version 1.12, 2014/04/01 11:58:32 |
version 1.19, 2014/04/03 10:17:14 |
|
|
#include <unistd.h> |
#include <unistd.h> |
|
|
/* |
/* |
* In what section can we find Perl manuals? |
* In what section can we find Perl module manuals? |
|
* Sometimes (Mac OS X) it's 3pm, sometimes (OpenBSD, etc.) 3p. |
|
* XXX IF YOU CHANGE THIS, CHANGE POD2MDOC.1 AS WELL. |
*/ |
*/ |
#define PERL_SECTION "3p" |
#define PERL_SECTION "3p" |
|
|
Line 170 formatescape(const char *buf, size_t *start, size_t en |
|
Line 172 formatescape(const char *buf, size_t *start, size_t en |
|
|
|
/* |
/* |
* Run some heuristics to intuit a link format. |
* Run some heuristics to intuit a link format. |
* I recognise L<foo::bar> as a Perl manpage, printing it in section 3p; |
* I set "start" to be the end of the sequence (last right-carrot) so |
* or a general UNIX foo(5) manpage. |
|
* If I recognise one, I set "start" to be the end of the sequence so |
|
* that the caller can safely just continue processing. |
* that the caller can safely just continue processing. |
* Otherwise, I don't touch "start". |
* If this is just an empty tag, I'll return 0. |
*/ |
*/ |
static int |
static int |
trylink(const char *buf, size_t *start, size_t end, size_t dsz) |
trylink(const char *buf, size_t *start, size_t end, size_t dsz) |
{ |
{ |
size_t sv, nstart, nend, i, j; |
size_t linkstart, realend, linkend, i, j, textsz; |
int hasdouble; |
const char *text; |
|
|
/* |
/* |
* Scan to the start of the terminus. |
* Scan to the start of the terminus. |
* This function is more or less replicated in the formatcode() |
* This function is more or less replicated in the formatcode() |
* for null or index formatting codes. |
* for null or index formatting codes. |
*/ |
*/ |
hasdouble = 0; |
for (linkstart = realend = *start; realend < end; realend++) { |
for (sv = nstart = *start; nstart < end; nstart++) { |
if ('>' != buf[realend]) |
/* Do we have a double-colon? */ |
|
if (':' == buf[nstart] && |
|
nstart > sv && |
|
':' == buf[nstart - 1]) |
|
hasdouble = 1; |
|
if ('>' != buf[nstart]) |
|
continue; |
continue; |
else if (dsz == 1) |
else if (dsz == 1) |
break; |
break; |
assert(nstart > 0); |
assert(realend > 0); |
if (' ' != buf[nstart - 1]) |
if (' ' != buf[realend - 1]) |
continue; |
continue; |
i = nstart; |
for (i = realend, j = 0; i < end && j < dsz; j++) |
for (j = 0; i < end && j < dsz; j++) |
|
if ('>' != buf[i++]) |
if ('>' != buf[i++]) |
break; |
break; |
if (dsz == j) |
if (dsz == j) |
break; |
break; |
} |
} |
|
|
/* We don't care about stubs. */ |
/* Ignore stubs. */ |
if (nstart == end || nstart == *start) |
if (realend == end || realend == *start) |
return(0); |
return(0); |
|
|
/* Set nend to the end of content. */ |
/* Set linkend to the end of content. */ |
nend = nstart; |
linkend = dsz > 1 ? realend - 1 : realend; |
if (dsz > 1) |
|
nend--; |
|
|
|
/* |
/* Re-scan to see if we have a title or section. */ |
* Provide for some common invocations of the link primitive. |
text = &buf[*start]; |
* First, allow us to link to other Perl manuals. |
for (textsz = *start; textsz < linkend; textsz++) |
*/ |
if ('|' == buf[textsz] || '/' == buf[textsz]) |
if (hasdouble) |
break; |
|
|
|
/* If we have a title, find the section. */ |
|
if (textsz < linkend && '|' == buf[textsz]) { |
|
linkstart = textsz + 1; |
|
textsz = textsz - *start; |
|
for (i = linkstart; i < linkend; i++) |
|
if ('/' == buf[i]) |
|
break; |
|
if (i < linkend) |
|
linkend = i; |
|
} else { |
|
textsz = 0; |
|
if (textsz < linkend && '/' == buf[textsz]) |
