360         struct eqn_box   *next; /* node sibling */
 361         struct eqn_box   *prev; /* node sibling */
 362         struct eqn_box   *parent; /* node sibling */
 363         char             *text; /* text (or NULL) */
 364         char             *left; /* fence left-hand */
 365         char             *right; /* fence right-hand */
 366         char             *top; /* expression over-symbol */
 367         char             *bottom; /* expression under-symbol */
 368         size_t            args; /* arguments in parent */
 369         size_t            expectargs; /* max arguments in parent */
 370         enum eqn_post     pos; /* position of next box */
 371         enum eqn_fontt    font; /* font of box */
 372         enum eqn_pilet    pile; /* equation piling */
 373 };
 374
 375 /*
 376  * An equation consists of a tree of expressions starting at a given
 377  * line and position.
 378  */
 379 struct  eqn {
 380         char             *name; /* identifier (or NULL) */
 381         struct eqn_box   *root; /* root mathematical expression */
 382         int               ln; /* invocation line */
 383         int               pos; /* invocation position */
 384 };
 385
 386 /*
 387  * Parse options.
 388  */
 389 #define MPARSE_MDOC     1  /* assume -mdoc */
 390 #define MPARSE_MAN      2  /* assume -man */
 391 #define MPARSE_SO       4  /* honour .so requests */
 392 #define MPARSE_QUICK    8  /* abort the parse early */
 393
 394 enum    mandoc_esc {
 395         ESCAPE_ERROR = 0, /* bail! unparsable escape */
 396         ESCAPE_IGNORE, /* escape to be ignored */
 397         ESCAPE_SPECIAL, /* a regular special character */
 398         ESCAPE_FONT, /* a generic font mode */
 399         ESCAPE_FONTBOLD, /* bold font mode */
 400         ESCAPE_FONTITALIC, /* italic font mode */
 401         ESCAPE_FONTBI, /* bold italic font mode */
 402         ESCAPE_FONTROMAN, /* roman font mode */
 403         ESCAPE_FONTPREV, /* previous font mode */
 404         ESCAPE_NUMBERED, /* a numbered glyph */
 405         ESCAPE_UNICODE, /* a unicode codepoint */
 406         ESCAPE_NOSPACE, /* suppress space if the last on a line */
 407         ESCAPE_SKIPCHAR /* skip the next character */
 408 };
 409
 410 typedef void    (*mandocmsg)(enum mandocerr, enum mandoclevel,
 411                         const char *, int, int, const char *);
 412
 413 struct  mparse;
 414 struct  mchars;
 415 struct  mdoc;
 416 struct  man;
 417
 418 __BEGIN_DECLS
 419
 420 enum mandoc_esc   mandoc_escape(const char **, const char **, int *);
 421 struct mchars    *mchars_alloc(void);
 422 void              mchars_free(struct mchars *);
 423 char              mchars_num2char(const char *, size_t);
 424 int               mchars_num2uc(const char *, size_t);
 425 int               mchars_spec2cp(const struct mchars *,
 426                         const char *, size_t);
 427 const char       *mchars_spec2str(const struct mchars *,
 428                         const char *, size_t, size_t *);
 429 struct mparse    *mparse_alloc(int, enum mandoclevel, mandocmsg,
 430                         const char *);
 431 void              mparse_free(struct mparse *);
 432 void              mparse_keep(struct mparse *);
 433 enum mandoclevel  mparse_open(struct mparse *, int *, const char *,
 434                         pid_t *);
 435 enum mandoclevel  mparse_readfd(struct mparse *, int, const char *);
 436 enum mandoclevel  mparse_readmem(struct mparse *, const void *, size_t,
 437                         const char *);
 438 void              mparse_reset(struct mparse *);
 439 void              mparse_result(struct mparse *,
 440                         struct mdoc **, struct man **, char **);
 441 const char       *mparse_getkeep(const struct mparse *);
 442 const char       *mparse_strerror(enum mandocerr);
 443 const char       *mparse_strlevel(enum mandoclevel);
 444 enum mandoclevel  mparse_wait(struct mparse *, pid_t);
 445
 446 __END_DECLS
 447
 448 #endif /*!MANDOC_H*/