version 1.8, 2014/03/23 23:57:38 |
version 1.13, 2014/04/01 13:05:50 |
|
|
#include <string.h> |
#include <string.h> |
#include <unistd.h> |
#include <unistd.h> |
|
|
|
/* |
|
* In what section can we find Perl manuals? |
|
*/ |
|
#define PERL_SECTION "3p" |
|
|
struct args { |
struct args { |
const char *title; /* override "Dt" title */ |
const char *title; /* override "Dt" title */ |
const char *date; /* override "Dd" date */ |
const char *date; /* override "Dd" date */ |
|
|
LIST__MAX |
LIST__MAX |
}; |
}; |
|
|
|
enum sect { |
|
SECT_NONE = 0, |
|
SECT_NAME, /* NAME section */ |
|
SECT_SYNOPSIS, /* SYNOPSIS section */ |
|
}; |
|
|
struct state { |
struct state { |
int parsing; /* after =cut of before command */ |
int parsing; /* after =cut of before command */ |
int paused; /* in =begin and before =end */ |
int paused; /* in =begin and before =end */ |
int haspar; /* in paragraph: do we need Pp? */ |
int haspar; /* in paragraph: do we need Pp? */ |
int isname; /* are we the NAME section? */ |
enum sect sect; /* which section are we in? */ |
const char *fname; /* file being parsed */ |
const char *fname; /* file being parsed */ |
#define LIST_STACKSZ 128 |
#define LIST_STACKSZ 128 |
enum list lstack[LIST_STACKSZ]; /* open lists */ |
enum list lstack[LIST_STACKSZ]; /* open lists */ |
Line 158 formatescape(const char *buf, size_t *start, size_t en |
|
Line 169 formatescape(const char *buf, size_t *start, size_t en |
|
} |
} |
|
|
/* |
/* |
|
* Run some heuristics to intuit a link format. |
|
* I recognise L<foo::bar> as a Perl manpage, printing it in section 3p; |
|
* 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. |
|
* Otherwise, I don't touch "start". |
|
*/ |
|
static int |
|
trylink(const char *buf, size_t *start, size_t end, size_t dsz) |
|
{ |
|
size_t sv, nstart, nend, i, j; |
|
int hasdouble; |
|
|
|
/* |
|
* Scan to the start of the terminus. |
|
* This function is more or less replicated in the formatcode() |
|
* for null or index formatting codes. |
|
*/ |
|
hasdouble = 0; |
|
for (sv = nstart = *start; nstart < end; nstart++) { |
|
/* Do we have a double-colon? */ |
|
if (':' == buf[nstart] && |
|
nstart > sv && |
|
':' == buf[nstart - 1]) |
|
hasdouble = 1; |
|
if ('>' != buf[nstart]) |
|
continue; |
|
else if (dsz == 1) |
|
break; |
|
assert(nstart > 0); |
|
if (' ' != buf[nstart - 1]) |
|
continue; |
|
i = nstart; |
|
for (j = 0; i < end && j < dsz; j++) |
|
if ('>' != buf[i++]) |
|
break; |
|
if (dsz == j) |
|
break; |
|
} |
|
|
|
/* We don't care about stubs. */ |
|
if (nstart == end || nstart == *start) |
|
return(0); |
|
|
|
/* Set nend to the end of content. */ |
|
nend = nstart; |
|
if (dsz > 1) |
|
nend--; |
|
|
|
/* |
|
* Provide for some common invocations of the link primitive. |
|
* First, allow us to link to other Perl manuals. |
|
*/ |
|
if (hasdouble) |
|
printf("Xr %.*s " PERL_SECTION, |
|
(int)(nend - sv), &buf[sv]); |
|
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 |
|
return(0); |
|
|
|
*start = nstart; |
|
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: |
|
(*start)++; |
|
for (i = *start; i < end; i++) |
|
if (isalnum((int)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 171 formatescape(const char *buf, size_t *start, size_t en |
|
Line 307 formatescape(const char *buf, size_t *start, size_t en |
|
* reentrant). |
* reentrant). |
*/ |
*/ |
static int |
static int |
formatcode(const char *buf, size_t *start, |
formatcode(struct state *st, const char *buf, |
size_t end, int reentrant, int nomacro) |
size_t *start, size_t end, int reentrant, int nomacro) |
{ |
{ |
enum fmt fmt; |
enum fmt fmt; |
size_t i, j, dsz; |
size_t i, j, dsz; |
Line 278 formatcode(const char *buf, size_t *start, |
|
Line 414 formatcode(const char *buf, size_t *start, |