|
linkend = textsz; |
|
} |
|
|
|
*start = realend; |
|
|
|
j = linkend - linkstart; |
|
|
|
if (0 == j) |
|
return(0); |
|
|
|
/* See if we qualify as being a link or not. */ |
|
if ((j > 5 && 0 == memcmp("http:", &buf[linkstart], j)) || |
|
(j > 6 && 0 == memcmp("https:", &buf[linkstart], j)) || |
|
(j > 4 && 0 == memcmp("ftp:", &buf[linkstart], j)) || |
|
(j > 5 && 0 == memcmp("sftp:", &buf[linkstart], j)) || |
|
(j > 4 && 0 == memcmp("smb:", &buf[linkstart], j)) || |
|
(j > 4 && 0 == memcmp("afs:", &buf[linkstart], j))) { |
|
printf("Lk %.*s", (int)j, &buf[linkstart]); |
|
return(1); |
|
} |
|
|
|
/* See if we qualify as a mailto. */ |
|
if (j > 7 && 0 == memcmp("mailto:", &buf[linkstart], j)) { |
|
printf("Mt %.*s", (int)j, &buf[linkstart]); |
|
return(1); |
|
} |
|
|
|
/* See if we're a foo(5), foo(5x), or foo(5xx) manpage. */ |
|
if ((j > 3 && ')' == buf[linkend - 1]) && |
|
('(' == buf[linkend - 3])) { |
|
printf("Xr %.*s %c", (int)(j - 3), |
|
&buf[linkstart], buf[linkend - 2]); |
|
return(1); |
|
} else if ((j > 4 && ')' == buf[linkend - 1]) && |
|
('(' == buf[linkend - 4])) { |
|
printf("Xr %.*s %.*s", (int)(j - 4), |
|
&buf[linkstart], 2, &buf[linkend - 3]); |
|
return(1); |
|
} else if ((j > 5 && ')' == buf[linkend - 1]) && |
|
('(' == buf[linkend - 5])) { |
|
printf("Xr %.*s %.*s", (int)(j - 5), |
|
&buf[linkstart], 3, &buf[linkend - 4]); |
|
return(1); |
|
} |
|
|
|
/* Last try: do we have a double-colon? */ |
|
for (i = linkstart + 1; i < linkend; i++) |
|
if (':' == buf[i] && ':' == buf[i - 1]) |
|
break; |
|
|
|
if (i < linkend) |
printf("Xr %.*s " PERL_SECTION, |
printf("Xr %.*s " PERL_SECTION, |
(int)(nend - sv), &buf[sv]); |
(int)j, &buf[linkstart]); |
else if (nend - sv > 3 && isalnum(buf[sv]) && |
|
')' == buf[nend - 1] && |
|
isdigit((int)buf[nend - 2]) && |
|
'(' == buf[nend - 3]) |
|
printf("Xr %.*s %c", |
|
(int)(nend - 3 - sv), |
|
&buf[sv], buf[nend - 2]); |
|
else |
else |
return(0); |
printf("Xr %.*s 1", (int)j, &buf[linkstart]); |
|
|
*start = nstart; |
|
return(1); |
return(1); |
} |
} |
|
|
/* |
/* |
|
* Doclifting: if we're a bold "-xx" and we're in the SYNOPSIS section, |
|
* then it's likely that we're a flag. |
|
* Our flag might be followed by an argument, so make sure that we're |
|
* accounting for that, too. |
|
* If we don't have a flag at all, however, then assume we're an "Ar". |
|
*/ |
|
static void |
|
dosynopsisfl(const char *buf, size_t *start, size_t end) |
|
{ |
|
size_t i; |
|
again: |
|
assert(*start + 1 < end); |
|
assert('-' == buf[*start]); |
|
|
|
if ( ! isalnum((int)buf[*start + 1]) && |
|
'?' != buf[*start + 1] && |
|
'-' != buf[*start + 1]) { |
|
(*start)--; |
|
fputs("Ar ", stdout); |
|
return; |
|
} |
|
|
|
(*start)++; |
|
for (i = *start; i < end; i++) |
|
if (isalnum((int)buf[i])) |
|
continue; |
|
else if ('?' == buf[i]) |
|
continue; |
|
else if ('-' == buf[i]) |
|
continue; |
|
else if ('_' == buf[i]) |
|
continue; |
|
else |
|
break; |
|
|
|
assert(i < end); |
|
|
|
if ( ! (' ' == buf[i] || '>' == buf[i])) { |
|
printf("Ar "); |
|
return; |
|
} |
|
|
|
printf("Fl "); |
|
if (end - *start > 1 && |
|
isupper((int)buf[*start]) && |
|
islower((int)buf[*start + 1]) && |
|
(end - *start == 2 || |
|
' ' == buf[*start + 2])) |
|
printf("\\&"); |
|
printf("%.*s ", (int)(i - *start), &buf[*start]); |
|
*start = i; |
|
|
|
if (' ' == buf[i]) { |
|
while (i < end && ' ' == buf[i]) |
|
i++; |
|
assert(i < end); |
|
if ('-' == buf[i]) { |
|
*start = i; |
|
goto again; |
|
} |
|
