=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.5 retrieving revision 1.13 diff -u -p -r1.5 -r1.13 --- texi2mdoc/main.c 2015/02/18 11:01:56 1.5 +++ texi2mdoc/main.c 2015/02/19 07:50:51 1.13 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.5 2015/02/18 11:01:56 kristaps Exp $ */ +/* $Id: main.c,v 1.13 2015/02/19 07:50:51 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -27,6 +27,8 @@ #include #include #include +#include +#include /* * This defines each one of the Texinfo commands that we understand. @@ -42,7 +44,9 @@ enum texicmd { TEXICMD_ASTERISK, TEXICMD_AT, TEXICMD_AUTHOR, + TEXICMD_B, TEXICMD_BANG, + TEXICMD_BULLET, TEXICMD_BYE, TEXICMD_CENTER, TEXICMD_CHAPTER, @@ -88,6 +92,7 @@ enum texicmd { TEXICMD_IFSET, TEXICMD_IMAGE, TEXICMD_INCLUDE, + TEXICMD_INDENTBLOCK, TEXICMD_INSERTCOPYING, TEXICMD_ITEM, TEXICMD_ITEMIZE, @@ -98,25 +103,34 @@ enum texicmd { TEXICMD_NEWLINE, TEXICMD_NODE, TEXICMD_NOINDENT, + TEXICMD_PXREF, TEXICMD_QUESTIONMARK, TEXICMD_QUOTATION, TEXICMD_PAGE, TEXICMD_PARINDENT, TEXICMD_PRINTINDEX, + TEXICMD_R, TEXICMD_REF, TEXICMD_SAMP, + TEXICMD_SANSSERIF, + TEXICMD_SC, TEXICMD_SECTION, TEXICMD_SET, TEXICMD_SETCHAPNEWPAGE, TEXICMD_SETFILENAME, TEXICMD_SETTITLE, + TEXICMD_SLANTED, TEXICMD_SP, TEXICMD_SPACE, + TEXICMD_SMALLDISPLAY, TEXICMD_SMALLEXAMPLE, + TEXICMD_SMALLINDENTBLOCK, TEXICMD_SQUIGGLE_LEFT, TEXICMD_SQUIGGLE_RIGHT, + TEXICMD_STRONG, TEXICMD_SUBSECTION, TEXICMD_SUBTITLE, + TEXICMD_T, TEXICMD_TAB, TEXICMD_TABLE, TEXICMD_TEX, @@ -130,7 +144,9 @@ enum texicmd { TEXICMD_UREF, TEXICMD_URL, TEXICMD_VAR, + TEXICMD_VSKIP, TEXICMD_W, + TEXICMD_XREF, TEXICMD__MAX }; @@ -179,43 +195,44 @@ 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 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 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 doquotation(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dotable(struct texi *, enum texicmd, const char *, size_t, size_t *); @@ -224,25 +241,28 @@ 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 dotitle(struct texi *, enum texicmd, const char *, size_t, size_t *); static const struct texitok texitoks[TEXICMD__MAX] = { - { doarg1, "acronym", 7 }, /* TEXICMD_ACRONYM */ + { doignargn, "acronym", 7 }, /* TEXICMD_ACRONYM */ { 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 */ { 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 */ + { dofont, "code", 4 }, /* TEXICMD_CODE */ + { dofont, "cite", 4 }, /* TEXICMD_CITE */ { dosymbol, ":", 1 }, /* TEXICMD_COLON */ - { docommand, "command", 7 }, /* TEXICMD_COMMAND */ + { doinline, "command", 7 }, /* TEXICMD_COMMAND */ { doignline, "c", 1 }, /* TEXICMD_COMMENT */ { doignline, "comment", 7 }, /* TEXICMD_COMMENT_LONG */ { doignline, "contents", 8 }, /* TEXICMD_CONTENTS */ @@ -255,23 +275,23 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { dodeftypevar, "deftypevar", 10 }, /* TEXICMD_DEFTYPEVAR */ { dodeftypevar, "deftypevr", 9 }, /* TEXICMD_DEFTYPEVR */ { 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 */ { doexample, "example", 7 }, /* TEXICMD_EXAMPLE */ - { dofile, "file", 4 }, /* TEXICMD_FILE */ + { doinline, "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 */ + { dofont, "i", 1 }, /* TEXICMD_I */ { doignblock, "ifclear", 7 }, /* TEXICMD_IFCLEAR */ { doignblock, "ifhtml", 6 }, /* TEXICMD_IFHTML */ { doignblock, "ifinfo", 6 }, /* TEXICMD_IFINFO */ @@ -280,35 +300,45 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { doignblock, "ifset", 5 }, /* TEXICMD_IFSET */ { 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 */ + { dofont, "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 */ + { 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 */ + { 