From: Thomas Hochstein <thh@inter.net>
Date: Sat, 29 May 2021 08:21:13 +0000 (+0200)
Subject: Merge branch 'thh-parsedb' into pu
X-Git-Url: https://code.th-h.de/?p=usenet%2Fnewsstats.git;a=commitdiff_plain;h=refs%2Fheads%2Fpu;hp=84e9923abeadcfb488985b9e60eb78d4eaa950fd

Merge branch 'thh-parsedb' into pu

* thh-parsedb:
  Forcibly decode headers with unencoded 8bit chars.
  Add MID to error message to make it more useful.
---

diff --git a/bin/parsedb.pl b/bin/parsedb.pl
index b4c2056..8004cb1 100755
--- a/bin/parsedb.pl
+++ b/bin/parsedb.pl
@@ -198,6 +198,11 @@ while (my $HeadersR = $DBQuery->fetchrow_hashref) {
         $Headers{$HeaderName.'_parsed'} = decode('MIME-Header',$Headers{$_})
           if (exists($LegalEncodings{$Encoding}));
       }
+      # forcibly modify headers with un-encoded 8bit data assuming utf-8
+      # TODO: try to guess correct enconding
+      elsif ($Headers{$_} =~ /[^\x00-\x7F]/) {
+        $Headers{$_} = decode('utf-8',$Headers{$_});
+      }
       # extract name(s) and mail(s) from From: / Sender: / Reply-To:
       # in parsed form, if available
       if ($_ ne 'subject') {
@@ -258,9 +263,9 @@ while (my $HeadersR = $DBQuery->fetchrow_hashref) {
                                        split(/ /,'? ' x scalar(@DBFields)))
                                 ));
   $DBWrite->execute(@SQLBindVars)
-      or &Bleat(2,sprintf("Can't write parsed data for %s to %s.%s: ".
+      or &Bleat(2,sprintf("Can't write parsed data for %s to %s.%s for %s: ".
                           "$DBI::errstr\n",$Period,
-                          $Conf{'DBDatabase'},$Conf{'DBTableParse'}));
+                          $Conf{'DBDatabase'},$Conf{'DBTableParse'}, $Headers{'mid'}));
     $DBWrite->finish;
   }
 };