=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.10 retrieving revision 1.41 diff -u -p -r1.10 -r1.41 --- texi2mdoc/main.c 2015/02/18 15:30:31 1.10 +++ texi2mdoc/main.c 2015/02/24 14:35:40 1.41 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.10 2015/02/18 15:30:31 kristaps Exp $ */ +/* $Id: main.c,v 1.41 2015/02/24 14:35:40 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -14,12 +14,11 @@ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ -#include -#include - +#if defined(__linux__) || defined(__MINT__) +# define _GNU_SOURCE /* memmem */ +#endif #include #include -#include #include #include #include @@ -28,1083 +27,749 @@ #include #include #include -#include -/* - * 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_ACRONYM, - TEXICMD_A4PAPER, - TEXICMD_ANCHOR, - TEXICMD_APPENDIX, - TEXICMD_APPENDIXSEC, - TEXICMD_ASTERISK, - TEXICMD_AT, - TEXICMD_AUTHOR, - TEXICMD_BANG, - TEXICMD_BULLET, - TEXICMD_BYE, - TEXICMD_CENTER, - TEXICMD_CHAPTER, - TEXICMD_CINDEX, - TEXICMD_CITE, - TEXICMD_CODE, - TEXICMD_COLON, - TEXICMD_COMMAND, - TEXICMD_COMMENT, - TEXICMD_COMMENT_LONG, - TEXICMD_CONTENTS, - TEXICMD_COPYING, - TEXICMD_COPYRIGHT, - TEXICMD_DEFTYPEFN, - TEXICMD_DEFTYPEFNX, - TEXICMD_DEFTYPEFUN, - TEXICMD_DEFTYPEFUNX, - TEXICMD_DEFTYPEVAR, - TEXICMD_DEFTYPEVR, - TEXICMD_DETAILMENU, - TEXICMD_DFN, - TEXICMD_DIRCATEGORY, - TEXICMD_DIRENTRY, - TEXICMD_DISPLAY, - TEXICMD_DOTS, - TEXICMD_EMAIL, - TEXICMD_EMPH, - TEXICMD_END, - TEXICMD_ENUMERATE, - TEXICMD_ENV, - TEXICMD_EXAMPLE, - TEXICMD_FILE, - TEXICMD_GROUP, - TEXICMD_HEADING, - TEXICMD_HEADINGS, - TEXICMD_HYPHEN, - TEXICMD_I, - TEXICMD_IFCLEAR, - TEXICMD_IFHTML, - TEXICMD_IFINFO, - TEXICMD_IFNOTTEX, - TEXICMD_IFTEX, - TEXICMD_IFSET, - TEXICMD_IMAGE, - TEXICMD_INCLUDE, - TEXICMD_INSERTCOPYING, - TEXICMD_ITEM, - TEXICMD_ITEMIZE, - TEXICMD_KBD, - TEXICMD_LATEX, - TEXICMD_MATH, - TEXICMD_MENU, - TEXICMD_NEWLINE, - TEXICMD_NODE, - TEXICMD_NOINDENT, - TEXICMD_PXREF, - TEXICMD_QUESTIONMARK, - TEXICMD_QUOTATION, - TEXICMD_PAGE, - TEXICMD_PARINDENT, - TEXICMD_PRINTINDEX, - TEXICMD_REF, - TEXICMD_SAMP, - TEXICMD_SC, - TEXICMD_SECTION, - TEXICMD_SET, - TEXICMD_SETCHAPNEWPAGE, - TEXICMD_SETFILENAME, - TEXICMD_SETTITLE, - TEXICMD_SP, - TEXICMD_SPACE, - TEXICMD_SMALLEXAMPLE, - TEXICMD_SQUIGGLE_LEFT, - TEXICMD_SQUIGGLE_RIGHT, - TEXICMD_STRONG, - TEXICMD_SUBSECTION, - TEXICMD_SUBTITLE, - TEXICMD_TAB, - TEXICMD_TABLE, - TEXICMD_TEX, - TEXICMD_TEXSYM, - TEXICMD_TITLE, - TEXICMD_TITLEFONT, - TEXICMD_TITLEPAGE, - TEXICMD_TOP, - TEXICMD_UNNUMBERED, - TEXICMD_UNNUMBEREDSEC, - TEXICMD_UREF, - TEXICMD_URL, - TEXICMD_VAR, - TEXICMD_VSKIP, - TEXICMD_W, - TEXICMD_XREF, - TEXICMD__MAX -}; +#include "extern.h" -/* - * The file currently being parsed. - * This keeps track of our location within that file. - */ -struct texifile { - 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 */ +#define SECTSZ 4 +static const char *const sects[SECTSZ] = { + "Sh", + "Ss", + "Em", + "No", }; -struct texi; - -/* - * Callback for functions implementing texi commands. - */ -typedef void (*texicmdfp)(struct texi *, - enum texicmd, const char *, size_t, size_t *); - -/* - * Describes Texinfo commands, whether native or overriden. - */ -struct texitok { - texicmdfp fp; /* callback (or NULL if none) */ - const char *tok; /* name of the token */ - size_t len; /* strlen(tok) */ -}; - -enum texilist { - TEXILIST_NONE = 0, - TEXILIST_ITEM, - TEXILIST_NOITEM, -}; - -/* - * 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 */ - char **dirs; /* texi directories */ - size_t dirsz; /* number of texi directories */ - enum texilist list; /* current list (set recursively) */ - int outmacro; /* if >0, output is in line macro */ - int seenws; /* ws has been seen (and ignored) */ - int ign; /* if >0, don't print anything */ - int literal; /* if >0, literal context */ - char *title; /* title of document */ - char *subtitle; /* subtitle of document */ -}; - -/* FIXME: FIND A BETTER WAY. */ -#define ismpunct(_x) \ - ('.' == (_x) || \ - ',' == (_x) || \ - ';' == (_x)) -/* Texi disregards spaces and tabs. */ -#define isws(_x) \ - (' ' == (_x) || '\t' == (_x)) -#define ismspace(_x) \ - (isws((_x)) || '\n' == (_x)) - +static void doaccent(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doblock(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dobracket(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dobye(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dochapter(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void docommand(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dodeftypefun(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dodeftypevar(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dodefindex(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dodefn(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dodisplay(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doemph(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doenumerate(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doenv(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doexample(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dofile(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doignargn(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doignblock(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doignbracket(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doignline(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void doinline(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doinclude(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doitalic(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doitem(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doitemize(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dolink(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doliteral(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void domacro(struct texi *, enum texicmd, const char *, size_t, size_t *); static void domath(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void domultitable(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doquotation(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dotable(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dotop(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dosecoffs(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dosection(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dosp(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dosubsection(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dosubsubsection(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dosymbol(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dotab(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dotitle(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dovalue(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void doverb(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void doverbatim(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void doverbinclude(struct texi *, enum texicmd, const char *, size_t, size_t *); -static const struct texitok texitoks[TEXICMD__MAX] = { +static const struct texitok __texitoks[TEXICMD__MAX] = { + /* TEXICMD__BEGIN */ { doignargn, "acronym", 7 }, /* TEXICMD_ACRONYM */ + { doaccent, "'", 1 }, /* TEXICMD_ACUTE */ { doignline, "afourpaper", 10 }, /* TEXICMD_A4PAPER */ { doignbracket, "anchor", 6 }, /* TEXICMD_ANCHOR */ - { dochapter, "appendix", 8 }, /* TEXICMD_APPENDIX */ - { dochapter, "appendixsec", 11 }, /* TEXICMD_APPENDIXSEC */ + { dosection, "appendix", 8 }, /* TEXICMD_APPENDIX */ + { dosection, "appendixsec", 11 }, /* TEXICMD_APPENDIXSEC */ + { dosubsection, "appendixsubsec", 14 }, /* TEXICMD_APPENDIXSUBSEC */ + { dosubsubsection, "appendixsubsubsec", 17 }, /* TEXICMD_APPENDIXSUBSUBSEC */ + { doinline, "asis", 4 }, /* TEXICMD_ASIS */ { dosymbol, "*", 1 }, /* TEXICMD_ASTERISK */ { dosymbol, "@", 1 }, /* TEXICMD_AT */ { doignline, "author", 6 }, /* TEXICMD_AUTHOR */ + { doinline, "b", 1 }, /* TEXICMD_BOLD */ { dosymbol, "!", 1 }, /* TEXICMD_BANG */ { dosymbol, "bullet", 6 }, /* TEXICMD_BULLET */ { dobye, "bye", 3 }, /* TEXICMD_BYE */ - { doignline, "center", 5 }, /* TEXICMD_CENTER */ - { dochapter, "chapter", 