version 1.24, 2014/04/07 11:58:35 |
version 1.38, 2015/02/13 12:40:54 |
|
|
/* $Id$ */ |
/* $Id$ */ |
/* |
/* |
* Copyright (c) 2014 Kristaps Dzonsons <kristaps@bsd.lv> |
* Copyright (c) 2014 Kristaps Dzonsons <kristaps@bsd.lv> |
|
* Copyright (c) 2014, 2015 Ingo Schwarze <schwarze@openbsd.org> |
* |
* |
* Permission to use, copy, modify, and distribute this software for any |
* Permission to use, copy, modify, and distribute this software for any |
* purpose with or without fee is hereby granted, provided that the above |
* purpose with or without fee is hereby granted, provided that the above |
|
|
#include <string.h> |
#include <string.h> |
#include <unistd.h> |
#include <unistd.h> |
|
|
|
#include "dict.h" |
|
|
/* |
/* |
* In what section can we find Perl module manuals? |
* In what section can we find Perl module manuals? |
* Sometimes (Mac OS X) it's 3pm, sometimes (OpenBSD, etc.) 3p. |
* Sometimes (Mac OS X) it's 3pm, sometimes (OpenBSD, etc.) 3p. |
|
|
SECT_SYNOPSIS, /* SYNOPSIS section */ |
SECT_SYNOPSIS, /* SYNOPSIS section */ |
}; |
}; |
|
|
|
enum outstate { |
|
OUST_NL = 0, /* just started a new output line */ |
|
OUST_TXT, /* text line output in progress */ |
|
OUST_MAC /* macro line output in progress */ |
|
}; |
|
|
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? */ |
enum sect sect; /* which section are we in? */ |
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? */ |
|
enum outstate oust; /* state of the mdoc output stream */ |
|
int wantws; /* let mdoc(7) output whitespace here */ |
|
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 123 static const char fmts[FMT__MAX] = { |
|
Line 137 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'; |
|
st->wantws = 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); |
|
st->outbuflen += slen; |
|
last = str[slen - 1]; |
|
st->wantws = 0; |
|
} |
|
|
|
static void |
|
outbuf_flush(struct state *st) |
|
{ |
|
|
|
if (0 == st->outbuflen) |
|
return; |
|
|
|
fputs(st->outbuf, stdout); |
|
*st->outbuf = '\0'; |
|
st->outbuflen = 0; |
|
|
|
if (OUST_NL == st->oust) |
|
st->oust = OUST_TXT; |
|
} |
|
|
|
static void |
|
mdoc_newln(struct state *st) |
|
{ |
|
|
|
if (OUST_NL == st->oust) |
|
return; |
|
|
|
putchar('\n'); |
|
last = '\n'; |
|
st->oust = OUST_NL; |
|
st->wantws = 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 '>'. |
|
* |
|
* This function does not care about output modes, |
|
* it merely appends text to the output buffer, |
|
* which can then be used in any mode. |
*/ |
*/ |
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 157 formatescape(const char *buf, size_t *start, size_t en |
|
Line 243 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, "verbar")) |
printf("\\(ba"); |
outbuf_addstr(st, "\\(ba"); |
else if (0 == strcmp(esc, "sol")) |
else if (0 == strcmp(esc, "sol")) |
printf("\\(sl"); |
outbuf_addstr(st, "\\(sl"); |
else |
|
return; |
|
|
|
last = 'a'; |
|
} |
} |
|
|
/* |
/* |
Line 175 formatescape(const char *buf, size_t *start, size_t en |
|
Line 257 formatescape(const char *buf, size_t *start, size_t en |
|
* I set "start" to be the end of the sequence (last right-carrot) so |
* I set "start" to be the end of the sequence (last right-carrot) so |
* that the caller can safely just continue processing. |
* that the caller can safely just continue processing. |
* If this is just an empty tag, I'll return 0. |
* If this is just an empty tag, I'll return 0. |
|
* |
|
* Always operates in OUST_MAC mode. |
|
* Mode handling is done by the caller. |
*/ |
*/ |
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 linkstart, realend, linkend, |
size_t linkstart, realend, linkend, |
i, j, textsz, stack; |
i, j, textsz, stack; |
