mandoc.h

   1 /*      $Id: mandoc.h,v 1.191 2015/01/28 17:32:07 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2010, 2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010-2015 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
  19 #define ASCII_NBRSP      31  /* non-breaking space */
  20 #define ASCII_HYPH       30  /* breakable hyphen */
  21 #define ASCII_BREAK      29  /* breakable zero-width space */
  22
  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  */
  29 enum    mandoclevel {
  30         MANDOCLEVEL_OK = 0,
  31         MANDOCLEVEL_RESERVED,
  32         MANDOCLEVEL_WARNING, /* warnings: syntax, whitespace, etc. */
  33         MANDOCLEVEL_ERROR, /* input has been thrown away */
  34         MANDOCLEVEL_UNSUPP, /* input needs unimplemented features */
  35         MANDOCLEVEL_BADARG, /* bad argument in invocation */
  36         MANDOCLEVEL_SYSERR, /* system error */
  37         MANDOCLEVEL_MAX
  38 };
  39
  40 /*
  41  * All possible things that can go wrong within a parse, be it libroff,
  42  * libmdoc, or libman.
  43  */
  44 enum    mandocerr {
  45         MANDOCERR_OK,
  46
  47         MANDOCERR_WARNING, /* ===== start of warnings ===== */
  48
  49         /* related to the prologue */
  50         MANDOCERR_DT_NOTITLE, /* missing manual title, using UNTITLED: line */
  51         MANDOCERR_TH_NOTITLE, /* missing manual title, using "": [macro] */
  52         MANDOCERR_TITLE_CASE, /* lower case character in document title */
  53         MANDOCERR_MSEC_MISSING, /* missing manual section, using "": macro */
  54         MANDOCERR_MSEC_BAD, /* unknown manual section: Dt ... section */
  55         MANDOCERR_DATE_MISSING, /* missing date, using today's date */
  56         MANDOCERR_DATE_BAD, /* cannot parse date, using it verbatim: date */
  57         MANDOCERR_OS_MISSING, /* missing Os macro, using "" */
  58         MANDOCERR_PROLOG_REP, /* duplicate prologue macro: macro */
  59         MANDOCERR_PROLOG_LATE, /* late prologue macro: macro */
  60         MANDOCERR_DT_LATE, /* skipping late title macro: Dt args */
  61         MANDOCERR_PROLOG_ORDER, /* prologue macros out of order: macros */
  62
  63         /* related to document structure */
  64         MANDOCERR_SO, /* .so is fragile, better use ln(1): so path */
  65         MANDOCERR_DOC_EMPTY, /* no document body */
  66         MANDOCERR_SEC_BEFORE, /* content before first section header: macro */
  67         MANDOCERR_NAMESEC_FIRST, /* first section is not NAME: Sh title */
  68         MANDOCERR_NAMESEC_BAD, /* bad NAME section contents: macro */
  69         MANDOCERR_SEC_ORDER, /* sections out of conventional order: Sh title */
  70         MANDOCERR_SEC_REP, /* duplicate section title: Sh title */
  71         MANDOCERR_SEC_MSEC, /* unexpected section: Sh title for ... only */
  72         MANDOCERR_XR_ORDER, /* unusual Xr order: ... after ... */
  73         MANDOCERR_XR_PUNCT, /* unusual Xr punctuation: ... after ... */
  74         MANDOCERR_AN_MISSING, /* AUTHORS section without An macro */
  75
  76         /* related to macros and nesting */
  77         MANDOCERR_MACRO_OBS, /* obsolete macro: macro */
  78         MANDOCERR_MACRO_CALL, /* macro neither callable nor escaped: macro */
  79         MANDOCERR_PAR_SKIP, /* skipping paragraph macro: macro ... */
  80         MANDOCERR_PAR_MOVE, /* moving paragraph macro out of list: macro */
  81         MANDOCERR_NS_SKIP, /* skipping no-space macro */
  82         MANDOCERR_BLK_NEST, /* blocks badly nested: macro ... */
  83         MANDOCERR_BD_NEST, /* nested displays are not portable: macro ... */
