=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.31 retrieving revision 1.39 diff -u -p -r1.31 -r1.39 --- texi2mdoc/main.c 2015/02/23 11:56:39 1.31 +++ texi2mdoc/main.c 2015/02/23 20:54:53 1.39 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.31 2015/02/23 11:56:39 kristaps Exp $ */ +/* $Id: main.c,v 1.39 2015/02/23 20:54:53 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -14,6 +14,9 @@ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ +#if defined(__linux__) || defined(__MINT__) +# define _GNU_SOURCE /* memmem */ +#endif #include #include @@ -74,6 +77,7 @@ static void dotab(struct texi *, enum texicmd, const c static void dotitle(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dovalue(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doverb(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void doverbatim(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doverbinclude(struct texi *, enum texicmd, const char *, size_t, size_t *); static const struct texitok __texitoks[TEXICMD__MAX] = { @@ -94,6 +98,8 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { dosymbol, "!", 1 }, /* TEXICMD_BANG */ { dosymbol, "bullet", 6 }, /* TEXICMD_BULLET */ { dobye, "bye", 3 }, /* TEXICMD_BYE */ + { doblock, "cartouche", 9 }, /* TEXICMD_CARTOUCHE */ + { doaccent, ",", 1 }, /* TEXICMD_CEDILLA */ { doignline, "center", 6 }, /* TEXICMD_CENTER */ { dosection, "chapter", 7 }, /* TEXICMD_CHAPTER */ { doignline, "cindex", 6 }, /* TEXICMD_CINDEX */ @@ -103,6 +109,7 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { doinline, "code", 4 }, /* TEXICMD_CODE */ { dosymbol, ":", 1 }, /* TEXICMD_COLON */ { NULL, "columnfractions", 15 }, /* TEXICMD_COLUMNFRACTIONS */ + { dosymbol, "comma", 5 }, /* TEXICMD_COMMA */ { doinline, "command", 7 }, /* TEXICMD_COMMAND */ { doignline, "c", 1 }, /* TEXICMD_COMMENT */ { doignline, "comment", 7 }, /* TEXICMD_COMMENT_LONG */ @@ -121,6 +128,8 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { dodefn, "deftypefnx", 10 }, /* TEXICMD_DEFTYPEFNX */ { dodefn, "deftypefun", 10 }, /* TEXICMD_DEFTYPEFUN */ { dodefn, "deftypefunx", 11 }, /* TEXICMD_DEFTYPEFUNX */ + { dodefn, "deftypemethod", 13 }, /* TEXICMD_DEFTYPEMETHOD */ + { dodefn, "deftypemethodx", 14 }, /* TEXICMD_DEFTYPEMETHODX */ { dodefn, "deftypevar", 10 }, /* TEXICMD_DEFTYPEVAR */ { dodefn, "deftypevarx", 11 }, /* TEXICMD_DEFTYPEVARX */ { dodefn, "deftypevr", 9 }, /* TEXICMD_DEFTYPEVR */ @@ -141,6 +150,7 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { dolink, "email", 5 }, /* TEXICMD_EMAIL */ { doinline, "emph", 4 }, /* TEXICMD_EMPH */ { NULL, "end", 3 }, /* TEXICMD_END */ + { dosymbol, "enddots", 7 }, /* TEXICMD_ENDDOTS */ { doenumerate, "enumerate", 9 }, /* TEXICMD_ENUMERATE */ { doinline, "env", 3 }, /* TEXICMD_ENV */ { dosymbol, "equiv", 5 }, /* TEXICMD_EQUIV */ @@ -151,6 +161,7 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { doinline, "file", 4 }, /* TEXICMD_FILE */ { doignline, "finalout", 8 }, /* TEXICMD_FINALOUT */ { doignline, "findex", 6 }, /* TEXICMD_FINDEX */ + { doignbracket, "footnote", 8 }, /* TEXICMD_FOOTNOTE */ { dotable, "ftable", 6 }, /* TEXICMD_FTABLE */ { dodisplay, "format", 6 }, /* TEXICMD_FORMAT */ { doaccent, "`", 1 }, /* TEXICMD_GRAVE */ @@ -179,7 +190,7 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { doignbracket, "image", 5 }, /* TEXICMD_IMAGE */ { doinclude, "include", 