version 1.7, 2014/03/23 23:35:59 |
version 1.31, 2014/07/15 19:03:07 |
|
|
#include <string.h> |
#include <string.h> |
#include <unistd.h> |
#include <unistd.h> |
|
|
|
/* |
|
* 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" |
|
|
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 { |
|
const char *fname; /* file being parsed */ |
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? */ |
enum sect sect; /* which section are we in? */ |
int isname; /* are we the NAME section? */ |
|
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 */ |
size_t lpos; /* where in list stack */ |
size_t lpos; /* where in list stack */ |
|
int haspar; /* in paragraph: do we need Pp? */ |
|
int hasnl; /* in text: just started a new line */ |
|
char *outbuf; /* text buffered for output */ |
|
size_t outbufsz; /* allocated size of outbuf */ |
|
size_t outbuflen; /* current length of outbuf */ |
}; |
}; |
|
|
enum fmt { |
enum fmt { |
Line 110 static const char fmts[FMT__MAX] = { |
|
Line 127 static const char fmts[FMT__MAX] = { |
|
|
|
static int last; |
static int last; |
|
|
|
|
|
static void |
|
outbuf_grow(struct state *st, size_t by) |
|
{ |
|
|
|
st->outbufsz += (by / 128 + 1) * 128; |
|
st->outbuf = realloc(st->outbuf, st->outbufsz); |
|
if (NULL == st->outbuf) { |
|
perror(NULL); |
|
exit(EXIT_FAILURE); |
|
} |
|
} |
|
|
|
static void |
|
outbuf_addchar(struct state *st) |
|
{ |
|
|
|
if (st->outbuflen + 2 >= st->outbufsz) |
|
outbuf_grow(st, 1); |
|
st->outbuf[st->outbuflen++] = last; |
|
if ('\\' == last) |
|
st->outbuf[st->outbuflen++] = 'e'; |
|
st->outbuf[st->outbuflen] = '\0'; |
|
} |
|
|
|
static void |
|
outbuf_addstr(struct state *st, const char *str) |
|
{ |
|
size_t slen; |
|
|
|
slen = strlen(str); |
|
if (st->outbuflen + slen >= st->outbufsz) |
|
outbuf_grow(st, slen); |
|
memcpy(st->outbuf + st->outbuflen, str, slen+1); |
|
last = str[slen - 1]; |
|
} |
|
|
|
static void |
|
outbuf_flush(struct state *st) |
|
{ |
|
|
|
if (0 == st->outbuflen) |
|
return; |
|
|
|
fputs(st->outbuf, stdout); |
|
*st->outbuf = '\0'; |
|
st->outbuflen = 0; |
|
st->hasnl = 0; |
|
} |
|
|
|
static void |
|
outbuf_newln(struct state *st) |
|
{ |
|
|
|
if ('\n' == last) |
|
return; |
|
outbuf_flush(st); |
|
putchar('\n'); |
|
last = '\n'; |
|
st->hasnl = 1; |
|
} |
|
|
/* |
/* |
* Given buf[*start] is at the start of an escape name, read til the end |
* Given buf[*start] is at the start of an escape name, read til the end |
* of the escape ('>') then try to do something with it. |
* of the escape ('>') then try to do something with it. |
* Sets start to be one after the '>'. |
* Sets start to be one after the '>'. |
*/ |
*/ |
static void |
static void |
formatescape(const char *buf, size_t *start, size_t end) |
formatescape(struct state *st, const char *buf, size_t *start, size_t end) |
{ |
{ |
char esc[16]; /* no more needed */ |
char esc[16]; /* no more needed */ |
size_t i, max; |
size_t i, max; |
Line 144 formatescape(const char *buf, size_t *start, size_t en |
|
Line 223 formatescape(const char *buf, size_t *start, size_t en |
|
* Just let the rest of them go. |
* Just let the rest of them go. |
*/ |
*/ |
if (0 == strcmp(esc, "lt")) |
if (0 == strcmp(esc, "lt")) |
printf("\\(la"); |
outbuf_addstr(st, "\\(la"); |
else if (0 == strcmp(esc, "gt")) |
else if (0 == strcmp(esc, "gt")) |
printf("\\(ra"); |
outbuf_addstr(st, "\\(ra"); |
else if (0 == strcmp(esc, "vb")) |
else if (0 == strcmp(esc, "vb")) |
printf("\\(ba"); |
outbuf_addstr(st, "\\(ba"); |
else if (0 == strcmp(esc, "sol")) |
else if (0 == strcmp(esc, "sol")) |
printf("\\(sl"); |
outbuf_addstr(st, "\\(sl"); |
|
} |
|
|
|
/* |
|
* Run some heuristics to intuit a link format. |
|