  84         MANDOCERR_BL_MOVE, /* moving content out of list: macro */
  85         MANDOCERR_VT_CHILD, /* .Vt block has child macro: macro */
  86         MANDOCERR_FI_SKIP, /* fill mode already enabled, skipping: fi */
  87         MANDOCERR_NF_SKIP, /* fill mode already disabled, skipping: nf */
  88         MANDOCERR_BLK_LINE, /* line scope broken: macro breaks macro */
  89
  90         /* related to missing arguments */
  91         MANDOCERR_REQ_EMPTY, /* skipping empty request: request */
  92         MANDOCERR_COND_EMPTY, /* conditional request controls empty scope */
  93         MANDOCERR_MACRO_EMPTY, /* skipping empty macro: macro */
  94         MANDOCERR_ARG_EMPTY, /* empty argument, using 0n: macro arg */
  95         MANDOCERR_ARGCWARN, /* argument count wrong */
  96         MANDOCERR_BD_NOTYPE, /* missing display type, using -ragged: Bd */
  97         MANDOCERR_BL_LATETYPE, /* list type is not the first argument: Bl arg */
  98         MANDOCERR_BL_NOWIDTH, /* missing -width in -tag list, using 8n */
  99         MANDOCERR_EX_NONAME, /* missing utility name, using "": Ex */
 100         MANDOCERR_IT_NOHEAD, /* empty head in list item: Bl -type It */
 101         MANDOCERR_IT_NOBODY, /* empty list item: Bl -type It */
 102         MANDOCERR_BF_NOFONT, /* missing font type, using \fR: Bf */
 103         MANDOCERR_BF_BADFONT, /* unknown font type, using \fR: Bf font */
 104         MANDOCERR_PF_SKIP, /* nothing follows prefix: Pf arg */
 105         MANDOCERR_ARG_STD, /* missing -std argument, adding it: macro */
 106         MANDOCERR_EQN_NOBOX, /* missing eqn box, using "": op */
 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_FN_PAREN, /* parenthesis in function name: arg */
 118         MANDOCERR_RS_BAD, /* invalid content in Rs block: macro */
 119         MANDOCERR_SM_BAD, /* invalid Boolean argument: macro arg */
 120         MANDOCERR_FT_BAD, /* unknown font, skipping request: ft font */
 121
 122         /* related to plain text */
 123         MANDOCERR_FI_BLANK, /* blank line in fill mode, using .sp */
 124         MANDOCERR_FI_TAB, /* tab in filled text */
 125         MANDOCERR_SPACE_EOL, /* whitespace at end of input line */
 126         MANDOCERR_COMMENT_BAD, /* bad comment style */
 127         MANDOCERR_ESC_BAD, /* invalid escape sequence: esc */
 128         MANDOCERR_STR_UNDEF, /* undefined string, using "": name */
 129
 130         /* related to tables */
 131         MANDOCERR_TBLLAYOUT_SPAN, /* tbl line starts with span */
 132         MANDOCERR_TBLLAYOUT_DOWN, /* tbl column starts with span */
 133         MANDOCERR_TBLLAYOUT_VERT, /* skipping vertical bar in tbl layout */
 134
 135         MANDOCERR_ERROR, /* ===== start of errors ===== */
 136
 137         /* related to equations */
 138         MANDOCERR_EQNNSCOPE, /* unexpected equation scope closure*/
 139         MANDOCERR_EQNSCOPE, /* equation scope open on exit */
 140         MANDOCERR_EQNBADSCOPE, /* overlapping equation scopes */
 141         MANDOCERR_EQNEOF, /* unexpected end of equation */
 142
 143         /* related to tables */
 144         MANDOCERR_TBLOPT_ALPHA, /* non-alphabetic character in tbl options */
 145         MANDOCERR_TBLOPT_BAD, /* skipping unknown tbl option: option */
 146         MANDOCERR_TBLOPT_NOARG, /* missing tbl option argument: option */
 147         MANDOCERR_TBLOPT_ARGSZ, /* wrong tbl option argument size: option */
 148         MANDOCERR_TBLLAYOUT_NONE, /* empty tbl layout */
 149         MANDOCERR_TBLLAYOUT_CHAR, /* invalid character in tbl layout: char */
 150         MANDOCERR_TBLLAYOUT_PAR, /* unmatched parenthesis in tbl layout */
 151         MANDOCERR_TBLDATA_NONE, /* tbl without any data cells */
