Re: What happened to strict_mime?
* Tue Mar 20 2007 Thomas Roessler <roessler@xxxxxxxxxxxxxxxxxx>
> On 2007-03-20 12:01:49 +0900, TAKAHASHI Tamotsu wrote:
> >strict_mime=yes:
> > unset ignore_linear_white_space
> >strict_mime=no:
> > set ignore_linear_white_space
> Why on earth do we need two different behaviors for encoding or
> decoding subject headers as far as whitespace is concerned? There's
> a spec-compliant way to do this, let's do that, and be done with.
Personally I don't need $ignore_linear_space.
But I understand the situation.
For example, RFC2047 says that
| Subject: (a
| =?ISO-8859-1?Q?b?=)
should be displayed as the following:
| Subject: (a
| b)
But many Japanese MUAs insert spaces and a CRLF between alphabet-word
and kanjis because kanji becomes very long, and because RFC limits
header length. For example,
| Subject: (a <kanji>)
is encoded to
| Subject: (a
| =?ISO-2022-JP?B?GyRCNEE7ehsoQgo=?=)
by such MUAs. We don't like this:
| Subject: you
| are
| annoyed
| by
| this
| subject
This is what we feel from such bogus encoded headers
when $ignore_linear_white_space="no".
> If the other approach is really needed to deal with widespread
> breakage (I don't think it is), implement that, and implement it
> always.
You are generous.
Most Japanese users say "It is really needed!"
> As an aside, the code used to implement this (in the instance around
> line 816 of rfc2047.c) looks like an incredibly convoluted and
> inefficient way of saying something like this:
(snip)
LWS code is not written by me. Takashi, please review it.
Well, safe_strdup can return NULL.
And I found a comment: "/* LWS doesn't end with CRLF */"
These _may_ be the reason, but I don't know what Takashi thought.
> The way in which this code iterates through AssumedCharset in
> convert_nonmime_string is clumsy at best. (A better way to go
> through it would be to safe_strdup() the entire thing, use strtok on
> the copy, and then free the temporary buffer.)
Thanks for your advice. I like your implementation.
I didn't know of strtok.
> So, from a casual glance at this code, two suggestions:
> - Please clean up this code.
I'm going to try.
Beta patch is attached.
--
tamo
diff -r d0b9363c79aa charset.c
--- a/charset.c Fri Mar 16 17:14:54 2007 +0900
+++ b/charset.c Wed Mar 21 01:15:09 2007 +0900
@@ -290,16 +290,15 @@ int mutt_chscmp (const char *s, const ch
char *mutt_get_default_charset ()
{
- static char fcharset[SHORT_STRING];
- const char *c = AssumedCharset;
- const char *c1;
-
- if (c && *c) {
- c1 = strchr (c, ':');
- strfcpy (fcharset, c, c1 ? (c1 - c + 1) : sizeof (fcharset));
- return fcharset;
- }
- return strcpy (fcharset, "us-ascii"); /* __STRCPY_CHECKED__ */
+ char *s;
+
+ if (AssumedCharset && *AssumedCharset)
+ {
+ s = strdup (AssumedCharset);
+ strtok (s, ":");
+ return s;
+ }
+ return strdup ("us-ascii");
}
#ifndef HAVE_ICONV
diff -r d0b9363c79aa handler.c
--- a/handler.c Fri Mar 16 17:14:54 2007 +0900
+++ b/handler.c Wed Mar 21 00:53:53 2007 +0900
@@ -1445,10 +1445,12 @@ void mutt_decode_attachment (BODY *b, ST
if (istext && s->flags & M_CHARCONV)
{
char *charset = mutt_get_parameter ("charset", b->parameter);
+ char *assumed_charset = NULL;
if (!charset && AssumedCharset && *AssumedCharset)
- charset = mutt_get_default_charset ();
- if (charset && Charset)
- cd = mutt_iconv_open (Charset, charset, M_ICONV_HOOK_FROM);
+ assumed_charset = mutt_get_default_charset ();
+ if ((charset || assumed_charset) && Charset)
+ cd = mutt_iconv_open (Charset, charset ? charset : assumed_charset,
M_ICONV_HOOK_FROM);
+ FREE (&assumed_charset);
}
else if (istext && b->charset)
cd = mutt_iconv_open (Charset, b->charset, M_ICONV_HOOK_FROM);
diff -r d0b9363c79aa parse.c
--- a/parse.c Fri Mar 16 17:14:54 2007 +0900
+++ b/parse.c Wed Mar 21 00:55:27 2007 +0900
@@ -402,9 +402,11 @@ void mutt_parse_content_type (char *s, B
if (ct->type == TYPETEXT)
{
if (!(pc = mutt_get_parameter ("charset", ct->parameter)))
- mutt_set_parameter ("charset", (AssumedCharset && *AssumedCharset) ?