7 }, /* TEXICMD_CHAPTER */ + { doblock, "cartouche", 9 }, /* TEXICMD_CARTOUCHE */ + { doaccent, ",", 1 }, /* TEXICMD_CEDILLA */ + { doignline, "center", 6 }, /* TEXICMD_CENTER */ + { dosection, "chapter", 7 }, /* TEXICMD_CHAPTER */ { doignline, "cindex", 6 }, /* TEXICMD_CINDEX */ - { doliteral, "code", 4 }, /* TEXICMD_CODE */ - { doitalic, "cite", 4 }, /* TEXICMD_CITE */ + { doaccent, "^", 1 }, /* TEXICMD_CIRCUMFLEX */ + { doinline, "cite", 4 }, /* TEXICMD_CITE */ + { dovalue, "clear", 5 }, /* TEXICMD_CLEAR */ + { doinline, "code", 4 }, /* TEXICMD_CODE */ { dosymbol, ":", 1 }, /* TEXICMD_COLON */ - { docommand, "command", 7 }, /* TEXICMD_COMMAND */ + { NULL, "columnfractions", 15 }, /* TEXICMD_COLUMNFRACTIONS */ + { dosymbol, "comma", 5 }, /* TEXICMD_COMMA */ + { doinline, "command", 7 }, /* TEXICMD_COMMAND */ { doignline, "c", 1 }, /* TEXICMD_COMMENT */ { doignline, "comment", 7 }, /* TEXICMD_COMMENT_LONG */ { doignline, "contents", 8 }, /* TEXICMD_CONTENTS */ { doignblock, "copying", 7 }, /* TEXICMD_COPYING */ { dosymbol, "copyright", 9 }, /* TEXICMD_COPYRIGHT */ - { dodeftypefun, "deftypefn", 9 }, /* TEXICMD_DEFTYPEFN */ - { dodeftypefun, "deftypefnx", 10 }, /* TEXICMD_DEFTYPEFNX */ - { dodeftypefun, "deftypefun", 10 }, /* TEXICMD_DEFTYPEFUN */ - { dodeftypefun, "deftypefunx", 11 }, /* TEXICMD_DEFTYPEFUNX */ - { dodeftypevar, "deftypevar", 10 }, /* TEXICMD_DEFTYPEVAR */ - { dodeftypevar, "deftypevr", 9 }, /* TEXICMD_DEFTYPEVR */ + { dodefindex, "defcodeindex", 12 }, /* TEXICMD_DEFCODEINDEX */ + { dodefn, "deffn", 5 }, /* TEXICMD_DEFFN */ + { dodefn, "deffnx", 6 }, /* TEXICMD_DEFFNX */ + { dodefindex, "defindex", 8 }, /* TEXICMD_DEFINDEX */ + { dodefn, "defmac", 6 }, /* TEXICMD_DEFMAC */ + { dodefn, "defmacx", 7 }, /* TEXICMD_DEFMACX */ + { dodefn, "deftp", 5 }, /* TEXICMD_DEFTP */ + { dodefn, "deftpx", 6 }, /* TEXICMD_DEFTPX */ + { dodefn, "deftypefn", 9 }, /* TEXICMD_DEFTYPEFN */ + { dodefn, "deftypefnx", 10 }, /* TEXICMD_DEFTYPEFNX */ + { dodefn, "deftypefun", 10 }, /* TEXICMD_DEFTYPEFUN */ + { dodefn, "deftypefunx", 11 }, /* TEXICMD_DEFTYPEFUNX */ + { dodefn, "deftypemethod", 13 }, /* TEXICMD_DEFTYPEMETHOD */ + { dodefn, "deftypemethodx", 14 }, /* TEXICMD_DEFTYPEMETHODX */ + { dodefn, "deftypevar", 10 }, /* TEXICMD_DEFTYPEVAR */ + { dodefn, "deftypevarx", 11 }, /* TEXICMD_DEFTYPEVARX */ + { dodefn, "deftypevr", 9 }, /* TEXICMD_DEFTYPEVR */ + { dodefn, "deftypevrx", 10 }, /* TEXICMD_DEFTYPEVRX */ + { dodefn, "defun", 5 }, /* TEXICMD_DEFUN */ + { dodefn, "defunx", 6 }, /* TEXICMD_DEFUNX */ + { dodefn, "defvar", 6 }, /* TEXICMD_DEFVAR */ + { dodefn, "defvarx", 7 }, /* TEXICMD_DEFVARX */ + { dodefn, "defvr", 5 }, /* TEXICMD_DEFVR */ + { dodefn, "defvrx", 6 }, /* TEXICMD_DEFVRX */ { doignblock, "detailmenu", 10 }, /* TEXICMD_DETAILMENU */ - { doitalic, "dfn", 3 }, /* TEXICMD_DFN */ + { doinline, "dfn", 3 }, /* TEXICMD_DFN */ { doignline, "dircategory", 11 }, /* TEXICMD_DIRCATEGORY */ { doignblock, "direntry", 8 }, /* TEXICMD_DIRENTRY */ { dodisplay, "display", 7 }, /* TEXICMD_DISPLAY */ + { doignbracket, "dmn", 3 }, /* TEXICMD_DMN */ + { doignblock, "documentdescription", 19 }, /* TEXICMD_DOCUMENTDESCRIPTION */ + { doignline, "documentencoding", 16 }, /* TEXICMD_DOCUMENTENCODING */ { dosymbol, "dots", 4 }, /* TEXICMD_DOTS */ { dolink, "email", 5 }, /* TEXICMD_EMAIL */ - { doemph, "emph", 4 }, /* TEXICMD_EMPH */ + { doinline, "emph", 4 }, /* TEXICMD_EMPH */ { NULL, "end", 3 }, /* TEXICMD_END */ + { dosymbol, "enddots", 7 }, /* TEXICMD_ENDDOTS */ { doenumerate, "enumerate", 9 }, /* TEXICMD_ENUMERATE */ - { doenv, "env", 3 }, /* TEXICMD_ENV */ + { doinline, "env", 3 }, /* TEXICMD_ENV */ + { dosymbol, "equiv", 5 }, /* TEXICMD_EQUIV */ + { dosymbol, "error", 5 }, /* TEXICMD_ERROR */ { doexample, "example", 7 }, /* TEXICMD_EXAMPLE */ - { dofile, "file", 4 }, /* TEXICMD_FILE */ + { doignline, "exdent", 6 }, /* TEXICMD_EXDENT */ + { dosymbol, "expansion", 9 }, /* TEXICMD_EXPANSION */ + { doinline, "file", 4 }, /* TEXICMD_FILE */ + { doignline, "finalout", 8 }, /* TEXICMD_FINALOUT */ + { doignline, "findex", 6 }, /* TEXICMD_FINDEX */ + { doignbracket, "footnote", 8 }, /* TEXICMD_FOOTNOTE */ + { dotable, "ftable", 6 }, /* TEXICMD_FTABLE */ + { dodisplay, "format", 6 }, /* TEXICMD_FORMAT */ + { doaccent, "`", 1 }, /* TEXICMD_GRAVE */ { doblock, "group", 5 }, /* TEXICMD_GROUP */ { dosection, "heading", 7 }, /* TEXICMD_HEADING */ { doignline, "headings", 8 }, /* TEXICMD_HEADINGS */ + { doitem, "headitem", 8 }, /* TEXICMD_HEADITEM */ + { doignblock, "html", 4 }, /* TEXICMD_HTML */ { dosymbol, "-", 1 }, /* TEXICMD_HYPHEN */ - { doitalic, "i", 1 }, /* TEXICMD_I */ - { doignblock, "ifclear", 7 }, /* TEXICMD_IFCLEAR */ + { doinline, "i", 1 }, /* TEXICMD_I */ + { dovalue, "ifclear", 7 }, /* TEXICMD_IFCLEAR */ + { doignblock, "ifdocbook", 9 }, /* TEXICMD_IFDOCBOOK */ { doignblock, "ifhtml", 6 }, /* TEXICMD_IFHTML */ - { doignblock, "ifinfo", 6 }, /* TEXICMD_IFINFO */ + { doblock, "ifinfo", 6 }, /* TEXICMD_IFINFO */ + { doblock, "ifnotdocbook", 12 }, /* TEXICMD_IFNOTDOCBOOK */ + { doblock, "ifnothtml", 9 }, /* TEXICMD_IFNOTHTML */ + { doblock, "ifnotinfo", 9 }, /* TEXICMD_IFNOTINFO */ + { doignblock, "ifnotplaintext", 14 }, /* TEXICMD_IFNOTPLAINTEXT */ { doblock, "ifnottex", 8 }, /* TEXICMD_IFNOTTEX */ + { doblock, "ifnotxml", 8 }, /* TEXICMD_IFNOTXML */ + { doblock, "ifplaintext", 11 }, /* TEXICMD_IFPLAINTEXT */ { doignblock, "iftex", 5 }, /* TEXICMD_IFTEX */ - { doignblock, "ifset", 5 }, /* TEXICMD_IFSET */ + { dovalue, "ifset", 5 }, /* TEXICMD_IFSET */ + { doignblock, "ifxml", 5 }, /* TEXICMD_IFXML */ + { doignblock, "ignore", 6 }, /* TEXICMD_IGNORE */ { doignbracket, "image", 5 }, /* TEXICMD_IMAGE */ { doinclude, "include", 7 }, /* TEXICMD_INCLUDE */ + { dodisplay, "indentblock", 11 }, /* TEXICMD_INDENTBLOCK */ + { dolink, "inforef", 7 }, /* TEXICMD_INDENTBLOCK */ { doignline, "insertcopying", 13 }, /* TEXICMD_INSERTCOPYING */ { doitem, "item", 4 }, /* TEXICMD_ITEM */ { doitemize, "itemize", 7 }, /* TEXICMD_ITEMIZE */ - { doliteral, "kbd", 3 }, /* TEXICMD_KBD */ + { doitem, "itemx", 5 }, /* TEXICMD_ITEMX */ + { doinline, "kbd", 3 }, /* TEXICMD_KBD */ + { dobracket, "key", 3 }, /* TEXICMD_KEY */ + { doignline, "kindex", 6 }, /* TEXICMD_KINDEX */ { dosymbol, "LaTeX", 5 }, /* TEXICMD_LATEX */ + { dosecoffs, "lowersections", 13 }, /* TEXICMD_LOWERSECTIONS */ + { domacro, "macro", 5 }, /* TEXICMD_MACRO */ { domath, "math", 4 }, /* TEXICMD_MATH */ { doignblock, "menu", 4 }, /* TEXICMD_MENU */ + { dosymbol, "minus", 5 }, /* TEXICMD_MINUS */ + { domultitable, "multitable", 10 }, /* TEXICMD_MULTITABLE */ + { doignline, "need", 4 }, /* TEXICMD_NEED */ { dosymbol, "\n", 1 }, /* TEXICMD_NEWLINE */ { doignline, "node", 4 }, /* TEXICMD_NODE */ { doignline, "noindent", 8 }, /* TEXICMD_NOINDENT */ + { doinline, "option", 6 }, /* TEXICMD_OPTION */ { dolink, "pxref", 5 }, /* TEXICMD_PXREF */ { dosymbol, "?", 1 }, /* TEXICMD_QUESTIONMARK */ { doquotation, "quotation", 9 }, /* TEXICMD_QUOTATION */ { doignline, "page", 4 }, /* TEXICMD_PAGE */ - { doignline, "paragraphindent", 14 }, /* TEXICMD_PARINDENT */ + { doignline, "paragraphindent", 15 }, /* TEXICMD_PARINDENT */ + { dosymbol, ".", 1 }, /* TEXICMD_PERIOD */ + { doignline, "pindex", 6 }, /* TEXICMD_PINDEX */ { doignline, "printindex", 10 }, /* TEXICMD_PRINTINDEX */ + { doinline, "r", 1 }, /* TEXICMD_R */ + { dosecoffs, "raisesections", 13 }, /* TEXICMD_RAISESECTIONS */ { dobracket, "ref", 3 }, /* TEXICMD_REF */ - { doliteral, "samp", 4 }, /* TEXICMD_SAMP */ + { doignline, "refill", 6 }, /* TEXICMD_REFILL */ + { dosymbol, "result", 6 }, /* TEXICMD_RESULT */ + { doinline, "samp", 4 }, /* TEXICMD_SAMP */ + { doinline, "sansserif", 9 }, /* TEXICMD_SANSSERIF */ { dobracket, "sc", 2 }, /* TEXICMD_SC */ { dosection, "section", 7 }, /* TEXICMD_SECTION */ - { doignline, "set", 3 }, /* TEXICMD_SET */ + { dovalue, "set", 3 }, /* TEXICMD_SET */ { doignline, "setchapternewpage", 17 }, /* TEXICMD_SETCHAPNEWPAGE */ + { doignline, "setcontentsaftertitlepage", 25 }, /* TEXICMD_SETCONTENTSAFTER */ { doignline, "setfilename", 11 }, /* TEXICMD_SETFILENAME */ { dotitle, "settitle", 8 }, /* TEXICMD_SETTITLE */ + { doignline, "shortcontents", 13 }, /* TEXICMD_SHORTCONTENTS */ + { doinline, "slanted", 7 }, /* TEXICMD_SLANTED */ + { dosymbol, "/", 1 }, /* TEXICMD_SLASH */ { dosp, "sp", 2 }, /* TEXICMD_SP */ { dosymbol, " ", 1 }, /* TEXICMD_SPACE */ + { doignline, "smallbook", 9 }, /* TEXICMD_SMALLBOOK */ + { dodisplay, "smalldisplay", 12 }, /* TEXICMD_SMALLDISPLAY */ { doexample, "smallexample", 12 }, /* TEXICMD_SMALLEXAMPLE */ + { dodisplay, "smallformat", 11 }, /* TEXICMD_SMALLFORMAT */ + { dodisplay, "smallindentblock", 16 }, /* TEXICMD_SMALLINDENTBLOCK */ { dosymbol, "{", 1 }, /* TEXICMD_SQUIGGLE_LEFT */ { dosymbol, "}", 1 }, /* TEXICMD_SQUIGGLE_RIGHT */ - { doemph, "strong", 6 }, /* TEXICMD_STRONG */ + { dosymbol, "ss", 2 }, /* TEXICMD_SS */ + { doinline, "strong", 6 }, /* TEXICMD_STRONG */ + { dosubsection, "subheading", 10 }, /* TEXICMD_SUBHEADING */ { dosubsection, "subsection", 10 }, /* TEXICMD_SUBSECTION */ + { dosubsubsection, "subsubheading", 13 }, /* TEXICMD_SUBSUBHEADING */ + { dosubsubsection, "subsubsection", 13 }, /* TEXICMD_SUBSUBSECTION */ { doignline, "subtitle", 8 }, /* TEXICMD_SUBTITLE */ - { dosymbol, "\t", 1 }, /* TEXICMD_TAB */ + { doignline, "summarycontents", 15 }, /* TEXICMD_SUMMARYCONTENTS */ + { doignline, "synindex", 8 }, /* TEXICMD_SYNINDEX */ + { doignline, "syncodeindex", 12 }, /* TEXICMD_SYNCODEINDEX */ + { doinline, "t", 1 }, /* TEXICMD_T */ + { dotab, "tab", 3 }, /* TEXICMD_TAB */ + { dosymbol, "\t", 1 }, /* TEXICMD_TABSYM */ { dotable, "table", 5 }, /* TEXICMD_TABLE */ { doignblock, "tex", 3 }, /* TEXICMD_TEX */ { dosymbol, "TeX", 3 }, /* TEXICMD_TEXSYM */ + { dosymbol, "tie", 3 }, /* TEXICMD_TIE */ + { doaccent, "~", 1 }, /* TEXICMD_TILDE */ + { doignline, "tindex", 6 }, /* TEXICMD_TINDEX */ { doignline, "title", 5 }, /* TEXICMD_TITLE */ { dobracket, "titlefont", 9 }, /* TEXICMD_TITLEFONT */ { doignblock, "titlepage", 9 }, /* TEXICMD_TITLEPAGE */ { dotop, "top", 3 }, /* TEXICMD_TOP */ - { dochapter, "unnumbered", 10 }, /* TEXICMD_UNNUMBERED */ + { doaccent, "\"", 1 }, /* TEXICMD_UMLAUT */ + { dosection, "unnumbered", 10 }, /* TEXICMD_UNNUMBERED */ { dosection, "unnumberedsec", 13 }, /* TEXICMD_UNNUMBEREDSEC */ + { dosubsection, "unnumberedsubsec", 16 }, /* TEXICMD_UNNUMBEREDSUBSEC */ + { dosubsubsection, "unnumberedsubsubsec", 19 }, /* TEXICMD_UNNUMBEREDSUBSUBSEC */ { dolink, "uref", 4 }, /* TEXICMD_UREF */ { dolink, "url", 3 }, /* TEXICMD_URL */ - { doliteral, "var", 3 }, /* TEXICMD_VAR */ + { doignline, "", 0 }, /* TEXICMD_USER_INDEX */ + { dovalue, "value", 5 }, /* TEXICMD_VALUE */ + { doinline, "var", 3 }, /* TEXICMD_VAR */ + { doverb, "verb", 4 }, /* TEXICMD_VERB */ + { doverbatim, "verbatim", 8 }, /* TEXICMD_VERBATIM */ + { doverbinclude, "verbatiminclude", 15 }, /* TEXICMD_VERBATIMINCLUDE */ + { doignline, "vindex", 6 }, /* TEXICMD_VINDEX */ { dosp, "vskip", 5 }, /* TEXICMD_VSKIP */ + { dotable, "vtable", 6 }, /* TEXICMD_VTABLE */ { dobracket, "w", 1 }, /* TEXICMD_W */ { dolink, "xref", 4 }, /* TEXICMD_XREF */ + /* TEXICMD__END */ }; -/* - * Unmap the top-most file that we're using. - */ -static void -texifilepop(struct texi *p) -{ - struct texifile *f; +const struct texitok *const texitoks = __texitoks; - assert(p->filepos > 0); - f = &p->files[--p->filepos]; - munmap(f->map, f->mapsz); -} - -/* - * Unmap all files that we're currently using and free all resources - * that we've allocated during the parse. - * The utility should exit(...) after this is called. - */ static void -texiexit(struct texi *p) +dodefindex(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) { - size_t i; + size_t start, end; + char *cp; - if (p->outcol) - putchar('\n'); + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); - while (p->filepos > 0) - texifilepop(p); + start = end = *pos; + while (end < sz && ! ismspace(buf[end])) + end++; - for (i = 0; i < p->dirsz; i++) - free(p->dirs[i]); + if (start == end) { + advanceeoln(p, buf, sz, pos, 1); + return; + } else if (NULL == (cp = malloc(end - start + 1))) + texiabort(p, NULL); - free(p->dirs); - free(p->subtitle); - free(p->title); -} + memcpy(cp, &buf[start], end - start); + cp[end - start] = '\0'; -/* - * Fatal error: unmap all files and exit. - * The "errstring" is passed to perror(3). - */ -static void -texiabort(struct texi *p, const char *errstring) -{ + p->indexs = realloc(p->indexs, + sizeof(char *) * (p->indexsz + 1)); - perror(errstring); - texiexit(p); - exit(EXIT_FAILURE); + if (NULL == p->indexs) + texiabort(p, NULL); + p->indexs[p->indexsz++] = cp; } -/* - * Print a generic warning message (to stderr) tied to our current - * location in the parse sequence. - */ static void -texiwarn(const struct texi *p, const char *fmt, ...) +dodefn(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) { - va_list ap; + const char *blk; - fprintf(stderr, "%s:%zu:%zu: warning: ", - p->files[p->filepos - 1].name, - p->files[p->filepos - 1].line + 1, - p->files[p->filepos - 1].col + 1); - va_start(ap, fmt); - vfprintf(stderr, fmt, ap); - va_end(ap); - fputc('\n', stderr); -} + blk = NULL; + switch (cmd) { + case (TEXICMD_DEFFN): + case (TEXICMD_DEFMAC): + case (TEXICMD_DEFTP): + case (TEXICMD_DEFTYPEFN): + case (TEXICMD_DEFTYPEFUN): + case (TEXICMD_DEFTYPEMETHOD): + case (TEXICMD_DEFTYPEVAR): + case (TEXICMD_DEFTYPEVR): + case (TEXICMD_DEFUN): + case (TEXICMD_DEFVAR): + case (TEXICMD_DEFVR): + blk = texitoks[cmd].tok; + break; + default: + break; + } -/* - * Print an error message (to stderr) tied to our current location in - * the parse sequence, invoke texiexit(), then die. - */ -static void -texierr(struct texi *p, const char *fmt, ...) -{ - va_list ap; - - fprintf(stderr, "%s:%zu:%zu: error: ", - p->files[p->filepos - 1].name, - p->files[p->filepos - 1].line + 1, - p->files[p->filepos - 1].col + 1); - va_start(ap, fmt); - vfprintf(stderr, fmt, ap); - va_end(ap); - fputc('\n', stderr); - texiexit(p); - exit(EXIT_FAILURE); -} - -/* - * Put a single data character to the output if we're not ignoring. - * Adjusts our output status. - */ -static void -texiputchar(struct texi *p, char c) -{ - - if (p->ign) + if (p->ign) { + NULL != blk ? + parseto(p, buf, sz, pos, blk) : + parseeoln(p, buf, sz, pos); return; - putchar(c); - if ('\n' == c) { - p->outcol = 0; - p->seenws = 0; - } else - p->outcol++; -} + } -/* - * Put multiple characters (see texiputchar()). - */ -static void -texiputchars(struct texi *p, const char *s) -{ + if (NULL != blk) + texivspace(p); - while ('\0' != *s) - texiputchar(p, *s++); -} + switch (cmd) { + case (TEXICMD_DEFTYPEMETHOD): + case (TEXICMD_DEFTYPEMETHODX): + texiputchars(p, "Method"); + break; + case (TEXICMD_DEFMAC): + case (TEXICMD_DEFMACX): + texiputchars(p, "Macro"); + break; + case (TEXICMD_DEFTYPEVAR): + case (TEXICMD_DEFTYPEVARX): + case (TEXICMD_DEFVAR): + case (TEXICMD_DEFVARX): + texiputchars(p, "Variable"); + break; + case (TEXICMD_DEFTYPEFUN): + case (TEXICMD_DEFTYPEFUNX): + case (TEXICMD_DEFUN): + case (TEXICMD_DEFUNX): + texiputchars(p, "Function"); + break; + default: + parselinearg(p, buf, sz, pos); + break; + } -/* - * Close an mdoc(7) macro opened with teximacroopen(). - * If there are no more macros on the line, prints a newline. - */ -static void -teximacroclose(struct texi *p) -{ + texiputchars(p, ":\n"); - /* FIXME: punctuation. */ - if (0 == --p->outmacro) - texiputchar(p, '\n'); -} + switch (cmd) { + case (TEXICMD_DEFMAC): + case (TEXICMD_DEFMACX): + teximacroopen(p, "Dv"); + while (parselinearg(p, buf, sz, pos)) + /* Spin. */ ; + teximacroclose(p); + break; + case (TEXICMD_DEFFN): + case (TEXICMD_DEFFNX): + case (TEXICMD_DEFUN): + case (TEXICMD_DEFUNX): + teximacroopen(p, "Fo"); + parselinearg(p, buf, sz, pos); + teximacroclose(p); + teximacroopen(p, "Fa"); + while (parselinearg(p, buf, sz, pos)) + /* Spin. */ ; + teximacroclose(p); + teximacro(p, "Fc"); + break; + case (TEXICMD_DEFTYPEFUN): + case (TEXICMD_DEFTYPEFUNX): + case (TEXICMD_DEFTYPEFN): + case (TEXICMD_DEFTYPEFNX): + case (TEXICMD_DEFTYPEMETHOD): + case (TEXICMD_DEFTYPEMETHODX): + teximacroopen(p, "Ft"); + parselinearg(p, buf, sz, pos); + teximacroclose(p); + teximacroopen(p, "Fo"); + parselinearg(p, buf, sz, pos); + teximacroclose(p); + teximacroopen(p, "Fa"); + while (parselinearg(p, buf, sz, pos)) + /* Spin. */ ; + teximacroclose(p); + teximacro(p, "Fc"); + break; + case (TEXICMD_DEFTP): + case (TEXICMD_DEFTPX): + case (TEXICMD_DEFTYPEVAR): + case (TEXICMD_DEFTYPEVARX): + case (TEXICMD_DEFTYPEVR): + case (TEXICMD_DEFTYPEVRX): + teximacroopen(p, "Vt"); + while (parselinearg(p, buf, sz, pos)) + /* Spin. */ ; + teximacroclose(p); + break; + case (TEXICMD_DEFVAR): + case (TEXICMD_DEFVARX): + case (TEXICMD_DEFVR): + case (TEXICMD_DEFVRX): + teximacroopen(p, "Va"); + while (parselinearg(p, buf, sz, pos)) + /* Spin. */ ; + teximacroclose(p); + break; + default: + abort(); + } -/* - * Open a mdoc(7) macro. - * This is used for line macros, e.g., Qq [foo bar baz]. - * It can be invoked for nested macros, e.g., Qq Li foo . - */ -static void -teximacroopen(struct texi *p, const char *s) -{ - - if (p->outcol && 0 == p->outmacro) - texiputchar(p, '\n'); - if (0 == p->outmacro) - texiputchar(p, '.'); - else - texiputchar(p, ' '); - texiputchars(p, s); - texiputchar(p, ' '); - p->outmacro++; - p->seenws = 0; + texivspace(p); + if (NULL != blk) + parseto(p, buf, sz, pos, blk); } -/* - * Put a stadnalone mdoc(7) command with the trailing newline. - */ static void -teximacro(struct texi *p, const char *s) +domacro(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) { + size_t start, end, endtoksz, len; + struct teximacro m; + const char *endtok, *blk; - if (p->outmacro) - texierr(p, "\"%s\" in open line scope!?", s); - else if (p->literal) - texierr(p, "\"%s\" in a literal scope!?", s); + memset(&m, 0, sizeof(struct teximacro)); - if (p->outcol) - texiputchar(p, '\n'); + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); - texiputchar(p, '.'); - texiputchars(p, s); - texiputchar(p, '\n'); -} + for (start = end = *pos; end < sz; end++) + if (ismspace(buf[end]) || '{' == buf[end]) + break; -/* - * Advance by a single byte in the input stream. - */ -static void -advance(struct texi *p, const char *buf, size_t *pos) -{ + if (start == end) + texierr(p, "zero-length macro name"); - if ('\n' == buf[*pos]) { - p->files[p->filepos - 1].line++; - p->files[p->filepos - 1].col = 0; - } else - p->files[p->filepos - 1].col++; + advanceto(p, buf, pos, end); - (*pos)++; -} + m.key = malloc(end - start + 1); + if (NULL == m.key) + texiabort(p, NULL); + memcpy(m.key, &buf[start], end - start); + m.key[end - start] = '\0'; -/* - * Advance to the next non-whitespace word in the input stream. - * If we're in literal mode, then print all of the whitespace as we're - * doing so. - */ -static size_t -advancenext(struct texi *p, const char *buf, size_t sz, size_t *pos) -{ - - if (p->literal) { - while (*pos < sz && ismspace(buf[*pos])) { - if (*pos && '\n' == buf[*pos] && - '\\' == buf[*pos - 1]) - texiputchar(p, 'e'); - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - return(*pos); - } + m.args = argparse(p, buf, sz, pos, &m.argsz, 0); + advanceeoln(p, buf, sz, pos, 0); - while (*pos < sz && ismspace(buf[*pos])) { - p->seenws = 1; - /* - * If it looks like we've printed a double-line, then - * output a paragraph. - * FIXME: this is stupid. - */ - if (*pos && '\n' == buf[*pos] && - '\n' == buf[*pos - 1]) - teximacro(p, "Pp"); + start = *pos; + endtok = "\n@end macro\n"; + endtoksz = strlen(endtok); + blk = memmem(&buf[start], sz, endtok, endtoksz); + if (NULL == blk) + texierr(p, "unterminated macro body"); + while (&buf[*pos] != blk) advance(p, buf, pos); - } - return(*pos); -} + assert('\n' == buf[*pos]); + advance(p, buf, pos); + len = blk - &buf[start]; + m.value = malloc(len + 1); + if (NULL == m.value) + texiabort(p, NULL); + memcpy(m.value, &buf[start], len); + m.value[len] = '\0'; -/* - * Advance to the EOLN in the input stream. - */ -static size_t -advanceeoln(struct texi *p, const char *buf, - size_t sz, size_t *pos, int consumenl) -{ + p->macros = realloc + (p->macros, + (p->macrosz + 1) * + sizeof(struct teximacro)); + if (NULL == p->macros) + texiabort(p, NULL); - /* FIXME: disregards @NEWLINE. */ - while (*pos < sz && '\n' != buf[*pos]) - advance(p, buf, pos); - if (*pos < sz && consumenl) - advance(p, buf, pos); - return(*pos); + p->macros[p->macrosz++] = m; + advanceeoln(p, buf, sz, pos, 1); } -/* - * Advance to position "end", which is an absolute position in the - * current buffer greater than or equal to the current position. - */ static void -advanceto(struct texi *p, const char *buf, size_t *pos, size_t end) +doignblock(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) { + char end[32]; + const char *term; + size_t endsz, endpos; - assert(*pos <= end); - while (*pos < end) + /* + * We want to completely ignore everything in these blocks, so + * simply jump to the @end block. + */ + endsz = snprintf(end, sizeof(end), + "\n@end %s\n", texitoks[cmd].tok); + assert(endsz < sizeof(end)); + + /* + * Look up where our end token occurs. + * Set our end position based on the relative offset of that + * from our current position, or the EOF if we don't have a + * proper ending point. + */ + term = memmem(&buf[*pos], sz, end, endsz); + endpos = NULL == term ? sz : + *pos + (size_t)(term - &buf[*pos]); + assert(endpos <= sz); + while (*pos < endpos) advance(p, buf, pos); + + /* Only do this if we're not already at the end. */ + if (endpos < sz) + advanceto(p, buf, pos, endpos + endsz); } -/* - * Output a free-form word in the input stream, progressing to the next - * command or white-space. - * This also will advance the input stream. - */ static void -texiword(struct texi *p, const char *buf, - size_t sz, size_t *pos, char extra) +doblock(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) { - - if (0 == p->outmacro && p->outcol > 72 && 0 == p->literal) - texiputchar(p, '\n'); - /* FIXME: abstract this: we use it elsewhere. */ - if (p->seenws && p->outcol && 0 == p->literal) - texiputchar(p, ' '); - - p->seenws = 0; - - while (*pos < sz && ! ismspace(buf[*pos])) { - switch (buf[*pos]) { - case ('@'): - case ('}'): - case ('{'): - return; - } - if ('\0' != extra && buf[*pos] == extra) - return; - if (*pos < sz - 1 && - '`' == buf[*pos] && - '`' == buf[*pos + 1]) { - texiputchars(p, "\\(lq"); - advance(p, buf, pos); - } else if (*pos < sz - 1 && - '\'' == buf[*pos] && - '\'' == buf[*pos + 1]) { - texiputchars(p, "\\(rq"); - advance(p, buf, pos); - } else - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } + + parseto(p, buf, sz, pos, texitoks[cmd].tok); } -/* - * Look up the command at position "pos" in the buffer, returning it (or - * TEXICMD__MAX if none found) and setting "end" to be the absolute - * index after the command name. - */ -static enum texicmd -texicmd(struct texi *p, const char *buf, - size_t pos, size_t sz, size_t *end) +static void +doinline(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) { - size_t i, len; + const char *macro = NULL; - assert('@' == buf[pos]); - - if ((*end = pos) == sz) - return(TEXICMD__MAX); - else if ((*end = ++pos) == sz) - return(TEXICMD__MAX); - - /* Alphabetic commands are special. */ - if ( ! isalpha(buf[pos])) { - if ((*end = pos + 1) == sz) - return(TEXICMD__MAX); - for (i = 0; i < TEXICMD__MAX; i++) { - if (1 != texitoks[i].len) - continue; - if (0 == strncmp(texitoks[i].tok, &buf[pos], 1)) - return(i); - } - texiwarn(p, "bad command: @%c", buf[pos]); - return(TEXICMD__MAX); + switch (cmd) { + case (TEXICMD_CODE): + case (TEXICMD_KBD): + case (TEXICMD_SAMP): + case (TEXICMD_T): + macro = "Li"; + break; + case (TEXICMD_CITE): + case (TEXICMD_DFN): + case (TEXICMD_EMPH): + case (TEXICMD_I): + case (TEXICMD_SLANTED): + macro = "Em"; + break; + case (TEXICMD_B): + case (TEXICMD_STRONG): + macro = "Sy"; + break; + case (TEXICMD_COMMAND): + macro = "Xr"; + break; + case (TEXICMD_ENV): + macro = "Ev"; + break; + case (TEXICMD_FILE): + macro = "Pa"; + break; + case (TEXICMD_OPTION): + macro = "Op"; + break; + case (TEXICMD_VAR): + macro = "Va"; + break; + default: + break; } - for (*end = pos; *end < sz && ! ismspace(buf[*end]); (*end)++) - if ((*end > pos && ('@' == buf[*end] || - '{' == buf[*end] || '}' == buf[*end]))) - break; - - len = *end - pos; - for (i = 0; i < TEXICMD__MAX; i++) { - if (len != texitoks[i].len) - continue; - if (0 == strncmp(texitoks[i].tok, &buf[pos], len)) - return(i); + if (NULL == macro || p->literal || TEXILIST_TABLE == p->list) { + parsebracket(p, buf, sz, pos); + return; } - texiwarn(p, "bad command: @%.