const char *text; |
|
|
|
/* |
/* |
* Scan to the start of the terminus. |
* Scan to the start of the terminus. |
Line 218 trylink(const char *buf, size_t *start, size_t end, si |
|
Line 302 trylink(const char *buf, size_t *start, size_t end, si |
|
linkend = dsz > 1 ? realend - 1 : realend; |
linkend = dsz > 1 ? realend - 1 : realend; |
|
|
/* Re-scan to see if we have a title or section. */ |
/* Re-scan to see if we have a title or section. */ |
text = &buf[*start]; |
|
for (textsz = *start; textsz < linkend; textsz++) |
for (textsz = *start; textsz < linkend; textsz++) |
if ('|' == buf[textsz] || '/' == buf[textsz]) |
if ('|' == buf[textsz] || '/' == buf[textsz]) |
break; |
break; |
Line 311 trylink(const char *buf, size_t *start, size_t end, si |
|
Line 394 trylink(const char *buf, size_t *start, size_t end, si |
|
* Our flag might be followed by an argument, so make sure that we're |
* Our flag might be followed by an argument, so make sure that we're |
* accounting for that, too. |
* accounting for that, too. |
* If we don't have a flag at all, however, then assume we're an "Ar". |
* If we don't have a flag at all, however, then assume we're an "Ar". |
|
* |
|
* Always operates in OUST_MAC mode. |
|
* Mode handlinf is done by the caller. |
*/ |
*/ |
static void |
static void |
dosynopsisfl(const char *buf, size_t *start, size_t end) |
dosynopsisfl(const char *buf, size_t *start, size_t end) |
|
|
* 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 |
* the end of matched production. |
* the end of matched production. |
* If "reentrant", then we're being called after a macro has already |
|
* 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 |
* "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"), |
* as the first format code on a line (for decoration as an "Nm"), |
* non-zero otherwise. |
* non-zero otherwise. |
* Return whether we've printed a macro or not--in other words, whether |
* |
* this should trigger a subsequent newline (this should be ignored when |
* Output mode handling is most complicated here. |
* reentrant). |
* We may enter in any mode. |
|
* We usually exit in OUST_MAC mode, except when |
|
* entering without OUST_MAC and the code is invalid. |
*/ |
*/ |
static int |
static int |
formatcode(struct state *st, const char *buf, size_t *start, |
formatcode(struct state *st, const char *buf, size_t *start, |
size_t end, int reentrant, int nomacro, int pos) |
size_t end, int nomacro, int pos) |
{ |
{ |
enum fmt fmt; |
enum fmt fmt; |
size_t i, j, dsz; |
size_t i, j, dsz; |
Line 399 formatcode(struct state *st, const char *buf, size_t * |
|
Line 485 formatcode(struct state *st, const char *buf, size_t * |
|
|
|
/* |
/* |
* 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)++]); |
|
if ('\\' == last) |
|
putchar('e'); |
|
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 436 formatcode(struct state *st, const char *buf, size_t * |
|
Line 515 formatcode(struct state *st, const char *buf, size_t * |
|
* 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 474 formatcode(struct state *st, const char *buf, size_t * |
|
Line 553 formatcode(struct state *st, const char *buf, size_t * |
|
* 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) { |
|
|
/* |
/* |
* Print out the macro describing this format code. |
* We may already have wantws if there was whitespace |
* If we're not "reentrant" (not yet on a macro line) |
* before the code ("text B<text"), but initial |
* then print a newline, if necessary, and the macro |
* whitespace inside our scope ("textB< text") |
* indicator. |
* allows to break at this point as well. |
* Otherwise, offset us with a space. |
|
*/ |
*/ |
if ( ! reentrant) { |
|
if (last != '\n') |
