Add script to split an mbox into individual messages labelled by Message-Id

git-svn-id: file:///Users/dpage/pgweb/svn-repo/trunk@2492 8f5c7a92-453e-0410-a47f-ad33c8a6b003

1 files changed, 65 insertions, 0 deletions

diff --git a/archives/bin/splitmbox b/archives/bin/splitmbox
new file mode 100755
index 00000000..7f679897
--- /dev/null
+++ b/archives/bin/splitmbox
@@ -0,0 +1,65 @@
+#!/bin/bash
+#
+# Split an mbox into message files, and sort them into directories according
+# to message-id splitting.
+#
+# This needs to run under bash because some of the ${} stuff is a bash-ism.
+#
+# $Id$
+
+mbox=$1
+destdir=messages
+export destdir
+
+# *() and other patterns need this set
+shopt -s extglob
+
+if [ ! -f $mbox ]; then
+  echo "mbox $mbox not found" >&2
+  exit 1
+fi
+
+if [ ! -d $destdir ]; then
+  echo "destination directory $destdir does not exist" >&2
+  exit 1
+fi
+
+if [ $(basename $mbox) = $(basename $mbox .gz) ]; then
+  cat=cat
+else
+  cat=zcat
+fi
+
+tempdir=$(mktemp -d)
+export tempdir
+$cat $mbox | formail -s sh -c 'cat - > $tempdir/msg.$FILENO'
+
+for message in $tempdir/msg.*; do
+  # Fetch the Message-Id.  Note: due to braindamage at Microsoft we need to
+  # cope with messages having more than one Message-Id, so we loop here.
+  formail -x Message-Id < $message | while read messageid; do
+    # strip the initial " <", the trailing ">", and replace /'s with _
+    messageid=${messageid##*([< ])}
+    messageid=${messageid%%>}
+    messageid=${messageid//\//_}
+    # fetch the part after the first @ (inclusive)
+    dir=${messageid##*([^@])}
+    # escape [, ? and * from it (these are special chars for bash ${} expansion)
+    diresc=${dir/[/\\[}
+    diresc=${diresc/\*/\\*}
+    diresc=${diresc/\?/\\?}
+    # and fetch the part before the @ (i.e. strip $dir)
+    file=${messageid%%$diresc}
+    # create the directory if needed
+    if [ ! -d "$destdir/$dir" ]; then
+      mkdir -v "$destdir/$dir" || echo "failed to create $destdir/$dir"
+    fi
+    # and link the message into place
+    if [ ! -f "$destdir/$dir/$file" ]; then
+      ln $message "$destdir/$dir/$file" || echo "failed to link $message to $destdir/$dir/$file"
+    fi
+  done
+  rm $message
+done
+rmdir $tempdir # unless something remained
+