*s", (int)len, &buf[pos]); - return(TEXICMD__MAX); + teximacroopen(p, macro); + p->seenws = 0; + parsebracket(p, buf, sz, pos); + texipunctuate(p, buf, sz, pos); + teximacroclose(p); } -/* - * Parse an argument from a bracketed command, e.g., @url{foo, baz}. - * Num should be set to the argument we're currently parsing, although - * it suffixes for it to be zero or non-zero. - * This will return 1 if there are more arguments, 0 otherwise. - * This will stop (returning 0) in the event of EOF or if we're not at a - * bracket for the zeroth parse. - */ -static int -parsearg(struct texi *p, const char *buf, - size_t sz, size_t *pos, size_t num) -{ - size_t end; - enum texicmd cmd; - - while (*pos < sz && ismspace(buf[*pos])) - advance(p, buf, pos); - if (*pos == sz || (0 == num && '{' != buf[*pos])) - return(0); - if (0 == num) - advance(p, buf, pos); - - while ((*pos = advancenext(p, buf, sz, pos)) < sz) { - switch (buf[*pos]) { - case (','): - advance(p, buf, pos); - return(1); - case ('}'): - advance(p, buf, pos); - return(0); - case ('{'): - if (0 == p->ign) - texiwarn(p, "unexpected \"{\""); - advance(p, buf, pos); - continue; - case ('@'): - break; - default: - texiword(p, buf, sz, pos, ','); - continue; - } - - cmd = texicmd(p, buf, *pos, sz, &end); - advanceto(p, buf, pos, end); - if (TEXICMD__MAX == cmd) - continue; - if (NULL != texitoks[cmd].fp) - (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); - } - return(0); -} - -/* - * Parse until the end of a bracketed statement, e.g., @foo{bar baz}. - * This will stop in the event of EOF or if we're not at a bracket. - */ static void -parsebracket(struct texi *p, const char *buf, size_t sz, size_t *pos) +doverb(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) { - size_t end; - enum texicmd cmd; + char delim; + size_t start; - while (*pos < sz && ismspace(buf[*pos])) + while (*pos < sz && isws(buf[*pos])) advance(p, buf, pos); - if (*pos == sz || '{' != buf[*pos]) return; advance(p, buf, pos); + if (*pos == sz) + return; - while ((*pos = advancenext(p, buf, sz, pos)) < sz) { - switch (buf[*pos]) { - case ('}'): - advance(p, buf, pos); - return; - case ('{'): - if (0 == p->ign) - texiwarn(p, "unexpected \"{\""); - advance(p, buf, pos); - continue; - case ('@'): + delim = buf[*pos]; + advance(p, buf, pos); + /* Make sure we flush out our initial whitespace... */ + if (p->seenws && p->outcol && 0 == p->literal) + texiputchar(p, ' '); + p->seenws = 0; + start = *pos; + /* Read until we see the delimiter then end-brace. */ + while (*pos < sz - 1) { + if (buf[*pos] == delim && buf[*pos + 1] == '}') break; - default: - texiword(p, buf, sz, pos, '\0'); - continue; - } - - cmd = texicmd(p, buf, *pos, sz, &end); - advanceto(p, buf, pos, end); - if (TEXICMD__MAX == cmd) - continue; - if (NULL != texitoks[cmd].fp) - (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); + advance(p, buf, pos); } + if (*pos == sz - 1) + return; + texiputbuf(p, buf, start, *pos); + + /* Make sure we read after the end-brace. */ + assert(delim == buf[*pos]); + advance(p, buf, pos); + assert('}' == buf[*pos]); + advance(p, buf, pos); } -/* - * This should be invoked when we're on a macro line and want to process - * to the end of the current input line, doing all of our macros along - * the way. - */ static void -parseeoln(struct texi *p, const char *buf, size_t sz, size_t *pos) +doverbatim(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) { - size_t end; - enum texicmd cmd; + const char *end, *term; + size_t endsz, endpos; - assert(0 == p->literal); + advanceeoln(p, buf, sz, pos, 1); - while (*pos < sz && '\n' != buf[*pos]) { - while (*pos < sz && isws(buf[*pos])) { - p->seenws = 1; - advance(p, buf, pos); - } - switch (buf[*pos]) { - case ('}'): - if (0 == p->ign) - texiwarn(p, "unexpected \"}\""); - advance(p, buf, pos); - continue; - case ('{'): - if (0 == p->ign) - texiwarn(p, "unexpected \"{\""); - advance(p, buf, pos); - continue; - case ('@'): - break; - default: - texiword(p, buf, sz, pos, '\0'); - continue; - } + /* We end at exactly this token. */ + end = "\n@end verbatim\n"; + endsz = strlen(end); - cmd = texicmd(p, buf, *pos, sz, &end); - advanceto(p, buf, pos, end); - if (TEXICMD__MAX == cmd) - continue; - if (NULL != texitoks[cmd].fp) - (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); - } -} + /* + * Look up where our end token occurs. + * Set our end position based on the relative offset of that + * from our current position. + */ + term = memmem(&buf[*pos], sz, end, endsz); + endpos = NULL == term ? sz : + *pos + (size_t)(term - &buf[*pos]); -/* - * Parse a single word or command. - * This will return immediately at the EOF. - */ -static void -parsesingle(struct texi *p, const char *buf, size_t sz, size_t *pos) -{ - size_t end; - enum texicmd cmd; - - if ((*pos = advancenext(p, buf, sz, pos)) >= sz) - return; - - switch (buf[*pos]) { - case ('}'): - if (0 == p->ign) - texiwarn(p, "unexpected \"}\""); + teximacro(p, "Bd -literal -offset indent"); + assert(endpos <= sz); + while (*pos < endpos) { + texiputchar(p, buf[*pos]); advance(p, buf, pos); - return; - case ('{'): - if (0 == p->ign) - texiwarn(p, "unexpected \"{\""); - advance(p, buf, pos); - return; - case ('@'): - break; - default: - texiword(p, buf, sz, pos, '\0'); - return; } - - cmd = texicmd(p, buf, *pos, sz, &end); - advanceto(p, buf, pos, end); - if (TEXICMD__MAX == cmd) - return; - if (NULL != texitoks[cmd].fp) - (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); + teximacro(p, "Ed"); + advanceto(p, buf, pos, endpos + endsz); } -/* - * Parse til the end of the buffer. - */ static void -parseeof(struct texi *p, const char *buf, size_t sz) +doverbinclude(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) { - size_t pos; + char fname[PATH_MAX], path[PATH_MAX]; + int rc; + size_t i, end; + const char *v; + enum texicmd type; - for (pos = 0; pos < sz; ) - parsesingle(p, buf, sz, &pos); -} + while (*pos < sz && ' ' == buf[*pos]) + advance(p, buf, pos); -/* - * Parse a block sequence until we have the "@end endtoken" command - * invocation. - * This will return immediately at EOF. - */ -static void -parseto(struct texi *p, const char *buf, - size_t sz, size_t *pos, const char *endtoken) -{ - size_t end; - enum texicmd cmd; - size_t endtoksz; - - endtoksz = strlen(endtoken); - assert(endtoksz > 0); - - while ((*pos = advancenext(p, buf, sz, pos)) < sz) { - switch (buf[*pos]) { - case ('}'): - if (0 == p->ign) - texiwarn(p, "unexpected \"}\""); + for (i = 0; *pos < sz && '\n' != buf[*pos]; ) { + if (i == sizeof(fname) - 1) + break; + if ('@' != buf[*pos]) { + fname[i++] = buf[*pos]; advance(p, buf, pos); continue; - case ('{'): - if (0 == p->ign) - texiwarn(p, "unexpected \"{\""); - advance(p, buf, pos); - continue; - case ('@'): - break; - default: - texiword(p, buf, sz, pos, '\0'); - continue; } - - cmd = texicmd(p, buf, *pos, sz, &end); + type = texicmd(p, buf, *pos, sz, &end, NULL); advanceto(p, buf, pos, end); - if (TEXICMD_END == cmd) { - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); - /* - * FIXME: check the full word, not just its - * initial substring! - */ - if (sz - *pos >= endtoksz && 0 == strncmp - (&buf[*pos], endtoken, endtoksz)) { - advanceeoln(p, buf, sz, pos, 0); - break; - } - if (0 == p->ign) - texiwarn(p, "unexpected \"end\""); - advanceeoln(p, buf, sz, pos, 0); + if (TEXICMD_VALUE != type) + texierr(p, "unknown verbatiminclude command"); + v = valueblookup(p, buf, sz, pos); + if (NULL == v) continue; - } else if (TEXICMD__MAX != cmd) - if (NULL != texitoks[cmd].fp) - (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); + while ('\0' != *v) { + if (i == sizeof(fname) - 1) + break; + fname[i++] = *v++; + } + if ('\0' != *v) + break; } -} -/* - * Memory-map the file "fname" and begin parsing it. - * This can be called in a nested context. - */ -static void -parsefile(struct texi *p, const char *fname) -{ - struct texifile *f; - int fd; - struct stat st; + if (i == 0) + texierr(p, "path too short"); + else if ('\n' != buf[*pos]) + texierr(p, "path too long"); + else if ('/' == fname[0]) + texierr(p, "no absolute paths"); + fname[i] = '\0'; - assert(p->filepos < 64); - f = &p->files[p->filepos]; - memset(f, 0, sizeof(struct texifile)); + if (strstr(fname, "../") || strstr(fname, "/..")) + texierr(p, "insecure path"); - f->name = fname; - if (-1 == (fd = open(fname, O_RDONLY, 0))) { - texiabort(p, fname); - } else if (-1 == fstat(fd, &st)) { - close(fd); - texiabort(p, fname); - } + rc = snprintf(path, sizeof(path), + "%s/%s", p->dirs[0], fname); + if (rc < 0) + texierr(p, "couldn't format path"); + else if ((size_t)rc >= sizeof(path)) + texierr(p, "path too long"); - f->mapsz = st.st_size; - f->map = mmap(NULL, f->mapsz, - PROT_READ, MAP_SHARED, fd, 0); - close(fd); - - if (MAP_FAILED == f->map) - texiabort(p, fname); - - p->filepos++; - parseeof(p, f->map, f->mapsz); - texifilepop(p); + parsefile(p, path, 0); } static void -dodeftypevar(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ - const char *blk; - - blk = TEXICMD_DEFTYPEVR == cmd ? - "deftypevr" : "deftypevar"; - - if (p->ign) { - parseto(p, buf, sz, pos, blk); - return; - } - - teximacro(p, "Pp"); - if (TEXICMD_DEFTYPEVR == cmd) { - parsebracket(p, buf, sz, pos); - texiputchars(p, ":\n"); - } - teximacroopen(p, "Vt"); - parseeoln(p, buf, sz, pos); - teximacroclose(p); - teximacro(p, "Pp"); - parseto(p, buf, sz, pos, blk); -} - -static void -dodeftypefun(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ - const char *blk; - - blk = NULL; - switch (cmd) { - case (TEXICMD_DEFTYPEFN): - case (TEXICMD_DEFTYPEFUN): - blk = texitoks[cmd].tok; - break; - default: - break; - } - - if (p->ign) { - if (NULL != blk) - parseto(p, buf, sz, pos, blk); - return; - } - - switch (cmd) { - case (TEXICMD_DEFTYPEFN): - case (TEXICMD_DEFTYPEFUN): - teximacro(p, "Pp"); - break; - default: - break; - } - if (TEXICMD_DEFTYPEFN == cmd || - TEXICMD_DEFTYPEFNX == cmd) { - parsebracket(p, buf, sz, pos); - texiputchars(p, ":\n"); - } - teximacroopen(p, "Ft"); - parsesingle(p, buf, sz, pos); - teximacroclose(p); - teximacroopen(p, "Fn"); - parsesingle(p, buf, sz, pos); - teximacroclose(p); - teximacroopen(p, "Li"); - parseeoln(p, buf, sz, pos); - teximacroclose(p); - teximacro(p, "Pp"); - if (NULL != blk) - parseto(p, buf, sz, pos, blk); -} - -static void -doignblock(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ - - p->ign++; - parseto(p, buf, sz, pos, texitoks[cmd].tok); - p->ign--; -} - -static void -doblock(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ - - parseto(p, buf, sz, pos, texitoks[cmd].tok); -} - -static void -doinline(struct texi *p, const char *buf, - size_t sz, size_t *pos, const char *macro) -{ - - teximacroopen(p, macro); - p->seenws = 0; - parsebracket(p, buf, sz, pos); - if (*pos < sz - 1 && - ismpunct(buf[*pos]) && - ismspace(buf[*pos + 1])) { - texiputchar(p, ' '); - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - teximacroclose(p); -} - -static void doinclude(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - char fname[PATH_MAX], path[PATH_MAX]; - size_t i; - int rc; + char fname[PATH_MAX], path[PATH_MAX]; + size_t i, end; + int rc; + const char *v; + enum texicmd type; while (*pos < sz && ' ' == buf[*pos]) advance(p, buf, pos); /* Read in the filename. */ - for (i = 0; *pos < sz && '\n' != buf[*pos]; i++) { + for (i = 0; *pos < sz && '\n' != buf[*pos]; ) { if (i == sizeof(fname) - 1) break; - fname[i] = buf[*pos]; - advance(p, buf, pos); + if ('@' != buf[*pos]) { + fname[i++] = buf[*pos]; + advance(p, buf, pos); + continue; + } + type = texicmd(p, buf, *pos, sz, &end, NULL); + advanceto(p, buf, pos, end); + if (TEXICMD_VALUE != type) + texierr(p, "unknown include command"); + v = valueblookup(p, buf, sz, pos); + if (NULL == v) + continue; + while ('\0' != *v) { + if (i == sizeof(fname) - 1) + break; + fname[i++] = *v++; + } + if ('\0' != *v) + break; } if (i == 0) @@ -1128,7 +793,7 @@ doinclude(struct texi *p, enum texicmd cmd, else if (-1 == access(path, R_OK)) continue; - parsefile(p, path); + parsefile(p, path, 1); return; } @@ -1136,57 +801,6 @@ doinclude(struct texi *p, enum texicmd cmd, } static void -doitalic(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ - - texiputchars(p, "\\fI"); - parsebracket(p, buf, sz, pos); - texiputchars(p, "\\fP"); -} - -static void -doenv(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ - - if (p->literal) - parsebracket(p, buf, sz, pos); - else - doinline(p, buf, sz, pos, "Ev"); -} - -static void -doliteral(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ - - if (p->literal) - parsebracket(p, buf, sz, pos); - else - doinline(p, buf, sz, pos, "Li"); -} - -static void -doemph(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ - - if (p->literal) - doitalic(p, cmd, buf, sz, pos); - else - doinline(p, buf, sz, pos, "Em"); -} - -static void -docommand(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ - - doinline(p, buf, sz, pos, "Xr"); -} - -static void dobracket(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { @@ -1195,24 +809,24 @@ dobracket(struct texi *p, enum texicmd cmd, } static void -dofile(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ - - if (p->literal) - parsebracket(p, buf, sz, pos); - else - doinline(p, buf, sz, pos, "Pa"); -} - -static void dodisplay(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - teximacro(p, "Bd -display -offset indent"); + switch (cmd) { + case (TEXICMD_FORMAT): + case (TEXICMD_SMALLFORMAT): + teximacro(p, "Bd -filled"); + break; + default: + teximacro(p, "Bd -filled -offset indent"); + break; + } + + p->seenvs = 1; + /* FIXME: ignore and parseeoln. */ advanceeoln(p, buf, sz, pos, 1); - parseto(p, buf, sz, pos, "display"); + parseto(p, buf, sz, pos, texitoks[cmd].tok); teximacro(p, "Ed"); } @@ -1220,14 +834,12 @@ static void doexample(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - const char *blk; - blk = TEXICMD_EXAMPLE == cmd ? "example" : "smallexample"; - teximacro(p, "Bd -literal -offset indent"); + /* FIXME: ignore and parseeoln. */ advanceeoln(p, buf, sz, pos, 1); p->literal++; - parseto(p, buf, sz, pos, blk); + parseto(p, buf, sz, pos, texitoks[cmd].tok); p->literal--; teximacro(p, "Ed"); } @@ -1252,13 +864,123 @@ dotitle(struct texi *p, enum texicmd cmd, start = end = *pos; while (end < sz && '\n' != buf[end]) end++; + advanceeoln(p, buf, sz, pos, 1); free(p->subtitle); p->subtitle = malloc(end - start + 1); + if (NULL == p->subtitle) + texiabort(p, NULL); memcpy(p->subtitle, &buf[start], end - start); p->subtitle[end - start] = '\0'; } static void +doaccent(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + + if (*pos == sz) + return; + if (p->seenws && p->outcol && 0 == p->literal) + texiputchar(p, ' '); + p->seenws = 0; + switch (cmd) { + case (TEXICMD_CEDILLA): + /* Strange rules... */ + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + if (*pos == sz || '{' != buf[*pos]) + return; + advance(p, buf, pos); + switch (buf[*pos]) { + case ('c'): case ('C'): + texiputchars(p, "\\(,"); + texiputchar(p, buf[*pos]); + break; + default: + texiputchar(p, buf[*pos]); + break; + } + advance(p, buf, pos); + break; + case (TEXICMD_ACUTE): + switch (buf[*pos]) { + case ('a'): case ('A'): + case ('e'): case ('E'): + case ('i'): case ('I'): + case ('o'): case ('O'): + case ('u'): case ('U'): + texiputchars(p, "\\(\'"); + texiputchar(p, buf[*pos]); + break; + default: + texiputchar(p, buf[*pos]); + } + break; + case (TEXICMD_CIRCUMFLEX): + switch (buf[*pos]) { + case ('a'): case ('A'): + case ('e'): case ('E'): + case ('i'): case ('I'): + case ('o'): case ('O'): + case ('u'): case ('U'): + texiputchars(p, "\\(^"); + texiputchar(p, buf[*pos]); + break; + default: + texiputchar(p, buf[*pos]); + break; + } + break; + case (TEXICMD_GRAVE): + switch (buf[*pos]) { + case ('a'): case ('A'): + case ('e'): case ('E'): + case ('i'): case ('I'): + case ('o'): case ('O'): + case ('u'): case ('U'): + texiputchars(p, "\\(`"); + texiputchar(p, buf[*pos]); + break; + default: + texiputchar(p, buf[*pos]); + } + break; + case (TEXICMD_TILDE): + switch (buf[*pos]) { + case ('a'): case ('A'): + case ('n'): case ('N'): + case ('o'): case ('O'): + texiputchars(p, "\\(~"); + texiputchar(p, buf[*pos]); + break; + default: + texiputchar(p, buf[*pos]); + break; + } + break; + case (TEXICMD_UMLAUT): + switch (buf[*pos]) { + case ('a'): case ('A'): + case ('e'): case ('E'): + case ('i'): case ('I'): + case ('o'): case ('O'): + case ('u'): case ('U'): + case ('y'): + texiputchars(p, "\\(:"); + texiputchar(p, buf[*pos]); + break; + default: + texiputchar(p, buf[*pos]); + break; + } + break; + default: + abort(); + } + advance(p, buf, pos); +} + +static void dosymbol(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { @@ -1272,7 +994,7 @@ dosymbol(struct texi *p, enum texicmd cmd, case (TEXICMD_ASTERISK): case (TEXICMD_NEWLINE): case (TEXICMD_SPACE): - case (TEXICMD_TAB): + case (TEXICMD_TABSYM): texiputchar(p, ' '); break; case (TEXICMD_AT): @@ -1284,18 +1006,46 @@ dosymbol(struct texi *p, enum texicmd cmd, case (TEXICMD_BULLET): texiputchars(p, "\\(bu"); break; + case (TEXICMD_COMMA): + texiputchar(p, ','); + break; case (TEXICMD_COPYRIGHT): texiputchars(p, "\\(co"); break; case (TEXICMD_DOTS): + case (TEXICMD_ENDDOTS): texiputchars(p, "..."); break; + case (TEXICMD_EQUIV): + texiputchars(p, "\\(=="); + break; + case (TEXICMD_ERROR): + texiputchars(p, "error\\(->"); + break; + case (TEXICMD_EXPANSION): + texiputchars(p, "\\(->"); + break; case (TEXICMD_LATEX): texiputchars(p, "LaTeX"); break; + case (TEXICMD_MINUS): + texiputchars(p, "\\-"); + break; + case (TEXICMD_PERIOD): + texiputchar(p, '.'); + break; case (TEXICMD_QUESTIONMARK): texiputchar(p, '?'); break; + case (TEXICMD_RESULT): + texiputchars(p, "\\(rA"); + break; + case (TEXICMD_SLASH): + texiputchar(p, '/'); + break; + case (TEXICMD_SS): + texiputchars(p, "\\(ss"); + break; case (TEXICMD_SQUIGGLE_LEFT): texiputchars(p, "{"); break; @@ -1305,11 +1055,13 @@ dosymbol(struct texi *p, enum texicmd cmd, case (TEXICMD_TEXSYM): texiputchars(p, "TeX"); break; + case (TEXICMD_TIE): + texiputchars(p, "\\ "); + break; case (TEXICMD_COLON): case (TEXICMD_HYPHEN): break; default: - texiwarn(p, "sym: %d", cmd); abort(); } @@ -1331,7 +1083,7 @@ static void domath(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - size_t nest; + size_t nest, start; /* * Math handling is different from everything else. @@ -1347,22 +1099,92 @@ domath(struct texi *p, enum texicmd cmd, if (p->seenws && p->outcol && 0 == p->literal) texiputchar(p, ' '); p->seenws = 0; - for (nest = 1; *pos < sz && nest > 0; ) { + for (nest = 1, start = *pos; *pos < sz && nest > 0; ) { if ('{' == buf[*pos]) nest++; else if ('}' == buf[*pos]) if (0 == --nest) continue; - texiputchar(p, buf[*pos]); advance(p, buf, pos); } if (*pos == sz) return; assert('}' == buf[*pos]); + texiputbuf(p, buf, start, *pos); advance(p, buf, pos); } static void +dovalue(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + size_t start, end, i; + char *key, *val; + const char *cp; + + if (TEXICMD_SET == cmd) { + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + for (start = end = *pos; end < sz; end++) + if (ismspace(buf[end])) + break; + /* We don't allow empty keys. */ + if (start == end) + return; + advanceto(p, buf, pos, end); + + key = malloc(end - start + 1); + if (NULL == key) + texiabort(p, NULL); + memcpy(key, &buf[start], end - start); + key[end - start] = '\0'; + + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + for (start = end = *pos; end < sz; end++) + if ('\n' == buf[end]) + break; + /* We do allow empty values. */ + advanceeoln(p, buf, sz, pos, 1); + + val = malloc(end - start + 1); + if (NULL == val) + texiabort(p, NULL); + memcpy(val, &buf[start], end - start); + val[end - start] = '\0'; + valueadd(p, key, val); + } else if (TEXICMD_VALUE == cmd) { + if (p->seenws) + texiputchar(p, ' '); + p->seenws = 0; + if (NULL != (cp = valueblookup(p, buf, sz, pos))) { + for (i = 0; i < p->valstackpos; i++) + if (cp == p->valstack[i]) + break; + if (i < p->valstackpos) + texierr(p, "recursive value"); + if (64 == p->valstackpos) + texierr(p, "too many nested values"); + p->valstack[p->valstackpos++] = cp; + parsemembuf(p, cp, strlen(cp)); + p->valstackpos--; + } else + texiputchars(p, "{No value}"); + } else if (TEXICMD_IFCLEAR == cmd) { + if (NULL != valuellookup(p, buf, sz, pos)) + doignblock(p, cmd, buf, sz, pos); + else + parseto(p, buf, sz, pos, texitoks[cmd].tok); + } else if (TEXICMD_IFSET == cmd) { + if (NULL == valuellookup(p, buf, sz, pos)) + doignblock(p, cmd, buf, sz, pos); + else + parseto(p, buf, sz, pos, texitoks[cmd].tok); + } else if (TEXICMD_CLEAR == cmd) + valuelclear(p, buf, sz, pos); +} + +static void dolink(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { @@ -1378,12 +1200,16 @@ dolink(struct texi *p, enum texicmd cmd, break; case (TEXICMD_XREF): texiputchars(p, "See Section"); - teximacroopen(p, "Qq"); + teximacroopen(p, "Dq"); break; case (TEXICMD_PXREF): texiputchars(p, "see Section"); - teximacroopen(p, "Qq"); + teximacroopen(p, "Dq"); break; + case (TEXICMD_INFOREF): + texiputchars(p, "See Info file node"); + teximacroopen(p, "Dq"); + break; default: abort(); } @@ -1394,14 +1220,7 @@ dolink(struct texi *p, enum texicmd cmd, c = parsearg(p, buf, sz, pos, 1); p->ign--; - if (*pos < sz - 1 && - ismpunct(buf[*pos]) && - ismspace(buf[*pos + 1])) { - texiputchar(p, ' '); - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - + texipunctuate(p, buf, sz, pos); teximacroclose(p); } @@ -1418,58 +1237,124 @@ doignargn(struct texi *p, enum texicmd cmd, p->ign--; } +/* + * Sections can be made subsections and so on by way of the + * @raiseections and @lowersections commands. + * Perform this check here and return the actual section number adjusted + * to the raise level. + */ +static int +sectioner(struct texi *p, int sec) +{ + + if ((sec -= p->secoffs) < 0) { + texiwarn(p, "section below minimum, clamping"); + return(0); + } else if (sec >= SECTSZ) { + texiwarn(p, "section above maximum, clamping"); + return(SECTSZ - 1); + } + return(sec); +} + static void -dosubsection(struct texi *p, enum texicmd cmd, +dosubsubsection(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + int sec; - teximacro(p, "Pp"); - teximacroopen(p, "Em"); + sec = sectioner(p, 3); + + /* We don't have a subsubsubsection, so make one up. */ + texivspace(p); + teximacroopen(p, sects[sec]); parseeoln(p, buf, sz, pos); teximacroclose(p); - teximacro(p, "Pp"); + texivspace(p); } static void -dosection(struct texi *p, enum texicmd cmd, +dosubsection(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + int sec; + sec = sectioner(p, 2); + if (p->outmacro) - texierr(p, "subsection in open line scope!?"); + texierr(p, "\"%s\" in open line scope!?", sects[sec]); else if (p->literal) - texierr(p, "subsection in a literal scope!?"); + texierr(p, "\"%s\" in a literal scope!?", sects[sec]); - teximacroopen(p, "Ss"); + /* We don't have a subsubsection, so make one up. */ + if (sec > 1) + texivspace(p); + teximacroopen(p, sects[sec]); parseeoln(p, buf, sz, pos); teximacroclose(p); + if (sec > 1) + texivspace(p); } static void -dosp(struct texi *p, enum texicmd cmd, +dosecoffs(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - teximacro(p, "Pp"); - advanceeoln(p, buf, sz, pos, 1); + if (TEXICMD_RAISESECTIONS == cmd) + p->secoffs++; + else + p->secoffs--; } static void -dochapter(struct texi *p, enum texicmd cmd, +dosection(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + int sec; + switch (cmd) { + case (TEXICMD_APPENDIX): + case (TEXICMD_CHAPTER): + case (TEXICMD_TOP): + case (TEXICMD_UNNUMBERED): + sec = sectioner(p, 0); + break; + case (TEXICMD_APPENDIXSEC): + case (TEXICMD_HEADING): + case (TEXICMD_SECTION): + case (TEXICMD_UNNUMBEREDSEC): + sec = sectioner(p, 1); + break; + default: + abort(); + } + if (p->outmacro) - texierr(p, "section in open line scope!?"); + texierr(p, "\"%s\" in open line scope!?", sects[sec]); else if (p->literal) - texierr(p, "section in a literal scope!?"); + texierr(p, "\"%s\" in a literal scope!?", sects[sec]); - teximacroopen(p, "Sh"); + teximacroopen(p, sects[sec]); parseeoln(p, buf, sz, pos); teximacroclose(p); + p->seenvs = 1; } static void +dosp(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + + if (p->literal) + texiputchar(p, '\n'); + else + texivspace(p); + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); +} + +static void dotop(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { @@ -1477,27 +1362,42 @@ dotop(struct texi *p, enum texicmd cmd, time_t t; char date[32]; + if (--p->ign) + texierr(p, "@top command while ignoring (%d)", p->ign); + + /* + * Here we print our standard mdoc(7) prologue. + * We use the title set with @settitle for the `Nd' description + * and the source document filename (the first one as invoked on + * the command line) for the title. + * The date is set to the current date. + */ t = time(NULL); strftime(date, sizeof(date), "%F", localtime(&t)); - p->ign--; - advanceeoln(p, buf, sz, pos, 1); teximacroopen(p, "Dd"); texiputchars(p, date); teximacroclose(p); teximacroopen(p, "Dt"); for (cp = p->title; '\0' != *cp; cp++) texiputchar(p, toupper(*cp)); + texiputchars(p, " 7"); teximacroclose(p); teximacro(p, "Os"); teximacro(p, "Sh NAME"); teximacroopen(p, "Nm"); - texiputchars(p, p->title); + for (cp = p->title; '\0' != *cp; cp++) + texiputchar(p, *cp); teximacroclose(p); teximacroopen(p, "Nd"); - texiputchars(p, NULL != p->subtitle ? - p->subtitle : "Unknown description"); + if (NULL != p->subtitle) + for (cp = p->subtitle; '\0' != *cp; cp++) + texiputchar(p, *cp); + else + texiputchars(p, "Unknown description"); teximacroclose(p); + p->seenvs = 1; + dosection(p, cmd, buf, sz, pos); } static void @@ -1505,6 +1405,12 @@ doitem(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + /* Multitable is using raw tbl(7). */ + if (TEXILIST_TABLE == p->list) { + texiputchar(p, '\n'); + return; + } + if (p->outmacro) texierr(p, "item in open line scope!?"); else if (p->literal) @@ -1518,10 +1424,12 @@ doitem(struct texi *p, enum texicmd cmd, teximacro(p, "It"); break; default: - teximacro(p, "Pp"); + texivspace(p); break; } + /* Trick so we don't start with Pp. */ + p->seenvs = 1; parseeoln(p, buf, sz, pos); if (TEXILIST_ITEM == p->list) @@ -1531,6 +1439,87 @@ doitem(struct texi *p, enum texicmd cmd, } static void +dotab(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + + /* This command is only useful in @multitable. */ + if (TEXILIST_TABLE == p->list) + texiputchar(p, '\t'); +} + +static void +domultitable(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + enum texilist sv = p->list; + int svliteral = p->literal; + enum texicmd type; + size_t i, end, columns; + + p->list = TEXILIST_TABLE; + /* + * TS/TE blocks aren't "in mdoc(7)", so we can disregard the + * fact that we're in literal mode right now. + */ + p->literal = 0; + teximacro(p, "TS"); + columns = 0; + + /* Advance to the first argument... */ + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + + /* Make sure we don't print anything when scanning. */ + p->ign++; + if ('@' == buf[*pos]) { + /* + * Look for @columnfractions. + * We ignore these, but we do use the number of + * arguments to set the number of columns that we'll + * have. + */ + type = texicmd(p, buf, *pos, sz, &end, NULL); + advanceto(p, buf, pos, end); + if (TEXICMD_COLUMNFRACTIONS != type) + texierr(p, "unknown multitable command"); + while (*pos < sz && '\n' != buf[*pos]) { + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + while (*pos < sz && ! isws(buf[*pos])) { + if ('\n' == buf[*pos]) + break; + advance(p, buf, pos); + } + columns++; + } + } else + /* + * We have arguments. + * We could parse these, but it's easier to just let + * tbl(7) figure it out. + * So use this only to count arguments. + */ + while (parselinearg(p, buf, sz, pos) > 0) + columns++; + p->ign--; + + /* Left-justify each table entry. */ + for (i = 0; i < columns; i++) { + if (i > 0) + texiputchar(p, ' '); + texiputchar(p, 'l'); + } + texiputchars(p, ".\n"); + p->outmacro++; + parseto(p, buf, sz, pos, texitoks[cmd].tok); + p->outmacro--; + teximacro(p, "TE"); + p->literal = svliteral; + p->list = sv; +} + +static void dotable(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { @@ -1538,7 +1527,10 @@ dotable(struct texi *p, enum texicmd cmd, p->list = TEXILIST_ITEM; teximacro(p, "Bl -tag -width Ds"); - parseto(p, buf, sz, pos, "table"); + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); + p->seenvs = 1; + parseto(p, buf, sz, pos, texitoks[cmd].tok); teximacro(p, "El"); p->list = sv; } @@ -1551,6 +1543,9 @@ doenumerate(struct texi *p, enum texicmd cmd, p->list = TEXILIST_NOITEM; teximacro(p, "Bl -enum"); + p->seenvs = 1; + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); parseto(p, buf, sz, pos, "enumerate"); teximacro(p, "El"); p->list = sv; @@ -1562,8 +1557,11 @@ doitemize(struct texi *p, enum texicmd cmd, { enum texilist sv = p->list; - p->list = TEXILIST_ITEM; + p->list = TEXILIST_NOITEM; teximacro(p, "Bl -bullet"); + p->seenvs = 1; + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); parseto(p, buf, sz, pos, "itemize"); teximacro(p, "El"); p->list = sv; @@ -1584,50 +1582,44 @@ doignline(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + /* FIXME: ignore and parseeoln. */ advanceeoln(p, buf, sz, pos, 1); } /* * Parse colon-separated directories from "cp" (if not NULL) and returns * the array of pointers. - * Prepends "base" to the array. + * Prepends "base" to the array, if found. * This does NOT sanitise the directories! */ static char ** -parsedirs(const char *base, const char *cp, size_t *sz) +parsedirs(struct texi *p, const char *base, const char *cp, size_t *sz) { char *tok, *str, *tofree; const char *cpp; - size_t i; + size_t i = 0; char **dirs; - *sz = NULL != (cpp = cp) ? 2 : 1; - if (*sz > 1) - for ( ; NULL != (cpp = strchr(cpp, ':')); (*sz)++) + /* Count up our expected arguments. */ + *sz = NULL != base; + if (NULL != (cpp = cp)) + for ((*sz)++; NULL != (cpp = strchr(cpp, ':')); (*sz)++) cpp++; - dirs = calloc(*sz, sizeof(char *)); - if (NULL == dirs) { - perror(NULL); - exit(EXIT_FAILURE); - } else if (NULL == (dirs[0] = strdup(base))) { - perror(NULL); - exit(EXIT_FAILURE); - } - + if (0 == *sz) + return(NULL); + if (NULL == (dirs = calloc(*sz, sizeof(char *)))) + texiabort(p, NULL); + if (NULL != base && NULL == (dirs[i++] = strdup(base))) + texiabort(p, NULL); if (NULL == cp) return(dirs); - - if (NULL == (tofree = tok = str = strdup(cp))) { - perror(NULL); - exit(EXIT_FAILURE); - } + if (NULL == (tofree = tok = str = strdup(cp))) + texiabort(p, NULL); - for (i = 1; NULL != (tok = strsep(&str, ":")); i++) - if (NULL == (dirs[i] = strdup(tok))) { - perror(NULL); - exit(EXIT_FAILURE); - } + for ( ; NULL != (tok = strsep(&str, ":")); i++) + if (NULL == (dirs[i] = strdup(tok))) + texiabort(p, NULL); free(tofree); return(dirs); @@ -1638,7 +1630,7 @@ main(int argc, char *argv[]) { struct texi texi; int c; - char *path, *dir; + char *dirpath, *dir, *ccp; const char *progname, *Idir, *cp; progname = strrchr(argv[0], '/'); @@ -1648,6 +1640,7 @@ main(int argc, char *argv[]) ++progname; memset(&texi, 0, sizeof(struct texi)); + texi.ign = 1; Idir = NULL; while (-1 != (c = getopt(argc, argv, "I:"))) @@ -1660,37 +1653,33 @@ main(int argc, char *argv[]) } argv += optind; - if (0 == (argc -= optind)) - goto usage; + argc -= optind; - if (NULL == (path = strdup(argv[0]))) { - perror(NULL); - exit(EXIT_FAILURE); - } else if (NULL == (dir = dirname(path))) { - perror(argv[0]); - free(path); - exit(EXIT_FAILURE); + if (argc > 0) { + if (NULL == (dirpath = strdup(argv[0]))) + texiabort(&texi, NULL); + if (NULL == (dir = dirname(dirpath))) + texiabort(&texi, NULL); + if (NULL != (cp = strrchr(argv[0], '/'))) + texi.title = strdup(cp + 1); + else + texi.title = strdup(argv[0]); + if (NULL == texi.title) + texiabort(&texi, NULL); + else if (NULL != (ccp = strchr(texi.title, '.'))) + *ccp = '\0'; + texi.dirs = parsedirs(&texi, dir, Idir, &texi.dirsz); + free(dirpath); + parsefile(&texi, argv[0], 1); + } else { + texi.title = strdup("Unknown Manual"); + texi.dirs = parsedirs(&texi, NULL, Idir, &texi.dirsz); + parsestdin(&texi); } - free(path); - if (NULL != (cp = strrchr(argv[0], '/'))) - texi.title = strdup(cp + 1); - else - texi.title = strdup(argv[0]); - - if (NULL == texi.title) { - perror(NULL); - exit(EXIT_FAILURE); - } else if (NULL != (path = strchr(texi.title, '.'))) - *path = '\0'; - - texi.ign = 1; - texi.dirs = parsedirs(dir, Idir, &texi.dirsz); - parsefile(&texi, argv[0]); - /* We shouldn't get here. */ texiexit(&texi); return(EXIT_FAILURE); usage: - fprintf(stderr, "usage: %s [-Idirs] file\n", progname); + fprintf(stderr, "usage: %s [-Idirs] [file]\n", progname); return(EXIT_FAILURE); }