st->wantws |= ' ' == buf[*start]; |
|
|
|
/* |
|
* If we are on a text line and there is no |
|
* whitespace before our content, we have to make |
|
* the previous word a prefix to the macro line. |
|
* In the following, mdoc_newln() must not be used |
|
* lest we clobber out output state. |
|
*/ |
|
|
|
if (OUST_MAC != st->oust && !st->wantws) { |
|
if (OUST_NL != st->oust) |
putchar('\n'); |
putchar('\n'); |
|
printf(".Pf "); |
|
} |
|
|
|
outbuf_flush(st); |
|
|
|
/* Whitespace is easier to suppress on macro lines. */ |
|
|
|
if (OUST_MAC == st->oust && !st->wantws) |
|
printf(" Ns "); |
|
|
|
/* Unless we are on a macro line, start one. */ |
|
|
|
if (OUST_MAC != st->oust && st->wantws) { |
|
if (OUST_NL != st->oust) |
|
putchar('\n'); |
putchar('.'); |
putchar('.'); |
} else |
} else |
putchar(' '); |
putchar(' '); |
|
|
/* |
/* |
* If we don't have whitespace before us (and none after |
* Print the macro corresponding to this format code, |
* the opening delimiter), then suppress macro |
* and update the output state afterwards. |
* whitespace with Pf. |
|
*/ |
*/ |
if (' ' != last && '\n' != last && ' ' != buf[*start]) |
|
printf("Pf "); |
|
|
|
switch (fmt) { |
switch (fmt) { |
case (FMT_ITALIC): |
case (FMT_ITALIC): |
Line 511 formatcode(struct state *st, const char *buf, size_t * |
|
Line 613 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 "); |
|
break; |
|
} |
|
i = 0; |
|
while (isalnum((unsigned char)buf[*start + i]) || |
|
'_' == buf[*start + i]) |
|
i++; |
|
if (i && MDOC_Fa == dict_get(buf + *start, i)) |
|
printf("Fa "); |
|
else |
|
printf("Sy "); |
break; |
break; |
case (FMT_CODE): |
case (FMT_CODE): |
printf("Qo Li "); |
printf("Qo Li "); |
Line 530 formatcode(struct state *st, const char *buf, size_t * |
|
Line 645 formatcode(struct state *st, const char *buf, size_t * |
|
default: |
default: |
abort(); |
abort(); |
} |
} |
} |
st->oust = OUST_MAC; |
|
st->wantws = 1; |
|
} 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 559 formatcode(struct state *st, const char *buf, size_t * |
|
Line 677 formatcode(struct state *st, const char *buf, size_t * |
|
break; |
break; |
} |
} |
} |
} |
if (*start + 1 < end && '<' == buf[*start + 1]) { |
if (*start + 1 < end && '<' == buf[*start + 1] && |
formatcode(st, buf, start, end, 1, nomacro, 1); |
'A' <= buf[*start] && 'Z' >= buf[*start]) { |
|
formatcode(st, buf, start, end, nomacro, 1); |
continue; |
continue; |
} |
} |
|
|
/* |
/* Suppress newlines and multiple spaces. */ |
* Make sure that any macro-like words (or |
|
* really any word starting with a capital |
|
* letter) is assumed to be a macro that must be |
|
* escaped. |
|
* This matches "Xx " and "XxEOLN". |
|
*/ |
|
if ((' ' == last || '\n' == last) && |
|
end - *start > 1 && |
|
isupper((int)buf[*start]) && |
|
islower((int)buf[*start + 1]) && |
|
(end - *start == 2 || |
|
' ' == buf[*start + 2])) |
|
printf("\\&"); |
|
|
|
/* Suppress newline. */ |
last = buf[(*start)++]; |
if ('\n' == buf[*start]) |
if (' ' == last || '\n' == last) { |
putchar(last = ' '); |
putchar(' '); |
else |
while (*start < end && ' ' == buf[*start]) |
putchar(last = buf[*start]); |
(*start)++; |
|
continue; |
|
} |
|
|
|
if (OUST_MAC == st->oust && FMT__MAX != fmt) { |
|
if ( ! st->wantws) { |
|
printf(" Ns "); |
|
st->wantws = 1; |
|
} |
|
|
|
/* |
|
* Escape macro-like words. |
|
* This matches "Xx " and "XxEOLN". |
|
*/ |
|
|
|
if (end - *start > 0 && |
|
isupper((unsigned char)last) && |
|
islower((unsigned char)buf[*start]) && |
|
(end - *start == 1 || |
|
' ' == buf[*start + 1] || |
|
'>' == buf[*start + 1])) |
|
printf("\\&"); |
|
} |
|
|
|
putchar(last); |
|
|
/* Protect against character escapes. */ |
/* Protect against character escapes. */ |
