=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.6 retrieving revision 1.20 diff -u -p -r1.6 -r1.20 --- texi2mdoc/main.c 2015/02/18 11:03:04 1.6 +++ texi2mdoc/main.c 2015/02/19 16:15:17 1.20 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.6 2015/02/18 11:03:04 kristaps Exp $ */ +/* $Id: main.c,v 1.20 2015/02/19 16:15:17 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -27,6 +27,7 @@ #include #include #include +#include #include /* @@ -36,33 +37,53 @@ */ enum texicmd { TEXICMD_ACRONYM, + TEXICMD_ACUTE, TEXICMD_A4PAPER, TEXICMD_ANCHOR, TEXICMD_APPENDIX, TEXICMD_APPENDIXSEC, + TEXICMD_APPENDIXSUBSEC, TEXICMD_ASTERISK, TEXICMD_AT, TEXICMD_AUTHOR, + TEXICMD_B, TEXICMD_BANG, + TEXICMD_BULLET, TEXICMD_BYE, TEXICMD_CENTER, TEXICMD_CHAPTER, TEXICMD_CINDEX, + TEXICMD_CIRCUMFLEX, TEXICMD_CITE, TEXICMD_CODE, TEXICMD_COLON, + TEXICMD_COLUMNFRACTIONS, TEXICMD_COMMAND, TEXICMD_COMMENT, TEXICMD_COMMENT_LONG, TEXICMD_CONTENTS, TEXICMD_COPYING, TEXICMD_COPYRIGHT, + TEXICMD_DEFFN, + TEXICMD_DEFFNX, + TEXICMD_DEFMAC, + TEXICMD_DEFMACX, + TEXICMD_DEFTP, + TEXICMD_DEFTPX, TEXICMD_DEFTYPEFN, TEXICMD_DEFTYPEFNX, TEXICMD_DEFTYPEFUN, TEXICMD_DEFTYPEFUNX, TEXICMD_DEFTYPEVAR, + TEXICMD_DEFTYPEVARX, TEXICMD_DEFTYPEVR, + TEXICMD_DEFTYPEVRX, + TEXICMD_DEFUN, + TEXICMD_DEFUNX, + TEXICMD_DEFVAR, + TEXICMD_DEFVARX, + TEXICMD_DEFVR, + TEXICMD_DEFVRX, TEXICMD_DETAILMENU, TEXICMD_DFN, TEXICMD_DIRCATEGORY, @@ -74,64 +95,111 @@ enum texicmd { TEXICMD_END, TEXICMD_ENUMERATE, TEXICMD_ENV, + TEXICMD_ERROR, TEXICMD_EXAMPLE, + TEXICMD_EXPANSION, TEXICMD_FILE, + TEXICMD_FINALOUT, + TEXICMD_FINDEX, + TEXICMD_FTABLE, + TEXICMD_FORMAT, + TEXICMD_GRAVE, TEXICMD_GROUP, TEXICMD_HEADING, TEXICMD_HEADINGS, + TEXICMD_HEADITEM, TEXICMD_HYPHEN, TEXICMD_I, TEXICMD_IFCLEAR, + TEXICMD_IFDOCBOOK, TEXICMD_IFHTML, TEXICMD_IFINFO, + TEXICMD_IFNOTDOCBOOK, + TEXICMD_IFNOTHTML, + TEXICMD_IFNOTINFO, + TEXICMD_IFNOTPLAINTEXT, TEXICMD_IFNOTTEX, + TEXICMD_IFNOTXML, + TEXICMD_IFPLAINTEXT, TEXICMD_IFTEX, TEXICMD_IFSET, + TEXICMD_IFXML, + TEXICMD_IGNORE, TEXICMD_IMAGE, TEXICMD_INCLUDE, + TEXICMD_INDENTBLOCK, TEXICMD_INSERTCOPYING, TEXICMD_ITEM, TEXICMD_ITEMIZE, + TEXICMD_ITEMX, TEXICMD_KBD, + TEXICMD_KEY, + TEXICMD_KINDEX, TEXICMD_LATEX, TEXICMD_MATH, TEXICMD_MENU, + TEXICMD_MULTITABLE, + TEXICMD_NEED, TEXICMD_NEWLINE, TEXICMD_NODE, TEXICMD_NOINDENT, + TEXICMD_OPTION, + TEXICMD_PXREF, TEXICMD_QUESTIONMARK, TEXICMD_QUOTATION, TEXICMD_PAGE, TEXICMD_PARINDENT, TEXICMD_PRINTINDEX, + TEXICMD_R, TEXICMD_REF, + TEXICMD_RESULT, TEXICMD_SAMP, + TEXICMD_SANSSERIF, + TEXICMD_SC, TEXICMD_SECTION, TEXICMD_SET, TEXICMD_SETCHAPNEWPAGE, TEXICMD_SETFILENAME, TEXICMD_SETTITLE, + TEXICMD_SLANTED, TEXICMD_SP, TEXICMD_SPACE, + TEXICMD_SMALLBOOK, + TEXICMD_SMALLDISPLAY, TEXICMD_SMALLEXAMPLE, + TEXICMD_SMALLFORMAT, + TEXICMD_SMALLINDENTBLOCK, TEXICMD_SQUIGGLE_LEFT, TEXICMD_SQUIGGLE_RIGHT, + TEXICMD_STRONG, + TEXICMD_SUBHEADING, TEXICMD_SUBSECTION, TEXICMD_SUBTITLE, + TEXICMD_SYNCODEINDEX, + TEXICMD_T, TEXICMD_TAB, + TEXICMD_TABSYM, TEXICMD_TABLE, TEXICMD_TEX, TEXICMD_TEXSYM, + TEXICMD_TILDE, TEXICMD_TITLE, TEXICMD_TITLEFONT, TEXICMD_TITLEPAGE, TEXICMD_TOP, + TEXICMD_UMLAUT, TEXICMD_UNNUMBERED, TEXICMD_UNNUMBEREDSEC, + TEXICMD_UNNUMBEREDSUBSEC, TEXICMD_UREF, TEXICMD_URL, TEXICMD_VAR, + TEXICMD_VERBATIMINCLUDE, + TEXICMD_VINDEX, + TEXICMD_VSKIP, + TEXICMD_VTABLE, TEXICMD_W, + TEXICMD_XREF, TEXICMD__MAX }; @@ -168,6 +236,7 @@ enum texilist { TEXILIST_NONE = 0, TEXILIST_ITEM, TEXILIST_NOITEM, + TEXILIST_TABLE }; /* @@ -180,44 +249,46 @@ struct texi { size_t outcol; /* column in output line */ char **dirs; /* texi directories */ size_t dirsz; /* number of texi directories */ - enum texilist list; - int outmacro; /* whether output is in line macro */ - int seenws; /* whitespace has been ignored */ - int ign; /* don't print anything */ - int literal; + char *title; /* title of document */ + char *subtitle; /* subtitle of document */ + /* + * 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 */ + 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 */ }; -/* FIXME: don't use this crap. */ -#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 doarg1(struct texi *, enum texicmd, const char *, size_t, size_t *); +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 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 dofont(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 doliteral(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 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 *); @@ -225,105 +296,177 @@ static void dosection(struct texi *, enum texicmd, con 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 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 doverbinclude(struct texi *, enum texicmd, const char *, size_t, size_t *); static const struct texitok texitoks[TEXICMD__MAX] = { - { doarg1, "acronym", 7 }, /* TEXICMD_ACRONYM */ + /* 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 */ { dosymbol, "*", 1 }, /* TEXICMD_ASTERISK */ { dosymbol, "@", 1 }, /* TEXICMD_AT */ { doignline, "author", 6 }, /* TEXICMD_AUTHOR */ + { dofont, "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 */ + { 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 */ + { dofont, "code", 4 }, /* TEXICMD_CODE */ + { dofont, "cite", 4 }, /* TEXICMD_CITE */ { dosymbol, ":", 1 }, /* TEXICMD_COLON */ - { docommand, "command", 7 }, /* TEXICMD_COMMAND */ + { NULL, "columnfractions", 15 }, /* TEXICMD_COLUMNFRACTIONS */ + { 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 */ + { dodefn, "deffn", 5 }, /* TEXICMD_DEFFN */ + { dodefn, "deffnx", 6 }, /* TEXICMD_DEFFNX */ + { 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, "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 */ + { dofont, "dfn", 3 }, /* TEXICMD_DFN */ { doignline, "dircategory", 11 }, /* TEXICMD_DIRCATEGORY */ { doignblock, "direntry", 8 }, /* TEXICMD_DIRENTRY */ { dodisplay, "display", 7 }, /* TEXICMD_DISPLAY */ { dosymbol, "dots", 4 }, /* TEXICMD_DOTS */ - { doarg1, "email", 5 }, /* TEXICMD_EMAIL */ - { doemph, "emph", 4 }, /* TEXICMD_EMPH */ + { dolink, "email", 5 }, /* TEXICMD_EMAIL */ + { dofont, "emph", 4 }, /* TEXICMD_EMPH */ { NULL, "end", 3 }, /* TEXICMD_END */ { doenumerate, "enumerate", 9 }, /* TEXICMD_ENUMERATE */ - { doenv, "env", 3 }, /* TEXICMD_ENV */ + { doinline, "env", 3 }, /* TEXICMD_ENV */ + { dosymbol, "error", 5 }, /* TEXICMD_ERROR */ { doexample, "example", 7 }, /* TEXICMD_EXAMPLE */ - { dofile, "file", 4 }, /* TEXICMD_FILE */ + { dosymbol, "expansion", 9 }, /* TEXICMD_EXPANSION */ + { doinline, "file", 4 }, /* TEXICMD_FILE */ + { doignline, "finalout", 8 }, /* TEXICMD_FINALOUT */ + { doignline, "findex", 6 }, /* TEXICMD_FINDEX */ + { 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 */ { dosymbol, "-", 1 }, /* TEXICMD_HYPHEN */ - { doitalic, "i", 1 }, /* TEXICMD_I */ + { dofont, "i", 1 }, /* TEXICMD_I */ { doignblock, "ifclear", 7 }, /* TEXICMD_IFCLEAR */ + { doignblock, "ifdocbook", 9 }, /* TEXICMD_IFDOCBOOK */ { doignblock, "ifhtml", 6 }, /* TEXICMD_IFHTML */ { doignblock, "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 */ + { 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 */ { 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 */ + { dofont, "kbd", 3 }, /* TEXICMD_KBD */ + { dobracket, "key", 3 }, /* TEXICMD_KEY */ + { doignline, "kindex", 6 }, /* TEXICMD_KINDEX */ { dosymbol, "LaTeX", 5 }, /* TEXICMD_LATEX */ { domath, "math", 4 }, /* TEXICMD_MATH */ { doignblock, "menu", 4 }, /* TEXICMD_MENU */ + { 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, "printindex", 10 }, /* TEXICMD_PRINTINDEX */ + { dofont, "r", 1 }, /* TEXICMD_R */ { dobracket, "ref", 3 }, /* TEXICMD_REF */ - { doliteral, "samp", 4 }, /* TEXICMD_SAMP */ + { dosymbol, "result", 6 }, /* TEXICMD_RESULT */ + { dofont, "samp", 4 }, /* TEXICMD_SAMP */ + { dofont, "sansserif", 9 }, /* TEXICMD_SANSSERIF */ + { dobracket, "sc", 2 }, /* TEXICMD_SC */ { dosection, "section", 7 }, /* TEXICMD_SECTION */ { doignline, "set", 3 }, /* TEXICMD_SET */ { doignline, "setchapternewpage", 17 }, /* TEXICMD_SETCHAPNEWPAGE */ { doignline, "setfilename", 11 }, /* TEXICMD_SETFILENAME */ - { doignline, "settitle", 8 }, /* TEXICMD_SETTITLE */ + { dotitle, "settitle", 8 }, /* TEXICMD_SETTITLE */ + { dofont, "slanted", 7 }, /* TEXICMD_SLANTED */ { 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 */ + { dofont, "strong", 6 }, /* TEXICMD_STRONG */ + { dosubsection, "subheading", 10 }, /* TEXICMD_SUBHEADING */ { dosubsection, "subsection", 10 }, /* TEXICMD_SUBSECTION */ { doignline, "subtitle", 8 }, /* TEXICMD_SUBTITLE */ - { dosymbol, "\t", 1 }, /* TEXICMD_TAB */ + { doignline, "syncodeindex", 12 }, /* TEXICMD_SYNCODEINDEX */ + { dofont, "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 */ + { doaccent, "~", 1 }, /* TEXICMD_TILDE */ { 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 */ - { doarg1, "uref", 4 }, /* TEXICMD_UREF */ - { doarg1, "url", 3 }, /* TEXICMD_URL */ - { doliteral, "var", 3 }, /* TEXICMD_VAR */ + { dosubsection, "unnumberedsubsec", 16 }, /* TEXICMD_UNNUMBEREDSUBSEC */ + { dolink, "uref", 4 }, /* TEXICMD_UREF */ + { dolink, "url", 3 }, /* TEXICMD_URL */ + { doinline, "var", 3 }, /* TEXICMD_VAR */ + { 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 */ }; /* @@ -340,7 +483,8 @@ texifilepop(struct texi *p) } /* - * Unmap all files that we're currently using. + * 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 @@ -356,7 +500,10 @@ texiexit(struct texi *p) for (i = 0; i < p->dirsz; i++) free(p->dirs[i]); + free(p->dirs); + free(p->subtitle); + free(p->title); } /* @@ -391,6 +538,10 @@ texiwarn(const struct texi *p, const char *fmt, ...) fputc('\n', stderr); } +/* + * 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, ...) { @@ -409,9 +560,9 @@ texierr(struct texi *p, const char *fmt, ...) } /* - * Put a single data character. - * This MUST NOT be a mdoc(7) command: it should be free text that's - * outputted to the screen. + * Put a single data character to the output if we're not ignoring. + * Adjusts our output status. + * This shouldn't be called for macros: just for ordinary text. */ static void texiputchar(struct texi *p, char c) @@ -419,7 +570,12 @@ texiputchar(struct texi *p, char c) if (p->ign) return; + + if ('.' == c && 0 == p->outcol) + fputs("\\&", stdout); + putchar(c); + p->seenvs = 0; if ('\n' == c) { p->outcol = 0; p->seenws = 0; @@ -429,6 +585,7 @@ texiputchar(struct texi *p, char c) /* * Put multiple characters (see texiputchar()). + * This shouldn't be called for macros: just for ordinary text. */ static void texiputchars(struct texi *p, const char *s) @@ -439,58 +596,87 @@ texiputchars(struct texi *p, const char *s) } /* - * Put an mdoc(7) command without the trailing newline. - * This should ONLY be used for mdoc(7) commands! + * 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) { - if (0 == --p->outmacro) - texiputchar(p, '\n'); + if (p->ign) + return; + + if (0 == --p->outmacro) { + putchar('\n'); + p->outcol = p->seenws = 0; + } } /* - * Put an mdoc(7) command without the trailing newline. - * This should ONLY be used for mdoc(7) commands! + * 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) { + int rc; - if (p->outcol && 0 == p->outmacro) - texiputchar(p, '\n'); + if (p->ign) + return; + + if (p->outcol && 0 == p->outmacro) { + putchar('\n'); + p->outcol = 0; + } + if (0 == p->outmacro) - texiputchar(p, '.'); + putchar('.'); else - texiputchar(p, ' '); - texiputchars(p, s); - texiputchar(p, ' '); + putchar(' '); + + if (EOF != (rc = fputs(s, stdout))) + p->outcol += rc; + + putchar(' '); + p->outcol++; p->outmacro++; p->seenws = 0; } /* - * Put an mdoc(7) command with