=================================================================== RCS file: /cvs/texi2mdoc/extern.h,v retrieving revision 1.2 retrieving revision 1.22 diff -u -p -r1.2 -r1.22 --- texi2mdoc/extern.h 2015/02/20 12:25:25 1.2 +++ texi2mdoc/extern.h 2015/03/01 13:39:51 1.22 @@ -1,4 +1,4 @@ -/* $Id: extern.h,v 1.2 2015/02/20 12:25:25 kristaps Exp $ */ +/* $Id: extern.h,v 1.22 2015/03/01 13:39:51 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -17,15 +17,34 @@ #ifndef EXTERN_H #define EXTERN_H +#if !defined(__BEGIN_DECLS) +# ifdef __cplusplus +# define __BEGIN_DECLS extern "C" { +# else +# define __BEGIN_DECLS +# endif +#endif +#if !defined(__END_DECLS) +# ifdef __cplusplus +# define __END_DECLS } +# else +# define __END_DECLS +# endif +#endif + /* * This defines each one of the Texinfo commands that we understand. * Obviously this only refers to native commands; overriden names are a * different story. */ enum texicmd { + TEXICMD_A4PAPER, + TEXICMD_AA, + TEXICMD_AASMALL, TEXICMD_ACRONYM, TEXICMD_ACUTE, - TEXICMD_A4PAPER, + TEXICMD_AE, + TEXICMD_AESMALL, TEXICMD_ANCHOR, TEXICMD_APPENDIX, TEXICMD_APPENDIXSEC, @@ -39,6 +58,8 @@ enum texicmd { TEXICMD_BANG, TEXICMD_BULLET, TEXICMD_BYE, + TEXICMD_CARTOUCHE, + TEXICMD_CEDILLA, TEXICMD_CENTER, TEXICMD_CHAPTER, TEXICMD_CINDEX, @@ -48,14 +69,17 @@ enum texicmd { TEXICMD_CODE, TEXICMD_COLON, TEXICMD_COLUMNFRACTIONS, + TEXICMD_COMMA, TEXICMD_COMMAND, TEXICMD_COMMENT, TEXICMD_COMMENT_LONG, TEXICMD_CONTENTS, TEXICMD_COPYING, TEXICMD_COPYRIGHT, + TEXICMD_DEFCODEINDEX, TEXICMD_DEFFN, TEXICMD_DEFFNX, + TEXICMD_DEFINDEX, TEXICMD_DEFMAC, TEXICMD_DEFMACX, TEXICMD_DEFTP, @@ -64,6 +88,8 @@ enum texicmd { TEXICMD_DEFTYPEFNX, TEXICMD_DEFTYPEFUN, TEXICMD_DEFTYPEFUNX, + TEXICMD_DEFTYPEMETHOD, + TEXICMD_DEFTYPEMETHODX, TEXICMD_DEFTYPEVAR, TEXICMD_DEFTYPEVARX, TEXICMD_DEFTYPEVR, @@ -76,26 +102,51 @@ enum texicmd { TEXICMD_DEFVRX, TEXICMD_DETAILMENU, TEXICMD_DFN, + TEXICMD_DH, + TEXICMD_DHSMALL, TEXICMD_DIRCATEGORY, TEXICMD_DIRENTRY, TEXICMD_DISPLAY, + TEXICMD_DMN, + TEXICMD_DOCUMENTDESCRPITION, + TEXICMD_DOCUMENTENCODING, + TEXICMD_DOCUMENTLANGUAGE, + TEXICMD_DOTACCENT, + TEXICMD_DOTLESS, TEXICMD_DOTS, TEXICMD_EMAIL, TEXICMD_EMPH, TEXICMD_END, + TEXICMD_ENDDOTS, TEXICMD_ENUMERATE, TEXICMD_ENV, + TEXICMD_EQUIV, TEXICMD_ERROR, + TEXICMD_EURO, TEXICMD_EXAMPLE, + TEXICMD_EXCLAMDOWN, TEXICMD_EXDENT, TEXICMD_EXPANSION, TEXICMD_FILE, TEXICMD_FINALOUT, TEXICMD_FINDEX, + TEXICMD_FLUSHLEFT, + TEXICMD_FLUSHRIGHT, + TEXICMD_FIRSTPARAGRAPHINDENT, + TEXICMD_FOOTNOTE, + TEXICMD_FOOTNOTESTYLE, TEXICMD_FTABLE, TEXICMD_FORMAT, + TEXICMD_GEQ, TEXICMD_GRAVE, TEXICMD_GROUP, + TEXICMD_GUILLEMETLEFT, + TEXICMD_GUILLEMETRIGHT, + TEXICMD_GUILLEMOTLEFT, + TEXICMD_GUILLEMOTRIGHT, + TEXICMD_GUILSINGLLEFT, + TEXICMD_GUILSINGLRIGHT, + TEXICMD_H, TEXICMD_HEADING, TEXICMD_HEADINGS, TEXICMD_HEADITEM, @@ -120,6 +171,8 @@ enum texicmd { TEXICMD_IMAGE, TEXICMD_INCLUDE, TEXICMD_INDENTBLOCK, + TEXICMD_INDICATEURL, + TEXICMD_INFOREF, TEXICMD_INSERTCOPYING, TEXICMD_ITEM, TEXICMD_ITEMIZE, @@ -127,8 +180,13 @@ enum texicmd { TEXICMD_KBD, TEXICMD_KEY, TEXICMD_KINDEX, + TEXICMD_L, TEXICMD_LATEX, + TEXICMD_LEQ, TEXICMD_LOWERSECTIONS, + TEXICMD_LSMALL, + TEXICMD_MACRO, + TEXICMD_MACRON, TEXICMD_MATH, TEXICMD_MENU, TEXICMD_MINUS, @@ -137,28 +195,49 @@ enum texicmd { TEXICMD_NEWLINE, TEXICMD_NODE, TEXICMD_NOINDENT, + TEXICMD_O, + TEXICMD_OE, + TEXICMD_OESMALL, + TEXICMD_OGONEK, TEXICMD_OPTION, - TEXICMD_PXREF, - TEXICMD_QUESTIONMARK, - TEXICMD_QUOTATION, + TEXICMD_ORDF, + TEXICMD_ORDM, + TEXICMD_OSMALL, TEXICMD_PAGE, TEXICMD_PARINDENT, TEXICMD_PERIOD, + TEXICMD_PINDEX, + TEXICMD_POUNDS, TEXICMD_PRINTINDEX, + TEXICMD_PXREF, + TEXICMD_QUESTIONDOWN, + TEXICMD_QUESTIONMARK, + TEXICMD_QUOTATION, + TEXICMD_QUOTEDBLBASE, + TEXICMD_QUOTEDBLLEFT, + TEXICMD_QUOTEDBLRIGHT, + TEXICMD_QUOTESINGLBASE, + TEXICMD_QUOTELEFT, + TEXICMD_QUOTERIGHT, TEXICMD_R, TEXICMD_RAISESECTIONS, TEXICMD_REF, + TEXICMD_REFILL, + TEXICMD_REGISTEREDSYMBOL, TEXICMD_RESULT, + TEXICMD_RINGACCENT, TEXICMD_SAMP, TEXICMD_SANSSERIF, TEXICMD_SC, TEXICMD_SECTION, TEXICMD_SET, TEXICMD_SETCHAPNEWPAGE, + TEXICMD_SETCONTENTSAFTER, TEXICMD_SETFILENAME, TEXICMD_SETTITLE, TEXICMD_SHORTCONTENTS, TEXICMD_SLANTED, + TEXICMD_SLASH, TEXICMD_SP, TEXICMD_SPACE, TEXICMD_SMALLBOOK, @@ -168,12 +247,15 @@ enum texicmd { TEXICMD_SMALLINDENTBLOCK, TEXICMD_SQUIGGLE_LEFT, TEXICMD_SQUIGGLE_RIGHT, + TEXICMD_SS, TEXICMD_STRONG, TEXICMD_SUBHEADING, TEXICMD_SUBSECTION, + TEXICMD_SUBSUBHEADING, TEXICMD_SUBSUBSECTION, TEXICMD_SUBTITLE, TEXICMD_SUMMARYCONTENTS, + TEXICMD_SYNINDEX, TEXICMD_SYNCODEINDEX, TEXICMD_T, TEXICMD_TAB, @@ -181,12 +263,20 @@ enum texicmd { TEXICMD_TABLE, TEXICMD_TEX, TEXICMD_TEXSYM, + TEXICMD_TEXTDEGREE, + TEXICMD_TH, + TEXICMD_THSMALL, + TEXICMD_TIE, + TEXICMD_TIEACCENT, TEXICMD_TILDE, TEXICMD_TINDEX, TEXICMD_TITLE, TEXICMD_TITLEFONT, TEXICMD_TITLEPAGE, TEXICMD_TOP, + TEXICMD_U, + TEXICMD_UBARACCENT, + TEXICMD_UDOTACCENT, TEXICMD_UMLAUT, TEXICMD_UNNUMBERED, TEXICMD_UNNUMBEREDSEC, @@ -194,8 +284,11 @@ enum texicmd { TEXICMD_UNNUMBEREDSUBSUBSEC, TEXICMD_UREF, TEXICMD_URL, + TEXICMD_USER_INDEX, + TEXICMD_V, TEXICMD_VALUE, TEXICMD_VAR, + TEXICMD_VERB, TEXICMD_VERBATIM, TEXICMD_VERBATIMINCLUDE, TEXICMD_VINDEX, @@ -206,16 +299,25 @@ enum texicmd { TEXICMD__MAX }; +enum texisrc { + TEXISRC_FILE, + TEXISRC_STDIN +}; + /* * The file currently being parsed. * This keeps track of our location within that file. */ struct texifile { + enum texisrc type; /* type of file */ const char *name; /* name of the file */ size_t line; /* current line (from zero) */ size_t col; /* current column in line (from zero) */ - char *map; /* mmap'd file */ - size_t mapsz; /* size of mmap */ + char *map; /* allocated file buffer */ + size_t mapsz; /* size of map */ + size_t mapmaxsz; /* full size of map */ + size_t insplice; /* how many bytes