|
|
if ('\\' == last) |
if ('\\' == last) |
putchar('e'); |
putchar('e'); |
|
|
(*start)++; |
|
|
|
if (' ' == last) |
|
while (*start < end && ' ' == buf[*start]) |
|
(*start)++; |
|
} |
} |
|
|
|
if (FMT__MAX == fmt) |
|
return(0); |
|
|
if ( ! nomacro && FMT_CODE == fmt) |
if ( ! nomacro && FMT_CODE == fmt) |
printf(" Qc "); |
printf(" Qc "); |
|
|
/* |
st->wantws = ' ' == last; |
* We're now subsequent the format code. |
|
* If there isn't a space (or newline) here, and we haven't just |
|
* printed a space, then suppress space. |
|
*/ |
|
if ( ! nomacro && ' ' != last) |
|
if (' ' != buf[*start] && '\n' != buf[*start]) |
|
printf(" Ns "); |
|
|
|
return(1); |
return(1); |
} |
} |
|
|
/* |
/* |
* Calls formatcode() til the end of a paragraph. |
* Calls formatcode() til the end of a paragraph. |
|
* Goes to OUST_MAC mode and stays there when returning, |
|
* such that the caller can add arguments to the macro line |
|
* before closing it out. |
*/ |
*/ |
static void |
static void |
formatcodeln(struct state *st, const char *buf, |
formatcodeln(struct state *st, const char *linemac, |
size_t *start, size_t end, int nomacro) |
const char *buf, size_t *start, size_t end, int nomacro) |
{ |
{ |
|
int gotmacro, wantws; |
|
|
last = ' '; |
assert(OUST_NL == st->oust); |
|
assert(st->wantws); |
|
printf(".%s ", linemac); |
|
st->oust = OUST_MAC; |
|
|
|
gotmacro = 0; |
while (*start < end) { |
while (*start < end) { |
if (*start + 1 < end && '<' == buf[*start + 1]) { |
wantws = ' ' == buf[*start] || '\n' == buf[*start]; |
formatcode(st, buf, start, end, 1, nomacro, 1); |
if (wantws) { |
|
last = ' '; |
|
do { |
|
(*start)++; |
|
} while (*start < end && ' ' == buf[*start]); |
|
} |
|
|
|
if (*start + 1 < end && '<' == buf[*start + 1] && |
|
'A' <= buf[*start] && 'Z' >= buf[*start]) { |
|
st->wantws |= wantws; |
|
gotmacro = formatcode(st, buf, |
|
start, end, nomacro, 1); |
continue; |
continue; |
} |
} |
|
|
|
if (gotmacro) { |
|
if (*start < end || st->outbuflen) { |
|
if (st->wantws || |
|
(wantws && !st->outbuflen)) |
|
printf(" No "); |
|
else |
|
printf(" Ns "); |
|
} |
|
gotmacro = 0; |
|
} |
|
outbuf_flush(st); |
|
st->wantws = wantws; |
|
|
|
if (*start >= end) |
|
break; |
|
|
|
if (st->wantws) { |
|
putchar(' '); |
|
st->wantws = 0; |
|
} |
|
|
/* |
/* |
* Since we're already on a macro line, we want to make |
* Since we're already on a macro line, we want to make |
* sure that we don't inadvertently invoke a macro. |
* sure that we don't inadvertently invoke a macro. |
Line 633 formatcodeln(struct state *st, const char *buf, |
|
Line 795 formatcodeln(struct state *st, const char *buf, |
|
* something that needn't be escaped. |
* something that needn't be escaped. |
*/ |
*/ |
if (' ' == last && end - *start > 1 && |
if (' ' == last && end - *start > 1 && |
isupper((int)buf[*start]) && |
isupper((unsigned char)buf[*start]) && |
islower((int)buf[*start + 1]) && |
islower((unsigned char)buf[*start + 1]) && |
(end - *start == 2 || |
(end - *start == 2 || ' ' == buf[*start + 2])) |
' ' == buf[*start + 2])) |
|
printf("\\&"); |
printf("\\&"); |
|
|
if ('\n' == buf[*start]) |
putchar(last = buf[*start]); |
putchar(last = ' '); |
|
else |
|
putchar(last = buf[*start]); |
|
|
|
/* Protect against character escapes. */ |
/* Protect against character escapes. */ |
|
|
if ('\\' == last) |
if ('\\' == last) |
putchar('e'); |
putchar('e'); |
|
|
Line 678 listguess(const char *buf, size_t start, size_t end) |
|
Line 837 listguess(const char *buf, size_t start, size_t end) |
|
* A command paragraph, as noted in the perlpod manual, just indicates |
* A command paragraph, as noted in the perlpod manual, just indicates |