the trailing newline. - * This should ONLY be used for mdoc(7) commands! + * Put a stadnalone mdoc(7) command with the trailing newline. */ static void teximacro(struct texi *p, const char *s) { + if (p->ign) + return; + if (p->outmacro) texierr(p, "\"%s\" in open line scope!?", s); - else if (p->literal) + if (p->literal) texierr(p, "\"%s\" in a literal scope!?", s); if (p->outcol) - texiputchar(p, '\n'); + putchar('\n'); - texiputchar(p, '.'); - texiputchars(p, s); - texiputchar(p, '\n'); + putchar('.'); + puts(s); + p->outcol = p->seenws = 0; } +static void +texivspace(struct texi *p) +{ + + if (p->seenvs) + return; + teximacro(p, "Pp"); + p->seenvs = 1; +} + /* * Advance by a single byte in the input stream. */ @@ -507,6 +693,40 @@ advance(struct texi *p, const char *buf, size_t *pos) (*pos)++; } +static void +texipunctuate(struct texi *p, const char *buf, size_t sz, size_t *pos) +{ + size_t start, end; + + if (1 != p->outmacro) + return; + + for (start = end = *pos; end < sz; end++) { + switch (buf[end]) { + case (','): + case (')'): + case ('.'): + case ('"'): + case (':'): + case ('!'): + case ('?'): + continue; + default: + break; + } + break; + } + if (end == *pos) + return; + if (end + 1 == sz || ' ' == buf[end] || '\n' == buf[end]) { + for ( ; start < end; start++) { + texiputchar(p, ' '); + texiputchar(p, buf[start]); + advance(p, buf, pos); + } + } +} + /* * 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 @@ -517,7 +737,7 @@ advancenext(struct texi *p, const char *buf, size_t sz { if (p->literal) { - while (*pos < sz && isspace(buf[*pos])) { + while (*pos < sz && ismspace(buf[*pos])) { if (*pos && '\n' == buf[*pos] && '\\' == buf[*pos - 1]) texiputchar(p, 'e'); @@ -527,16 +747,15 @@ advancenext(struct texi *p, const char *buf, size_t sz return(*pos); } - while (*pos < sz && isspace(buf[*pos])) { + 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"); + if (*pos && '\n' == buf[*pos] && '\n' == buf[*pos - 1]) + texivspace(p); advance(p, buf, pos); } return(*pos); @@ -550,6 +769,7 @@ advanceeoln(struct texi *p, const char *buf, size_t sz, size_t *pos, int consumenl) { + /* FIXME: disregards @NEWLINE. */ while (*pos < sz && '\n' != buf[*pos]) advance(p, buf, pos); if (*pos < sz && consumenl) @@ -576,28 +796,27 @@ advanceto(struct texi *p, const char *buf, size_t *pos * This also will advance the input stream. */ static void -texiword(struct texi *p, const char *buf, size_t sz, size_t *pos) +texiword(struct texi *p, const char *buf, + size_t sz, size_t *pos, char extra) { - /* - * XXX: if we're in literal mode, then we shouldn't do any - * reflowing of text here. - */ - if (0 == p->outmacro && p->outcol > 72 && 0 == p->literal) + if (p->seenws && 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 && ! isspace(buf[*pos])) { + 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]) { @@ -614,6 +833,11 @@ texiword(struct texi *p, const char *buf, size_t sz, s } } +/* + * 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) @@ -622,12 +846,15 @@ texicmd(struct texi *p, const char *buf, assert('@' == buf[pos]); - if (++pos >= sz) + if ((*end = pos) == sz) return(TEXICMD__MAX); + else if ((*end = ++pos) == sz) + return(TEXICMD__MAX); /* Alphabetic commands are special. */ if ( ! isalpha(buf[pos])) { - *end = pos + 1; + if ((*end = pos + 1) == sz) + return(TEXICMD__MAX); for (i = 0; i < TEXICMD__MAX; i++) { if (1 != texitoks[i].len) continue; @@ -638,7 +865,7 @@ texicmd(struct texi *p, const char *buf, return(TEXICMD__MAX); } - for (*end = pos; *end < sz && ! isspace(buf[*end]); (*end)++) + for (*end = pos; *end < sz && ! ismspace(buf[*end]); (*end)++) if ((*end > pos && ('@' == buf[*end] || '{' == buf[*end] || '}' == buf[*end]))) break; @@ -655,48 +882,69 @@ texicmd(struct texi *p, const char *buf, return(TEXICMD__MAX); } -static void -parseeof(struct texi *p, const char *buf, size_t sz) +/* + * 