=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.7 retrieving revision 1.8 diff -u -p -r1.7 -r1.8 --- texi2mdoc/main.c 2015/02/18 11:08:58 1.7 +++ texi2mdoc/main.c 2015/02/18 12:03:21 1.8 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.7 2015/02/18 11:08:58 kristaps Exp $ */ +/* $Id: main.c,v 1.8 2015/02/18 12:03:21 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -100,6 +100,7 @@ enum texicmd { TEXICMD_NEWLINE, TEXICMD_NODE, TEXICMD_NOINDENT, + TEXICMD_PXREF, TEXICMD_QUESTIONMARK, TEXICMD_QUOTATION, TEXICMD_PAGE, @@ -118,6 +119,7 @@ enum texicmd { TEXICMD_SMALLEXAMPLE, TEXICMD_SQUIGGLE_LEFT, TEXICMD_SQUIGGLE_RIGHT, + TEXICMD_STRONG, TEXICMD_SUBSECTION, TEXICMD_SUBTITLE, TEXICMD_TAB, @@ -134,6 +136,7 @@ enum texicmd { TEXICMD_URL, TEXICMD_VAR, TEXICMD_W, + TEXICMD_XREF, TEXICMD__MAX }; @@ -182,22 +185,22 @@ 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; + 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 *); @@ -211,6 +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 *); @@ -218,6 +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 *); @@ -229,7 +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 */ @@ -264,7 +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 */ @@ -294,6 +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 */ @@ -312,6 +318,7 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { doexample, "smallexample", 12 }, /* TEXICMD_SMALLEXAMPLE */ { 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 */ @@ -324,10 +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 */ }; /* @@ -344,7 +352,8 @@ 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 @@ -395,6 +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, ...) { @@ -413,9 +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) @@ -443,20 +455,22 @@ 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) @@ -475,8 +489,7 @@ 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) @@ -554,6 +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) @@ -580,16 +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, ' '); @@ -602,6 +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]) { @@ -618,6 +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) @@ -659,6 +677,62 @@ texicmd(struct texi *p, const char *buf, return(TEXICMD__MAX); } +/* + * Parse an argument from a bracketed command, e.g., @url{foo, baz}. + * Num should be set to the argument we're currently parsing, although + * it suffixes for it to be zero or non-zero. + * This will return 1 if there are more arguments, 0 otherwise. + * This will stop (returning 0) in the event of EOF or if we're not at a + * bracket for the zeroth parse. + */ +static int +parsearg(struct texi *p, const char *buf, + size_t sz, size_t *pos, size_t num) +{ + size_t end; + enum texicmd cmd; + + while (*pos < sz && 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 ('}'): + advance(p, buf, pos); + return(0); + case ('{'): + if (0 == p->ign) + texiwarn(p, "unexpected \"{\""); + advance(p, buf, pos); + continue; + case ('@'): + break; + default: + texiword(p, buf, sz, pos, ','); + continue; + } + + cmd = texicmd(p, buf, *pos, sz, &end); + advanceto(p, buf, pos, end); + if (TEXICMD__MAX == cmd) + continue; + if (NULL != texitoks[cmd].fp) + (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); + } + return(0); +} + +/* + * Parse until the end of a bracketed statement, e.g., @foo{bar baz}. + * This will stop in the event of EOF or if we're not at a bracket. + */ static void parsebracket(struct texi *p, const char *buf, size_t sz, size_t *pos) { @@ -685,7 +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; } @@ -730,7 +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; } @@ -743,6 +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) { @@ -766,7 +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; } @@ -778,6 +856,9 @@ 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) { @@ -787,6 +868,11 @@ parseeof(struct texi *p, const char *buf, size_t 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) @@ -813,7 +899,7 @@ parseto(struct texi *p, const char *buf, case ('@'): break; default: - texiword(p, buf, sz, pos); + texiword(p, buf, sz, pos, '\0'); continue; } @@ -823,8 +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)) { @@ -841,6 +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) { @@ -1206,7 +1296,6 @@ 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) @@ -1242,15 +1331,12 @@ 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"); @@ -1259,18 +1345,24 @@ 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; + 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])) { @@ -1278,18 +1370,24 @@ doarg1(struct texi *p, enum texicmd cmd, texiputchar(p, buf[*pos]); advance(p, buf, pos); } - switch (cmd) { - case (TEXICMD_EMAIL): - case (TEXICMD_UREF): - case (TEXICMD_URL): - teximacroclose(p); - break; - default: - break; - } + + 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) { @@ -1442,6 +1540,12 @@ 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) { @@ -1528,6 +1632,6 @@ 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); }