* that we should do something, optionally with some text to print as |
* that we should do something, optionally with some text to print as |
* well. |
* well. |
|
* From the perspective of external callers, |
|
* always stays in OUST_NL/wantws mode, |
|
* but its children do use OUST_MAC. |
*/ |
*/ |
static void |
static void |
command(struct state *st, const char *buf, size_t start, size_t end) |
command(struct state *st, const char *buf, size_t start, size_t end) |
Line 721 command(struct state *st, const char *buf, size_t star |
|
Line 883 command(struct state *st, const char *buf, size_t star |
|
* The behaviour of head= follows from a quick glance at |
* The behaviour of head= follows from a quick glance at |
* how pod2man handles it. |
* how pod2man handles it. |
*/ |
*/ |
printf(".Sh "); |
|
st->sect = SECT_NONE; |
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)) |
Line 730 command(struct state *st, const char *buf, size_t star |
|
Line 891 command(struct state *st, const char *buf, size_t star |
|
if (0 == memcmp(&buf[start], "SYNOPSIS", 8)) |
if (0 == memcmp(&buf[start], "SYNOPSIS", 8)) |
st->sect = SECT_SYNOPSIS; |
st->sect = SECT_SYNOPSIS; |
} |
} |
formatcodeln(st, buf, &start, end, 1); |
formatcodeln(st, "Sh", buf, &start, end, 1); |
putchar('\n'); |
mdoc_newln(st); |
st->haspar = 1; |
st->haspar = 1; |
break; |
break; |
case (CMD_HEAD2): |
case (CMD_HEAD2): |
printf(".Ss "); |
formatcodeln(st, "Ss", buf, &start, end, 1); |
formatcodeln(st, buf, &start, end, 1); |
mdoc_newln(st); |
putchar('\n'); |
|
st->haspar = 1; |
st->haspar = 1; |
break; |
break; |
case (CMD_HEAD3): |
case (CMD_HEAD3): |
puts(".Pp"); |
puts(".Pp"); |
printf(".Em "); |
formatcodeln(st, "Em", buf, &start, end, 0); |
formatcodeln(st, buf, &start, end, 0); |
mdoc_newln(st); |
putchar('\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 "); |
formatcodeln(st, "No", buf, &start, end, 0); |
formatcodeln(st, buf, &start, end, 0); |
mdoc_newln(st); |
putchar('\n'); |
|
puts(".Pp"); |
puts(".Pp"); |
st->haspar = 1; |
st->haspar = 1; |
break; |
break; |
Line 804 command(struct state *st, const char *buf, size_t star |
|
Line 962 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 "); |
formatcodeln(st, "It", buf, &start, end, 0); |
formatcodeln(st, buf, &start, end, 0); |
mdoc_newln(st); |
putchar('\n'); |
|
break; |
break; |
case (LIST_ENUM): |
case (LIST_ENUM): |
/* FALLTHROUGH */ |
/* FALLTHROUGH */ |
Line 858 command(struct state *st, const char *buf, size_t star |
|
Line 1015 command(struct state *st, const char *buf, size_t star |
|
|
|
/* |
/* |
* Just pump out the line in a verbatim block. |
* Just pump out the line in a verbatim block. |
|
* From the perspective of external callers, |
|
* always stays in OUST_NL/wantws mode. |
*/ |
*/ |
static void |
static void |
verbatim(struct state *st, const char *buf, size_t start, size_t end) |
verbatim(struct state *st, char *buf, size_t start, size_t end) |
{ |
{ |
int last; |
size_t i, ift, ifo, ifa, ifc, inl; |
size_t i; |
char *cp, *cp2; |
|
int nopen; |
|
|
if ( ! st->parsing || st->paused) |
if ( ! st->parsing || st->paused || start == end) |
return; |
return; |
again: |
again: |
/* |
/* |
|
|
*/ |
*/ |
if (SECT_SYNOPSIS == st->sect) { |
if (SECT_SYNOPSIS == st->sect) { |
i = start; |
i = start; |
for (i = start; i < end && ' ' == buf[i]; i++) |
while (i < end && buf[i] == ' ') |
/* Spin. */ ; |
i++; |
if (i == end) |
if (i == end) |
return; |
return; |
|
|
/* We're an include block! */ |
/* We're an include block! */ |
if (end - i > 10 && |
if (end - i > 10 && |
0 == memcmp(&buf[i], "#include <", 10)) { |
0 == memcmp(&buf[i], "#include <", 10)) { |
|
|
goto again; |
goto again; |
return; |
return; |
} |
} |
|
|
|