 152         MANDOCERR_TBLDATA_SPAN, /* ignoring data in spanned tbl cell: data */
 153         MANDOCERR_TBLDATA_EXTRA, /* ignoring extra tbl data cells: data */
 154         MANDOCERR_TBLDATA_BLK, /* data block open at end of tbl: macro */
 155
 156         /* related to document structure and macros */
 157         MANDOCERR_FILE, /* cannot open file */
 158         MANDOCERR_ROFFLOOP, /* input stack limit exceeded, infinite loop? */
 159         MANDOCERR_CHAR_BAD, /* skipping bad character: number */
 160         MANDOCERR_MACRO, /* skipping unknown macro: macro */
 161         MANDOCERR_REQ_INSEC, /* skipping insecure request: request */
 162         MANDOCERR_IT_STRAY, /* skipping item outside list: It ... */
 163         MANDOCERR_TA_STRAY, /* skipping column outside column list: Ta */
 164         MANDOCERR_BLK_NOTOPEN, /* skipping end of block that is not open */
 165         MANDOCERR_RE_NOTOPEN, /* fewer RS blocks open, skipping: RE arg */
 166         MANDOCERR_BLK_BROKEN, /* inserting missing end of block: macro ... */
 167         MANDOCERR_BLK_NOEND, /* appending missing end of block: macro */
 168
 169         /* related to request and macro arguments */
 170         MANDOCERR_NAMESC, /* escaped character not allowed in a name: name */
 171         MANDOCERR_ARGCOUNT, /* argument count wrong */
 172         MANDOCERR_BD_FILE, /* NOT IMPLEMENTED: Bd -file */
 173         MANDOCERR_BL_NOTYPE, /* missing list type, using -item: Bl */
 174         MANDOCERR_NM_NONAME, /* missing manual name, using "": Nm */
 175         MANDOCERR_OS_UNAME, /* uname(3) system call failed, using UNKNOWN */
 176         MANDOCERR_ST_BAD, /* unknown standard specifier: St standard */
 177         MANDOCERR_IT_NONUM, /* skipping request without numeric argument */
 178         MANDOCERR_SO_PATH, /* NOT IMPLEMENTED: .so with absolute path or ".." */
 179         MANDOCERR_SO_FAIL, /* .so request failed */
 180         MANDOCERR_ARG_SKIP, /* skipping all arguments: macro args */
 181         MANDOCERR_ARG_EXCESS, /* skipping excess arguments: macro ... args */
 182         MANDOCERR_DIVZERO, /* divide by zero */
 183
 184         MANDOCERR_UNSUPP, /* ===== start of unsupported features ===== */
 185
 186         MANDOCERR_TOOLARGE, /* input too large */
 187         MANDOCERR_CHAR_UNSUPP, /* unsupported control character: number */
 188         MANDOCERR_REQ_UNSUPP, /* unsupported roff request: request */
 189         MANDOCERR_TBLOPT_EQN, /* eqn delim option in tbl: arg */
 190         MANDOCERR_TBLLAYOUT_MOD, /* unsupported tbl layout modifier: m */
 191         MANDOCERR_TBLMACRO, /* ignoring macro in table: macro */
 192
 193         MANDOCERR_MAX
 194 };
 195
 196 struct  tbl_opts {
 197         char              tab; /* cell-separator */
 198         char              decimal; /* decimal point */
 199         int               opts;
 200 #define TBL_OPT_CENTRE   (1 << 0)
 201 #define TBL_OPT_EXPAND   (1 << 1)
 202 #define TBL_OPT_BOX      (1 << 2)
 203 #define TBL_OPT_DBOX     (1 << 3)
 204 #define TBL_OPT_ALLBOX   (1 << 4)
 205 #define TBL_OPT_NOKEEP   (1 << 5)
 206 #define TBL_OPT_NOSPACE  (1 << 6)
 207 #define TBL_OPT_NOWARN   (1 << 7)
 208         int               cols; /* number of columns */
 209         int               lvert; /* width of left vertical line */
 210         int               rvert; /* width of right vertical line */
 211 };
 212
 213 /*
 214  * The head of a table specifies all of its columns.  When formatting a
 215  * tbl_span, iterate over these and plug in data from the tbl_span when
 216  * appropriate, using tbl_cell as a guide to placement.
