Changeset 3111

Timestamp:

2002-05-07T15:33:35+12:00 (22 years ago)

Author:

jrm21

Message:

Allow .eml extension (IE and mozilla default to this for individual files).

Tidy up the checks for testing for valid mail file format.

File:

• trunk/gsdl/perllib/plugins/EMAILPlug.pm (modified) (2 diffs)

trunk/gsdl/perllib/plugins/EMAILPlug.pm

@@ -94 +94 @@
     # mbx/email for mailbox file format, \d+ for maildir (each message is
     # in a separate file, with a unique number for filename)
-    return q@([\\/]\d+|\.(mbx|email))$@;
+    # mozilla and IE will save individual mbx format files with a ".eml" ext.
+    return q@([\\/]\d+|\.(mbx|email|eml))$@;
 }
 
 # This plugin splits the mbox mail files at lines starting with From<sp>
+# It is supposed to be "\n\nFrom ", but this isn't always used.
 sub get_default_split_exp {
     return q^\nFrom .*\n^;
@@ -111 +113 @@
 
     # Check that we're dealing with a valid mail file
-    return undef unless (($$textref =~ /From:/m) || ($$textref =~ /To:/m));
-
-    # slightly more strict validity check, to prevent us from matching 
-    # .so.x files ...
-    return undef unless (($$textref =~ /^From /) ||
-             ($$textref =~ /^[-A-Za-z]{2,100}:/m));
+    # mbox message files start with "From "
+    # maildir messages usually start with Return-Path and Delivered-To
+    # mh is very similar to maildir
+    my $startoffile=substr($$textref,0,256);
+    if (($startoffile !~ /^(From )/) &&
+    ($startoffile !~ /^(From|To|Envelope.*|Received|Return-Path|Date|Subject|Content\-.*|MIME-Version|Forwarded):/im)) {
+    return undef;
+    }
+
 
     print $outhandle "EMAILPlug: processing $file\n"