* I set "start" to be the end of the sequence (last right-carrot) so |
|
* that the caller can safely just continue processing. |
|
* If this is just an empty tag, I'll return 0. |
|
*/ |
|
static int |
|
trylink(const char *buf, size_t *start, size_t end, size_t dsz) |
|
{ |
|
size_t linkstart, realend, linkend, |
|
i, j, textsz, stack; |
|
|
|
/* |
|
* Scan to the start of the terminus. |
|
* This function is more or less replicated in the formatcode() |
|
* for null or index formatting codes. |
|
* However, we're slightly different because we might have |
|
* nested escapes we need to ignore. |
|
*/ |
|
stack = 0; |
|
for (linkstart = realend = *start; realend < end; realend++) { |
|
if ('<' == buf[realend]) |
|
stack++; |
|
if ('>' != buf[realend]) |
|
continue; |
|
else if (stack-- > 0) |
|
continue; |
|
if (dsz == 1) |
|
break; |
|
assert(realend > 0); |
|
if (' ' != buf[realend - 1]) |
|
continue; |
|
for (i = realend, j = 0; i < end && j < dsz; j++) |
|
if ('>' != buf[i++]) |
|
break; |
|
if (dsz == j) |
|
break; |
|
} |
|
|
|
/* Ignore stubs. */ |
|
if (realend == end || realend == *start) |
|
return(0); |
|
|
|
/* Set linkend to the end of content. */ |
|
linkend = dsz > 1 ? realend - 1 : realend; |
|
|
|
/* Re-scan to see if we have a title or section. */ |
|
for (textsz = *start; textsz < linkend; textsz++) |
|
if ('|' == buf[textsz] || '/' == buf[textsz]) |
|
break; |
|
|
|
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, |
|
(int)j, &buf[linkstart]); |
else |
else |
return; |
printf("Xr %.*s 1", (int)j, &buf[linkstart]); |
|
|
last = 'a'; |
return(1); |
} |
} |
|
|
/* |
/* |
* Skip space characters. |
* 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 int |
static void |
skipspace(const char *buf, size_t *start, size_t end) |
dosynopsisfl(const char *buf, size_t *start, size_t end) |
{ |
{ |
size_t sv = *start; |
size_t i; |
|
again: |
|
assert(*start + 1 < end); |
|
assert('-' == buf[*start]); |
|
|
while (*start < end && ' ' == buf[*start]) |
if ( ! isalnum((int)buf[*start + 1]) && |
(*start)++; |
'?' != buf[*start + 1] && |
|
'-' != buf[*start + 1]) { |
|
(*start)--; |
|
fputs("Ar ", stdout); |
|
return; |
|
} |
|
|
return(*start > sv); |
(*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; |
|
} |
} |
} |
|
|
/* |
/* |
Line 180 skipspace(const char *buf, size_t *start, size_t end) |
|
Line 440 skipspace(const char *buf, size_t *start, size_t end) |
|
* 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(const char *buf, size_t *start, |
formatcode(struct state *st, const char *buf, 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; |
|
int white; |
|
|
assert(*start + 1 < end); |
assert(*start + 1 < end); |
assert('<' == buf[*start + 1]); |
assert('<' == buf[*start + 1]); |
|
|
/* |
/* |
* First, look up the format code. |
* First, look up the format code. |
* If it's not valid, then exit immediately. |
* If it's not valid, treat it as a NOOP. |
*/ |
*/ |
for (fmt = 0; fmt < FMT__MAX; fmt++) |
for (fmt = 0; fmt < FMT__MAX; fmt++) |
if (buf[*start] == fmts[fmt]) |
if (buf[*start] == fmts[fmt]) |
break; |
break; |
|
|
if (FMT__MAX == fmt) { |
|
putchar(last = buf[(*start)++]); |
|
return(0); |
|
} |
|
|
|
/* |
/* |
* Determine whether we're overriding our delimiter. |
* Determine whether we're overriding our delimiter. |
* According to POD, if we have more than one '<' followed by a |
* According to POD, if we have more than one '<' followed by a |
Line 231 formatcode(const char *buf, size_t *start, |
|
Line 490 formatcode(const char *buf, size_t *start, |
|
* processing for real macros. |
* processing for real macros. |
*/ |
*/ |
if (FMT_ESCAPE == fmt) { |
if (FMT_ESCAPE == fmt) { |
formatescape(buf, start, end); |
