=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.2 retrieving revision 1.5 diff -u -p -r1.2 -r1.5 --- texi2mdoc/main.c 2015/02/17 10:27:18 1.2 +++ texi2mdoc/main.c 2015/02/18 11:01:56 1.5 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.2 2015/02/17 10:27:18 kristaps Exp $ */ +/* $Id: main.c,v 1.5 2015/02/18 11:01:56 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -39,60 +39,98 @@ enum texicmd { TEXICMD_ANCHOR, TEXICMD_APPENDIX, TEXICMD_APPENDIXSEC, + TEXICMD_ASTERISK, TEXICMD_AT, + TEXICMD_AUTHOR, + TEXICMD_BANG, 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_QUESTIONMARK, TEXICMD_QUOTATION, + TEXICMD_PAGE, TEXICMD_PARINDENT, TEXICMD_PRINTINDEX, TEXICMD_REF, TEXICMD_SAMP, TEXICMD_SECTION, + TEXICMD_SET, TEXICMD_SETCHAPNEWPAGE, TEXICMD_SETFILENAME, TEXICMD_SETTITLE, + TEXICMD_SP, + TEXICMD_SPACE, + TEXICMD_SMALLEXAMPLE, + TEXICMD_SQUIGGLE_LEFT, + TEXICMD_SQUIGGLE_RIGHT, 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_W, TEXICMD__MAX }; @@ -125,21 +163,27 @@ struct texitok { 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]; - size_t filepos; - unsigned flags; -#define TEXI_IGN 0x01 /* don't print anything */ -#define TEXI_HEADER (TEXI_IGN | 0x02) /* haven't seen @top yet */ -#define TEXI_LITERAL 0x04 /* output all whitespace */ - size_t outcol; /* column of output */ - int outmacro; /* whether output is in line macro */ - int seenws; /* whitespace has been ignored */ - char *dir; /* texi directory */ + 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; + int outmacro; /* whether output is in line macro */ + int seenws; /* whitespace has been ignored */ + int ign; /* don't print anything */ + int literal; }; /* FIXME: don't use this crap. */ @@ -151,14 +195,19 @@ struct texi { (' ' == (_x) || '\t' == (_x)) static void doarg1(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 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 doifnottex(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 *); @@ -167,11 +216,12 @@ static void doitalic(struct texi *, enum texicmd, cons 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 domath(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 dosection(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dosh(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 dosymbol(struct texi *, enum texicmd, const char *, size_t, size_t *); @@ -179,62 +229,100 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { doarg1, "acronym", 7 }, /* TEXICMD_ACRONYM */ { doignline, "afourpaper", 10 }, /* TEXICMD_A4PAPER */ { doignbracket, "anchor", 6 }, /* TEXICMD_ANCHOR */ - { dosh, "appendix", 8 }, /* TEXICMD_APPENDIX */ - { dosh, "appendixsec", 11 }, /* TEXICMD_APPENDIXSEC */ + { dochapter, "appendix", 8 }, /* TEXICMD_APPENDIX */ + { dochapter, "appendixsec", 11 }, /* TEXICMD_APPENDIXSEC */ + { dosymbol, "*", 1 }, /* TEXICMD_ASTERISK */ { dosymbol, "@", 1 }, /* TEXICMD_AT */ + { doignline, "author", 6 }, /* TEXICMD_AUTHOR */ + { dosymbol, "!", 1 }, /* TEXICMD_BANG */ { dobye, "bye", 3 }, /* TEXICMD_BYE */ - { dosh, "chapter", 7 }, /* TEXICMD_CHAPTER */ + { doignline, "center", 5 }, /* TEXICMD_CENTER */ + { dochapter, "chapter", 7 }, /* TEXICMD_CHAPTER */ { doignline, "cindex", 6 }, /* TEXICMD_CINDEX */ { doliteral, "code", 4 }, /* TEXICMD_CODE */ + { doitalic, "cite", 4 }, /* TEXICMD_CITE */ + { dosymbol, ":", 1 }, /* TEXICMD_COLON */ { docommand, "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 */ { doignblock, "detailmenu", 10 }, /* TEXICMD_DETAILMENU */ + { doitalic, "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 */ { NULL, "end", 3 }, /* TEXICMD_END */ { doenumerate, "enumerate", 9 }, /* TEXICMD_ENUMERATE */ + { doenv, "env", 3 }, /* TEXICMD_ENV */ { doexample, "example", 7 }, /* TEXICMD_EXAMPLE */ { dofile, "file", 4 }, /* TEXICMD_FILE */ + { doblock, "group", 5 }, /* TEXICMD_GROUP */ { dosection, "heading", 7 }, /* TEXICMD_HEADING */ + { doignline, "headings", 8 }, /* TEXICMD_HEADINGS */ + { dosymbol, "-", 1 }, /* TEXICMD_HYPHEN */ { doitalic, "i", 1 }, /* TEXICMD_I */ + { doignblock, "ifclear", 7 }, /* TEXICMD_IFCLEAR */ { doignblock, "ifhtml", 6 }, /* TEXICMD_IFHTML */ - { doifnottex, "ifnottex", 8 }, /* TEXICMD_IFNOTTEX */ + { doignblock, "ifinfo", 6 }, /* TEXICMD_IFINFO */ + { doblock, "ifnottex", 8 }, /* TEXICMD_IFNOTTEX */ { doignblock, "iftex", 5 }, /* TEXICMD_IFTEX */ + { doignblock, "ifset", 5 }, /* TEXICMD_IFSET */ { doignbracket, "image", 5 }, /* TEXICMD_IMAGE */ { doinclude, "include", 7 }, /* TEXICMD_INCLUDE */ + { doignline, "insertcopying", 13 }, /* TEXICMD_INSERTCOPYING */ { doitem, "item", 4 }, /* TEXICMD_ITEM */ { doitemize, "itemize", 7 }, /* TEXICMD_ITEMIZE */ { doliteral, "kbd", 3 }, /* TEXICMD_KBD */ { dosymbol, "LaTeX", 5 }, /* TEXICMD_LATEX */ + { domath, "math", 4 }, /* TEXICMD_MATH */ { doignblock, "menu", 4 }, /* TEXICMD_MENU */ + { dosymbol, "\n", 1 }, /* TEXICMD_NEWLINE */ { doignline, "node", 4 }, /* TEXICMD_NODE */ + { doignline, "noindent", 8 }, /* TEXICMD_NOINDENT */ + { dosymbol, "?", 1 }, /* TEXICMD_QUESTIONMARK */ { doquotation, "quotation", 9 }, /* TEXICMD_QUOTATION */ - { doignline, "printindex", 10 }, /* TEXICMD_PRINTINDEX */ + { doignline, "page", 4 }, /* TEXICMD_PAGE */ { doignline, "paragraphindent", 14 }, /* TEXICMD_PARINDENT */ + { doignline, "printindex", 10 }, /* TEXICMD_PRINTINDEX */ { dobracket, "ref", 3 }, /* TEXICMD_REF */ { doliteral, "samp", 4 }, /* TEXICMD_SAMP */ { 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 */ + { dosp, "sp", 2 }, /* TEXICMD_SP */ + { dosymbol, " ", 1 }, /* TEXICMD_SPACE */ + { doexample, "smallexample", 12 }, /* TEXICMD_SMALLEXAMPLE */ + { dosymbol, "{", 1 }, /* TEXICMD_SQUIGGLE_LEFT */ + { dosymbol, "}", 1 }, /* TEXICMD_SQUIGGLE_RIGHT */ { dosubsection, "subsection", 10 }, /* TEXICMD_SUBSECTION */ + { doignline, "subtitle", 8 }, /* TEXICMD_SUBTITLE */ + { dosymbol, "\t", 1 }, /* TEXICMD_TAB */ { dotable, "table", 5 }, /* TEXICMD_TABLE */ { doignblock, "tex", 3 }, /* TEXICMD_TEX */ { dosymbol, "TeX", 3 }, /* TEXICMD_TEXSYM */ + { doignline, "title", 5 }, /* TEXICMD_TITLE */ { dobracket, "titlefont", 9 }, /* TEXICMD_TITLEFONT */ { doignblock, "titlepage", 9 }, /* TEXICMD_TITLEPAGE */ { dotop, "top", 3 }, /* TEXICMD_TOP */ - { dosh, "unnumbered", 10 }, /* TEXICMD_UNNUMBERED */ + { dochapter, "unnumbered", 10 }, /* TEXICMD_UNNUMBERED */ { dosection, "unnumberedsec", 13 }, /* TEXICMD_UNNUMBEREDSEC */ + { doarg1, "uref", 4 }, /* TEXICMD_UREF */ { doarg1, "url", 3 }, /* TEXICMD_URL */ { doliteral, "var", 3 }, /* TEXICMD_VAR */ + { dobracket, "w", 1 }, /* TEXICMD_W */ }; /* @@ -257,10 +345,17 @@ texifilepop(struct texi *p) static void texiexit(struct texi *p) { + size_t i; + if (p->outcol) + putchar('\n'); + while (p->filepos > 0) texifilepop(p); - free(p->dir); + + for (i = 0; i < p->dirsz; i++) + free(p->dirs[i]); + free(p->dirs); } /* @@ -321,13 +416,11 @@ static void texiputchar(struct texi *p, char c) { - if (TEXI_IGN & p->flags) + if (p->ign) return; - putchar(c); if ('\n' == c) { p->outcol = 0; - p->outmacro = 0; p->seenws = 0; } else p->outcol++; @@ -349,19 +442,34 @@ texiputchars(struct texi *p, const