=================================================================== RCS file: /cvs/pod2mdoc/pod2mdoc.c,v retrieving revision 1.37 retrieving revision 1.42 diff -u -p -r1.37 -r1.42 --- pod2mdoc/pod2mdoc.c 2015/02/13 00:44:16 1.37 +++ pod2mdoc/pod2mdoc.c 2015/02/14 10:35:02 1.42 @@ -1,4 +1,4 @@ -/* $Id: pod2mdoc.c,v 1.37 2015/02/13 00:44:16 schwarze Exp $ */ +/* $Id: pod2mdoc.c,v 1.42 2015/02/14 10:35:02 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 +478,10 @@ 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); assert('<' == buf[*start + 1]); @@ -562,7 +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 @@ -572,22 +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('.'); @@ -612,13 +617,28 @@ formatcode(struct state *st, const char *buf, size_t * else printf("Ar "); break; - } - if (0 == strncmp(buf + *start, "NULL", 4) && - ('=' == buf[*start + 4] || - '>' == buf[*start + 4])) + } + i = 0; + uc = buf[*start]; + while (isalnum(uc) || '_' == uc || ' ' == uc) + uc = buf[*start + ++i]; + if ('=' != uc && '>' != uc) + i = 0; + if (4 == i && ! strncmp(buf + *start, "NULL", 4)) { printf("Dv "); - else + break; + } + switch (i ? dict_get(buf + *start, i) : MDOC_MAX) { + case MDOC_Fa: + printf("Fa "); + break; + case MDOC_Vt: + printf("Vt "); + break; + default: printf("Sy "); + break; + } break; case (FMT_CODE): printf("Qo Li "); @@ -671,7 +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; } @@ -713,14 +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); } /* @@ -1006,6 +1024,32 @@ command(struct state *st, const char *buf, size_t star } /* + * Put the type provided as an argument into the dictionary. + */ +static void +register_type(const char *ptype) +{ + const char *pname, *pend; + + pname = ptype; + while (isalnum((unsigned char)*pname) || '_' == *pname) + pname++; + if ((pname - ptype == 6 && ! strncmp(ptype, "struct", 6)) || + (pname - ptype == 4 && ! strncmp(ptype, "enum", 4))) { + while (' ' == *pname) + pname++; + pend = pname; + while (isalnum((unsigned char)*pend) || '_' == *pend) + pend++; + if (pend > pname) + dict_put(pname, pend - pname, MDOC_Vt); + } else + pend = pname; + if (pend > ptype) + dict_put(ptype, pend - ptype, MDOC_Vt); +} + +/* * Just pump out the line in a verbatim block. * From the perspective of external callers, * always stays in OUST_NL/wantws mode. @@ -1014,7 +1058,7 @@ static void verbatim(struct state *st, char *buf, size_t start, size_t end) { size_t i, ift, ifo, ifa, ifc, inl; - char *cp; + char *cp, *cp2; int nopen; if ( ! st->parsing || st->paused || start == end) @@ -1049,11 +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; @@ -1102,6 +1156,7 @@ again: if (buf[i] == '\n') buf[i] = ' '; buf[ifo++] = '\0'; + register_type(buf + ift); printf(".Ft %s", buf + ift); if (buf[ifo] == '*') { fputs(" *", stdout); @@ -1110,12 +1165,21 @@ again: putchar('\n'); buf[ifa++] = '\0'; printf(".Fo %s\n", buf + ifo); - dict_put(buf + ifo, MDOC_Fo); + dict_put(buf + ifo, 0, MDOC_Fo); buf[ifc++] = '\0'; for (;;) { cp = strchr(buf + ifa, ','); - if (cp != NULL) + if (cp != NULL) { + cp2 = cp; *cp++ = '\0'; + } else + cp2 = strchr(buf + ifa, '\0'); + while (isalnum((unsigned char)cp2[-1]) || + '_' == cp2[-1]) + cp2--; + if ('\0' != *cp2) + dict_put(cp2, 0, MDOC_Fa); + register_type(buf + ifa); printf(".Fa \"%s\"\n", buf + ifa); if (cp == NULL) break; @@ -1257,7 +1321,8 @@ 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 @@ -1330,32 +1395,52 @@ 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); continue; } - if ( ! strcmp(st->outbuf + st->outbuflen - 2, "()") && + /* Detect function names. */ + + if (st->outbuflen > 2 && + ')' == st->outbuf[st->outbuflen - 1] && + '(' == st->outbuf[st->outbuflen - 2] && dict_get(st->outbuf, st->outbuflen - 2) == MDOC_Fo) { st->outbuflen -= 2; st->outbuf[st->outbuflen] = '\0'; mdoc_newln(st); fputs(".Fn ", stdout); - outbuf_flush(st); - mdoc_newln(st); - continue; + st->oust = OUST_MAC; } + /* + * On whitespace, flush the output buffer + * and allow breaking to a macro line. + */ + outbuf_flush(st); - putchar(' '); + + /* + * End macro lines, and + * end text lines at the end of sentences. + */ + + if (OUST_MAC == st->oust || (start > 3 && + ('.' == buf[start - 2] || + '!' == buf[start - 2] || + '?' == buf[start - 2]) && + islower((unsigned char)buf[start - 3]) && + islower((unsigned char)buf[start - 4]))) + mdoc_newln(st); + + /* Advance to the next word. */ + + while (isspace((unsigned char)buf[start])) + start++; st->wantws = 1; }