left in splice */ + size_t depth; /* pasted macro recursive depth */ }; struct texi; @@ -223,8 +325,7 @@ struct texi; /* * Callback for functions implementing texi commands. */ -typedef void (*texicmdfp)(struct texi *, - enum texicmd, const char *, size_t, size_t *); +typedef void (*texicmdfp)(struct texi *, enum texicmd, size_t *); /* * Describes Texinfo commands, whether native or overriden. @@ -235,6 +336,10 @@ struct texitok { size_t len; /* strlen(tok) */ }; +/* + * These values instruct us on whether a list (or table) of some type is + * currently being parsed. + */ enum texilist { TEXILIST_NONE = 0, TEXILIST_ITEM, @@ -242,38 +347,68 @@ enum texilist { TEXILIST_TABLE }; +/* + * Hold values assigned with @set and retrieved with @value. + * These values can contain arbitrary Texinfo. + */ struct texivalue { - char *key; - char *value; + char *key; /* the nil-terminated value name */ + char *value; /* the nil-terminated value */ }; /* + * Macros are (possibly-recursive) Texinfo sequences created with @macro + * and filled in by arguments when invoked. + */ +struct teximacro { + char *key; /* nil-terminated macro name */ + char *value; /* nil-terminated value */ + char **args; /* array of argument names (or NULL) */ + size_t argsz; /* array size */ +}; + +/* * The main parse structure. * This keeps any necessary information handy. */ struct texi { - struct texifile files[64]; /* stack of open files */ - size_t filepos; /* number of open files */ - size_t outcol; /* column in output line */ + const char *chapters; /* are we splitting chapters */ + size_t chapnum; char **dirs; /* texi directories */ size_t dirsz; /* number of texi directories */ + FILE *outfile; + /* + * Run-time parameters. + */ + struct texifile files[64]; /* stack of open files */ + size_t filepos; /* number of open files */ + const char *valstack[64]; /* stack of opened values */ + size_t valstackpos; /* position in valstack */ char *title; /* title of document */ char *subtitle; /* subtitle of document */ int secoffs; /* see sectioner() */ - struct texivalue *vals; - size_t valsz; + char **indexs; /* @defindex indices */ + size_t indexsz; /* entries in indexs */ + struct texivalue *vals; /* @value entries */ + size_t valsz; /* entries in vals */ + struct teximacro *macros; /* @macro entries */ + size_t macrosz; /* entries in macros */ /* * The following control what we output to the screen. * The complexity is required to accomodate for mdoc(7). */ enum texilist list; /* current list (set recursively) */ int outmacro; /* if >0, output is in line macro */ + size_t outcol; /* column in output line */ int seenws; /* ws has been seen (and ignored) */ int seenvs; /* newline has been Pp'd */ int ign; /* if >0, don't print anything */ int literal; /* if >0, literal context */ }; +#define BUF(_p) ((_p)->files[(_p)->filepos - 1].map) +#define BUFSZ(_p) ((_p)->files[(_p)->filepos - 1].mapsz) + #define