formatescape(st, buf, start, end); |
return(0); |
return(0); |
} else if (FMT_NULL == fmt || FMT_INDEX == fmt) { |
} else if (FMT_NULL == fmt || FMT_INDEX == fmt) { |
/* |
/* |
Line 255 formatcode(const char *buf, size_t *start, |
|
Line 514 formatcode(const char *buf, size_t *start, |
|
(*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 262 formatcode(const char *buf, size_t *start, |
|
Line 528 formatcode(const char *buf, size_t *start, |
|
* Check whether we're supposed to print macro stuff (this is |
* Check whether we're supposed to print macro stuff (this is |
* suppressed in, e.g., "Nm" and "Sh" macros). |
* suppressed in, e.g., "Nm" and "Sh" macros). |
*/ |
*/ |
if ( ! nomacro) { |
if (FMT__MAX != fmt && !nomacro) { |
|
white = ' ' == last || '\n' == last || |
|
' ' == buf[*start]; |
|
|
/* |
/* |
* Print out the macro describing this format code. |
* If we are on a text line and there is no |
* If we're not "reentrant" (not yet on a macro line) |
* whitespace before our content, we have to make |
* then print a newline, if necessary, and the macro |
* the previous word a prefix to the macro line. |
* indicator. |
|
* Otherwise, offset us with a space. |
|
*/ |
*/ |
if ( ! reentrant) { |
|
|
if ( ! white && ! reentrant) { |
|
if ( ! st->hasnl) |
|
putchar('\n'); |
|
printf(".Pf "); |
|
} |
|
|
|
outbuf_flush(st); |
|
|
|
/* Whitespace is easier to suppress on macro lines. */ |
|
|
|
if ( ! white && reentrant) |
|
printf(" Ns"); |
|
|
|
/* Unless we are on a macro line, start one. */ |
|
|
|
if (white && ! reentrant) { |
if (last != '\n') |
if (last != '\n') |
putchar('\n'); |
putchar('\n'); |
putchar('.'); |
putchar('.'); |
} else |
} else |
putchar(' '); |
putchar(' '); |
|
|
/* |
|
* If we don't have whitespace before us (and none after |
|
* the opening delimiter), then suppress macro |
|
* whitespace with Pf. |
|
*/ |
|
if (' ' != last && '\n' != last && ' ' != buf[*start]) |
|
printf("Pf "); |
|
|
|
|
/* Print the macro corresponding to this format code. */ |
|
|
switch (fmt) { |
switch (fmt) { |
case (FMT_ITALIC): |
case (FMT_ITALIC): |
printf("Em "); |
printf("Em "); |
break; |
break; |
case (FMT_BOLD): |
case (FMT_BOLD): |
printf("Sy "); |
if (SECT_SYNOPSIS == st->sect) { |
|
if (1 == dsz && '-' == buf[*start]) |
|
dosynopsisfl(buf, start, end); |
|
else if (0 == pos) |
|
printf("Nm "); |
|
else |
|
printf("Ar "); |
|
break; |
|
} |
|
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): |
printf("Lk "); |
/* Try to link; use "No" if it's empty. */ |
|
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: |
abort(); |
abort(); |
} |
} |
} |
} else |
|
outbuf_flush(st); |
|
|
/* |
/* |
* Process until we reach the end marker (e.g., '>') or until we |
* Process until we reach the end marker (e.g., '>') or until we |
Line 338 formatcode(const char *buf, size_t *start, |
|
Line 631 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, 1); |
continue; |
continue; |
} |
} |
|
|
Line 363 formatcode(const char *buf, size_t *start, |
|
Line 656 formatcode(const char *buf, size_t *start, |
|
else |
else |
putchar(last = buf[*start]); |
putchar(last = buf[*start]); |
|
|
|
/* Protect against character escapes. */ |
|
if ('\\' == last) |
|
putchar('e'); |
|
|
(*start)++; |
(*start)++; |
|
|
if (' ' == last) |
if (' ' == last) |
Line 370 formatcode(const char *buf, size_t *start, |
|
Line 667 formatcode(const char *buf, size_t *start, |
|
(*start)++; |
(*start)++; |
} |
} |
|
|
|
if (FMT__MAX == fmt) |
|
return(0); |
|
|
if ( ! nomacro && FMT_CODE == fmt) |
if ( ! nomacro && FMT_CODE == fmt) |
printf(" Qc "); |
printf(" Qc "); |
|
|
Line 389 formatcode(const char *buf, size_t *start, |
|