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 */ + { dodisplay, "smalldisplay", 12 }, /* TEXICMD_SMALLDISPLAY */ { doexample, "smallexample", 12 }, /* TEXICMD_SMALLEXAMPLE */ + { dodisplay, "smallindentblock", 16 }, /* TEXICMD_SMALLINDENTBLOCK */ { dosymbol, "{", 1 }, /* TEXICMD_SQUIGGLE_LEFT */ { dosymbol, "}", 1 }, /* TEXICMD_SQUIGGLE_RIGHT */ + { dofont, "strong", 6 }, /* TEXICMD_STRONG */ { dosubsection, "subsection", 10 }, /* TEXICMD_SUBSECTION */ { doignline, "subtitle", 8 }, /* TEXICMD_SUBTITLE */ + { dofont, "t", 1 }, /* TEXICMD_T */ { dosymbol, "\t", 1 }, /* TEXICMD_TAB */ { dotable, "table", 5 }, /* TEXICMD_TABLE */ { doignblock, "tex", 3 }, /* TEXICMD_TEX */ @@ -317,12 +347,14 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { dobracket, "titlefont", 9 }, /* TEXICMD_TITLEFONT */ { doignblock, "titlepage", 9 }, /* TEXICMD_TITLEPAGE */ { dotop, "top", 3 }, /* TEXICMD_TOP */ - { dochapter, "unnumbered", 10 }, /* TEXICMD_UNNUMBERED */ + { 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 */ + { dolink, "uref", 4 }, /* TEXICMD_UREF */ + { dolink, "url", 3 }, /* TEXICMD_URL */ + { doinline, "var", 3 }, /* TEXICMD_VAR */ + { dosp, "vskip", 5 }, /* TEXICMD_VSKIP */ { dobracket, "w", 1 }, /* TEXICMD_W */ + { dolink, "xref", 4 }, /* TEXICMD_XREF */ }; /* @@ -339,7 +371,8 @@ 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 @@ -355,7 +388,10 @@ texiexit(struct texi *p) for (i = 0; i < p->dirsz; i++) free(p->dirs[i]); + free(p->dirs); + free(p->subtitle); + free(p->title); } /* @@ -390,6 +426,10 @@ 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, ...) { @@ -408,9 +448,8 @@ 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. */ static void texiputchar(struct texi *p, char c) @@ -419,6 +458,7 @@ texiputchar(struct texi *p, char c) if (p->ign) return; putchar(c); + p->seenvs = 0; if ('\n' == c) { p->outcol = 0; p->seenws = 0; @@ -438,20 +478,22 @@ 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) { + /* FIXME: punctuation. */ if (0 == --p->outmacro) texiputchar(p, '\n'); } /* - * 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) @@ -470,8 +512,7 @@ teximacroopen(struct texi *p, const char *s) } /* - * 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) @@ -490,6 +531,16 @@ teximacro(struct texi *p, const char *s) texiputchar(p, '\n'); } +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. */ @@ -506,6 +557,40 @@ 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 @@ -516,7 +601,7 @@ 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'); @@ -526,16 +611,15 @@ 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); @@ -549,6 +633,7 @@ 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) @@ -575,28 +660,27 @@ 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) 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]) { @@ -613,6 +697,11 @@ 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) @@ -621,12 +710,15 @@ 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; @@ -637,7 +729,7 @@ 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; @@ -654,48 +746,69 @@ 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]) @@ -715,7 +828,7 @@ 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; } @@ -760,7 +873,7 @@ 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; } @@ -773,6 +886,10 @@ 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) { @@ -796,7 +913,7 @@ 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; } @@ -808,7 +925,24 @@ parsesingle(struct texi *p, const char *buf, size_t sz (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); } +/* + * 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) { @@ -834,7 +968,7 @@ parseto(struct texi *p, const