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 pos = 0; + size_t end; enum texicmd cmd; - size_t end; - while ((pos = advancenext(p, buf, sz, &pos)) < sz) { - switch (buf[pos]) { + 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 ('}'): - if (0 == p->ign) - texiwarn(p, "unexpected \"}\""); - advance(p, buf, &pos); - continue; + advance(p, buf, pos); + return(0); case ('{'): if (0 == p->ign) texiwarn(p, "unexpected \"{\""); - advance(p, buf, &pos); + advance(p, buf, pos); continue; case ('@'): break; default: - texiword(p, buf, sz, &pos); + texiword(p, buf, sz, pos, ','); continue; } - cmd = texicmd(p, buf, pos, sz, &end); - advanceto(p, buf, &pos, 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); - } + (*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) { size_t end; enum texicmd cmd; - while (*pos < sz && isspace(buf[*pos])) + while (*pos < sz && ismspace(buf[*pos])) advance(p, buf, pos); if (*pos == sz || '{' != buf[*pos]) @@ -716,7 +964,7 @@ parsebracket(struct texi *p, const char *buf, size_t s case ('@'): break; default: - texiword(p, buf, sz, pos); + texiword(p, buf, sz, pos, '\0'); continue; } @@ -740,11 +988,11 @@ parseeoln(struct texi *p, const char *buf, size_t sz, size_t end; enum texicmd cmd; - assert(0 == p->literal); - while (*pos < sz && '\n' != buf[*pos]) { while (*pos < sz && isws(buf[*pos])) { p->seenws = 1; + if (p->literal) + texiputchar(p, buf[*pos]); advance(p, buf, pos); } switch (buf[*pos]) { @@ -761,7 +1009,7 @@ parseeoln(struct texi *p, const char *buf, size_t sz, case ('@'): break; default: - texiword(p, buf, sz, pos); + texiword(p, buf, sz, pos, '\0'); continue; } @@ -774,6 +1022,10 @@ parseeoln(struct texi *p, const char *buf, size_t sz, } } +/* + * 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) { @@ -797,7 +1049,7 @@ parsesingle(struct texi *p, const char *buf, size_t sz case ('@'): break; default: - texiword(p, buf, sz, pos); + texiword(p, buf, sz, pos, '\0'); return; } @@ -809,7 +1061,43 @@ parsesingle(struct texi *p, const char *buf, size_t sz (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); } +static int +parselinearg(struct texi *p, const char *buf, size_t sz, size_t *pos) +{ + + while (*pos < sz && isws(buf[*pos])) { + p->seenws = 1; + advance(p, buf, pos); + } + + if (*pos < sz && '{' == buf[*pos]) + parsebracket(p, buf, sz, pos); + else if ('\n' != buf[*pos]) + parsesingle(p, buf, sz, pos); + else + return(0); + + return(1); +} + +/* + * Parse til the end of the buffer. + */ static void +parseeof(struct texi *p, const char *buf, size_t sz) +{ + size_t pos; + + for (pos = 0; pos < sz; ) + parsesingle(p, buf, sz, &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) { @@ -835,7 +1123,7 @@ parseto(struct texi *p, const char *buf, case ('@'): break; default: - texiword(p, buf, sz, pos); + texiword(p, buf, sz, pos, '\0'); continue; } @@ -845,8 +1133,8 @@ parseto(struct texi *p, const char *buf, while (*pos < sz && isws(buf[*pos])) advance(p, buf, pos); /* - * FIXME: skip tabs and also check the full - * word, not just its initial substring! + * FIXME: check the full word, not just its + * initial substring! */ if (sz - *pos >= endtoksz && 0 == strncmp (&buf[*pos], endtoken, endtoksz)) { @@ -863,12 +1151,17 @@ parseto(struct texi *p, const char *buf, } } +/* + * 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) +parsefile(struct texi *p, const char *fname, int parse) { - struct texifile *f; - int fd; - struct stat st; + struct texifile *f; + int fd; + struct stat st; + size_t i; assert(p->filepos < 64); f = &p->files[p->filepos]; @@ -891,46 +1184,43 @@ parsefile(struct texi *p, const char *fname) texiabort(p, fname); p->filepos++; - parseeof(p, f->map, f->mapsz); + if ( ! parse) { + /* + * We're printing verbatim output. + * Make sure it doesn't get interpreted as mdoc by + * escaping escapes and making sure leading dots don't + * trigger