=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.30 retrieving revision 1.41 diff -u -p -r1.30 -r1.41 --- texi2mdoc/main.c 2015/02/23 11:44:30 1.30 +++ texi2mdoc/main.c 2015/02/24 14:35:40 1.41 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.30 2015/02/23 11:44:30 kristaps Exp $ */ +/* $Id: main.c,v 1.41 2015/02/24 14:35:40 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -14,12 +14,11 @@ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ -#include -#include - +#if defined(__linux__) || defined(__MINT__) +# define _GNU_SOURCE /* memmem */ +#endif #include #include -#include #include #include #include @@ -28,7 +27,6 @@ #include #include #include -#include #include "extern.h" @@ -74,6 +72,7 @@ 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 +93,8 @@ 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 +104,7 @@ 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 +123,8 @@ 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 */ @@ -137,10 +141,13 @@ static const struct texitok __texitoks[TEXICMD__MAX] = { doignblock, "direntry", 8 }, /* TEXICMD_DIRENTRY */ { dodisplay, "display", 7 }, /* TEXICMD_DISPLAY */ { doignbracket, "dmn", 3 }, /* TEXICMD_DMN */ + { doignblock, "documentdescription", 19 }, /* TEXICMD_DOCUMENTDESCRIPTION */ + { doignline, "documentencoding", 16 }, /* TEXICMD_DOCUMENTENCODING */ { dosymbol, "dots", 4 }, /* TEXICMD_DOTS */ { 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 +158,7 @@ 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 +187,7 @@ 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 +226,12 @@ 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 +241,7 @@ 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 +257,7 @@ 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 +271,11 @@ 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 +327,11 @@ 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 +354,10 @@ 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 +406,8 @@ 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); @@ -455,7 +477,7 @@ domacro(struct texi *p, enum texicmd cmd, memcpy(m.key, &buf[start], end - start); m.key[end - start] = '\0'; - m.args = argparse(p, buf, sz, pos, &m.argsz); + m.args = argparse(p, buf, sz, pos, &m.argsz, 0); advanceeoln(p, buf, sz, pos, 0); start = *pos; @@ -510,7 +532,7 @@ 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 +609,45 @@ 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; @@ -603,21 +664,12 @@ 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); while (*pos < endpos) { - if (buf[*pos] == '\n') - p->outcol = 0; - else - p->outcol++; - if (*pos > 0 && '.' == buf[*pos]) - if ('\n' == buf[*pos - 1]) - fputs("\\&", stdout); - putchar(buf[*pos]); - if ('\\' == buf[*pos]) - putchar('e'); + texiputchar(p, buf[*pos]); advance(p, buf, pos); } teximacro(p, "Ed"); @@ -828,8 +880,28 @@ 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 +928,7 @@ doaccent(struct texi *p, enum texicmd cmd, break; default: texiputchar(p, buf[*pos]); + break; } break; case (TEXICMD_GRAVE): @@ -882,6 +955,7 @@ doaccent(struct texi *p, enum texicmd cmd, break; default: texiputchar(p, buf[*pos]); + break; } break; case (TEXICMD_UMLAUT): @@ -897,11 +971,13 @@ 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 +1006,14 @@ 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 +1040,12 @@ 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 +1055,9 @@ 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 +1083,7 @@ 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 +1099,18 @@ 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); } @@ -1077,7 +1166,7 @@ dovalue(struct texi *p, enum texicmd cmd, if (64 == p->valstackpos) texierr(p, "too many nested values"); p->valstack[p->valstackpos++] = cp; - parseeof(p, cp, strlen(cp)); + parsemembuf(p, cp, strlen(cp)); p->valstackpos--; } else texiputchars(p, "{No