[Dovecot] Search with specified charset

[Chris Wakelin]

I've seen a problem with searching in the Prayer webmail client that 
turns out to be caused by Prayer insisting on a charset called 
"ISO-8859-1" and Dovecot, using Solaris iconv, not recognising it (all 
the Solaris 8 iconv modules appear to be "ISO8859-1" etc. or just "8859-1").

This isn't a problem in Linux:

% iconv --list | grep 8859.1/

8859_1//
ISO-8859-1//
ISO8859-1//
ISO885911//
ISO_8859-1//

I've hacked together a patch for Dovecot to try matching without the 
"ISO-" prefix if matching fails at first, which works for Solaris (as 
would creating an extra symbolic link in /usr/lib/iconv or hacking 
Prayer to use "ISO8859-1") and it doesn't break Linux, but it's not 
pretty :(

Better may be to try substituting underscores, hyphens, spaces(?) and 
nothing in the string, but that makes the change much more complicated!

Best Wishes,
Chris

--- dovecot-1.0-stable/src/lib-charset/charset-iconv.c.orig     Mon Jan 
31 16:37:55 2005
+++ dovecot-1.0-stable/src/lib-charset/charset-iconv.c  Wed Jul 13 
21:47:03 2005
@@ -36,9 +36,17 @@
                 ascii = FALSE;
                 cd = iconv_open("UTF-8", charset);
                 if (cd == (iconv_t)-1) {
-                       if (unknown_charset != NULL)
-                               *unknown_charset = TRUE;
-                       return NULL;
+                       /* Try matching charset without "ISO" prefix */
+                       if (strncasecmp(charset, "ISO-", 4) == 0 ||
+                           strncasecmp(charset, "ISO ", 4) == 0)
+                               cd = iconv_open("UTF-8", charset + 4);
+                       else if (strncasecmp(charset, "ISO", 3) == 0)
+                               cd = iconv_open("UTF-8", charset + 3);
+                       if (cd == (iconv_t)-1) {
+                               if (unknown_charset != NULL)
+                                       *unknown_charset = TRUE;
+                               return NULL;
+                       }
                 }
         }

@@ -137,9 +145,17 @@

         cd = iconv_open("UTF-8", charset);
         if (cd == (iconv_t)-1) {
-               if (unknown_charset != NULL)
-                       *unknown_charset = TRUE;
-               return NULL;
+               /* Try matching charset without "ISO" prefix */
+               if (strncasecmp(charset, "ISO-", 4) == 0 ||
+                   strncasecmp(charset, "ISO ", 4) == 0)
+                       cd = iconv_open("UTF-8", charset + 4);
+               else if (strncasecmp(charset, "ISO", 3) == 0)
+                       cd = iconv_open("UTF-8", charset + 3);
+               if (cd == (iconv_t)-1) {
+                       if (unknown_charset != NULL)
+                               *unknown_charset = TRUE;
+                       return NULL;
+               }
         }

         if (unknown_charset != NULL)



-- 
--+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+-
Christopher Wakelin,                           c.d.wakelin at reading.ac.uk
IT Services Centre, The University of Reading,  Tel: +44 (0)118 378 8439
Whiteknights, Reading, RG6 2AF, UK              Fax: +44 (0)118 975 3094