 217  */
 218 struct  tbl_head {
 219         int               ident; /* 0 <= unique id < cols */
 220         struct tbl_head  *next;
 221         struct tbl_head  *prev;
 222 };
 223
 224 enum    tbl_cellt {
 225         TBL_CELL_CENTRE, /* c, C */
 226         TBL_CELL_RIGHT, /* r, R */
 227         TBL_CELL_LEFT, /* l, L */
 228         TBL_CELL_NUMBER, /* n, N */
 229         TBL_CELL_SPAN, /* s, S */
 230         TBL_CELL_LONG, /* a, A */
 231         TBL_CELL_DOWN, /* ^ */
 232         TBL_CELL_HORIZ, /* _, - */
 233         TBL_CELL_DHORIZ, /* = */
 234         TBL_CELL_MAX
 235 };
 236
 237 /*
 238  * A cell in a layout row.
 239  */
 240 struct  tbl_cell {
 241         struct tbl_cell  *next;
 242         int               vert; /* width of subsequent vertical line */
 243         enum tbl_cellt    pos;
 244         size_t            spacing;
 245         int               flags;
 246 #define TBL_CELL_TALIGN  (1 << 0) /* t, T */
 247 #define TBL_CELL_BALIGN  (1 << 1) /* d, D */
 248 #define TBL_CELL_BOLD    (1 << 2) /* fB, B, b */
 249 #define TBL_CELL_ITALIC  (1 << 3) /* fI, I, i */
 250 #define TBL_CELL_EQUAL   (1 << 4) /* e, E */
 251 #define TBL_CELL_UP      (1 << 5) /* u, U */
 252 #define TBL_CELL_WIGN    (1 << 6) /* z, Z */
 253 #define TBL_CELL_WMAX    (1 << 7) /* x, X */
 254         struct tbl_head  *head;
 255 };
 256
 257 /*
 258  * A layout row.
 259  */
 260 struct  tbl_row {
 261         struct tbl_row   *next;
 262         struct tbl_cell  *first;
 263         struct tbl_cell  *last;
 264         int               vert; /* width of left vertical line */
 265 };
 266
 267 enum    tbl_datt {
 268         TBL_DATA_NONE, /* has no data */
 269         TBL_DATA_DATA, /* consists of data/string */
 270         TBL_DATA_HORIZ, /* horizontal line */
 271         TBL_DATA_DHORIZ, /* double-horizontal line */
 272         TBL_DATA_NHORIZ, /* squeezed horizontal line */
 273         TBL_DATA_NDHORIZ /* squeezed double-horizontal line */
 274 };
 275
 276 /*
 277  * A cell within a row of data.  The "string" field contains the actual
 278  * string value that's in the cell.  The rest is layout.