printf("Ar "); |
|
*start = i; |
|
} |
|
} |
|
|
|
/* |
* We're at the character in front of a format code, which is structured |
* We're at the character in front of a format code, which is structured |
* like X<...> and can contain nested format codes. |
* like X<...> and can contain nested format codes. |
* This consumes the whole format code, and any nested format codes, til |
* This consumes the whole format code, and any nested format codes, til |
Line 248 trylink(const char *buf, size_t *start, size_t end, si |
|
Line 361 trylink(const char *buf, size_t *start, size_t end, si |
|
* been printed to the current line. |
* been printed to the current line. |
* If "nomacro", then we don't print any macros, just contained data |
* If "nomacro", then we don't print any macros, just contained data |
* (e.g., following "Sh" or "Nm"). |
* (e.g., following "Sh" or "Nm"). |
|
* "pos" is only significant in SYNOPSIS, and should be 0 when invoked |
|
* as the first format code on a line (for decoration as an "Nm"), |
|
* non-zero otherwise. |
* Return whether we've printed a macro or not--in other words, whether |
* Return whether we've printed a macro or not--in other words, whether |
* this should trigger a subsequent newline (this should be ignored when |
* this should trigger a subsequent newline (this should be ignored when |
* reentrant). |
* reentrant). |
*/ |
*/ |
static int |
static int |
formatcode(struct state *st, const char *buf, |
formatcode(struct state *st, const char *buf, size_t *start, |
size_t *start, size_t end, int reentrant, int nomacro) |
size_t end, int reentrant, int nomacro, int pos) |
{ |
{ |
enum fmt fmt; |
enum fmt fmt; |
size_t i, j, dsz; |
size_t i, j, dsz; |
Line 360 formatcode(struct state *st, const char *buf, |
|
Line 476 formatcode(struct state *st, const char *buf, |
|
printf("Em "); |
printf("Em "); |
break; |
break; |
case (FMT_BOLD): |
case (FMT_BOLD): |
/* |
if (SECT_SYNOPSIS == st->sect) { |
* Doclifting: if we're a bold "-xx" and we're |
if (1 == dsz && '-' == buf[*start]) |
* in the SYNOPSIS section, then it's likely |
dosynopsisfl(buf, start, end); |
* that we're a flag. |
else if (0 == pos) |
* Be really strict: only do this when the dash |
printf("Nm "); |
* is followed by alnums til the end marker, |
else |
* which mustn't be a custom. |
printf("Ar "); |
*/ |
break; |
if (SECT_SYNOPSIS == st->sect && |
|
end - *start > 1 && |
|
'-' == buf[*start] && |
|
(isalnum((int)buf[*start + 1]) || |
|
'?' == buf[*start + 1])) { |
|
for (i = *start + 1; i < end; i++) |
|
if ( ! isalnum((int)buf[i])) |
|
break; |
|
if (i < end && '>' == buf[i]) { |
|
(*start)++; |
|
printf("Fl "); |
|
break; |
|
} |
|
} |
} |
printf("Sy "); |
printf("Sy "); |
break; |
break; |
Line 388 formatcode(struct state *st, const char *buf, |
|
Line 491 formatcode(struct state *st, const char *buf, |
|
printf("Qo Li "); |
printf("Qo Li "); |
break; |
break; |
case (FMT_LINK): |
case (FMT_LINK): |
|
/* Try to link; use "No" if it's empty. */ |
if ( ! trylink(buf, start, end, dsz)) |
if ( ! trylink(buf, start, end, dsz)) |
printf("No "); |
printf("No "); |
break; |
break; |
Line 430 formatcode(struct state *st, const char *buf, |
|
Line 534 formatcode(struct state *st, const char *buf, |
|
} |
} |
} |
} |
if (*start + 1 < end && '<' == buf[*start + 1]) { |
if (*start + 1 < end && '<' == buf[*start + 1]) { |
formatcode(st, buf, start, end, 1, nomacro); |
formatcode(st, buf, start, end, 1, nomacro, 1); |
continue; |
continue; |
} |
} |
|
|