isws(_x) \ (' ' == (_x) || '\t' == (_x)) #define ismspace(_x) \ @@ -281,42 +416,51 @@ struct texi { __BEGIN_DECLS -void advance(struct texi *, const char *, size_t *); -size_t advanceeoln(struct texi *, const char *, size_t, size_t *, int); -void advanceto(struct texi *, const char *, size_t *, size_t); +void advance(struct texi *, size_t *); +size_t advanceeoln(struct texi *, size_t *, int); +void advanceto(struct texi *, size_t *, size_t); -int parsearg(struct texi *, const char *, size_t, size_t *, size_t); -void parsebracket(struct texi *, const char *, size_t, size_t *); +char **argparse(struct texi *, size_t *, size_t *, size_t); + +int parsearg(struct texi *, size_t *, size_t); +void parsebracket(struct texi *, size_t *, int); +void parsestdin(struct texi *); void parsefile(struct texi *, const char *, int); -int parselinearg(struct texi *, const char *, size_t, size_t *); -void parseeof(struct texi *, const char *, size_t); -void parseeoln(struct texi *, const char *, size_t, size_t *); -void parsesingle(struct texi *, const char *, size_t, size_t *); -void parseto(struct texi *, const char *, size_t, size_t *, const char *); +int parselinearg(struct texi *, size_t *); +void parseeoln(struct texi *, size_t *); +void parseto(struct texi *, size_t *, const char *); -void texiabort(struct texi *, const char *); +void texiabort(struct texi *, const char *) + __attribute__((noreturn)); enum texicmd - texicmd(struct texi *, const char *, size_t, size_t, size_t *); + texicmd(const struct texi *, size_t, size_t *, struct teximacro **); +enum texicmd + peekcmd(const struct texi *, size_t); void texierr(struct texi *, const char *, ...) - __attribute__((format(printf, 2, 3))); + __attribute__((format(printf, 2, 3))) + __attribute__((noreturn)); void texiexit(struct texi *p); void texifilepop(struct texi *); void teximacro(struct texi *, const char *); void teximacroclose(struct texi *); void teximacroopen(struct texi *, const char *); -void texipunctuate(struct texi *, const char *, size_t, size_t *); +void teximdocopen(struct texi *, size_t *); +void teximdocclose(struct texi *, int); +void texipunctuate(struct texi *, size_t *); +void texiputbuf(struct texi *p, size_t, size_t); void texiputchar(struct texi *p, char); void texiputchars(struct texi *, const char *); void texivspace(struct texi *); void texiwarn(const struct texi *, const char *, ...) __attribute__((format(printf, 2, 3))); +void texisplice(struct texi *, const char *, size_t, size_t); void valueadd(struct texi *, char *, char *); const char * - valueblookup(struct texi *, const char *, size_t, size_t *); -void valuelclear(struct texi *, const char *, size_t, size_t *); + valueblookup(struct texi *, size_t *); +void valuelclear(struct texi *, size_t *); const char * - valuellookup(struct texi *, const char *, size_t, size_t *); + valuellookup(struct texi *, size_t *); extern const struct texitok *const texitoks;