version 1.15, 2014/04/02 14:50:09 |
version 1.29, 2014/07/11 20:45:55 |
|
|
#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, |
int hasdouble; |
i, j, textsz, stack; |
|
|
/* |
/* |
* 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. |
|
* However, we're slightly different because we might have |
|
* nested escapes we need to ignore. |
*/ |
*/ |
hasdouble = 0; |
stack = 0; |
for (sv = nstart = *start; nstart < end; nstart++) { |
for (linkstart = realend = *start; realend < end; realend++) { |
/* Do we have a double-colon? */ |
if ('<' == buf[realend]) |
if (':' == buf[nstart] && |
stack++; |
nstart > sv && |
if ('>' != buf[realend]) |
':' == buf[nstart - 1]) |
|
hasdouble = 1; |
|
if ('>' != buf[nstart]) |
|
continue; |
continue; |
else if (dsz == 1) |
else if (stack-- > 0) |
|
continue; |
|
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. |
for (textsz = *start; textsz < linkend; textsz++) |
* First, allow us to link to other Perl manuals. |
if ('|' == buf[textsz] || '/' == buf[textsz]) |
*/ |
break; |
if (hasdouble) |
|
|
if (textsz < linkend && '|' == buf[textsz]) { |
|
/* With title: set start, then end at section. */ |
|
linkstart = textsz + 1; |
|
textsz = textsz - *start; |
|
for (i = linkstart; i < linkend; i++) |
|
if ('/' == buf[i]) |
|
break; |
|
if (i < linkend) |
|
linkend = i; |
|
} else if (textsz < linkend && '/' == buf[textsz]) { |
|
/* With section: set end at section. */ |
|
linkend = textsz; |
|
textsz = 0; |
|
} else |
|
/* No title, no section. */ |
|
textsz = 0; |
|
|
|
*start = realend; |
|
j = linkend - linkstart; |
|
|
|
/* Do we have only subsection material? */ |
|
if (0 == j && '/' == buf[linkend]) { |
|
linkstart = linkend + 1; |
|
linkend = dsz > 1 ? realend - 1 : realend; |
|
if (0 == (j = linkend - linkstart)) |
|
return(0); |
|
printf("Sx %.*s", (int)j, &buf[linkstart]); |
|
return(1); |
|
} else if (0 == j) |
|
return(0); |
|
|
|
/* See if we qualify as being a link or not. */ |
|
if ((j > 4 && 0 == memcmp("http:", &buf[linkstart], j)) || |
|
(j > 5 && 0 == memcmp("https:", &buf[linkstart], j)) || |
|
(j > 3 && 0 == memcmp("ftp:", &buf[linkstart], j)) || |
|
(j > 4 && 0 == memcmp("sftp:", &buf[linkstart], j)) || |
|
(j > 3 && 0 == memcmp("smb:", &buf[linkstart], j)) || |
|
(j > 3 && 0 == memcmp("afs:", &buf[linkstart], j))) { |
|
/* Gross. */ |
|
printf("Lk %.*s", (int)((dsz > 1 ? realend - 1 : |
|
realend) - linkstart), &buf[linkstart]); |
|
return(1); |
|
} |
|
|
|
/* See if we qualify as a mailto. */ |
|
if (j > 6 && 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); |
} |
} |
|
|
Line 394 formatcode(struct state *st, const char *buf, size_t * |
|
Line 458 formatcode(struct state *st, const char *buf, size_t * |
|
(*start) += dsz; |
(*start) += dsz; |
break; |
break; |
} |
} |
|
if (*start < end) { |
|
assert('>' == buf[*start]); |
|
(*start)++; |
|
} |
|
if (isspace(last)) |
|
while (*start < end && isspace((int)buf[*start])) |
|
(*start)++; |
return(0); |
return(0); |
} |
} |
|
|
Line 438 formatcode(struct state *st, const char *buf, size_t * |
|
Line 509 formatcode(struct state *st, const char *buf, size_t * |
|
printf("Ar "); |
printf("Ar "); |
break; |
break; |
} |
} |
printf("Sy "); |
if (0 == strncmp(buf + *start, "NULL", 4) && |
|
('=' == buf[*start + 4] || |
|
'>' == buf[*start + 4])) |
|
printf("Dv "); |
|
else |
|
printf("Sy "); |
break; |
break; |
case (FMT_CODE): |
case (FMT_CODE): |
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; |
|
|
verbatim(struct state *st, const char *buf, size_t start, size_t end) |
verbatim(struct state *st, const char *buf, size_t start, size_t end) |
{ |
{ |
int last; |
int last; |
|
size_t i; |
|
|
if ( ! st->parsing || st->paused) |
if ( ! st->parsing || st->paused) |
return; |
return; |
|
again: |
|
/* |
|
* If we're in the SYNOPSIS, see if we're an #include block. |
|
* If we are, then print the "In" macro and re-loop. |
|
* This handles any number of inclusions, but only when they |
|
* come before the remaining parts... |
|
*/ |
|
if (SECT_SYNOPSIS == st->sect) { |
|
i = start; |
|
for (i = start; i < end && ' ' == buf[i]; i++) |
|
/* Spin. */ ; |
|
if (i == end) |
|
return; |
|
/* We're an include block! */ |
|
if (end - i > 10 && |
|
0 == memcmp(&buf[i], "#include <", 10)) { |
|
start = i + 10; |
|
while (start < end && ' ' == buf[start]) |
|
start++; |
|
fputs(".In ", stdout); |
|
/* Stop til the '>' marker or we hit eoln. */ |
|
while (start < end && |
|
'>' != buf[start] && '\n' != buf[start]) |