Line 492 formatcodeln(struct state *st, const char *buf, |
|
Line 596 formatcodeln(struct state *st, const char *buf, |
|
last = ' '; |
last = ' '; |
while (*start < end) { |
while (*start < end) { |
if (*start + 1 < end && '<' == buf[*start + 1]) { |
if (*start + 1 < end && '<' == buf[*start + 1]) { |
formatcode(st, buf, start, end, 1, nomacro); |
formatcode(st, buf, start, end, 1, nomacro, 1); |
continue; |
continue; |
} |
} |
/* |
/* |
Line 755 verbatim(struct state *st, const char *buf, size_t sta |
|
Line 859 verbatim(struct state *st, const char *buf, size_t sta |
|
} |
} |
|
|
/* |
/* |
|
* See dosynopsisop(). |
|
*/ |
|
static int |
|
hasmatch(const char *buf, size_t start, size_t end) |
|
{ |
|
size_t stack; |
|
|
|
for (stack = 0; start < end; start++) |
|
if (buf[start] == '[') |
|
stack++; |
|
else if (buf[start] == ']' && 0 == stack) |
|
return(1); |
|
else if (buf[start] == ']') |
|
stack--; |
|
return(0); |
|
} |
|
|
|
/* |
|
* If we're in the SYNOPSIS section and we've encounter braces in an |
|
* ordinary paragraph, then try to see whether we're an [-option]. |
|
* Do this, if we're an opening bracket, by first seeing if we have a |
|
* matching end via hasmatch(). |
|
* If we're an ending bracket, see if we have a stack already. |
|
*/ |
|
static int |
|
dosynopsisop(const char *buf, int *last, |
|
size_t *start, size_t end, size_t *opstack) |
|
{ |
|
|
|
assert('[' == buf[*start] || ']' == buf[*start]); |
|
|
|
if ('[' == buf[*start] && hasmatch(buf, *start + 1, end)) { |
|
if ('\n' != *last) |
|
putchar('\n'); |
|
puts(".Oo"); |
|
(*opstack)++; |
|
} else if ('[' == buf[*start]) |
|
return(0); |
|
|
|
if (']' == buf[*start] && *opstack > 0) { |
|
if ('\n' != *last) |
|
putchar('\n'); |
|
puts(".Oc"); |
|
(*opstack)--; |
|
} else if (']' == buf[*start]) |
|
return(0); |
|
|
|
(*start)++; |
|
*last = '\n'; |
|
while (' ' == buf[*start]) |
|
(*start)++; |
|
return(1); |
|
} |
|
|
|
/* |
|
* Format multiple "Nm" manpage names in the NAME section. |
|
*/ |
|
static void |
|
donamenm(struct state *st, const char *buf, size_t *start, size_t end) |
|
{ |
|
size_t word; |
|
|
|
while (*start < end && ' ' == buf[*start]) |
|
(*start)++; |
|
|
|
if (end == *start) { |
|
puts(".Nm unknown"); |
|
return; |
|
} |
|
|
|
while (*start < end) { |
|
fputs(".Nm ", stdout); |
|
for (word = *start; word < end; word++) |
|
if (',' == buf[word]) |
|
break; |
|
formatcodeln(st, buf, start, word, 1); |
|
if (*start == end) { |
|
putchar('\n'); |
|
continue; |
|
} |
|
assert(',' == buf[*start]); |
|
puts(" ,"); |
|
(*start)++; |
|
while (*start < end && ' ' == buf[*start]) |
|
(*start)++; |
|
} |
|
} |
|
|
|
/* |
* Ordinary paragraph. |
* Ordinary paragraph. |
* Well, this is really the hardest--POD seems to assume that, for |
* Well, this is really the hardest--POD seems to assume that, for |
* example, a leading space implies a newline, and so on. |
* example, a leading space implies a newline, and so on. |
Line 765 verbatim(struct state *st, const char *buf, size_t sta |
|
Line 958 verbatim(struct state *st, const char *buf, size_t sta |
|
static void |
static void |
ordinary(struct state *st, const char *buf, size_t start, size_t end) |
ordinary(struct state *st, const char *buf, size_t start, size_t end) |
{ |
{ |
size_t i, j; |
size_t i, j, opstack; |
|
int seq; |
|
|
if ( ! st->parsing || st->paused) |
if ( ! st->parsing || st->paused) |
return; |
return; |
Line 777 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 971 ordinary(struct state *st, const char *buf, size_t sta |
|