/* Parse function declaration. */ |
|
ifo = ifa = ifc = 0; |
|
inl = end; |
|
nopen = 0; |
|
for (ift = i; i < end; i++) { |
|
if (ifc) { |
|
if (buf[i] != '\n') |
|
continue; |
|
inl = i; |
|
break; |
|
} |
|
switch (buf[i]) { |
|
case ' ': |
|
if ( ! ifa) |
|
ifo = i; |
|
break; |
|
case '(': |
|
if (ifo) { |
|
nopen++; |
|
if ( ! ifa) |
|
ifa = i; |
|
} else |
|
i = end; |
|
break; |
|
case ')': |
|
switch (nopen) { |
|
case 0: |
|
i = end; |
|
break; |
|
case 1: |
|
ifc = i; |
|
break; |
|
default: |
|
nopen--; |
|
break; |
|
} |
|
break; |
|
default: |
|
break; |
|
} |
|
} |
|
|
|
/* Encode function declaration. */ |
|
if (ifc) { |
|
for (i = ifa; i < ifc; i++) |
|
if (buf[i] == '\n') |
|
buf[i] = ' '; |
|
buf[ifo++] = '\0'; |
|
printf(".Ft %s", buf + ift); |
|
if (buf[ifo] == '*') { |
|
fputs(" *", stdout); |
|
ifo++; |
|
} |
|
putchar('\n'); |
|
buf[ifa++] = '\0'; |
|
printf(".Fo %s\n", buf + ifo); |
|
dict_put(buf + ifo, MDOC_Fo); |
|
buf[ifc++] = '\0'; |
|
for (;;) { |
|
cp = strchr(buf + ifa, ','); |
|
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, MDOC_Fa); |
|
printf(".Fa \"%s\"\n", buf + ifa); |
|
if (cp == NULL) |
|
break; |
|
while (*cp == ' ') |
|
cp++; |
|
ifa = cp - buf; |
|
} |
|
puts(".Fc"); |
|
if (buf[ifc] == ';') |
|
ifc++; |
|
if (ifc < inl) { |
|
buf[inl] = '\0'; |
|
puts(buf + ifc); |
|
} |
|
start = inl + 1; |
|
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++) { |
/* |
/* |
|
|
if ('\\' == buf[start]) |
if ('\\' == buf[start]) |
printf("e"); |
printf("e"); |
} |
} |
putchar('\n'); |
putchar(last = '\n'); |
puts(".Ed"); |
puts(".Ed"); |
} |
} |
|
|
Line 947 hasmatch(const char *buf, size_t start, size_t end) |
|
Line 1196 hasmatch(const char *buf, size_t start, size_t end) |
|
* If we're an ending bracket, see if we have a stack already. |
* If we're an ending bracket, see if we have a stack already. |
*/ |
*/ |
static int |
static int |
dosynopsisop(const char *buf, int *last, |
dosynopsisop(struct state *st, const char *buf, |
size_t *start, size_t end, size_t *opstack) |
size_t *start, size_t end, size_t *opstack) |
{ |
{ |
|
|
assert('[' == buf[*start] || ']' == buf[*start]); |
assert('[' == buf[*start] || ']' == buf[*start]); |
|
|
if ('[' == buf[*start] && hasmatch(buf, *start + 1, end)) { |
if ('[' == buf[*start] && hasmatch(buf, *start + 1, end)) { |
if ('\n' != *last) |
mdoc_newln(st); |
putchar('\n'); |
|
puts(".Oo"); |
puts(".Oo"); |
(*opstack)++; |
(*opstack)++; |
} else if ('[' == buf[*start]) |
} else if ('[' == buf[*start]) |
return(0); |
return(0); |
|
|
if (']' == buf[*start] && *opstack > 0) { |
if (']' == buf[*start] && *opstack > 0) { |
if ('\n' != *last) |
mdoc_newln(st); |
putchar('\n'); |
|
puts(".Oc"); |
puts(".Oc"); |
(*opstack)--; |
(*opstack)--; |
} else if (']' == buf[*start]) |
} else if (']' == buf[*start]) |
return(0); |
return(0); |
|
|
(*start)++; |
(*start)++; |
*last = '\n'; |
last = '\n'; |
while (' ' == buf[*start]) |
while (' ' == buf[*start]) |
(*start)++; |
(*start)++; |
return(1); |
return(1); |
Line 978 dosynopsisop(const char *buf, int *last, |
|
Line 1225 dosynopsisop(const char *buf, int *last, |
|
|
|
/* |
/* |
* Format multiple "Nm" manpage names in the NAME section. |
* Format multiple "Nm" manpage names in the NAME section. |
|
* From the perspective of external callers, |
|
* always stays in OUST_NL/wantws mode, |
|
* but its children do use OUST_MAC. |
*/ |
*/ |
static void |
static void |
donamenm(struct state *st, const char *buf, size_t *start, size_t end) |
donamenm(struct state *st, const char *buf, size_t *start, size_t end) |
{ |
{ |
size_t word; |
size_t word; |
|
|
|
assert(OUST_NL == st->oust); |
|
assert(st->wantws); |
|
|