|
printf("Em "); |
printf("Em "); |
break; |
break; |
case (FMT_BOLD): |
case (FMT_BOLD): |
printf("Sy "); |
if (SECT_SYNOPSIS == st->sect && |
|
1 == dsz && |
|
'-' == buf[*start]) |
|
dosynopsisfl(buf, start, end); |
|
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): |
printf("Lk "); |
if ( ! trylink(buf, start, end, dsz)) |
|
printf("No "); |
break; |
break; |
case (FMT_FILE): |
case (FMT_FILE): |
printf("Pa "); |
printf("Pa "); |
break; |
break; |
case (FMT_NBSP): |
case (FMT_NBSP): |
/* TODO. */ |
|
printf("No "); |
printf("No "); |
break; |
break; |
default: |
default: |
Line 326 formatcode(const char *buf, size_t *start, |
|
Line 467 formatcode(const char *buf, size_t *start, |
|
} |
} |
} |
} |
if (*start + 1 < end && '<' == buf[*start + 1]) { |
if (*start + 1 < end && '<' == buf[*start + 1]) { |
formatcode(buf, start, end, 1, nomacro); |
formatcode(st, buf, start, end, 1, nomacro); |
continue; |
continue; |
} |
} |
|
|
Line 381 formatcode(const char *buf, size_t *start, |
|
Line 522 formatcode(const char *buf, size_t *start, |
|
* Calls formatcode() til the end of a paragraph. |
* Calls formatcode() til the end of a paragraph. |
*/ |
*/ |
static void |
static void |
formatcodeln(const char *buf, size_t *start, size_t end, int nomacro) |
formatcodeln(struct state *st, const char *buf, |
|
size_t *start, size_t end, int nomacro) |
{ |
{ |
|
|
last = ' '; |
last = ' '; |
while (*start < end) { |
while (*start < end) { |
if (*start + 1 < end && '<' == buf[*start + 1]) { |
if (*start + 1 < end && '<' == buf[*start + 1]) { |
formatcode(buf, start, end, 1, nomacro); |
formatcode(st, buf, start, end, 1, nomacro); |
continue; |
continue; |
} |
} |
/* |
/* |
Line 487 command(struct state *st, const char *buf, size_t star |
|
Line 629 command(struct state *st, const char *buf, size_t star |
|
* how pod2man handles it. |
* how pod2man handles it. |
*/ |
*/ |
printf(".Sh "); |
printf(".Sh "); |
st->isname = 0; |
st->sect = SECT_NONE; |
if (end - start == 4) |
if (end - start == 4) { |
if (0 == memcmp(&buf[start], "NAME", 4)) |
if (0 == memcmp(&buf[start], "NAME", 4)) |
st->isname = 1; |
st->sect = SECT_NAME; |
formatcodeln(buf, &start, end, 1); |
} else if (end - start == 8) { |
|
if (0 == memcmp(&buf[start], "SYNOPSIS", 8)) |
|
st->sect = SECT_SYNOPSIS; |
|
} |
|
formatcodeln(st, buf, &start, end, 1); |
putchar('\n'); |
putchar('\n'); |
st->haspar = 1; |
st->haspar = 1; |
break; |
break; |
case (CMD_HEAD2): |
case (CMD_HEAD2): |
printf(".Ss "); |
printf(".Ss "); |
formatcodeln(buf, &start, end, 1); |
formatcodeln(st, buf, &start, end, 1); |
putchar('\n'); |
putchar('\n'); |
st->haspar = 1; |
st->haspar = 1; |
break; |
break; |
case (CMD_HEAD3): |
case (CMD_HEAD3): |
puts(".Pp"); |
puts(".Pp"); |
printf(".Em "); |
printf(".Em "); |
formatcodeln(buf, &start, end, 0); |
formatcodeln(st, buf, &start, end, 0); |
putchar('\n'); |
putchar('\n'); |
puts(".Pp"); |
puts(".Pp"); |
st->haspar = 1; |
st->haspar = 1; |
Line 512 command(struct state *st, const char *buf, size_t star |
|
Line 658 command(struct state *st, const char *buf, size_t star |
|
case (CMD_HEAD4): |
case (CMD_HEAD4): |
puts(".Pp"); |
puts(".Pp"); |
printf(".No "); |
printf(".No "); |
formatcodeln(buf, &start, end, 0); |
formatcodeln(st, buf, &start, end, 0); |
putchar('\n'); |
putchar('\n'); |
puts(".Pp"); |
puts(".Pp"); |
st->haspar = 1; |
st->haspar = 1; |
Line 566 command(struct state *st, const char *buf, size_t star |
|
Line 712 command(struct state *st, const char *buf, size_t star |
|
switch (st->lstack[st->lpos - 1]) { |
switch (st->lstack[st->lpos - 1]) { |
case (LIST_TAG): |
case (LIST_TAG): |
printf(".It "); |
printf(".It "); |
formatcodeln(buf, &start, end, 0); |
formatcodeln(st, buf, &start, end, 0); |