* To wit, print out a "Nm" and "Nd" in that format. |
* To wit, print out a "Nm" and "Nd" in that format. |
*/ |
*/ |
if (SECT_NAME == st->sect) { |
if (SECT_NAME == st->sect) { |
for (i = end - 1; i > start; i--) |
for (i = end - 2; i > start; i--) |
if ('-' == buf[i]) |
if ('-' == buf[i] && ' ' == buf[i + 1]) |
break; |
break; |
if ('-' == buf[i]) { |
if ('-' == buf[i]) { |
j = i; |
j = i; |
Line 786 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 980 ordinary(struct state *st, const char *buf, size_t sta |
|
for ( ; i > start; i--) |
for ( ; i > start; i--) |
if ('-' != buf[i]) |
if ('-' != buf[i]) |
break; |
break; |
printf(".Nm "); |
donamenm(st, buf, &start, i + 1); |
formatcodeln(st, buf, &start, i + 1, 1); |
|
putchar('\n'); |
|
start = j + 1; |
start = j + 1; |
printf(".Nd "); |
while (start < end && ' ' == buf[start]) |
|
start++; |
|
fputs(".Nd ", stdout); |
formatcodeln(st, buf, &start, end, 1); |
formatcodeln(st, buf, &start, end, 1); |
putchar('\n'); |
putchar('\n'); |
return; |
return; |
Line 802 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 996 ordinary(struct state *st, const char *buf, size_t sta |
|
|
|
st->haspar = 0; |
st->haspar = 0; |
last = '\n'; |
last = '\n'; |
|
opstack = 0; |
|
|
while (start < end) { |
for (seq = 0; start < end; seq++) { |
/* |
/* |
* Loop til we get either to a newline or escape. |
* Loop til we get either to a newline or escape. |
* Escape initial control characters. |
* Escape initial control characters. |
Line 817 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1012 ordinary(struct state *st, const char *buf, size_t sta |
|
printf("\\&"); |
printf("\\&"); |
else if ('\n' == last && '\'' == buf[start]) |
else if ('\n' == last && '\'' == buf[start]) |
printf("\\&"); |
printf("\\&"); |
#if notyet |
|
/* |
/* |
* If we're in the SYNOPSIS, have square |
* If we're in the SYNOPSIS, have square |
* brackets indicate that we're opening and |
* brackets indicate that we're opening and |
* closing an optional context. |
* closing an optional context. |
*/ |
*/ |
if (SECT_SYNOPSIS == st->sect) { |
if (SECT_SYNOPSIS == st->sect && |
if ('[' == buf[start] || |
('[' == buf[start] || |
']' == buf[start]) { |
']' == buf[start]) && |
if (last != '\n') |
dosynopsisop(buf, &last, |
putchar('\n'); |
&start, end, &opstack)) |
if ('[' == buf[start]) |
continue; |
printf(".Oo\n"); |
|
else |
|
printf(".Oc\n"); |
|
start++; |
|
continue; |
|
} |
|
} |
|
#endif |
|
putchar(last = buf[start++]); |
putchar(last = buf[start++]); |
if ('\\' == last) |
if ('\\' == last) |
putchar('e'); |
putchar('e'); |
Line 851 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1037 ordinary(struct state *st, const char *buf, size_t sta |
|
* following that, a newline. |
* following that, a newline. |
* Consume all whitespace so we don't |
* Consume all whitespace so we don't |
* accidentally start an implicit literal line. |
* accidentally start an implicit literal line. |
|
* If the macro ends with a flush comma or |
|
* period, let mdoc(7) handle it for us. |
*/ |
*/ |
if (formatcode(st, buf, &start, end, 0, 0)) { |
if (formatcode(st, buf, &start, end, 0, 0, seq)) { |
|
if ((start == end - 1 || |
|
(start < end - 1 && |
|
(' ' == buf[start + 1] || |
|
'\n' == buf[start + 1]))) && |
|
('.' == buf[start] || |
|
',' == buf[start])) { |
|
putchar(' '); |
|
putchar(buf[start++]); |
|
} |
putchar(last = '\n'); |
putchar(last = '\n'); |
while (start < end && ' ' == buf[start]) |
while (start < end && ' ' == buf[start]) |
start++; |
start++; |