Added a check to search the output directories for duplicate video files

TVEncoder.py

@@ -31,6 +31,7 @@ def showhelp():
           'files that will be processed without actually encoding them'
     print 'TVEncoder.py -e - encode the files that have been processed'
     print 'TVEncoder.py -e -l - list the files that would be encoded'
+    print 'TVEncoder.py -c - check the output directories for duplicates'
 
 
 def print_shows(shows):
@@ -74,9 +75,11 @@ def processarguments(options):
     """
 
     inputoptions = namedtuple("inputoptions",
-                              "numfiles doencode readonly dolist")
+                              "numfiles doencode readonly dolist "
+                              "checkduplicates")
 
     inputoptions.readonly = False
+    inputoptions.checkduplicates = False
 
     for opt, arg in options:
         if opt == '-h':
@@ -90,6 +93,8 @@ def processarguments(options):
             inputoptions.numfiles = arg
         elif opt == "-l":
             inputoptions.readonly = True
+        elif opt == "-c":
+            inputoptions.checkduplicates = True
 
     return inputoptions
 
@@ -99,7 +104,7 @@ def main(argv):
     The main program for TVEncoder.
     """
     try:
-        opts, _ = getopt.getopt(argv, "hlpen:")
+        opts, _ = getopt.getopt(argv, "hlpecn:")
     except getopt.GetoptError:
         showhelp()
         sys.exit(2)
@@ -108,6 +113,16 @@ def main(argv):
     settings = Settings(SETTINGS)
     filemanager = FileManager(settings)
 
+    if inputoptions.checkduplicates:
+        print "Searching for duplicates..."
+        duplicates = filemanager.checkexistingduplicates()
+        if duplicates:
+            for duplicate in duplicates:
+                print duplicate
+        else:
+            print "No duplicates found."
+        return
+
     if inputoptions.readonly:
         if inputoptions.doencode:
             #Generate the list of files that would be encoded

libfilemanager.py

@@ -112,6 +112,26 @@ class FileManager:
         #will reach here if there were less than numberofFiles found
         return showstoprocess
 
+    def checkexistingduplicates(self):
+        """
+        Check the existing files in the output directories for duplicate
+        files, typically in different formats
+        """
+
+        duplicates = []
+        for show in self.__settings.getshownames():
+            outputdir = self.__settings.getshowoutputdirectory(show)
+
+            for rootdir, dirnames, filenames in os.walk(outputdir):
+                for fle in filenames:
+                    filename = os.path.join(rootdir, fle)
+                    if os.path.splitext(fle)[1] in [".avi", ".mpg", ".mpeg",
+                                                    "mp4", ".mkv"]:
+                        if self.checkduplicates(filename):
+                            duplicates.append(filename)
+
+        return duplicates
+
     @staticmethod
     def checkduplicates(filename):
         """
@@ -120,13 +140,16 @@ class FileManager:
         """
 
         dirname = os.path.dirname(filename)
-        filename = os.path.basename(filename)[:6]
+        filename = os.path.basename(filename)
+        fileseasonepisode = filename[:6]
+        fileextension = os.path.splitext(filename)[1]
 
         for _, _, filenames in os.walk(dirname):
             for show in filenames:
                 extension = os.path.splitext(show)[1]
-                if (extension in [".avi", ".mpg", ".mpeg", "mp4"] and
-                        show[:6] == filename):
+                if (extension in [".avi", ".mpg", ".mpeg", "mp4", ".mkv"] and
+                        show[:6] == fileseasonepisode
+                        and fileextension != extension):
                     return True
 
         return False

libsettings.py

@@ -8,19 +8,6 @@ Created on Fri Jul  5 20:14:15 2013
 from configobj import ConfigObj
 
 
-#==============================================================================
-# class ShowSettings:
-#     """
-#     Container for the settings for a show
-#     """
-#
-#     def __init__(self, name, inputdirectory, outputdirectory):
-#         self.name = name
-#         self.inputdirectory = inputdirectory
-#         self.outputdirectory = outputdirectory
-#==============================================================================
-
-
 class Settings:
     """
     Accessor for the configuration file

tests/libfilemanagertest.py

@@ -66,6 +66,16 @@ class libfilemanagertest(unittest.TestCase):
 
         self.assertFalse(result)
 
+    def test_checkduplicatesameextension(self):
+        settings = Mock('libsettings.Settings')
+        filemanager = FileManager(settings)
+
+        os.walk = dummywalk
+
+        result = filemanager.checkduplicates("/path/to/S03E14 - Test - SD TV.avi")
+
+        self.assertFalse(result)
+
 
 def dummywalk(arg):
     return [("/path/to/", [], ["S03E14 - Test - SD TV.avi"])]