while (*start < end && ' ' == buf[*start]) |
while (*start < end && ' ' == buf[*start]) |
(*start)++; |
(*start)++; |
|
|
Line 993 donamenm(struct state *st, const char *buf, size_t *st |
|
Line 1246 donamenm(struct state *st, const char *buf, size_t *st |
|
} |
} |
|
|
while (*start < end) { |
while (*start < end) { |
fputs(".Nm ", stdout); |
|
for (word = *start; word < end; word++) |
for (word = *start; word < end; word++) |
if (',' == buf[word]) |
if (',' == buf[word]) |
break; |
break; |
formatcodeln(st, buf, start, word, 1); |
formatcodeln(st, "Nm", buf, start, word, 1); |
if (*start == end) { |
if (*start == end) { |
putchar('\n'); |
mdoc_newln(st); |
continue; |
break; |
} |
} |
assert(',' == buf[*start]); |
assert(',' == buf[*start]); |
puts(" ,"); |
printf(" ,"); |
|
mdoc_newln(st); |
(*start)++; |
(*start)++; |
while (*start < end && ' ' == buf[*start]) |
while (*start < end && ' ' == buf[*start]) |
(*start)++; |
(*start)++; |
Line 1017 donamenm(struct state *st, const char *buf, size_t *st |
|
Line 1270 donamenm(struct state *st, const char *buf, size_t *st |
|
* Lots of other snakes in the grass: escaping a newline followed by a |
* Lots of other snakes in the grass: escaping a newline followed by a |
* period (accidental mdoc(7) control), double-newlines after macro |
* period (accidental mdoc(7) control), double-newlines after macro |
* passages, etc. |
* passages, etc. |
|
* |
|
* 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. |
|
* Must make sure to go back to OUST_NL/wantws mode before returning. |
*/ |
*/ |
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) |
Line 1047 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1305 ordinary(struct state *st, const char *buf, size_t sta |
|
start = j + 1; |
start = j + 1; |
while (start < end && ' ' == buf[start]) |
while (start < end && ' ' == buf[start]) |
start++; |
start++; |
fputs(".Nd ", stdout); |
formatcodeln(st, "Nd", buf, &start, end, 1); |
formatcodeln(st, buf, &start, end, 1); |
mdoc_newln(st); |
putchar('\n'); |
|
return; |
return; |
} |
} |
} |
} |
Line 1067 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1324 ordinary(struct state *st, const char *buf, size_t sta |
|
* Escape initial control characters. |
* Escape initial control characters. |
*/ |
*/ |
while (start < end) { |
while (start < end) { |
if (start < end - 1 && '<' == buf[start + 1]) |
if (start < end - 1 && '<' == buf[start + 1] && |
|
'A' <= buf[start] && 'Z' >= buf[start]) |
break; |
break; |
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, "\\&"); |
/* |
/* |
* 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 && |
('[' == buf[start] || |
('[' == buf[start] || |
']' == buf[start]) && |
']' == buf[start]) && |
dosynopsisop(buf, &last, |
dosynopsisop(st, buf, |
&start, end, &opstack)) |
&start, end, &opstack)) |
continue; |
continue; |
putchar(last = buf[start++]); |
|
if ('\\' == last) |
|
putchar('e'); |
|
} |
|
|
|
if (start < end - 1 && '<' == buf[start + 1]) { |
|
/* |
/* |
* We've encountered a format code. |
* On whitespace, flush the output buffer |
* This is going to trigger a macro no matter |
* and allow breaking to a macro line. |
* what, so print a newline now. |
* Otherwise, buffer text and clear wantws. |
* Then print the (possibly nested) macros and |
|
* following that, a newline. |
|
* Consume all whitespace so we don't |
|
* 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)) { |
|
|
last = buf[start++]; |
|
if (' ' != last) { |
|
outbuf_addchar(st); |
|
continue; |
|
} |
|
|
|
if ( ! strcmp(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; |
|
} |
|
|
|
outbuf_flush(st); |
|
putchar(' '); |
|
st->wantws = 1; |
|
} |
|
|
|
if (start < end - 1 && '<' == buf[start + 1] && |