value}"); @@ -1111,12 +1200,16 @@ 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 +1287,13 @@ 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 +1346,10 @@ 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); } @@ -1288,11 +1386,15 @@ dotop(struct texi *p, enum texicmd cmd, teximacro(p, "Os"); teximacro(p, "Sh NAME"); teximacroopen(p, "Nm"); - texiputchars(p, p->title); + for (cp = p->title; '\0' != *cp; cp++) + texiputchar(p, *cp); teximacroclose(p); teximacroopen(p, "Nd"); - texiputchars(p, NULL != p->subtitle ? - p->subtitle : "Unknown description"); + if (NULL != p->subtitle) + for (cp = p->subtitle; '\0' != *cp; cp++) + texiputchar(p, *cp); + else + texiputchars(p, "Unknown description"); teximacroclose(p); p->seenvs = 1; dosection(p, cmd, buf, sz, pos); @@ -1351,10 +1453,16 @@ 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 +1515,7 @@ 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; } @@ -1480,7 +1589,7 @@ doignline(struct texi *p, enum texicmd cmd, /* * Parse colon-separated directories from "cp" (if not NULL) and returns * the array of pointers. - * Prepends "base" to the array. + * Prepends "base" to the array, if found. * This does NOT sanitise the directories! */ static char ** @@ -1488,26 +1597,27 @@ parsedirs(struct texi *p, const char *base, const char { char *tok, *str, *tofree; const char *cpp; - size_t i; + size_t i = 0; char **dirs; - *sz = NULL != (cpp = cp) ? 2 : 1; - if (*sz > 1) - for ( ; NULL != (cpp = strchr(cpp, ':')); (*sz)++) + /* Count up our expected arguments. */ + *sz = NULL != base; + if (NULL != (cpp = cp)) + for ((*sz)++; NULL != (cpp = strchr(cpp, ':')); (*sz)++) cpp++; + if (0 == *sz) + return(NULL); if (NULL == (dirs = calloc(*sz, sizeof(char *)))) texiabort(p, NULL); - else if (NULL == (dirs[0] = strdup(base))) + if (NULL != base && NULL == (dirs[i++] = strdup(base))) texiabort(p, NULL); - if (NULL == cp) return(dirs); - if (NULL == (tofree = tok = str = strdup(cp))) texiabort(p, NULL); - for (i = 1; NULL != (tok = strsep(&str, ":")); i++) + for ( ; NULL != (tok = strsep(&str, ":")); i++) if (NULL == (dirs[i] = strdup(tok))) texiabort(p, NULL); @@ -1520,7 +1630,7 @@ 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], '/'); @@ -1530,6 +1640,7 @@ main(int argc, char *argv[]) ++progname; memset(&texi, 0, sizeof(struct texi)); + texi.ign = 1; Idir = NULL; while (-1 != (c = getopt(argc, argv, "I:"))) @@ -1542,33 +1653,33 @@ main(int argc, char *argv[]) } argv += optind; - if (0 == (argc -= optind)) - goto usage; + argc -= optind; - if (NULL == (path = strdup(argv[0]))) - texiabort(&texi, NULL); - else if (NULL == (dir = dirname(path))) - texiabort(&texi, NULL); + if (argc > 0) { + if (NULL == (dirpath = strdup(argv[0]))) + texiabort(&texi, NULL); + if (NULL == (dir = dirname(dirpath))) + texiabort(&texi, NULL); + if (NULL != (cp = strrchr(argv[0], '/'))) + texi.title = strdup(cp + 1); + else + texi.title = strdup(argv[0]); + if (NULL == texi.title) + texiabort(&texi, NULL); + else if (NULL != (ccp = strchr(texi.title, '.'))) + *ccp = '\0'; + texi.dirs = parsedirs(&texi, dir, Idir, &texi.dirsz); + free(dirpath); + parsefile(&texi, argv[0], 1); + } else { + texi.title = strdup("Unknown Manual"); + texi.dirs = parsedirs(&texi, NULL, Idir, &texi.dirsz); + parsestdin(&texi); + } - free(path); - - if (NULL != (cp = strrchr(argv[0], '/'))) - texi.title = strdup(cp + 1); - else - texi.title = strdup(argv[0]); - - if (NULL == texi.title) - texiabort(&texi, NULL); - else if (NULL != (path = strchr(texi.title, '.'))) - *path = '\0'; - - texi.ign = 1; - texi.dirs = parsedirs(&texi, dir, Idir, &texi.dirsz); - parsefile(&texi, argv[0], 1); - /* We shouldn't get here. */ texiexit(&texi); return(EXIT_FAILURE); usage: - fprintf(stderr, "usage: %s [-Idirs] file\n", progname); + fprintf(stderr, "usage: %s [-Idirs] [file]\n", progname); return(EXIT_FAILURE); }