 279  */
 280 struct  tbl_dat {
 281         struct tbl_cell  *layout; /* layout cell */
 282         int               spans; /* how many spans follow */
 283         struct tbl_dat   *next;
 284         char             *string; /* data (NULL if not TBL_DATA_DATA) */
 285         enum tbl_datt     pos;
 286 };
 287
 288 enum    tbl_spant {
 289         TBL_SPAN_DATA, /* span consists of data */
 290         TBL_SPAN_HORIZ, /* span is horizontal line */
 291         TBL_SPAN_DHORIZ /* span is double horizontal line */
 292 };
 293
 294 /*
 295  * A row of data in a table.
 296  */
 297 struct  tbl_span {
 298         struct tbl_opts  *opts;
 299         struct tbl_head  *head;
 300         struct tbl_row   *layout; /* layout row */
 301         struct tbl_dat   *first;
 302         struct tbl_dat   *last;
 303         struct tbl_span  *prev;
 304         struct tbl_span  *next;
 305         int               line; /* parse line */
 306         int               flags;
 307 #define TBL_SPAN_FIRST   (1 << 0)
 308 #define TBL_SPAN_LAST    (1 << 1)
 309         enum tbl_spant    pos;
 310 };
 311
 312 enum    eqn_boxt {
 313         EQN_ROOT, /* root of parse tree */
 314         EQN_TEXT, /* text (number, variable, whatever) */
 315         EQN_SUBEXPR, /* nested `eqn' subexpression */
 316         EQN_LIST, /* list (braces, etc.) */
 317         EQN_LISTONE, /* singleton list */
 318         EQN_PILE, /* vertical pile */
 319         EQN_MATRIX /* pile of piles */
 320 };
 321
 322 enum    eqn_fontt {
 323         EQNFONT_NONE = 0,
 324         EQNFONT_ROMAN,
 325         EQNFONT_BOLD,
 326         EQNFONT_FAT,
 327         EQNFONT_ITALIC,
 328         EQNFONT__MAX
 329 };
 330
 331 enum    eqn_post {
 332         EQNPOS_NONE = 0,
 333         EQNPOS_SUP,
 334         EQNPOS_SUBSUP,
 335         EQNPOS_SUB,
 336         EQNPOS_TO,
 337         EQNPOS_FROM,
 338         EQNPOS_FROMTO,
 339         EQNPOS_OVER,
 340         EQNPOS_SQRT,
 341         EQNPOS__MAX
 342 };
 343
 344 enum    eqn_pilet {
 345         EQNPILE_NONE = 0,
 346         EQNPILE_PILE,
 347         EQNPILE_CPILE,
 348         EQNPILE_RPILE,
 349         EQNPILE_LPILE,
 350         EQNPILE_COL,
 351         EQNPILE_CCOL,
 352         EQNPILE_RCOL,
 353         EQNPILE_LCOL,
 354         EQNPILE__MAX
 355 };
 356
 357  /*
 358  * A "box" is a parsed mathematical expression as defined by the eqn.7
 359  * grammar.
 360  */
 361 struct  eqn_box {
 362         int               size; /* font size of expression */
 363 #define EQN_DEFSIZE       INT_MIN
 364         enum eqn_boxt     type; /* type of node */
 365         struct eqn_box   *first; /* first child node */
 366         struct eqn_box   *last; /* last child node */
 367         struct eqn_box   *next; /* node sibling */
 368         struct eqn_box   *prev; /* node sibling */
 369         struct eqn_box   *parent; /* node sibling */
 370         char             *text; /* text (or NULL) */
 371         char             *left; /* fence left-hand */
 372         char             *right; /* fence right-hand */
 373         char             *top; /* expression over-symbol */
 374         char             *bottom; /* expression under-symbol */
 375         size_t            args; /* arguments in parent */
 376         size_t            expectargs; /* max arguments in parent */
 377         enum eqn_post     pos; /* position of next box */
 378         enum eqn_fontt    font; /* font of box */
 379         enum eqn_pilet    pile; /* equation piling */
 380 };
 381
 382 /*
 383  * An equation consists of a tree of expressions starting at a given
 384  * line and position.