7 }, /* TEXICMD_INCLUDE */ { dodisplay, "indentblock", 11 }, /* TEXICMD_INDENTBLOCK */ - { doignline, "", 0 }, /* TEXICMD_USER_INDEX */ + { dolink, "inforef", 7 }, /* TEXICMD_INDENTBLOCK */ { doignline, "insertcopying", 13 }, /* TEXICMD_INSERTCOPYING */ { doitem, "item", 4 }, /* TEXICMD_ITEM */ { doitemize, "itemize", 7 }, /* TEXICMD_ITEMIZE */ @@ -218,10 +229,12 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { dosection, "section", 7 }, /* TEXICMD_SECTION */ { dovalue, "set", 3 }, /* TEXICMD_SET */ { doignline, "setchapternewpage", 17 }, /* TEXICMD_SETCHAPNEWPAGE */ + { doignline, "setcontentsaftertitlepage", 25 }, /* TEXICMD_SETCONTENTSAFTER */ { doignline, "setfilename", 11 }, /* TEXICMD_SETFILENAME */ { dotitle, "settitle", 8 }, /* TEXICMD_SETTITLE */ { doignline, "shortcontents", 13 }, /* TEXICMD_SHORTCONTENTS */ { doinline, "slanted", 7 }, /* TEXICMD_SLANTED */ + { dosymbol, "/", 1 }, /* TEXICMD_SLASH */ { dosp, "sp", 2 }, /* TEXICMD_SP */ { dosymbol, " ", 1 }, /* TEXICMD_SPACE */ { doignline, "smallbook", 9 }, /* TEXICMD_SMALLBOOK */ @@ -231,6 +244,7 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { dodisplay, "smallindentblock", 16 }, /* TEXICMD_SMALLINDENTBLOCK */ { dosymbol, "{", 1 }, /* TEXICMD_SQUIGGLE_LEFT */ { dosymbol, "}", 1 }, /* TEXICMD_SQUIGGLE_RIGHT */ + { dosymbol, "ss", 2 }, /* TEXICMD_SS */ { doinline, "strong", 6 }, /* TEXICMD_STRONG */ { dosubsection, "subheading", 10 }, /* TEXICMD_SUBHEADING */ { dosubsection, "subsection", 10 }, /* TEXICMD_SUBSECTION */ @@ -246,6 +260,7 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { dotable, "table", 5 }, /* TEXICMD_TABLE */ { doignblock, "tex", 3 }, /* TEXICMD_TEX */ { dosymbol, "TeX", 3 }, /* TEXICMD_TEXSYM */ + { dosymbol, "tie", 3 }, /* TEXICMD_TIE */ { doaccent, "~", 1 }, /* TEXICMD_TILDE */ { doignline, "tindex", 6 }, /* TEXICMD_TINDEX */ { doignline, "title", 5 }, /* TEXICMD_TITLE */ @@ -259,9 +274,11 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { dosubsubsection, "unnumberedsubsubsec", 19 }, /* TEXICMD_UNNUMBEREDSUBSUBSEC */ { dolink, "uref", 4 }, /* TEXICMD_UREF */ { dolink, "url", 3 }, /* TEXICMD_URL */ + { doignline, "", 0 }, /* TEXICMD_USER_INDEX */ { dovalue, "value", 5 }, /* TEXICMD_VALUE */ { doinline, "var", 3 }, /* TEXICMD_VAR */ - { doverb, "verbatim", 8 }, /* TEXICMD_VERBATIM */ + { doverb, "verb", 4 }, /* TEXICMD_VERB */ + { doverbatim, "verbatim", 8 }, /* TEXICMD_VERBATIM */ { doverbinclude, "verbatiminclude", 15 }, /* TEXICMD_VERBATIMINCLUDE */ { doignline, "vindex", 6 }, /* TEXICMD_VINDEX */ { dosp, "vskip", 5 }, /* TEXICMD_VSKIP */ @@ -313,9 +330,11 @@ dodefn(struct texi *p, enum texicmd cmd, blk = NULL; switch (cmd) { case (TEXICMD_DEFFN): + case (TEXICMD_DEFMAC): case (TEXICMD_DEFTP): case (TEXICMD_DEFTYPEFN): case (TEXICMD_DEFTYPEFUN): + case (TEXICMD_DEFTYPEMETHOD): case (TEXICMD_DEFTYPEVAR): case (TEXICMD_DEFTYPEVR): case (TEXICMD_DEFUN): @@ -338,6 +357,10 @@ dodefn(struct texi *p, enum texicmd cmd, texivspace(p); switch (cmd) { + case (TEXICMD_DEFTYPEMETHOD): + case (TEXICMD_DEFTYPEMETHODX): + texiputchars(p, "Method"); + break; case (TEXICMD_DEFMAC): case (TEXICMD_DEFMACX): texiputchars(p, "Macro"); @@ -386,6 +409,8 @@ dodefn(struct texi *p, enum texicmd cmd, case (TEXICMD_DEFTYPEFUNX): case (TEXICMD_DEFTYPEFN): case (TEXICMD_DEFTYPEFNX): + case (TEXICMD_DEFTYPEMETHOD): + case (TEXICMD_DEFTYPEMETHODX): teximacroopen(p, "Ft"); parselinearg(p, buf, sz, pos); teximacroclose(p); @@ -510,7 +535,7 @@ doignblock(struct texi *p, enum texicmd cmd, */ term = memmem(&buf[*pos], sz, end, endsz); endpos = NULL == term ? sz : - *pos + term - &buf[*pos]; + *pos + (size_t)(term - &buf[*pos]); assert(endpos <= sz); while (*pos < endpos) advance(p, buf, pos); @@ -587,6 +612,45 @@ static void doverb(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + char delim; + size_t start; + + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + if (*pos == sz || '{' != buf[*pos]) + return; + advance(p, buf, pos); + if (*pos == sz) + return; + + delim = buf[*pos]; + advance(p, buf, pos); + /* Make sure we flush out our initial whitespace... */ + if (p->seenws && p->outcol && 0 == p->literal) + texiputchar(p, ' '); + p->seenws = 0; + start = *pos; + /* Read until we see the delimiter then end-brace. */ + while (*pos < sz - 1) { + if (buf[*pos] == delim && buf[*pos + 1] == '}') + break; + advance(p, buf, pos); + } + if (*pos == sz - 1) + return; + texiputbuf(p, buf, start, *pos); + + /* Make sure we read after the end-brace. */ + assert(delim == buf[*pos]); + advance(p, buf, pos); + assert('}' == buf[*pos]); + advance(p, buf, pos); +} + +static void +doverbatim(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ const char *end, *term; size_t endsz, endpos; @@ -603,7 +667,7 @@ doverb(struct texi *p, enum texicmd cmd, */ term = memmem(&buf[*pos], sz, end, endsz); endpos = NULL == term ? sz : - *pos + term - &buf[*pos]; + *pos + (size_t)(term - &buf[*pos]); teximacro(p, "Bd -literal -offset indent"); assert(endpos <= sz); @@ -828,8 +892,28 @@ doaccent(struct texi *p, enum texicmd cmd, if (*pos == sz) return; - advance(p, buf, pos); + if (p->seenws && p->outcol && 0 == p->literal) + texiputchar(p, ' '); + p->seenws = 0; switch (cmd) { + case (TEXICMD_CEDILLA): + /* Strange rules... */ + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + if (*pos == sz || '{' != buf[*pos]) + return; + advance(p, buf, pos); + switch (buf[*pos]) { + case ('c'): case ('C'): + texiputchars(p, "\\(,"); + texiputchar(p, buf[*pos]); + break; + default: + texiputchar(p, buf[*pos]); + break; + } + advance(p, buf, pos); + break; case (TEXICMD_ACUTE): switch (buf[*pos]) { case ('a'): case ('A'): @@ -856,6 +940,7 @@ doaccent(struct texi *p, enum texicmd cmd, break; default: texiputchar(p, buf[*pos]); + break; } break; case (TEXICMD_GRAVE): @@ -882,6 +967,7 @@ doaccent(struct texi *p, enum texicmd cmd, break; default: texiputchar(p, buf[*pos]); + break; } break; case (TEXICMD_UMLAUT): @@ -897,11 +983,13 @@ doaccent(struct texi *p, enum texicmd cmd, break; default: texiputchar(p, buf[*pos]); + break; } break; default: abort(); } + advance(p, buf, pos); } static void @@ -930,10 +1018,14 @@ dosymbol(struct texi *p, enum texicmd cmd, case (TEXICMD_BULLET): texiputchars(p, "\\(bu"); break; + case (TEXICMD_COMMA): + texiputchar(p, ','); + break; case (TEXICMD_COPYRIGHT): texiputchars(p, "\\(co"); break; case (TEXICMD_DOTS): + case (TEXICMD_ENDDOTS): texiputchars(p, "..."); break; case (TEXICMD_EQUIV): @@ -960,6 +1052,12 @@ dosymbol(struct texi *p, enum texicmd cmd, case (TEXICMD_RESULT): texiputchars(p, "\\(rA"); break; + case (TEXICMD_SLASH): + texiputchar(p, '/'); + break; + case (TEXICMD_SS): + texiputchars(p, "\\(ss"); + break; case (TEXICMD_SQUIGGLE_LEFT): texiputchars(p, "{"); break; @@ -969,6 +1067,9 @@ dosymbol(struct texi *p, enum texicmd cmd, case (TEXICMD_TEXSYM): texiputchars(p, "TeX"); break; + case (TEXICMD_TIE): + texiputchars(p, "\\ "); + break; case (TEXICMD_COLON): case (TEXICMD_HYPHEN): break; @@ -994,7 +1095,7 @@ static void domath(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - size_t nest; + size_t nest, start; /* * Math handling is different from everything else. @@ -1010,18 +1111,18 @@ domath(struct texi *p, enum texicmd cmd, if (p->seenws && p->outcol && 0 == p->literal) texiputchar(p, ' '); p->seenws = 0; - for (nest = 1; *pos < sz && nest > 0; ) { + for (nest = 1, start = *pos; *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]); + texiputbuf(p, buf, start, *pos); advance(p, buf, pos); } @@ -1111,12 +1212,16 @@ dolink(struct texi *p, enum texicmd cmd, break; case (TEXICMD_XREF): texiputchars(p, "See Section"); - teximacroopen(p, "Qq"); + teximacroopen(p, "Dq"); break; case (TEXICMD_PXREF): texiputchars(p, "see Section"); - teximacroopen(p, "Qq"); + teximacroopen(p, "Dq"); break; + case (TEXICMD_INFOREF): + texiputchars(p, "See Info file node"); + teximacroopen(p, "Dq"); + break; default: abort(); } @@ -1194,11 +1299,13 @@ dosubsection(struct texi *p, enum texicmd cmd, texierr(p, "\"%s\" in a literal scope!?", sects[sec]); /* We don't have a subsubsection, so make one up. */ - texivspace(p); + if (sec > 1) + texivspace(p); teximacroopen(p, sects[sec]); parseeoln(p, buf, sz, pos); teximacroclose(p); - texivspace(p); + if (sec > 1) + texivspace(p); } static void @@ -1251,7 +1358,10 @@ dosp(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - texivspace(p); + if (p->literal) + texiputchar(p, '\n'); + else + texivspace(p); /* FIXME: ignore and parseeoln. */ advanceeoln(p, buf, sz, pos, 1); } @@ -1351,10 +1461,16 @@ domultitable(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { enum texilist sv = p->list; + int svliteral = p->literal; enum texicmd type; size_t i, end, columns; p->list = TEXILIST_TABLE; + /* + * TS/TE blocks aren't "in mdoc(7)", so we can disregard the + * fact that we're in literal mode right now. + */ + p->literal = 0; teximacro(p, "TS"); columns = 0; @@ -1407,6 +1523,7 @@ domultitable(struct texi *p, enum texicmd cmd, parseto(p, buf, sz, pos, texitoks[cmd].tok); p->outmacro--; teximacro(p, "TE"); + p->literal = svliteral; p->list = sv; } @@ -1520,7 +1637,7 @@ main(int argc, char *argv[]) { struct texi texi; int c; - char *path, *dir; + char *dirpath, *dir, *ccp; const char *progname, *Idir, *cp; progname = strrchr(argv[0], '/'); @@ -1545,13 +1662,11 @@ main(int argc, char *argv[]) if (0 == (argc -= optind)) goto usage; - if (NULL == (path = strdup(argv[0]))) + if (NULL == (dirpath = strdup(argv[0]))) texiabort(&texi, NULL); - else if (NULL == (dir = dirname(path))) + if (NULL == (dir = dirname(dirpath))) texiabort(&texi, NULL); - free(path); - if (NULL != (cp = strrchr(argv[0], '/'))) texi.title = strdup(cp + 1); else @@ -1559,11 +1674,12 @@ main(int argc, char *argv[]) if (NULL == texi.title) texiabort(&texi, NULL); - else if (NULL != (path = strchr(texi.title, '.'))) - *path = '\0'; + else if (NULL != (ccp = strchr(texi.title, '.'))) + *ccp = '\0'; texi.ign = 1; texi.dirs = parsedirs(&texi, dir, Idir, &texi.dirsz); + free(dirpath); parsefile(&texi, argv[0], 1); /* We shouldn't get here. */ texiexit(&texi);