char *s) * This should ONLY be used for mdoc(7) commands! */ static void -texifputs(struct texi *p, const char *s) +teximacroclose(struct texi *p) { - int rc; - if (TEXI_IGN & p->flags) - return; - if (p->outcol) + if (0 == --p->outmacro) texiputchar(p, '\n'); - if (EOF != (rc = fputs(s, stdout))) - p->outcol += rc; } /* + * Put an mdoc(7) command without the trailing newline. + * This should ONLY be used for mdoc(7) commands! + */ +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; +} + +/* * Put an mdoc(7) command with the trailing newline. * This should ONLY be used for mdoc(7) commands! */ @@ -369,13 +477,17 @@ static void teximacro(struct texi *p, const char *s) { - if (TEXI_IGN & p->flags) - return; + if (p->outmacro) + texierr(p, "\"%s\" in open line scope!?", s); + else if (p->literal) + texierr(p, "\"%s\" in a literal scope!?", s); + if (p->outcol) texiputchar(p, '\n'); - puts(s); - p->outcol = 0; - p->seenws = 0; + + texiputchar(p, '.'); + texiputchars(p, s); + texiputchar(p, '\n'); } /* @@ -403,8 +515,11 @@ static size_t advancenext(struct texi *p, const char *buf, size_t sz, size_t *pos) { - if (TEXI_LITERAL & p->flags) { + if (p->literal) { while (*pos < sz && isspace(buf[*pos])) { + if (*pos && '\n' == buf[*pos] && + '\\' == buf[*pos - 1]) + texiputchar(p, 'e'); texiputchar(p, buf[*pos]); advance(p, buf, pos); } @@ -418,8 +533,9 @@ advancenext(struct texi *p, const char *buf, size_t sz * 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]) + teximacro(p, "Pp"); advance(p, buf, pos); } return(*pos); @@ -429,11 +545,14 @@ advancenext(struct texi *p, const char *buf, size_t sz * Advance to the EOLN in the input stream. */ static size_t -advanceeoln(struct texi *p, const char *buf, size_t sz, size_t *pos) +advanceeoln(struct texi *p, const char *buf, + size_t sz, size_t *pos, int consumenl) { while (*pos < sz && '\n' != buf[*pos]) advance(p, buf, pos); + if (*pos < sz && consumenl) + advance(p, buf, pos); return(*pos); } @@ -463,10 +582,10 @@ texiword(struct texi *p, const char *buf, size_t sz, s * XXX: if we're in literal mode, then we shouldn't do any * reflowing of text here. */ - if (p->outcol > 72 && ! (TEXI_LITERAL & p->flags)) + if (0 == p->outmacro && p->outcol > 72 && 0 == p->literal) texiputchar(p, '\n'); - if (p->seenws && p->outcol && ! (TEXI_LITERAL & p->flags)) + if (p->seenws && p->outcol && 0 == p->literal) texiputchar(p, ' '); p->seenws = 0; @@ -501,8 +620,26 @@ texicmd(struct texi *p, const char *buf, size_t i, len; assert('@' == buf[pos]); - for (*end = ++pos; *end < sz && ! isspace(buf[*end]); (*end)++) - if ((*end > pos && '@' == buf[*end]) || '{' == buf[*end]) + + if (++pos >= sz) + return(TEXICMD__MAX); + + /* Alphabetic commands are special. */ + if ( ! isalpha(buf[pos])) { + *end = pos + 1; + 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); + } + + for (*end = pos; *end < sz && ! isspace(buf[*end]); (*end)++) + if ((*end > pos && ('@' == buf[*end] || + '{' == buf[*end] || '}' == buf[*end]))) break; len = *end - pos; @@ -513,7 +650,7 @@ texicmd(struct texi *p, const char *buf, return(i); } - texiwarn(p, "bad command: %.*s", (int)len, &buf[pos]); + texiwarn(p, "bad command: @%.