Line 689 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, 1); |
continue; |
continue; |
} |
} |
/* |
/* |
Line 412 formatcodeln(const char *buf, size_t *start, size_t en |
|
Line 713 formatcodeln(const char *buf, size_t *start, size_t en |
|
' ' == buf[*start + 2])) |
' ' == buf[*start + 2])) |
printf("\\&"); |
printf("\\&"); |
|
|
if ('\n' != buf[*start]) |
if ('\n' == buf[*start]) |
putchar(last = buf[*start]); |
|
else |
|
putchar(last = ' '); |
putchar(last = ' '); |
|
else |
|
putchar(last = buf[*start]); |
|
|
|
/* Protect against character escapes. */ |
|
if ('\\' == last) |
|
putchar('e'); |
|
|
(*start)++; |
(*start)++; |
} |
} |
} |
} |
Line 471 command(struct state *st, const char *buf, size_t star |
|
Line 777 command(struct state *st, const char *buf, size_t star |
|
return; |
return; |
|
|
start += csz; |
start += csz; |
skipspace(buf, &start, end); |
while (start < end && ' ' == buf[start]) |
|
start++; |
|
|
len = end - start; |
len = end - start; |
|
|
if (st->paused) { |
if (st->paused) { |
Line 488 command(struct state *st, const char *buf, size_t star |
|
Line 796 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) { |
putchar('\n'); |
if (0 == memcmp(&buf[start], "SYNOPSIS", 8)) |
|
st->sect = SECT_SYNOPSIS; |
|
} |
|
formatcodeln(st, buf, &start, end, 1); |
|
putchar(last = '\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(last = '\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(last = '\n'); |
puts(".Pp"); |
puts(".Pp"); |
st->haspar = 1; |
st->haspar = 1; |
break; |
break; |
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(last = '\n'); |
puts(".Pp"); |
puts(".Pp"); |
st->haspar = 1; |
st->haspar = 1; |
break; |
break; |
Line 567 command(struct state *st, const char *buf, size_t star |
|
Line 879 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(last = '\n'); |
break; |
break; |
case (LIST_ENUM): |
case (LIST_ENUM): |
/* FALLTHROUGH */ |
/* FALLTHROUGH */ |
Line 624 command(struct state *st, const char *buf, size_t star |
|
Line 936 command(struct state *st, const char *buf, size_t star |
|
static void |
static void |
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) |
{ |
{ |
size_t sv = start; |
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"); |
while (start < end) { |
for (last = ' '; start < end; start++) { |
if (start > sv && '\n' == buf[start - 1]) |
/* |
|
* Handle accidental macros (newline starting with |
|
* control character) and escapes. |
|
*/ |
|
if ('\n' == last) |
if ('.' == buf[start] || '\'' == buf[start]) |
if ('.' == buf[start] || '\'' == buf[start]) |
printf("\\&"); |
printf("\\&"); |
putchar(buf[start++]); |
putchar(last = buf[start]); |
|
if ('\\' == buf[start]) |
|
printf("e"); |
} |
} |
putchar('\n'); |
putchar(last = '\n'); |
puts(".Ed"); |
puts(".Ed"); |
} |
} |
|
|
/* |
/* |
|
* 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, 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(last = '\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 651 verbatim(struct state *st, const char *buf, size_t sta |
|
Line 1093 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 662 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1105 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 - 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 672 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1115 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(buf, &start, i + 1, 1); |
|
putchar('\n'); |
|
start = j + 1; |
start = j + 1; |
printf(".Nd "); |
while (start < end && ' ' == buf[start]) |
formatcodeln(buf, &start, end, 1); |
start++; |
putchar('\n'); |
fputs(".Nd ", stdout); |
|
formatcodeln(st, buf, &start, end, 1); |
|
putchar(last = '\n'); |
return; |
return; |