mdoc(7) expansion. + */ + for (i = 0; i < f->mapsz; i++) { + if (i > 0 && '.' == f->map[i]) + if ('\n' == f->map[i - 1]) + fputs("\\&", stdout); + putchar(f->map[i]); + if ('\\' == f->map[i]) + putchar('e'); + } + } else + parseeof(p, f->map, f->mapsz); texifilepop(p); } static void -dodeftypevar(struct texi *p, enum texicmd cmd, +dodefn(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_DEFFN): + case (TEXICMD_DEFTP): case (TEXICMD_DEFTYPEFN): case (TEXICMD_DEFTYPEFUN): + case (TEXICMD_DEFTYPEVAR): + case (TEXICMD_DEFTYPEVR): + case (TEXICMD_DEFUN): + case (TEXICMD_DEFVAR): + case (TEXICMD_DEFVR): blk = texitoks[cmd].tok; break; default: @@ -938,34 +1228,101 @@ dodeftypefun(struct texi *p, enum texicmd cmd, } if (p->ign) { - if (NULL != blk) - parseto(p, buf, sz, pos, blk); + NULL != blk ? + parseto(p, buf, sz, pos, blk) : + parseeoln(p, buf, sz, pos); return; } + if (NULL != blk) + texivspace(p); + switch (cmd) { - case (TEXICMD_DEFTYPEFN): + 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): - teximacro(p, "Pp"); + case (TEXICMD_DEFTYPEFUNX): + case (TEXICMD_DEFUN): + case (TEXICMD_DEFUNX): + texiputchars(p, "Function"); break; default: + parselinearg(p, buf, sz, pos); break; } - if (TEXICMD_DEFTYPEFN == cmd || - TEXICMD_DEFTYPEFNX == cmd) { - parsebracket(p, buf, sz, pos); - texiputchars(p, ":\n"); + + texiputchars(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): + 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(); } - 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"); + + texivspace(p); if (NULL != blk) parseto(p, buf, sz, pos, blk); } @@ -989,21 +1346,81 @@ doblock(struct texi *p, enum texicmd cmd, } static void -doinline(struct texi *p, const char *buf, - size_t sz, size_t *pos, const char *macro) +doinline(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) { + const char *macro; + switch (cmd) { + 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: + abort(); + } + + if (p->literal) { + parsebracket(p, buf, sz, pos); + return; + } + teximacroopen(p, macro); p->seenws = 0; parsebracket(p, buf, sz, pos); - if (*pos < sz - 1 && - ismpunct(buf[*pos]) && - isspace(buf[*pos + 1])) { - texiputchar(p, ' '); - texiputchar(p, buf[*pos]); + texipunctuate(p, buf, sz, pos); + teximacroclose(p); +} + +static void +doverbinclude(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + char fname[PATH_MAX], path[PATH_MAX]; + int rc; + size_t i; + + while (*pos < sz && ' ' == buf[*pos]) advance(p, buf, pos); + + /* Read in the filename. */ + for (i = 0; *pos < sz && '\n' != buf[*pos]; i++) { + if (i == sizeof(fname) - 1) + break; + fname[i] = buf[*pos]; + advance(p, buf, pos); } - teximacroclose(p); + + 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'; + + if (strstr(fname, "../") || strstr(fname, "/..")) + texierr(p, "insecure path"); + + 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"); + + parsefile(p, path, 0); } static void @@ -1039,7 +1456,6 @@ doinclude(struct texi *p, enum texicmd cmd, for (i = 0; i < p->dirsz; i++) { rc = snprintf(path, sizeof(path), "%s/%s", p->dirs[i], fname); - texiwarn(p, "trying: %s", path); if (rc < 0) texierr(p, "couldn't format path"); else if ((size_t)rc >= sizeof(path)) @@ -1047,7 +1463,7 @@ doinclude(struct texi *p, enum texicmd cmd, else if (-1 == access(path, R_OK)) continue; - parsefile(p, path); + parsefile(p, path, 1); return; } @@ -1055,57 +1471,45 @@ doinclude(struct texi *p, enum texicmd cmd, } static void -doitalic(struct texi *p, enum texicmd cmd, +dofont(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + const char *font; - texiputchars(p, "\\fI"); + switch (cmd) { + case (TEXICMD_B): + case (TEXICMD_STRONG): + font = "\\fB"; + break; + case (TEXICMD_CITE): + case (TEXICMD_DFN): + case (TEXICMD_EMPH): + case (TEXICMD_I): + case (TEXICMD_SLANTED): + font = "\\fI"; + break; + case (TEXICMD_CODE): + case (TEXICMD_KBD): + case (TEXICMD_R): + case (TEXICMD_SAMP): + case (TEXICMD_SANSSERIF): + case (TEXICMD_T): + font = "\\fR"; + break; + default: + abort(); + } + + if (p->seenws) { + texiputchar(p, ' '); + p->seenws = 0; + } + texiputchars(p, font); 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) { @@ -1114,24 +1518,24 @@ 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"); } @@ -1139,14 +1543,12 @@ 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"); } @@ -1161,6 +1563,106 @@ dobye(struct texi *p, enum texicmd cmd, } static void +dotitle(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + size_t start, end; + + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + start = end = *pos; + while (end < sz && '\n' != buf[end]) + end++; + free(p->subtitle); + p->subtitle = malloc(end - start + 1); + 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; + advance(p, buf, pos); + switch (cmd) { + 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; + 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; + 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; + default: + abort(); + } +} + +static void dosymbol(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { @@ -1174,7 +1676,7 @@ 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): @@ -1183,18 +1685,30 @@ dosymbol(struct texi *p, enum texicmd cmd, case (TEXICMD_BANG): texiputchar(p, '!'); break; + case (TEXICMD_BULLET): + texiputchars(p, "\\(bu"); + break; case (TEXICMD_COPYRIGHT): texiputchars(p, "\\(co"); break; case (TEXICMD_DOTS): 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_QUESTIONMARK): texiputchar(p, '?'); break; + case (TEXICMD_RESULT): + texiputchars(p, "\\(rA"); + break; case (TEXICMD_SQUIGGLE_LEFT): texiputchars(p, "{"); break; @@ -1226,7 +1740,6 @@ doquotation(struct texi *p, enum texicmd cmd, teximacro(p, "Qc"); } -/* FIXME */ static void domath(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) @@ -1262,15 +1775,12 @@ domath(struct texi *p, enum texicmd cmd, advance(p, buf, pos); } -/* FIXME */ static void -doarg1(struct texi *p, enum texicmd cmd, +dolink(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + int c; - if (*pos == sz || '{' != buf[*pos]) - return; - advance(p, buf, pos); switch (cmd) { case (TEXICMD_EMAIL): teximacroopen(p, "Mt"); @@ -1279,61 +1789,90 @@ doarg1(struct texi *p, enum texicmd cmd, case (TEXICMD_URL): teximacroopen(p, "Lk"); break; - default: + case (TEXICMD_XREF): + texiputchars(p, "See Section"); + teximacroopen(p, "Qq"); break; - } - /* FIXME: this is ugly */ - while (*pos < sz && '}' != buf[*pos] && ',' != buf[*pos]) { - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - while (*pos < sz && '}' != buf[*pos]) - advance(p, buf, pos); - if (*pos < sz) - advance(p, buf, pos); - if (*pos < sz - 1 && - ismpunct(buf[*pos]) && - isspace(buf[*pos + 1])) { - texiputchar(p, ' '); - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - switch (cmd) { - case (TEXICMD_EMAIL): - case (TEXICMD_UREF): - case (TEXICMD_URL): - teximacroclose(p); + case (TEXICMD_PXREF): + texiputchars(p, "see Section"); + teximacroopen(p, "Qq"); break; default: - break; + abort(); } + + c = parsearg(p, buf, sz, pos, 0); + p->ign++; + while (c > 0) + c = parsearg(p, buf, sz, pos, 1); + p->ign--; + + texipunctuate(p, buf, sz, pos); + teximacroclose(p); } static void +doignargn(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + int c; + + c = parsearg(p, buf, sz, pos, 0); + p->ign++; + while (c > 0) + c = parsearg(p, buf, sz, pos, 1); + p->ign--; +} + +static void dosubsection(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - teximacro(p, "Pp"); + if (p->outmacro) + texierr(p, "\"Em\" in open line scope!?"); + else if (p->literal) + texierr(p, "\"Em\" in a literal scope!?"); + + texivspace(p); teximacroopen(p, "Em"); parseeoln(p, buf, sz, pos); teximacroclose(p); - teximacro(p, "Pp"); + texivspace(p); } static void