|
putchar(buf[start++]); |
|
putchar('\n'); |
|
if (start < end && '>' == buf[start]) |
|
start++; |
|
if (start < end && '\n' == buf[start]) |
|
start++; |
|
if (start < end) |
|
goto again; |
|
return; |
|
} |
|
} |
|
|
|
if (start == end) |
|
return; |
puts(".Bd -literal"); |
puts(".Bd -literal"); |
for (last = ' '; start < end; start++) { |
for (last = ' '; start < end; start++) { |
/* |
/* |
Line 866 dosynopsisop(const char *buf, int *last, |
|
Line 980 dosynopsisop(const char *buf, int *last, |
|
} |
} |
|
|
/* |
/* |
|
* 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 898 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1046 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; |
fputs(".Nm ", stdout); |
donamenm(st, buf, &start, i + 1); |
formatcodeln(st, buf, &start, i + 1, 1); |
|
putchar('\n'); |
|
start = j + 1; |
start = j + 1; |
|
while (start < end && ' ' == buf[start]) |
|
start++; |
fputs(".Nd ", stdout); |
fputs(".Nd ", stdout); |
formatcodeln(st, buf, &start, end, 1); |
formatcodeln(st, buf, &start, end, 1); |
putchar('\n'); |
putchar('\n'); |
Line 955 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1103 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, seq)) { |
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++; |
|
|
dofile(const struct args *args, const char *fname, |
dofile(const struct args *args, const char *fname, |
const struct tm *tm, const char *buf, size_t sz) |
const struct tm *tm, const char *buf, size_t sz) |
{ |
{ |
size_t sup, end, i, cur = 0; |
|
struct state st; |
|
const char *section, *date; |
|
char datebuf[64]; |
char datebuf[64]; |
|
struct state st; |
|
const char *fbase, *fext, *section, *date; |
char *title, *cp; |
char *title, *cp; |
|
size_t sup, end, i, cur = 0; |
|
|
if (0 == sz) |
if (0 == sz) |
return; |
return; |
|
|
/* Title is last path component of the filename. */ |
/* |
|
* Parsing the filename is almost always required, |
|
* except when both the title and the section |
|
* are provided on the command line. |
|
*/ |
|
|
if (NULL != args->title) |
if (NULL == args->title || NULL == args->section) { |
title = strdup(args->title); |
fbase = strrchr(fname, '/'); |
else if (NULL != (cp = strrchr(fname, '/'))) |
if (NULL == fbase) |
title = strdup(cp + 1); |
fbase = fname; |
else |
else |
title = strdup(fname); |
fbase++; |
|
fext = strrchr(fbase, '.'); |
|
} else |
|
fext = NULL; |
|
|
|
/* |
|
* The title will be converted to uppercase, |
|
* so it needs to be copied. |
|
*/ |
|
|
|
title = (NULL != args->title) ? strdup(args->title) : |
|
(NULL != fext) ? strndup(fbase, fext - fbase) : |
|
strdup(fbase); |
|
|
if (NULL == title) { |
if (NULL == title) { |
perror(NULL); |
perror(NULL); |
exit(EXIT_FAILURE); |
exit(EXIT_FAILURE); |
Line 1044 dofile(const struct args *args, const char *fname, |
|
Line 1219 dofile(const struct args *args, const char *fname, |
|
|
|
/* Section is 1 unless suffix is "pm". */ |
/* Section is 1 unless suffix is "pm". */ |
|
|
if (NULL == (section = args->section)) { |
section = (NULL != args->section) ? args->section : |
section = "1"; |
(NULL == fext || strcmp(fext + 1, "pm")) ? "1" : |
if (NULL != (cp = strrchr(title, '.'))) { |
PERL_SECTION; |
*cp++ = '\0'; |
|
if (0 == strcmp(cp, "pm")) |
|
section = PERL_SECTION; |
|
} |
|
} |
|
|
|
/* Date. Or the given "tm" if not supplied. */ |
/* Date. Or the given "tm" if not supplied. */ |
|
|
Line 1114 readfile(const struct args *args, const char *fname) |
|
Line 1284 readfile(const struct args *args, const char *fname) |
|
time_t ttm; |
time_t ttm; |
struct stat st; |
struct stat st; |
|
|
assert(NULL != fname); |
|
|
|
fd = 0 != strcmp("-", fname) ? |
fd = 0 != strcmp("-", fname) ? |
open(fname, O_RDONLY, 0) : STDIN_FILENO; |
open(fname, O_RDONLY, 0) : STDIN_FILENO; |
|
|
Line 1221 main(int argc, char *argv[]) |
|
Line 1389 main(int argc, char *argv[]) |
|
|
|
/* Accept only a single input file. */ |
/* Accept only a single input file. */ |
|
|
if (argc > 2) |
if (argc > 1) |
return(EXIT_FAILURE); |
goto usage; |
else if (1 == argc) |
else if (1 == argc) |
fname = *argv; |
fname = *argv; |
|
|
Line 1231 main(int argc, char *argv[]) |
|
Line 1399 main(int argc, char *argv[]) |
|
|
|
usage: |
usage: |
fprintf(stderr, "usage: %s [-d date] " |
fprintf(stderr, "usage: %s [-d date] " |
"[-n title] [-s section]\n", name); |
"[-n title] [-s section] [file]\n", name); |
|
|
return(EXIT_FAILURE); |
return(EXIT_FAILURE); |
} |
} |