=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.4 retrieving revision 1.8 diff -u -p -r1.4 -r1.8 --- texi2mdoc/main.c 2015/02/17 20:27:44 1.4 +++ texi2mdoc/main.c 2015/02/18 12:03:21 1.8 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.4 2015/02/17 20:27:44 kristaps Exp $ */ +/* $Id: main.c,v 1.8 2015/02/18 12:03:21 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -27,6 +27,7 @@ #include #include #include +#include /* * This defines each one of the Texinfo commands that we understand. @@ -43,7 +44,9 @@ enum texicmd { TEXICMD_AT, TEXICMD_AUTHOR, TEXICMD_BANG, + TEXICMD_BULLET, TEXICMD_BYE, + TEXICMD_CENTER, TEXICMD_CHAPTER, TEXICMD_CINDEX, TEXICMD_CITE, @@ -87,6 +90,7 @@ enum texicmd { TEXICMD_IFSET, TEXICMD_IMAGE, TEXICMD_INCLUDE, + TEXICMD_INSERTCOPYING, TEXICMD_ITEM, TEXICMD_ITEMIZE, TEXICMD_KBD, @@ -96,6 +100,7 @@ enum texicmd { TEXICMD_NEWLINE, TEXICMD_NODE, TEXICMD_NOINDENT, + TEXICMD_PXREF, TEXICMD_QUESTIONMARK, TEXICMD_QUOTATION, TEXICMD_PAGE, @@ -103,6 +108,7 @@ enum texicmd { TEXICMD_PRINTINDEX, TEXICMD_REF, TEXICMD_SAMP, + TEXICMD_SC, TEXICMD_SECTION, TEXICMD_SET, TEXICMD_SETCHAPNEWPAGE, @@ -113,6 +119,7 @@ enum texicmd { TEXICMD_SMALLEXAMPLE, TEXICMD_SQUIGGLE_LEFT, TEXICMD_SQUIGGLE_RIGHT, + TEXICMD_STRONG, TEXICMD_SUBSECTION, TEXICMD_SUBTITLE, TEXICMD_TAB, @@ -129,6 +136,7 @@ enum texicmd { TEXICMD_URL, TEXICMD_VAR, TEXICMD_W, + TEXICMD_XREF, TEXICMD__MAX }; @@ -172,26 +180,27 @@ enum texilist { * This keeps any necessary information handy. */ struct texi { - struct texifile files[64]; - size_t filepos; - size_t outcol; /* column of output */ - int outmacro; /* whether output is in line macro */ - int seenws; /* whitespace has been ignored */ - int ign; /* don't print anything */ - int literal; - char *dir; /* texi directory */ - enum texilist list; + struct texifile files[64]; /* stack of open files */ + size_t filepos; /* number of open files */ + size_t outcol; /* column in output line */ + char **dirs; /* texi directories */ + size_t dirsz; /* number of texi directories */ + enum texilist list; /* current list (set recursively) */ + int outmacro; /* if >0, output is in line macro */ + int seenws; /* ws has been seen (and ignored) */ + int ign; /* if >0, don't print anything */ + int literal; /* if >0, literal context */ }; -/* FIXME: don't use this crap. */ +/* FIXME: FIND A BETTER WAY. */ #define ismpunct(_x) \ ('.' == (_x) || \ ',' == (_x) || \ ';' == (_x)) +/* Texi disregards spaces and tabs. */ #define isws(_x) \ (' ' == (_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 *); @@ -205,6 +214,7 @@ static void doenumerate(struct texi *, enum texicmd, c static void doenv(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doexample(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dofile(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void doignargn(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doignblock(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doignbracket(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doignline(struct texi *, enum texicmd, const char *, size_t, size_t *); @@ -212,6 +222,7 @@ static void doinclude(struct texi *, enum texicmd, con static void doitalic(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doitem(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doitemize(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dolink(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doliteral(struct