Delete fdupes duplicates by directory

fdupes_delete_by_directory.py

#!/usr/bin/env python

import fileinput
import os.path as path
import re

class Processor(object):

    def __init__(self):
        self.condemnedFiles = []
        self.condemnedDirs = {}

    def prompt_for_choice(self, validChoices):
        choice = None
        condemnDir = False

        while choice is None:
            choice = raw_input('Enter choice [%d-%d][d]:' % (min(validChoices.keys()), max(validChoices.keys())))

            choiceMatch = re.match(r'^(\d+)(d?)$', choice)

            if not choiceMatch:
                choice = None
                continue

            dupeIndex = int( choiceMatch.group(1) )

            if dupeIndex not in validChoices:
                choice = None
                continue

            if len( choiceMatch.group(2) ):
                condemnDir = True

        toDelete = validChoices[dupeIndex]

        del validChoices[dupeIndex]

        self.condemnedFiles.append(toDelete)

        print 'Will delete %s' % toDelete
        if condemnDir:
            directory = path.split( toDelete )[ 0 ]
            self.condemnedDirs[directory] = True
            print 'Will delete all subsequent duplicates found in %s' % directory

    def run(self):

        group = []
        groupIndex = 1

        for line in fileinput.input():

            # Trim trailing newline
            line = line[:-1]

            if not len( line ):

                validChoices = {}

                print ''
                print '--- Group %d ---' % groupIndex

                condemnedInGroup = 0

                for index, dupe in enumerate(group):

                    index += 1

                    directory = path.split( dupe )[ 0 ]

                    if directory in self.condemnedDirs:

                        if condemnedInGroup == len(group) - 1:
                            print 'All duplicates in group are in condemned directories, preserving %s so there is a file left from the group' % dupe
                            validChoices[index] = dupe
                            break

                        print '%s is in a condemned directory so this duplicate will be deleted' % dupe

                        self.condemnedFiles.append(dupe)
                        condemnedInGroup += 1
                    else:
                        validChoices[index] = dupe
                        print '[%d] %s' % ( index, dupe )

                while len(validChoices) > 1:

                    self.prompt_for_choice(validChoices)

                print 'No duplicates of %s left to delete' % validChoices.values()[0]

                group = []

                groupIndex += 1

                continue

            group.append(line)

if __name__ == '__main__':

    OUTPUT_FILENAME = 'toDelete.log'

    print 'fdupes output processor to select duplicates to delete and remember which directories are unwanted.'
    print 'Usage:'
    print ''
    print 'fdupes -r . > fdupes_out.log'
    print 'python fdupes_delete_by_directory.py fdupes_out.log'
    print ''
    print 'Entering "d" following a duplicate name will cause all subsequent duplicates in that directory to be condemned for deletion.'
    print 'Files will not actually be deleted by this script, it will save %s containing the condemned files.' % OUTPUT_FILENAME
    print ''
    print 'Examples of how to process %s in shell:' % OUTPUT_FILENAME
    print ''
    print '# Delete all duplicates'
    print 'while read f; do rm "$f"; done <%s' % OUTPUT_FILENAME
    print ''
    print '# Move all files to a "dead" directory while preserving the original directory structure'
    print 'while read f; do mkdir -p "../dead/$(dirname $f)" && mv -v "$f" "../dead/$(dirname $f)/"; done <%s' % OUTPUT_FILENAME
    print ''

    processor = Processor()
    processor.run()

    with open( OUTPUT_FILENAME, 'w') as fp:
        fp.write('\n'.join(processor.condemnedFiles) + '\n')