|
'A' <= buf[start] && 'Z' >= buf[start]) { |
|
formatcode(st, buf, &start, end, 0, seq); |
|
if (OUST_MAC == st->oust) { |
|
/* |
|
* Let mdoc(7) handle trailing punctuation. |
|
* XXX Some punctuation characters |
|
* are not handled yet. |
|
*/ |
if ((start == end - 1 || |
if ((start == end - 1 || |
(start < end - 1 && |
(start < end - 1 && |
(' ' == buf[start + 1] || |
(' ' == buf[start + 1] || |
Line 1113 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1393 ordinary(struct state *st, const char *buf, size_t sta |
|
putchar(' '); |
putchar(' '); |
putchar(buf[start++]); |
putchar(buf[start++]); |
} |
} |
putchar(last = '\n'); |
|
|
if (st->wantws || |
|
' ' == buf[start] || |
|
'\n' == buf[start]) |
|
mdoc_newln(st); |
|
|
|
/* |
|
* 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++; |
|
|
|
/* |
|
* Some text is following. |
|
* Implement requested spacing. |
|
*/ |
|
|
|
if ( ! st->wantws && start < end && |
|
('<' != buf[start + 1] || |
|
'A' > buf[start] || |
|
'Z' < buf[start])) { |
|
printf(" Ns "); |
|
st->wantws = 1; |
|
} |
} |
} |
} else if (start < end && '\n' == buf[start]) { |
} else if (start < end && '\n' == buf[start]) { |
/* |
outbuf_flush(st); |
* Print the newline only if we haven't already |
mdoc_newln(st); |
* printed a newline. |
|
*/ |
|
if (last != '\n') |
|
putchar(last = buf[start]); |
|
if (++start >= end) |
if (++start >= end) |
continue; |
continue; |
/* |
/* |
Line 1134 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1434 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; |
} |
} |
} |
} |
|
outbuf_flush(st); |
if (last != '\n') |
mdoc_newln(st); |
putchar('\n'); |
|
} |
} |
|
|
/* |
/* |
Line 1154 ordinary(struct state *st, const char *buf, size_t sta |
|
Line 1451 ordinary(struct state *st, const char *buf, size_t sta |
|
* (default: starts with "="). |
* (default: starts with "="). |
*/ |
*/ |
static void |
static void |
dopar(struct state *st, const char *buf, size_t start, size_t end) |
dopar(struct state *st, char *buf, size_t start, size_t end) |
{ |
{ |
|
|
|
assert(OUST_NL == st->oust); |
|
assert(st->wantws); |
|
|
if (end == start) |
if (end == start) |
return; |
return; |
if (' ' == buf[start] || '\t' == buf[start]) |
if (' ' == buf[start] || '\t' == buf[start]) |
Line 1173 dopar(struct state *st, const char *buf, size_t start, |
|
Line 1473 dopar(struct state *st, const char *buf, size_t start, |
|
*/ |
*/ |
static void |
static void |
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, 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 1200 dofile(const struct args *args, const char *fname, |
|
Line 1516 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 1227 dofile(const struct args *args, const char *fname, |
|
Line 1538 dofile(const struct args *args, const char *fname, |
|
|
|
free(title); |
free(title); |
|
|
|
dict_init(); |
memset(&st, 0, sizeof(struct state)); |
memset(&st, 0, sizeof(struct state)); |
|
st.oust = OUST_NL; |
|
st.wantws = 1; |
|
|
assert(sz > 0); |
assert(sz > 0); |
|
|
/* Main loop over file contents. */ |
/* Main loop over file contents. */ |
Line 1251 dofile(const struct args *args, const char *fname, |
|
Line 1566 dofile(const struct args *args, const char *fname, |
|
dopar(&st, buf, cur, end); |
dopar(&st, buf, cur, end); |
cur = sup; |
cur = sup; |
} |
} |
|
dict_destroy(); |
} |
} |
|
|
/* |
/* |
Line 1270 readfile(const struct args *args, const char *fname) |
|
Line 1586 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 1377 main(int argc, char *argv[]) |
|
Line 1691 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 1387 main(int argc, char *argv[]) |
|
Line 1701 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); |
} |
} |