From 0cef6ebbb5fd6924f80022c3cb0f0eb77dc8244f Mon Sep 17 00:00:00 2001
From: Raymond Wagner <rwagner@mythtv.org>
Date: Fri, 29 Apr 2011 00:55:49 -0400
Subject: [PATCH] Add two more scripts.  One to hash any content with a hash of
 'NULL', and the other to remove any files with duplicate hashes. (cherry
 picked from commit 71a489768c83b80b328fb3d0665bcbf3591fbccc)

---
 python/hash_videos.py             | 24 ++++++++++++
 python/remove_duplicate_videos.py | 64 +++++++++++++++++++++++++++++++
 2 files changed, 88 insertions(+)
 create mode 100755 python/hash_videos.py
 create mode 100755 python/remove_duplicate_videos.py

diff --git a/python/hash_videos.py b/python/hash_videos.py
new file mode 100755
index 0000000..3322f7a
--- /dev/null
+++ b/python/hash_videos.py
@@ -0,0 +1,24 @@
+#!/usr/bin/env python
+# -*- coding: UTF-8 -*-
+#---------------------------
+# Name: hash_videos.py
+# Python Script
+# Author: Raymond Wagner
+# Purpose
+#   This script will regenerate the hash values associated with
+#   videos in the MythVideo database.
+#---------------------------
+
+from MythTV import Video
+
+def print_aligned(left, right):
+    indent = 100 - len(left)
+    print left, indent*' ', right
+
+for vid in Video.getAllEntries():
+    if vid.hash != 'NULL':
+        print_aligned(vid.filename, 'skipped')
+        continue
+    vid.hash = vid.getHash()
+    vid.update()
+    print_aligned(vid.filename, vid.hash)
diff --git a/python/remove_duplicate_videos.py b/python/remove_duplicate_videos.py
new file mode 100755
index 0000000..67b51e4
--- /dev/null
+++ b/python/remove_duplicate_videos.py
@@ -0,0 +1,64 @@
+#!/usr/bin/env python
+# -*- coding: UTF-8 -*-
+#---------------------------
+# Name: remove_duplicate_videos.py
+# Python Script
+# Author: Raymond Wagner
+# Purpose
+#   For reasons unknown, some people continue to get duplicate file
+#   entries in their MythVideo database. This script will detail those
+#   duplicate files based off hash number, and optionally delete them.
+#---------------------------
+__title__  = "Remove Duplicate Videos"
+__author__ = "Raymond Wagner"
+__version__= "v0.5.0"
+
+from optparse import OptionParser
+from MythTV import Video
+
+def format_name(vid):
+    # returns a string in the format 'TITLE[ - SEASONxEPISODE][ - SUBTITLE]'
+    s = vid.title
+    if vid.season:
+        s += ' - %dx%02d' % (vid.season, vid.episode)
+    if vid.subtitle:
+        s += ' - '+vid.subtitle
+    return s
+
+def FindDuplicates(dodelete):
+    dupvids = []
+    vids = sorted(Video.getAllEntries(), key=lambda v: v.hash)
+
+    for i in range(len(vids)-1):
+        if vids[i].hash == 'NULL':
+            continue
+        if vids[i].hash == vids[i+1].hash:
+            dupvids.append(vids[i+1])
+
+    if dodelete:
+        for vid in dupvids:
+            vid.delete()
+
+    return dupvids
+
+def main():
+    parser = OptionParser(usage="usage: %prog [options] [jobid]")
+
+    parser.add_option("-s", "--dry-run", action="store_true", default=False,
+            dest="dryrun", help="Print out duplicates but do not delete.")
+
+    opts, args = parser.parse_args()
+
+    dups = FindDuplicates(not opts.dryrun)
+
+    if len(dups):
+        print len(dups), 'Duplicates Found!'
+        print '----------------------'
+        for vid in dups:
+            print '  '+format_name(vid)
+    else:
+        print 'No Duplicates Found!'
+
+if __name__ == "__main__":
+    main()
+