- (const char *) mutt_get_default_charset ()
- : "us-ascii", &ct->parameter);
+ {
+ char *assumed_charset = mutt_get_default_charset ();
+ mutt_set_parameter ("charset", assumed_charset, &ct->parameter);
+ FREE (&assumed_charset);
+ }
}
}
diff -r d0b9363c79aa rfc2047.c
--- a/rfc2047.c Fri Mar 16 17:14:54 2007 +0900
+++ b/rfc2047.c Wed Mar 21 02:12:02 2007 +0900
@@ -89,38 +89,42 @@ static size_t convert_string (ICONV_CONS
int convert_nonmime_string (char **ps)
{
- const char *c, *c1;
-
- for (c = AssumedCharset; c; c = c1 ? c1 + 1 : 0)
- {
- char *u = *ps;
- char *s;
- char *fromcode;
- size_t m, n;
- size_t ulen = mutt_strlen (*ps);
- size_t slen;
-
- if (!u || !*u)
- return 0;
-
- c1 = strchr (c, ':');
- n = c1 ? c1 - c : mutt_strlen (c);
- if (!n)
- return 0;
- fromcode = safe_malloc (n + 1);
- strfcpy (fromcode, c, n + 1);
- m = convert_string (u, ulen, fromcode, Charset, &s, &slen);
- FREE (&fromcode);
+ char *c, *c1;
+ char *u = *ps;
+ char *s;
+ size_t ulen = mutt_strlen (*ps);
+ size_t slen;
+ size_t m;
+
+ if (!u || !*u)
+ return 0;
+
+ c = safe_strdup (AssumedCharset);
+ if (!c)
+ c = safe_calloc (1, 1);
+ for (c1 = strtok (c, ":"); c1;
+ c1 = strtok (NULL, ":"))
+ {
+ m = convert_string (u, ulen, c1, Charset, &s, &slen);
if (m != (size_t)(-1))
{
+ FREE (&c);
FREE (ps); /* __FREE_CHECKED__ */
*ps = s;
return 0;
}
}
- mutt_convert_string (ps,
- (const char *)mutt_get_default_charset (AssumedCharset),
+ FREE (&c);
+
+ /* pass-thru if the last entry is empty */
+ if (AssumedCharset && *AssumedCharset &&
+ AssumedCharset[strlen (AssumedCharset) - 1] == ':')
+ return 0;
+
+ /* going to do my best */
+ mutt_convert_string (ps, c = mutt_get_default_charset (),
Charset, M_ICONV_HOOK_FROM);
+ FREE (&c);
return -1;
}
@@ -827,11 +829,11 @@ void rfc2047_decode (char **pd)
char *t;
size_t tlen;
- n = mutt_strlen (s);
- t = safe_malloc (n + 1);
- strfcpy (t, s, n + 1);
+ t = safe_strdup (s);
convert_nonmime_string (&t);
tlen = mutt_strlen (t);
+ if (dlen < tlen) /* truncation */
+ tlen = dlen;
strncpy (d, t, tlen);
d += tlen;
FREE (&t);