texi *, enum texicmd, const char *, size_t, size_t *); static void domath(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doquotation(struct texi *, enum texicmd, const char *, size_t, size_t *); @@ -223,7 +234,7 @@ static void dosubsection(struct texi *, enum texicmd, static void dosymbol(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 */ @@ -232,7 +243,9 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { dosymbol, "@", 1 }, /* TEXICMD_AT */ { doignline, "author", 6 }, /* TEXICMD_AUTHOR */ { 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, "cindex", 6 }, /* TEXICMD_CINDEX */ { doliteral, "code", 4 }, /* TEXICMD_CODE */ @@ -256,7 +269,7 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { doignblock, "direntry", 8 }, /* TEXICMD_DIRENTRY */ { dodisplay, "display", 7 }, /* TEXICMD_DISPLAY */ { dosymbol, "dots", 4 }, /* TEXICMD_DOTS */ - { doarg1, "email", 5 }, /* TEXICMD_EMAIL */ + { dolink, "email", 5 }, /* TEXICMD_EMAIL */ { doemph, "emph", 4 }, /* TEXICMD_EMPH */ { NULL, "end", 3 }, /* TEXICMD_END */ { doenumerate, "enumerate", 9 }, /* TEXICMD_ENUMERATE */ @@ -276,6 +289,7 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { 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 */ @@ -285,6 +299,7 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { 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 */ @@ -292,16 +307,18 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { doignline, "printindex", 10 }, /* TEXICMD_PRINTINDEX */ { dobracket, "ref", 3 }, /* TEXICMD_REF */ { doliteral, "samp", 4 }, /* TEXICMD_SAMP */ + { 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 */ { dosp, "sp", 2 }, /* TEXICMD_SP */ { dosymbol, " ", 1 }, /* TEXICMD_SPACE */ { doexample, "smallexample", 12 }, /* TEXICMD_SMALLEXAMPLE */ - { doignline, "settitle", 8 }, /* TEXICMD_SETTITLE */ { dosymbol, "{", 1 }, /* TEXICMD_SQUIGGLE_LEFT */ { dosymbol, "}", 1 }, /* TEXICMD_SQUIGGLE_RIGHT */ + { doemph, "strong", 6 }, /* TEXICMD_STRONG */ { dosubsection, "subsection", 10 }, /* TEXICMD_SUBSECTION */ { doignline, "subtitle", 8 }, /* TEXICMD_SUBTITLE */ { dosymbol, "\t", 1 }, /* TEXICMD_TAB */ @@ -314,10 +331,11 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { dotop, "top", 3 }, /* TEXICMD_TOP */ { dochapter, "unnumbered", 10 }, /* TEXICMD_UNNUMBERED */ { dosection, "unnumberedsec", 13 }, /* TEXICMD_UNNUMBEREDSEC */ - { doarg1, "uref", 4 }, /* TEXICMD_UREF */ - { doarg1, "url", 3 }, /* TEXICMD_URL */ + { dolink, "uref", 4 }, /* TEXICMD_UREF */ + { dolink, "url", 3 }, /* TEXICMD_URL */ { doliteral, "var", 3 }, /* TEXICMD_VAR */ { dobracket, "w", 1 }, /* TEXICMD_W */ + { dolink, "xref", 4 }, /* TEXICMD_XREF */ }; /* @@ -334,18 +352,24 @@ 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 texiexit(struct texi *p) { + size_t i; - while (p->filepos > 0) - texifilepop(p); - free(p->dir); if (p->outcol) putchar('\n'); + + while (p->filepos > 0) + texifilepop(p); + + for (i = 0; i < p->dirsz; i++) + free(p->dirs[i]); + free(p->dirs); } /* @@ -380,6 +404,10 @@ 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, ...) { @@ -398,9 +426,8 @@ 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) @@ -428,40 +455,41 @@ 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) { - p->outmacro--; - if (p->ign) - return; - texiputchar(p, '\n'); + /* 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) { - int rc; - p->outmacro++; - if (p->ign) - return; - if (p->outcol) + if (p->outcol && 0 == p->outmacro) texiputchar(p, '\n'); - if (EOF != (rc = fputs(s, stdout))) - p->outcol += rc; + 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! + * Put a stadnalone mdoc(7) command with the trailing newline. */ static void teximacro(struct texi *p, const char *s) @@ -472,13 +500,12 @@ teximacro(struct texi *p, const char *s) else if (p->literal) texierr(p, "\"%s\" in a literal scope!?", s); - if (p->ign) - return; if (p->outcol) texiputchar(p, '\n'); - puts(s); - p->outcol = 0; - p->seenws = 0; + + texiputchar(p, '.'); + texiputchars(p, s); + texiputchar(p, '\n'); } /* @@ -508,6 +535,9 @@ advancenext(struct texi *p, const char *buf, size_t sz 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); } @@ -521,8 +551,9 @@ 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); @@ -536,6 +567,7 @@ 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) @@ -562,16 +594,13 @@ 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, ' '); @@ -584,6 +613,8 @@ texiword(struct texi *p, const char *buf, size_t sz, s case ('{'): return; } + if ('\0' != extra && buf[*pos] == extra) + return; if (*pos < sz - 1 && '`' == buf[*pos] && '`' == buf[*pos + 1]) { @@ -600,6 +631,11 @@ 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) @@ -641,41 +677,62 @@ 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 && isspace(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) { @@ -702,7 +759,7 @@ 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; } @@ -747,7 +804,7 @@ 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; } @@ -760,6 +817,10 @@ 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) { @@ -783,7 +844,7 @@ 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; } @@ -795,7 +856,24 @@ 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) { @@ -821,7 +899,7 @@ parseto(struct texi *p, const char *buf, case ('@'): break; default: - texiword(p, buf, sz, pos); + texiword(p, buf, sz, pos, '\0'); continue; } @@ -831,8 +909,8 @@ 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)) { @@ -849,6 +927,10 @@ 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) { @@ -895,15 +977,15 @@ dodeftypevar(struct texi *p, enum texicmd cmd, return; } - teximacro(p, ".Pp"); + teximacro(p, "Pp"); if (TEXICMD_DEFTYPEVR == cmd) { parsebracket(p, buf, sz, pos); texiputchars(p, ":\n"); } - teximacroopen(p, ".Vt "); + teximacroopen(p, "Vt"); parseeoln(p, buf, sz, pos); teximacroclose(p); - teximacro(p, ".Pp"); + teximacro(p, "Pp"); parseto(p, buf, sz, pos, blk); } @@ -913,19 +995,14 @@ dodeftypefun(struct texi *p, enum texicmd cmd, { const char *blk; + blk = NULL; switch (cmd) { case (TEXICMD_DEFTYPEFN): - blk = "deftypefn"; - break; case (TEXICMD_DEFTYPEFUN): - blk = "deftypefun"; + blk = texitoks[cmd].tok; break; - case (TEXICMD_DEFTYPEFNX): - case (TEXICMD_DEFTYPEFUNX): - blk = NULL; - break; default: - abort(); + break; } if (p->ign) { @@ -937,7 +1014,7 @@ dodeftypefun(struct texi *p, enum texicmd cmd, switch (cmd) { case (TEXICMD_DEFTYPEFN): case (TEXICMD_DEFTYPEFUN): - teximacro(p, ".Pp"); + teximacro(p, "Pp"); break; default: break; @@ -947,16 +1024,16 @@ dodeftypefun(struct texi *p, enum texicmd cmd, parsebracket(p, buf, sz, pos); texiputchars(p, ":\n"); } - teximacroopen(p, ".Ft "); + teximacroopen(p, "Ft"); parsesingle(p, buf, sz, pos); teximacroclose(p); - teximacroopen(p, ".Fn "); + teximacroopen(p, "Fn"); parsesingle(p, buf, sz, pos); teximacroclose(p); - teximacroopen(p, ".Li "); + teximacroopen(p, "Li"); parseeoln(p, buf, sz, pos); teximacroclose(p); - teximacro(p, ".Pp"); + teximacro(p, "Pp"); if (NULL != blk) parseto(p, buf, sz, pos, blk); } @@ -965,47 +1042,9 @@ static void doignblock(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - const char *blk; - switch (cmd) { - case (TEXICMD_COPYING): - blk = "copying"; - break; - case (TEXICMD_DETAILMENU): - blk = "detailmenu"; - break; - case (TEXICMD_DIRENTRY): - blk = "direntry"; - break; - case (TEXICMD_IFCLEAR): - blk = "ifclear"; - break; - case (TEXICMD_IFHTML): - blk = "ifhtml"; - break; - case (TEXICMD_IFINFO): - blk = "ifinfo"; - break; - case (TEXICMD_IFSET): - blk = "ifset"; - break; - case (TEXICMD_IFTEX): - blk = "iftex"; - break; - case (TEXICMD_MENU): - blk = "menu"; - break; - case (TEXICMD_TEX): - blk = "tex"; - break; - case (TEXICMD_TITLEPAGE): - blk = "titlepage"; - break; - default: - abort(); - } p->ign++; - parseto(p, buf, sz, pos, blk); + parseto(p, buf, sz, pos, texitoks[cmd].tok); p->ign--; } @@ -1013,36 +1052,16 @@ static void doblock(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - const char *blk; - - switch (cmd) { - case (TEXICMD_GROUP): - blk = "group"; - break; - case (TEXICMD_IFNOTTEX): - blk = "ifnottex"; - break; - default: - abort(); - } - parseto(p, buf, sz, pos, blk); + 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) { - int open = 0; - if (0 == p->outmacro) { - open = 1; - teximacroopen(p, "."); - } else - texiputchar(p, ' '); - - texiputchars(p, macro); - texiputchar(p, ' '); + teximacroopen(p, macro); p->seenws = 0; parsebracket(p, buf, sz, pos); if (*pos < sz - 1 && @@ -1052,8 +1071,7 @@ doinline(struct texi *p, const char *buf, texiputchar(p, buf[*pos]); advance(p, buf, pos); } - if (open) - teximacroclose(p); + teximacroclose(p); } static void @@ -1086,15 +1104,21 @@ 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); + 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 @@ -1172,10 +1196,10 @@ dodisplay(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - teximacro(p, ".Bd -display"); + teximacro(p, "Bd -display -offset indent"); advanceeoln(p, buf, sz, pos, 1); parseto(p, buf, sz, pos, "display"); - teximacro(p, ".Ed"); + teximacro(p, "Ed"); } static void @@ -1186,12 +1210,12 @@ doexample(struct texi *p, enum texicmd cmd, blk = TEXICMD_EXAMPLE == cmd ? "example" : "smallexample"; - teximacro(p, ".Bd -literal"); + 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"); + teximacro(p, "Ed"); } static void @@ -1226,6 +1250,9 @@ 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; @@ -1251,10 +1278,12 @@ dosymbol(struct texi *p, enum texicmd cmd, 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 @@ -1262,12 +1291,11 @@ 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) @@ -1303,44 +1331,38 @@ 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 open = 0; + int c; - if (*pos == sz || '{' != buf[*pos]) - return; - advance(p, buf, pos); switch (cmd) { case (TEXICMD_EMAIL): - if ( ! p->outmacro) { - open = 1; - teximacroopen(p, "."); - } - texiputchars(p, "Lk "); + teximacroopen(p, "Mt"); break; case (TEXICMD_UREF): case (TEXICMD_URL): - if ( ! p->outmacro) { - open = 1; - teximacroopen(p, "."); - } - texiputchars(p, "Mt "); + teximacroopen(p, "Lk"); break; - default: + case (TEXICMD_XREF): + texiputchars(p, "See Section"); + teximacroopen(p, "Qq"); break; + case (TEXICMD_PXREF): + texiputchars(p, "see Section"); + teximacroopen(p, "Qq"); + break; + default: + abort(); } - /* 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); + + c = parsearg(p, buf, sz, pos, 0); + p->ign++; + while (c > 0) + c = parsearg(p, buf, sz, pos, 1); + p->ign--; + if (*pos < sz - 1 && ismpunct(buf[*pos]) && isspace(buf[*pos + 1])) { @@ -1348,18 +1370,33 @@ doarg1(struct texi *p, enum texicmd cmd, texiputchar(p, buf[*pos]); advance(p, buf, pos); } - if (open) - teximacroclose(p); + + 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"); + teximacro(p, "Pp"); + teximacroopen(p, "Em"); parseeoln(p, buf, sz, pos); - teximacro(p, ".Pp"); + teximacroclose(p); + teximacro(p, "Pp"); } static void @@ -1372,7 +1409,7 @@ dosection(struct texi *p, enum texicmd cmd, else if (p->literal) texierr(p, "subsection in a literal scope!?"); - teximacroopen(p, ".Ss "); + teximacroopen(p, "Ss"); parseeoln(p, buf, sz, pos); teximacroclose(p); } @@ -1382,7 +1419,7 @@ dosp(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - teximacro(p, ".Pp"); + teximacro(p, "Pp"); advanceeoln(p, buf, sz, pos, 1); } @@ -1396,7 +1433,7 @@ dochapter(struct texi *p, enum texicmd cmd, else if (p->literal) texierr(p, "section in a literal scope!?"); - teximacroopen(p, ".Sh "); + teximacroopen(p, "Sh"); parseeoln(p, buf, sz, pos); teximacroclose(p); } @@ -1408,12 +1445,12 @@ dotop(struct texi *p, enum texicmd cmd, p->ign--; advanceeoln(p, buf, sz, pos, 1); - 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"); + 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 @@ -1428,13 +1465,13 @@ doitem(struct texi *p, enum texicmd cmd, switch (p->list) { case (TEXILIST_ITEM): - teximacroopen(p, ".It"); + teximacroopen(p, "It"); break; case (TEXILIST_NOITEM): - teximacro(p, ".It"); + teximacro(p, "It"); break; default: - teximacro(p, ".Pp"); + teximacro(p, "Pp"); break; } @@ -1453,9 +1490,9 @@ dotable(struct texi *p, enum texicmd cmd, enum texilist sv = p->list; p->list = TEXILIST_ITEM; - teximacro(p, ".Bl -tag -width Ds"); + teximacro(p, "Bl -tag -width Ds"); parseto(p, buf, sz, pos, "table"); - teximacro(p, ".El"); + teximacro(p, "El"); p->list = sv; } @@ -1466,9 +1503,9 @@ doenumerate(struct texi *p, enum texicmd cmd, enum texilist sv = p->list; p->list = TEXILIST_NOITEM; - teximacro(p, ".Bl -enum"); + teximacro(p, "Bl -enum"); parseto(p, buf, sz, pos, "enumerate"); - teximacro(p, ".El"); + teximacro(p, "El"); p->list = sv; } @@ -1479,9 +1516,9 @@ doitemize(struct texi *p, enum texicmd cmd, enum texilist sv = p->list; p->list = TEXILIST_ITEM; - teximacro(p, ".Bl -bullet"); + teximacro(p, "Bl -bullet"); parseto(p, buf, sz, pos, "itemize"); - teximacro(p, ".El"); + teximacro(p, "El"); p->list = sv; } @@ -1503,13 +1540,59 @@ doignline(struct texi *p, enum texicmd cmd, 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) +{ + 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) @@ -1517,8 +1600,12 @@ 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; } @@ -1539,11 +1626,12 @@ main(int argc, char *argv[]) memset(&texi, 0, sizeof(struct texi)); texi.ign = 1; - texi.dir = strdup(dir); + texi.dirs = parsedirs(dir, Idir, &texi.dirsz); parsefile(&texi, argv[0]); + /* We shouldn't get here. */ texiexit(&texi); return(EXIT_FAILURE); usage: - fprintf(stderr, "usage: %s file\n", progname); + fprintf(stderr, "usage: %s [-Idirs] file\n", progname); return(EXIT_FAILURE); }