git: Manually decode git output from utf-8, avoids warning messages on invalidly...

diff --git a/IkiWiki/Plugin/git.pm b/IkiWiki/Plugin/git.pm
index 12f3a74cb7a11178b29fae3300f9e84db76c07b7..b386ab71bfa6b0d0feffcb53fb167f17c1ac016d 100644 (file)
--- a/IkiWiki/Plugin/git.pm
+++ b/IkiWiki/Plugin/git.pm
@@ -136,14 +136,16 @@ sub safe_git (&@) {
        }
        # In parent.
 
        }
        # In parent.
 

+       # git output is probably utf-8 encoded, but may contain
+       # other encodings or invalidly encoded stuff. So do not rely
+       # on the normal utf-8 IO layer, decode it by hand.
+       binmode($OUT);
+

        my @lines;
        while (<$OUT>) {
        my @lines;
        while (<$OUT>) {

+               $_=decode_utf8($_, 0);
+

                chomp;
                chomp;

-               
-               # check for invalid utf-8, and toss it back to avoid crashes
-               if (! utf8::valid($_)) {
-                       $_=encode_utf8($_);
-               }

 
                push @lines, $_;
        }
 
                push @lines, $_;
        }

diff --git a/debian/changelog b/debian/changelog
index a7e7544c6cf0a68af1cf2e603bbc0753a1c36204..77b5d9fa6acc705d51bdc20da215dccd896b4caf 100644 (file)
--- a/debian/changelog
+++ b/debian/changelog
@@ -1,6 +1,8 @@
 ikiwiki (3.08) UNRELEASED; urgency=low
 
   * git: Fix utf-8 encoding of author names.
 ikiwiki (3.08) UNRELEASED; urgency=low
 
   * git: Fix utf-8 encoding of author names.

+  * git: Manually decode git output from utf-8, avoids
+    warning messages on invalidly encoded output.

 
  -- Joey Hess <joeyh@debian.org>  Mon, 09 Mar 2009 14:00:21 -0400
 
 
  -- Joey Hess <joeyh@debian.org>  Mon, 09 Mar 2009 14:00:21 -0400