=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.43 retrieving revision 1.45 diff -u -p -r1.43 -r1.45 --- texi2mdoc/main.c 2015/02/24 20:34:43 1.43 +++ texi2mdoc/main.c 2015/02/25 10:01:54 1.45 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.43 2015/02/24 20:34:43 kristaps Exp $ */ +/* $Id: main.c,v 1.45 2015/02/25 10:01:54 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -45,6 +45,7 @@ static void dobye(struct texi *, enum texicmd, const c static void dodefindex(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dodefn(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 doend(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 doexample(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doignargn(struct texi *, enum texicmd, const char *, size_t, size_t *); @@ -149,12 +150,13 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { doignbracket, "dmn", 3 }, /* TEXICMD_DMN */ { doignblock, "documentdescription", 19 }, /* TEXICMD_DOCUMENTDESCRIPTION */ { doignline, "documentencoding", 16 }, /* TEXICMD_DOCUMENTENCODING */ + { doignline, "documentlanguage", 16 }, /* TEXICMD_DOCUMENTLANGUAGE */ { doaccent, "dotaccent", 9 }, /* TEXICMD_DOTACCENT */ { doaccent, "dotless", 7 }, /* TEXICMD_DOTLESS */ { dosymbol, "dots", 4 }, /* TEXICMD_DOTS */ { dolink, "email", 5 }, /* TEXICMD_EMAIL */ { doinline, "emph", 4 }, /* TEXICMD_EMPH */ - { NULL, "end", 3 }, /* TEXICMD_END */ + { doend, "end", 3 }, /* TEXICMD_END */ { dosymbol, "enddots", 7 }, /* TEXICMD_ENDDOTS */ { doenumerate, "enumerate", 9 }, /* TEXICMD_ENUMERATE */ { doinline, "env", 3 }, /* TEXICMD_ENV */ @@ -168,7 +170,11 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { doinline, "file", 4 }, /* TEXICMD_FILE */ { doignline, "finalout", 8 }, /* TEXICMD_FINALOUT */ { doignline, "findex", 6 }, /* TEXICMD_FINDEX */ + { doblock, "flushleft", 9 }, /* TEXICMD_FLUSHLEFT */ + { doblock, "flushright", 10 }, /* TEXICMD_FLUSHRIGHT */ + { doignline, "firstparagraphindent", 20 }, /* TEXICMD_FIRSTPARAGRAPHINDENT */ { doignbracket, "footnote", 8 }, /* TEXICMD_FOOTNOTE */ + { doignline, "footnotestyle", 13 }, /* TEXICMD_FOOTNOTESTYLE */ { dotable, "ftable", 6 }, /* TEXICMD_FTABLE */ { dodisplay, "format", 6 }, /* TEXICMD_FORMAT */ { dosymbol, "geq", 3 }, /* TEXICMD_GEQ */ @@ -205,7 +211,8 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { doignbracket, "image", 5 }, /* TEXICMD_IMAGE */ { doinclude, "include", 7 }, /* TEXICMD_INCLUDE */ { dodisplay, "indentblock", 11 }, /* TEXICMD_INDENTBLOCK */ - { dolink, "inforef", 7 }, /* TEXICMD_INDENTBLOCK */ + { dolink, "indicateurl", 11 }, /* TEXICMD_INDICATEURL */ + { dolink, "inforef", 7 }, /* TEXICMD_INFOREF */ { doignline, "insertcopying", 13 }, /* TEXICMD_INSERTCOPYING */ { doitem, "item", 4 }, /* TEXICMD_ITEM */ { doitemize, "itemize", 7 }, /* TEXICMD_ITEMIZE */ @@ -530,7 +537,7 @@ domacro(struct texi *p, enum texicmd cmd, start = *pos; endtok = "\n@end macro\n"; endtoksz = strlen(endtok); - blk = memmem(&buf[start], sz, endtok, endtoksz); + blk = memmem(&buf[start], sz - start, endtok, endtoksz); if (NULL == blk) texierr(p, "unterminated macro body"); while (&buf[*pos] != blk) @@ -559,34 +566,64 @@ static void doignblock(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - char end[32]; - const char *term; - size_t endsz, endpos; + char end[32], start[32]; + const char *endt, *startt; + size_t esz, ssz, newpos, stack; /* - * We want to completely ignore everything in these blocks, so - * simply jump to the @end block. + * FIXME: this is cheating. + * These tokens are supposed to begin on a newline. + * However, if we do that, then we would need to check within + * the loop for trailer (or leading, as the case may be) + * newline, and that's just a bit too complicated right now. + * This is becasue + * @ifset BAR + * @ifset FOO + * @end ifset + * @end ifset + * won't work right now: we'd read after the first "@end ifset" + * to the next line, then look for the next line after that. */ - endsz = snprintf(end, sizeof(end), - "\n@end %s\n", texitoks[cmd].tok); - assert(endsz < sizeof(end)); + ssz = snprintf(start, sizeof(start), + "@%s", texitoks[cmd].tok); + assert(ssz < sizeof(start)); + esz = snprintf(end, sizeof(end), + "@end %s\n", texitoks[cmd].tok); + assert(esz < sizeof(end)); + stack = 1; - /* - * Look up where our end token occurs. - * Set our end position based on the relative offset of that - * from our current position, or the EOF if we don't have a - * proper ending point. + /* + * Here we look for the end token "end" somewhere in the file in + * front of us. + * It's not that easy, of course: if we have a nested block, + * then there'll be an "end" token of the same kind between us. + * Thus, we keep track of scopes for matching "end" blocks. */ - term = memmem(&buf[*pos], sz, end, endsz); - endpos = NULL == term ? sz : - *pos + (size_t)(term - &buf[*pos]); - assert(endpos <= sz); - while (*pos < endpos) - advance(p, buf, pos); + while (stack > 0 && *pos < sz) { + if (stack > 10) + abort(); + endt = memmem(&buf[*pos], sz - *pos, end, esz); + startt = memmem(&buf[*pos], sz - *pos, start, ssz); + if (NULL == endt) { + texiwarn(p, "unterminated \"%s\" " + "block", texitoks[cmd].tok); + *pos = sz; + break; + } - /* Only do this if we're not already at the end. */ - if (endpos < sz) - advanceto(p, buf, pos, endpos + endsz); + newpos = *pos; + if (NULL == startt || startt > endt) { + newpos += esz + (size_t)(endt - &buf[*pos]); + stack--; + } else { + newpos += ssz + (size_t)(startt - &buf[*pos]); + stack++; + } + + assert(newpos <= sz); + while (*pos < newpos) + advance(p, buf, pos); + } } static void @@ -698,29 +735,33 @@ doverbatim(struct texi *p, enum texicmd cmd, const char *end, *term; size_t endsz, endpos; - advanceeoln(p, buf, sz, pos, 1); - - /* We end at exactly this token. */ + /* We read from the @verbatim\n newline inclusive! */ end = "\n@end verbatim\n"; endsz = strlen(end); + advanceeoln(p, buf, sz, pos, 0); + if (*pos == sz) { + texiwarn(p, "unexpected end of file"); + return; + } - /* - * Look up where our end token occurs. - * Set our end position based on the relative offset of that - * from our current position. - */ - term = memmem(&buf[*pos], sz, end, endsz); - endpos = NULL == term ? sz : - *pos + (size_t)(term - &buf[*pos]); + term = memmem(&buf[*pos], sz - *pos, end, endsz); + if (NULL == term) { + texiwarn(p, "unterminated verbatim block"); + endpos = sz; + } else + endpos = *pos + (size_t)(term - &buf[*pos]); - teximacro(p, "Bd -literal -offset indent"); assert(endpos <= sz); + assert('\n' == buf[*pos]); + advance(p, buf, pos); + teximacro(p, "Bd -literal -offset indent"); while (*pos < endpos) { texiputchar(p, buf[*pos]); advance(p, buf, pos); } teximacro(p, "Ed"); - advanceto(p, buf, pos, endpos + endsz); + if (*pos < sz) + advanceto(p, buf, pos, endpos + endsz); } static void @@ -1401,6 +1442,7 @@ dolink(struct texi *p, enum texicmd cmd, break; case (TEXICMD_UREF): case (TEXICMD_URL): + case (TEXICMD_INDICATEURL): teximacroopen(p, "Lk"); break; case (TEXICMD_XREF): @@ -1738,6 +1780,22 @@ dotable(struct texi *p, enum texicmd cmd, parseto(p, buf, sz, pos, texitoks[cmd].tok); teximacro(p, "El"); p->list = sv; +} + +static void +doend(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + size_t start; + + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + start = *pos; + while (*pos < sz && '\n' != buf[*pos]) + advance(p, buf, pos); + + texiwarn(p, "unexpected \"end\": %.*s", (int)(*pos - start), &buf[start]); + advanceeoln(p, buf, sz, pos, 1); } static void