putchar('\n'); |
putchar('\n'); |
break; |
break; |
case (LIST_ENUM): |
case (LIST_ENUM): |
Line 646 verbatim(struct state *st, const char *buf, size_t sta |
|
Line 792 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); |
|
} |
|
|
|
/* |
* 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 656 verbatim(struct state *st, const char *buf, size_t sta |
|
Line 857 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; |
|
|
if ( ! st->parsing || st->paused) |
if ( ! st->parsing || st->paused) |
return; |
return; |
Line 667 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 868 ordinary(struct state *st, const char *buf, size_t sta |
|
* we're in "name - description" format. |
* we're in "name - description" format. |
* To wit, print out a "Nm" and "Nd" in that format. |
* To wit, print out a "Nm" and "Nd" in that format. |
*/ |
*/ |
if (st->isname) { |
if (SECT_NAME == st->sect) { |
for (i = end - 1; i > start; i--) |
for (i = end - 1; i > start; i--) |
if ('-' == buf[i]) |
if ('-' == buf[i]) |
break; |
break; |
Line 678 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 879 ordinary(struct state *st, const char *buf, size_t sta |
|
if ('-' != buf[i]) |
if ('-' != buf[i]) |
break; |
break; |
printf(".Nm "); |
printf(".Nm "); |
formatcodeln(buf, &start, i + 1, 1); |
formatcodeln(st, buf, &start, i + 1, 1); |
putchar('\n'); |
putchar('\n'); |
start = j + 1; |
start = j + 1; |
printf(".Nd "); |
printf(".Nd "); |
formatcodeln(buf, &start, end, 1); |
formatcodeln(st, buf, &start, end, 1); |
putchar('\n'); |
putchar('\n'); |
return; |
return; |
} |
} |
Line 693 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 894 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) { |
while (start < end) { |
/* |
/* |
Line 708 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 910 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 we're in the SYNOPSIS, have square |
|
* brackets indicate that we're opening and |
|
* closing an optional context. |
|
*/ |
|
if (SECT_SYNOPSIS == st->sect && |
|
('[' == buf[start] || |
|
']' == buf[start]) && |
|
dosynopsisop(buf, &last, |
|
&start, end, &opstack)) |
|
continue; |
putchar(last = buf[start++]); |
putchar(last = buf[start++]); |
if ('\\' == last) |
if ('\\' == last) |
putchar('e'); |
putchar('e'); |
Line 723 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 936 ordinary(struct state *st, const char *buf, size_t sta |
|
* 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 (formatcode(buf, &start, end, 0, 0)) { |
if (formatcode(st, buf, &start, end, 0, 0)) { |
putchar(last = '\n'); |
putchar(last = '\n'); |
while (start < end && ' ' == buf[start]) |
while (start < end && ' ' == buf[start]) |
start++; |
start++; |
Line 752 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 965 ordinary(struct state *st, const char *buf, size_t sta |
|
for ( ; start < end; start++) |
for ( ; start < end; start++) |
if (' ' != buf[start] && '\t' != buf[start]) |
if (' ' != buf[start] && '\t' != buf[start]) |
break; |
break; |
} else if (start < end) { |
} |
/* |
|
* Default: print the character. |
|
* Escape initial control characters. |
|
*/ |
|
if ('\n' == last && '.' == buf[start]) |
|
printf("\\&"); |
|
else if ('\n' == last && '\'' == buf[start]) |
|
printf("\\&"); |
|
putchar(last = buf[start++]); |
|
if ('\\' == last) |
|
putchar('e'); |
|
} |
|
} |
} |
|
|
if (last != '\n') |
if (last != '\n') |
Line 828 dofile(const struct args *args, const char *fname, |
|
Line 1029 dofile(const struct args *args, const char *fname, |
|
if (NULL != (cp = strrchr(title, '.'))) { |
if (NULL != (cp = strrchr(title, '.'))) { |
*cp++ = '\0'; |
*cp++ = '\0'; |
if (0 == strcmp(cp, "pm")) |
if (0 == strcmp(cp, "pm")) |
section = "3p"; |
section = PERL_SECTION; |
} |
} |
} |
} |
|
|