dosection(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + const char *blk; + switch (cmd) { + case (TEXICMD_APPENDIX): + case (TEXICMD_CHAPTER): + case (TEXICMD_TOP): + case (TEXICMD_UNNUMBERED): + blk = "Sh"; + break; + case (TEXICMD_APPENDIXSEC): + case (TEXICMD_HEADING): + case (TEXICMD_SECTION): + case (TEXICMD_UNNUMBEREDSEC): + blk = "Ss"; + break; + default: + abort(); + } + if (p->outmacro) - texierr(p, "subsection in open line scope!?"); + texierr(p, "%s in open line scope!?", blk); else if (p->literal) - texierr(p, "subsection in a literal scope!?"); + texierr(p, "%s in a literal scope!?", blk); - teximacroopen(p, "Ss"); + teximacroopen(p, blk); parseeoln(p, buf, sz, pos); teximacroclose(p); + p->seenvs = 1; } static void @@ -1341,38 +1880,49 @@ dosp(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - teximacro(p, "Pp"); + texivspace(p); + /* FIXME: ignore and parseeoln. */ advanceeoln(p, buf, sz, pos, 1); } static void -dochapter(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ - - if (p->outmacro) - texierr(p, "section in open line scope!?"); - else if (p->literal) - texierr(p, "section in a literal scope!?"); - - teximacroopen(p, "Sh"); - parseeoln(p, buf, sz, pos); - teximacroclose(p); -} - -static void dotop(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + const char *cp; + time_t t; + char date[32]; + /* + * 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); - teximacro(p, "Dd $Mdocdate: February 18 2015 $"); - teximacro(p, "Dt SOMETHING 7"); + 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"); - teximacro(p, "Nm Something"); - teximacro(p, "Nd Something"); + teximacroopen(p, "Nm"); + texiputchars(p, p->title); + teximacroclose(p); + teximacroopen(p, "Nd"); + texiputchars(p, NULL != p->subtitle ? + p->subtitle : "Unknown description"); + teximacroclose(p); + p->seenvs = 1; + dosection(p, cmd, buf, sz, pos); } static void @@ -1380,6 +1930,12 @@ 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) @@ -1393,19 +1949,95 @@ 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) teximacroclose(p); - else + else if (p->outcol > 0) texiputchar(p, '\n'); } 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; + enum texicmd type; + size_t i, end, columns; + + p->list = TEXILIST_TABLE; + 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); + advanceto(p, buf, pos, end); + if (TEXICMD_COLUMNFRACTIONS != type) + texierr(p, "unknown multitable type"); + 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->list = sv; +} + +static void dotable(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { @@ -1413,7 +2045,10 @@ 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; } @@ -1426,6 +2061,9 @@ 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; @@ -1439,6 +2077,9 @@ doitemize(struct texi *p, enum texicmd cmd, p->list = TEXILIST_ITEM; 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; @@ -1459,9 +2100,16 @@ 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. + * This does NOT sanitise the directories! + */ static char ** parsedirs(const char *base, const char *cp, size_t *sz) { @@ -1508,7 +2156,7 @@ main(int argc, char *argv[]) struct texi texi; int c; char *path, *dir; - const char *progname, *Idir; + const char *progname, *Idir, *cp; progname = strrchr(argv[0], '/'); if (progname == NULL) @@ -1516,7 +2164,9 @@ main(int argc, char *argv[]) else ++progname; + memset(&texi, 0, sizeof(struct texi)); Idir = NULL; + while (-1 != (c = getopt(argc, argv, "I:"))) switch (c) { case ('I'): @@ -1540,14 +2190,24 @@ main(int argc, char *argv[]) } free(path); - memset(&texi, 0, sizeof(struct texi)); + 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]); + parsefile(&texi, argv[0], 1); /* We shouldn't get here. */ texiexit(&texi); return(EXIT_FAILURE); usage: - fprintf(stderr, "usage: %s file\n", progname); + fprintf(stderr, "usage: %s [-Idirs] file\n", progname); return(EXIT_FAILURE); }