char *buf, case ('@'): break; default: - texiword(p, buf, sz, pos); + texiword(p, buf, sz, pos, '\0'); continue; } @@ -844,8 +978,8 @@ 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)) { @@ -862,6 +996,10 @@ 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) { @@ -908,7 +1046,7 @@ dodeftypevar(struct texi *p, enum texicmd cmd, return; } - teximacro(p, "Pp"); + texivspace(p); if (TEXICMD_DEFTYPEVR == cmd) { parsebracket(p, buf, sz, pos); texiputchars(p, ":\n"); @@ -916,7 +1054,7 @@ dodeftypevar(struct texi *p, enum texicmd cmd, teximacroopen(p, "Vt"); parseeoln(p, buf, sz, pos); teximacroclose(p); - teximacro(p, "Pp"); + texivspace(p); parseto(p, buf, sz, pos, blk); } @@ -945,7 +1083,7 @@ dodeftypefun(struct texi *p, enum texicmd cmd, switch (cmd) { case (TEXICMD_DEFTYPEFN): case (TEXICMD_DEFTYPEFUN): - teximacro(p, "Pp"); + texivspace(p); break; default: break; @@ -964,7 +1102,7 @@ dodeftypefun(struct texi *p, enum texicmd cmd, teximacroopen(p, "Li"); parseeoln(p, buf, sz, pos); teximacroclose(p); - teximacro(p, "Pp"); + texivspace(p); if (NULL != blk) parseto(p, buf, sz, pos, blk); } @@ -988,20 +1126,37 @@ 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_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]); - advance(p, buf, pos); - } + texipunctuate(p, buf, sz, pos); teximacroclose(p); } @@ -1038,7 +1193,6 @@ 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)) @@ -1054,57 +1208,41 @@ 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(); + } + + 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) { @@ -1113,24 +1251,15 @@ 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"); + teximacro(p, "Bd -filled -offset indent"); + 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"); } @@ -1138,14 +1267,12 @@ 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"); } @@ -1160,6 +1287,23 @@ 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 dosymbol(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { @@ -1182,6 +1326,9 @@ 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; @@ -1225,7 +1372,6 @@ 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) @@ -1261,15 +1407,12 @@ 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"); @@ -1278,61 +1421,90 @@ 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 @@ -1340,38 +1512,42 @@ 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]; + 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 @@ -1392,15 +1568,16 @@ doitem(struct texi *p, enum texicmd cmd, teximacro(p, "It"); break; default: - teximacro(p, "Pp"); + texivspace(p); break; } + p->seenvs = 1; parseeoln(p, buf, sz, pos); if (TEXILIST_ITEM == p->list) teximacroclose(p); - else + else if (p->outcol > 0) texiputchar(p, '\n'); } @@ -1412,6 +1589,9 @@ dotable(struct texi *p, enum texicmd cmd, p->list = TEXILIST_ITEM; teximacro(p, "Bl -tag -width Ds"); + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); + p->seenvs = 1; parseto(p, buf, sz, pos, "table"); teximacro(p, "El"); p->list = sv; @@ -1425,6 +1605,9 @@ 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; @@ -1438,6 +1621,9 @@ 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; @@ -1458,9 +1644,16 @@ 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) { @@ -1507,7 +1700,7 @@ 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) @@ -1515,7 +1708,9 @@ 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'): @@ -1539,7 +1734,17 @@ 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]); @@ -1547,6 +1752,6 @@ main(int argc, char *argv[]) texiexit(&texi); return(EXIT_FAILURE); usage: - fprintf(stderr, "usage: %s file\n", progname); + fprintf(stderr, "usage: %s [-Idirs] file\n", progname); return(EXIT_FAILURE); }