 385  */
 386 struct  eqn {
 387         char             *name; /* identifier (or NULL) */
 388         struct eqn_box   *root; /* root mathematical expression */
 389         int               ln; /* invocation line */
 390         int               pos; /* invocation position */
 391 };
 392
 393 /*
 394  * Parse options.
 395  */
 396 #define MPARSE_MDOC     1  /* assume -mdoc */
 397 #define MPARSE_MAN      2  /* assume -man */
 398 #define MPARSE_SO       4  /* honour .so requests */
 399 #define MPARSE_QUICK    8  /* abort the parse early */
 400 #define MPARSE_UTF8     16 /* accept UTF-8 input */
 401 #define MPARSE_LATIN1   32 /* accept ISO-LATIN-1 input */
 402
 403 enum    mandoc_esc {
 404         ESCAPE_ERROR = 0, /* bail! unparsable escape */
 405         ESCAPE_IGNORE, /* escape to be ignored */
 406         ESCAPE_SPECIAL, /* a regular special character */
 407         ESCAPE_FONT, /* a generic font mode */
 408         ESCAPE_FONTBOLD, /* bold font mode */
 409         ESCAPE_FONTITALIC, /* italic font mode */
 410         ESCAPE_FONTBI, /* bold italic font mode */
 411         ESCAPE_FONTROMAN, /* roman font mode */
 412         ESCAPE_FONTPREV, /* previous font mode */
 413         ESCAPE_NUMBERED, /* a numbered glyph */
 414         ESCAPE_UNICODE, /* a unicode codepoint */
 415         ESCAPE_NOSPACE, /* suppress space if the last on a line */
 416         ESCAPE_SKIPCHAR, /* skip the next character */
 417         ESCAPE_OVERSTRIKE /* overstrike all chars in the argument */
 418 };
 419
 420 typedef void    (*mandocmsg)(enum mandocerr, enum mandoclevel,
 421                         const char *, int, int, const char *);
 422
 423 __BEGIN_DECLS
 424
 425 struct  mparse;
 426 struct  mchars;
 427 struct  mdoc;
 428 struct  man;
 429
 430 enum mandoc_esc   mandoc_escape(const char **, const char **, int *);
 431 struct mchars    *mchars_alloc(void);
 432 void              mchars_free(struct mchars *);
 433 int               mchars_num2char(const char *, size_t);
 434 const char       *mchars_uc2str(int);
 435 int               mchars_num2uc(const char *, size_t);
 436 int               mchars_spec2cp(const struct mchars *,
 437                         const char *, size_t);
 438 const char       *mchars_spec2str(const struct mchars *,
 439                         const char *, size_t, size_t *);
 440 struct mparse    *mparse_alloc(int, enum mandoclevel, mandocmsg,
 441                         const struct mchars *, const char *);
 442 void              mparse_free(struct mparse *);
 443 void              mparse_keep(struct mparse *);
 444 enum mandoclevel  mparse_open(struct mparse *, int *, const char *);
 445 enum mandoclevel  mparse_readfd(struct mparse *, int, const char *);
 446 enum mandoclevel  mparse_readmem(struct mparse *, void *, size_t,
 447                         const char *);
 448 void              mparse_reset(struct mparse *);
 449 void              mparse_result(struct mparse *,
 450                         struct mdoc **, struct man **, char **);
 451 const char       *mparse_getkeep(const struct mparse *);
 452 const char       *mparse_strerror(enum mandocerr);
 453 const char       *mparse_strlevel(enum mandoclevel);
 454 enum mandoclevel  mparse_wait(struct mparse *);
 455
 456 __END_DECLS