*s", (int)len, &buf[pos]); return(TEXICMD__MAX); } @@ -527,11 +664,13 @@ parseeof(struct texi *p, const char *buf, size_t sz) while ((pos = advancenext(p, buf, sz, &pos)) < sz) { switch (buf[pos]) { case ('}'): - texiwarn(p, "unexpected \"}\""); + if (0 == p->ign) + texiwarn(p, "unexpected \"}\""); advance(p, buf, &pos); continue; case ('{'): - texiwarn(p, "unexpected \"{\""); + if (0 == p->ign) + texiwarn(p, "unexpected \"{\""); advance(p, buf, &pos); continue; case ('@'): @@ -556,6 +695,9 @@ parsebracket(struct texi *p, const char *buf, size_t s size_t end; enum texicmd cmd; + while (*pos < sz && isspace(buf[*pos])) + advance(p, buf, pos); + if (*pos == sz || '{' != buf[*pos]) return; advance(p, buf, pos); @@ -566,7 +708,8 @@ parsebracket(struct texi *p, const char *buf, size_t s advance(p, buf, pos); return; case ('{'): - texiwarn(p, "unexpected \"{\""); + if (0 == p->ign) + texiwarn(p, "unexpected \"{\""); advance(p, buf, pos); continue; case ('@'): @@ -585,7 +728,87 @@ parsebracket(struct texi *p, const char *buf, size_t s } } +/* + * 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) +{ + 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; + 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); + 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); + } +} + +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 \"}\""); + 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); + 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); +} + +static void parseto(struct texi *p, const char *buf, size_t sz, size_t *pos, const char *endtoken) { @@ -599,11 +822,13 @@ parseto(struct texi *p, const char *buf, while ((*pos = advancenext(p, buf, sz, pos)) < sz) { switch (buf[*pos]) { case ('}'): - texiwarn(p, "unexpected \"}\""); + if (0 == p->ign) + texiwarn(p, "unexpected \"}\""); advance(p, buf, pos); continue; case ('{'): - texiwarn(p, "unexpected \"{\""); + if (0 == p->ign) + texiwarn(p, "unexpected \"{\""); advance(p, buf, pos); continue; case ('@'): @@ -624,11 +849,12 @@ parseto(struct texi *p, const char *buf, */ if (sz - *pos >= endtoksz && 0 == strncmp (&buf[*pos], endtoken, endtoksz)) { - advanceeoln(p, buf, sz, pos); + advanceeoln(p, buf, sz, pos, 0); break; } - texiwarn(p, "unexpected \"end\""); - advanceeoln(p, buf, sz, pos); + if (0 == p->ign) + texiwarn(p, "unexpected \"end\""); + advanceeoln(p, buf, sz, pos, 0); continue; } else if (TEXICMD__MAX != cmd) if (NULL != texitoks[cmd].fp) @@ -669,66 +895,106 @@ parsefile(struct texi *p, const char *fname) } static void -doignblock(struct texi *p, enum texicmd cmd, +dodeftypevar(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - unsigned int sv = p->flags; - const char *blockname; - - p->flags |= TEXI_IGN; + 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_COPYING): - blockname = "copying"; + case (TEXICMD_DEFTYPEFN): + case (TEXICMD_DEFTYPEFUN): + blk = texitoks[cmd].tok; break; - case (TEXICMD_DETAILMENU): - blockname = "detailmenu"; + default: break; - case (TEXICMD_DIRENTRY): - blockname = "direntry"; + } + + 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; - case (TEXICMD_IFHTML): - blockname = "ifhtml"; - break; - case (TEXICMD_IFTEX): - blockname = "iftex"; - break; - case (TEXICMD_MENU): - blockname = "menu"; - break; - case (TEXICMD_TEX): - blockname = "tex"; - break; - case (TEXICMD_TITLEPAGE): - blockname = "titlepage"; - break; default: - abort(); + break; } - parseto(p, buf, sz, pos, blockname); - p->flags = sv; + 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 -doifnottex(struct texi *p, enum texicmd cmd, +doignblock(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - parseto(p, buf, sz, pos, "ifnottex"); + 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) { - if ( ! p->outmacro) - texifputs(p, "."); - texiputchars(p, macro); - texiputchar(p, ' '); + teximacroopen(p, macro); p->seenws = 0; - p->outmacro++; parsebracket(p, buf, sz, pos); - p->outmacro--; if (*pos < sz - 1 && ismpunct(buf[*pos]) && isspace(buf[*pos + 1])) { @@ -736,8 +1002,7 @@ doinline(struct texi *p, const char *buf, texiputchar(p, buf[*pos]); advance(p, buf, pos); } - if ( ! p->outmacro) - texiputchar(p, '\n'); + teximacroclose(p); } static void @@ -770,15 +1035,22 @@ doinclude(struct texi *p, enum texicmd cmd, if (strstr(fname, "../") || strstr(fname, "/..")) texierr(p, "insecure path"); - /* Append filename to original name's directory. */ - rc = snprintf(path, sizeof(path), "%s/%s", p->dir, fname); - if (rc < 0) - texierr(p, "couldn't format filename"); - else if ((size_t)rc >= sizeof(path)) - texierr(p, "path too long"); + 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)) + texierr(p, "path too long"); + else if (-1 == access(path, R_OK)) + continue; - /* Pump through to parser. */ - parsefile(p, path); + parsefile(p, path); + return; + } + + texierr(p, "couldn't find %s in includes", fname); } static void @@ -792,11 +1064,22 @@ doitalic(struct texi *p, enum texicmd cmd, } 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 (TEXI_LITERAL & p->flags) + if (p->literal) parsebracket(p, buf, sz, pos); else doinline(p, buf, sz, pos, "Li"); @@ -807,7 +1090,7 @@ doemph(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - if (TEXI_LITERAL & p->flags) + if (p->literal) doitalic(p, cmd, buf, sz, pos); else doinline(p, buf, sz, pos, "Em"); @@ -834,27 +1117,37 @@ dofile(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - if (TEXI_LITERAL & p->flags) + 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"); + advanceeoln(p, buf, sz, pos, 1); + parseto(p, buf, sz, pos, "display"); + teximacro(p, "Ed"); +} + +static void doexample(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - unsigned int sv; + const char *blk; - teximacro(p, ".Bd -literal"); - advanceeoln(p, buf, sz, pos); - if ('\n' == buf[*pos]) - advance(p, buf, pos); - sv = p->flags; - p->flags |= TEXI_LITERAL; - parseto(p, buf, sz, pos, "example"); - p->flags = sv; - teximacro(p, ".Ed"); + blk = TEXICMD_EXAMPLE == cmd ? "example" : "smallexample"; + + teximacro(p, "Bd -literal -offset indent"); + advanceeoln(p, buf, sz, pos, 1); + p->literal++; + parseto(p, buf, sz, pos, blk); + p->literal--; + teximacro(p, "Ed"); } static void @@ -871,10 +1164,24 @@ dosymbol(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + if (p->seenws && p->outcol && 0 == p->literal) { + texiputchar(p, ' '); + p->seenws = 0; + } + switch (cmd) { + case (TEXICMD_ASTERISK): + case (TEXICMD_NEWLINE): + case (TEXICMD_SPACE): + case (TEXICMD_TAB): + texiputchar(p, ' '); + break; case (TEXICMD_AT): - texiputchars(p, "@"); + texiputchar(p, '@'); break; + case (TEXICMD_BANG): + texiputchar(p, '!'); + break; case (TEXICMD_COPYRIGHT): texiputchars(p, "\\(co"); break; @@ -884,14 +1191,28 @@ dosymbol(struct texi *p, enum texicmd cmd, case (TEXICMD_LATEX): texiputchars(p, "LaTeX"); break; + case (TEXICMD_QUESTIONMARK): + texiputchar(p, '?'); + break; + case (TEXICMD_SQUIGGLE_LEFT): + texiputchars(p, "{"); + break; + case (TEXICMD_SQUIGGLE_RIGHT): + texiputchars(p, "}"); + break; case (TEXICMD_TEXSYM): texiputchars(p, "TeX"); break; + case (TEXICMD_COLON): + case (TEXICMD_HYPHEN): + break; default: + texiwarn(p, "sym: %d", cmd); abort(); } - doignbracket(p, cmd, buf, sz, pos); + if (texitoks[cmd].len > 1) + doignbracket(p, cmd, buf, sz, pos); } static void @@ -899,12 +1220,49 @@ doquotation(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - teximacro(p, ".Qo"); + teximacro(p, "Qo"); parseto(p, buf, sz, pos, "quotation"); - teximacro(p, ".Qc"); + teximacro(p, "Qc"); } +/* FIXME */ static void +domath(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + size_t nest; + + /* + * Math handling is different from everything else. + * We don't allow any subcomponents, and we ignore the rules in + * terms of @-commands. + * This departs from GNU's rules, but whatever. + */ + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + if (*pos == sz || '{' != buf[*pos]) + return; + advance(p, buf, pos); + if (p->seenws && p->outcol && 0 == p->literal) + texiputchar(p, ' '); + p->seenws = 0; + for (nest = 1; *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]); + advance(p, buf, pos); +} + +/* FIXME */ +static void doarg1(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { @@ -912,18 +1270,18 @@ doarg1(struct texi *p, enum texicmd cmd, if (*pos == sz || '{' != buf[*pos]) return; advance(p, buf, pos); - if ( ! p->outmacro) - texifputs(p, "."); switch (cmd) { case (TEXICMD_EMAIL): - texiputchars(p, "Lk "); + teximacroopen(p, "Mt"); break; + case (TEXICMD_UREF): case (TEXICMD_URL): - texiputchars(p, "Mt "); + teximacroopen(p, "Lk"); break; default: - abort(); + break; } + /* FIXME: this is ugly */ while (*pos < sz && '}' != buf[*pos] && ',' != buf[*pos]) { texiputchar(p, buf[*pos]); advance(p, buf, pos); @@ -939,8 +1297,15 @@ doarg1(struct texi *p, enum texicmd cmd, texiputchar(p, buf[*pos]); advance(p, buf, pos); } - if ( ! p->outmacro) - texiputchar(p, '\n'); + switch (cmd) { + case (TEXICMD_EMAIL): + case (TEXICMD_UREF): + case (TEXICMD_URL): + teximacroclose(p); + break; + default: + break; + } } static void @@ -948,18 +1313,11 @@ dosubsection(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - if (TEXI_IGN & p->flags) { - advanceeoln(p, buf, sz, pos); - return; - } - while (*pos < sz && ' ' == buf[*pos]) - advance(p, buf, pos); - texifputs(p, ".Pp"); - while (*pos < sz && '\n' != buf[*pos]) { - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - texifputs(p, ".Pp"); + teximacro(p, "Pp"); + teximacroopen(p, "Em"); + parseeoln(p, buf, sz, pos); + teximacroclose(p); + teximacro(p, "Pp"); } static void @@ -967,117 +1325,132 @@ dosection(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - if (TEXI_IGN & p->flags) { - advanceeoln(p, buf, sz, pos); - return; - } - while (*pos < sz && isws(buf[*pos]) ) - advance(p, buf, pos); - texifputs(p, ".Ss "); - while (*pos < sz && '\n' != buf[*pos]) { - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - texiputchar(p, '\n'); + if (p->outmacro) + texierr(p, "subsection in open line scope!?"); + else if (p->literal) + texierr(p, "subsection in a literal scope!?"); + + teximacroopen(p, "Ss"); + parseeoln(p, buf, sz, pos); + teximacroclose(p); } static void -dosh(struct texi *p, enum texicmd cmd, +dosp(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - if (TEXI_IGN & p->flags) { - advanceeoln(p, buf, sz, pos); - return; - } - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); - texifputs(p, ".Sh "); - while (*pos < sz && '\n' != buf[*pos]) { - texiputchar(p, toupper(buf[*pos])); - advance(p, buf, pos); - } - texiputchar(p, '\n'); + teximacro(p, "Pp"); + 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) { - p->flags &= ~TEXI_HEADER; - advanceeoln(p, buf, sz, pos); - teximacro(p, ".Dd $Mdocdate: February 17 2015 $"); - teximacro(p, ".Dt SOMETHING 7"); - teximacro(p, ".Os"); - teximacro(p, ".Sh NAME"); - teximacro(p, ".Nm Something"); - teximacro(p, ".Nd Something"); + p->ign--; + advanceeoln(p, buf, sz, pos, 1); + teximacro(p, "Dd $Mdocdate: February 18 2015 $"); + teximacro(p, "Dt SOMETHING 7"); + teximacro(p, "Os"); + teximacro(p, "Sh NAME"); + teximacro(p, "Nm Something"); + teximacro(p, "Nd Something"); } static void doitem(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - size_t end; - /* See if we have arguments... */ - for (end = *pos; end < sz; end++) - if ( ! isws(buf[end])) - break; + if (p->outmacro) + texierr(p, "item in open line scope!?"); + else if (p->literal) + texierr(p, "item in a literal scope!?"); - /* If we have arguments, print them too. */ - if ('\n' != buf[end]) { - texifputs(p, ".It"); - /* FIXME: process commands. */ - while (*pos < sz && '\n' != buf[*pos]) { - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } + switch (p->list) { + case (TEXILIST_ITEM): + teximacroopen(p, "It"); + break; + case (TEXILIST_NOITEM): + teximacro(p, "It"); + break; + default: + teximacro(p, "Pp"); + break; + } + + parseeoln(p, buf, sz, pos); + + if (TEXILIST_ITEM == p->list) + teximacroclose(p); + else texiputchar(p, '\n'); - } else - teximacro(p, ".It"); } static void dotable(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + enum texilist sv = p->list; - teximacro(p, ".Bl -tag -width Ds"); + p->list = TEXILIST_ITEM; + teximacro(p, "Bl -tag -width Ds"); parseto(p, buf, sz, pos, "table"); - teximacro(p, ".El"); + teximacro(p, "El"); + p->list = sv; } static void doenumerate(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + enum texilist sv = p->list; - teximacro(p, ".Bl -enum"); + p->list = TEXILIST_NOITEM; + teximacro(p, "Bl -enum"); parseto(p, buf, sz, pos, "enumerate"); - teximacro(p, ".El"); + teximacro(p, "El"); + p->list = sv; } static void doitemize(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + enum texilist sv = p->list; - teximacro(p, ".Bl -bullet"); + p->list = TEXILIST_ITEM; + teximacro(p, "Bl -bullet"); parseto(p, buf, sz, pos, "itemize"); - teximacro(p, ".El"); + teximacro(p, "El"); + p->list = sv; } static void doignbracket(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - unsigned int sv = p->flags; - p->flags |= TEXI_IGN; + p->ign++; parsebracket(p, buf, sz, pos); - p->flags = sv; + p->ign--; } static void @@ -1085,18 +1458,56 @@ doignline(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - advanceeoln(p, buf, sz, pos); - if (*pos < sz) - advance(p, buf, pos); + advanceeoln(p, buf, sz, pos, 1); } +static char ** +parsedirs(const char *base, const char *cp, size_t *sz) +{ + char *tok, *str, *tofree; + const char *cpp; + size_t i; + char **dirs; + + *sz = NULL != (cpp = cp) ? 2 : 1; + if (*sz > 1) + for ( ; 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 (NULL == cp) + return(dirs); + + if (NULL == (tofree = tok = str = strdup(cp))) { + perror(NULL); + exit(EXIT_FAILURE); + } + + for (i = 1; NULL != (tok = strsep(&str, ":")); i++) + if (NULL == (dirs[i] = strdup(tok))) { + perror(NULL); + exit(EXIT_FAILURE); + } + + free(tofree); + return(dirs); +} + int main(int argc, char *argv[]) { struct texi texi; int c; char *path, *dir; - const char *progname; + const char *progname, *Idir; progname = strrchr(argv[0], '/'); if (progname == NULL) @@ -1104,8 +1515,12 @@ main(int argc, char *argv[]) else ++progname; - while (-1 != (c = getopt(argc, argv, ""))) + Idir = NULL; + while (-1 != (c = getopt(argc, argv, "I:"))) switch (c) { + case ('I'): + Idir = optarg; + break; default: goto usage; } @@ -1125,9 +1540,10 @@ main(int argc, char *argv[]) free(path); memset(&texi, 0, sizeof(struct texi)); - texi.flags = TEXI_HEADER; - texi.dir = strdup(dir); + 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: