=================================================================== RCS file: /cvs/pod2mdoc/pod2mdoc.c,v retrieving revision 1.39 retrieving revision 1.46 diff -u -p -r1.39 -r1.46 --- pod2mdoc/pod2mdoc.c 2015/02/13 15:35:15 1.39 +++ pod2mdoc/pod2mdoc.c 2015/02/14 15:34:39 1.46 @@ -1,4 +1,4 @@ -/* $Id: pod2mdoc.c,v 1.39 2015/02/13 15:35:15 schwarze Exp $ */ +/* $Id: pod2mdoc.c,v 1.46 2015/02/14 15:34:39 schwarze Exp $ */ /* * Copyright (c) 2014 Kristaps Dzonsons * Copyright (c) 2014, 2015 Ingo Schwarze @@ -135,7 +135,7 @@ static const char fmts[FMT__MAX] = { 'Z' /* FMT_NULL */ }; -static int last; +static unsigned char last; static void @@ -160,7 +160,6 @@ outbuf_addchar(struct state *st) if ('\\' == last) st->outbuf[st->outbuflen++] = 'e'; st->outbuf[st->outbuflen] = '\0'; - st->wantws = 0; } static void @@ -174,7 +173,6 @@ outbuf_addstr(struct state *st, const char *str) memcpy(st->outbuf + st->outbuflen, str, slen+1); st->outbuflen += slen; last = str[slen - 1]; - st->wantws = 0; } static void @@ -184,6 +182,9 @@ outbuf_flush(struct state *st) if (0 == st->outbuflen) return; + if (OUST_TXT == st->oust && st->wantws) + putchar(' '); + fputs(st->outbuf, stdout); *st->outbuf = '\0'; st->outbuflen = 0; @@ -477,8 +478,9 @@ static int formatcode(struct state *st, const char *buf, size_t *start, size_t end, int nomacro, int pos) { - enum fmt fmt; size_t i, j, dsz; + enum fmt fmt; + int wantws; unsigned char uc; assert(*start + 1 < end); @@ -563,7 +565,8 @@ formatcode(struct state *st, const char *buf, size_t * * allows to break at this point as well. */ - st->wantws |= ' ' == buf[*start]; + wantws = ' ' == buf[*start] || + (OUST_MAC == st->oust ? st->wantws : ! st->outbuflen); /* * If we are on a text line and there is no @@ -573,22 +576,23 @@ formatcode(struct state *st, const char *buf, size_t * * lest we clobber out output state. */ - if (OUST_MAC != st->oust && !st->wantws) { + if (OUST_MAC != st->oust && ! wantws) { if (OUST_NL != st->oust) putchar('\n'); printf(".Pf "); + st->wantws = 0; } outbuf_flush(st); /* Whitespace is easier to suppress on macro lines. */ - if (OUST_MAC == st->oust && !st->wantws) + if (OUST_MAC == st->oust && ! wantws) printf(" Ns "); /* Unless we are on a macro line, start one. */ - if (OUST_MAC != st->oust && st->wantws) { + if (OUST_MAC != st->oust && wantws) { if (OUST_NL != st->oust) putchar('\n'); putchar('.'); @@ -687,7 +691,8 @@ formatcode(struct state *st, const char *buf, size_t * } if (*start + 1 < end && '<' == buf[*start + 1] && 'A' <= buf[*start] && 'Z' >= buf[*start]) { - formatcode(st, buf, start, end, nomacro, 1); + if ( ! formatcode(st, buf, start, end, nomacro, 1)) + st->wantws = 1; continue; } @@ -729,14 +734,11 @@ formatcode(struct state *st, const char *buf, size_t * putchar('e'); } - if (FMT__MAX == fmt) - return(0); - if ( ! nomacro && FMT_CODE == fmt) printf(" Qc "); st->wantws = ' ' == last; - return(1); + return(FMT__MAX != fmt); } /* @@ -1091,11 +1093,21 @@ again: start++; if (start < end && '\n' == buf[start]) start++; - if (start < end) - goto again; - return; + goto again; } + /* Other preprocessor directives. */ + if ('#' == buf[i]) { + fputs(".Fd ", stdout); + start = i; + while(start < end && '\n' != buf[start]) + putchar(buf[start++]); + putchar('\n'); + if (start < end && '\n' == buf[start]) + start++; + goto again; + } + /* Parse function declaration. */ ifo = ifa = ifc = 0; inl = end; @@ -1108,6 +1120,8 @@ again: break; } switch (buf[i]) { + case '\t': + /* FALLTHROUGH */ case ' ': if ( ! ifa) ifo = i; @@ -1171,7 +1185,7 @@ again: printf(".Fa \"%s\"\n", buf + ifa); if (cp == NULL) break; - while (*cp == ' ') + while (*cp == ' ' || *cp == '\t') cp++; ifa = cp - buf; } @@ -1309,14 +1323,16 @@ donamenm(struct state *st, const char *buf, size_t *st * * Uses formatcode() to go to OUST_MAC mode * and outbuf_flush() to go to OUST_TXT mode. - * Main text mode wantws handling is in this function. + * In text mode, wantws requests white space before the text + * currently contained in the outbuf, not before upcoming text. * Must make sure to go back to OUST_NL/wantws mode before returning. */ static void ordinary(struct state *st, const char *buf, size_t start, size_t end) { - size_t i, j, opstack; - int seq; + size_t i, j, opstack, wend; + enum mdoc_type mtype; + int eos, noeos, seq; if ( ! st->parsing || st->paused) return; @@ -1382,32 +1398,88 @@ ordinary(struct state *st, const char *buf, size_t sta &start, end, &opstack)) continue; - /* - * On whitespace, flush the output buffer - * and allow breaking to a macro line. - * Otherwise, buffer text and clear wantws. - */ + /* Merely buffer non-whitespace. */ last = buf[start++]; - if (' ' != last) { + if ( ! isspace(last)) outbuf_addchar(st); + if (start < end && + ! isspace((unsigned char)buf[start])) continue; + + /* + * Found the end of a word. + * Rewind trailing delimiters. + */ + + eos = noeos = 0; + for (wend = st->outbuflen; wend; wend--) + if ('.' == st->outbuf[wend - 1] || + '!' == st->outbuf[wend - 1] || + '?' == st->outbuf[wend - 1]) + eos = 1; + else if ('|' == st->outbuf[wend - 1] || + ',' == st->outbuf[wend - 1] || + ';' == st->outbuf[wend - 1] || + ':' == st->outbuf[wend - 1]) + noeos = 1; + else if ('\'' != st->outbuf[wend - 1] && + '"' != st->outbuf[wend - 1] && + ')' != st->outbuf[wend - 1] && + ']' != st->outbuf[wend - 1]) + break; + eos &= ! noeos; + + /* + * Detect function names. + */ + + mtype = MDOC_Fa; + if (wend && ')' == st->outbuf[wend] && + '(' == st->outbuf[wend - 1]) { + mtype = dict_get(st->outbuf, --wend); + if (MDOC_Fo == mtype || MDOC_MAX == mtype) { + st->outbuflen = wend; + st->outbuf[wend] = '\0'; + mdoc_newln(st); + if (MDOC_Fo == mtype) + fputs(".Fn ", stdout); + else + fputs(".Xr ", stdout); + st->oust = OUST_MAC; + } } - if ( ! strcmp(st->outbuf + st->outbuflen - 2, "()") && - dict_get(st->outbuf, st->outbuflen - 2) == - MDOC_Fo) { - st->outbuflen -= 2; - st->outbuf[st->outbuflen] = '\0'; + /* + * On whitespace, flush the output buffer + * and allow breaking to a macro line. + */ + + outbuf_flush(st); + + /* + * End macro lines, and + * end text lines at the end of sentences. + */ + + if (OUST_MAC == st->oust || (eos && wend > 1 && + islower((unsigned char)st->outbuf[wend - 1]))) { + if (MDOC_MAX == mtype) + fputs(" 3", stdout); + if (MDOC_Fa != mtype) + for (wend += 2; + '\0' != st->outbuf[wend]; + wend++) + printf(" %c", + st->outbuf[wend]); mdoc_newln(st); - fputs(".Fn ", stdout); - outbuf_flush(st); - mdoc_newln(st); - continue; } - outbuf_flush(st); - putchar(' '); + /* Advance to the next word. */ + + while ('\n' != buf[start] && + isspace((unsigned char)buf[start])) + start++; st->wantws = 1; } @@ -1513,7 +1585,7 @@ dofile(const struct args *args, const char *fname, { char datebuf[64]; struct state st; - const char *fbase, *fext, *section, *date; + const char *fbase, *fext, *section, *date, *format; char *title, *cp; size_t sup, end, i, cur = 0; @@ -1558,8 +1630,12 @@ dofile(const struct args *args, const char *fname, /* Date. Or the given "tm" if not supplied. */ - if (NULL == (date = args->date)) { - strftime(datebuf, sizeof(datebuf), "%B %d, %Y", tm); + date = args->date; + format = (NULL == date) ? "%B %d, %Y" : + strcmp(date, "Mdocdate") ? NULL : "$Mdocdate: February 14 2015 $"; + + if (NULL != format) { + strftime(datebuf, sizeof(datebuf), format, tm); date = datebuf; }