} |
} |
} |
} |
Line 687 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1130 ordinary(struct state *st, const char *buf, size_t sta |
|
puts(".Pp"); |
puts(".Pp"); |
|
|
st->haspar = 0; |
st->haspar = 0; |
|
st->hasnl = 1; |
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 700 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1145 ordinary(struct state *st, const char *buf, size_t sta |
|
else if ('\n' == buf[start]) |
else if ('\n' == buf[start]) |
break; |
break; |
else if ('\n' == last && '.' == buf[start]) |
else if ('\n' == last && '.' == buf[start]) |
printf("\\&"); |
outbuf_addstr(st, "\\&"); |
else if ('\n' == last && '\'' == buf[start]) |
else if ('\n' == last && '\'' == buf[start]) |
printf("\\&"); |
outbuf_addstr(st, "\\&"); |
putchar(last = buf[start++]); |
/* |
|
* 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, &start, end, &opstack)) |
|
continue; |
|
last = buf[start++]; |
|
if (' ' == last) { |
|
outbuf_flush(st); |
|
putchar(' '); |
|
} else |
|
outbuf_addchar(st); |
} |
} |
|
|
if (start < end - 1 && '<' == buf[start + 1]) { |
if (start < end - 1 && '<' == buf[start + 1]) { |
/* |
if (formatcode(st, buf, &start, end, 0, 0, seq)) { |
* We've encountered a format code. |
/* |
* This is going to trigger a macro no matter |
* Let mdoc(7) handle trailing punctuation. |
* what, so print a newline now. |
* XXX Some punctuation characters |
* Then print the (possibly nested) macros and |
* are not handled yet. |
* following that, a newline. |
*/ |
*/ |
if ((start == end - 1 || |
if (formatcode(buf, &start, end, 0, 0)) { |
(start < end - 1 && |
|
(' ' == buf[start + 1] || |
|
'\n' == buf[start + 1]))) && |
|
('.' == buf[start] || |
|
',' == buf[start])) { |
|
putchar(' '); |
|
putchar(buf[start++]); |
|
} |
|
/* End the macro line. */ |
putchar(last = '\n'); |
putchar(last = '\n'); |
|
st->hasnl = 1; |
|
/* |
|
* Consume all whitespace |
|
* so we don't accidentally start |
|
* an implicit literal line. |
|
*/ |
while (start < end && ' ' == buf[start]) |
while (start < end && ' ' == buf[start]) |
start++; |
start++; |
} |
} |
} else if (start < end && '\n' == buf[start]) { |
} else if (start < end && '\n' == buf[start]) { |
/* |
outbuf_newln(st); |
* Print the newline only if we haven't already |
|
* printed a newline. |
|
*/ |
|
if (last != '\n') |
|
putchar(last = buf[start]); |
|
if (++start >= end) |
if (++start >= end) |
continue; |
continue; |
/* |
/* |
Line 736 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1205 ordinary(struct state *st, const char *buf, size_t sta |
|
* have a macro subsequent it, which may be |
* have a macro subsequent it, which may be |
* possible if we have an escape next. |
* possible if we have an escape next. |
*/ |
*/ |
if (' ' == buf[start] || '\t' == buf[start]) { |
if (' ' == buf[start] || '\t' == buf[start]) |
puts(".br"); |
puts(".br"); |
last = '\n'; |
|
} |
|
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++]); |
|
} |
|
} |
} |
|
outbuf_newln(st); |
if (last != '\n') |
|
putchar('\n'); |
|
} |
} |
|
|
/* |
/* |
|
|
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 812 dofile(const struct args *args, const char *fname, |
|
Line 1283 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 = "3p"; |
|
} |
|
} |
|
|
|
/* Date. Or the given "tm" if not supplied. */ |
/* Date. Or the given "tm" if not supplied. */ |
|
|
Line 882 readfile(const struct args *args, const char *fname) |
|
Line 1348 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 989 main(int argc, char *argv[]) |
|
Line 1453 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 999 main(int argc, char *argv[]) |
|
Line 1463 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); |
} |
} |