Merge pull request #1 from NYPD/v0.0.1

V0.0.1

Author: NYPD

.gitignore

@@ -102,3 +102,10 @@ venv.bak/
 
 # mypy
 .mypy_cache/
+
+### VisualStudioCode ###
+.vscode
+.history
+
+# Pycharm
+.idea

README.md

@@ -1 +1,20 @@
-# Phodupe
+# phodupe
+
+## What is it
+
+A simple little python script that finds duplicate file names (without the extension) across two directories and gives the user the option to delete them.
+
+## Why
+
+I used this as an opportunity to begin learning python. It was created for a very very specific scenario and might not be too useful to anyone else.
+
+## How do to install / run
+
+(There are probobly better ways to run this, but i'm a python noob)
+
+- Navigate to the folder you downloaded the git project to
+  - cd \development\git\Phodupe\
+- Install the python script
+  - python .\setup.py install
+- Run the script by passing two directories
+  - python phodupe "C:\Users\NYPD\Desktop\OG Phone Walls" "C:\Users\NYPD\Desktop\2Delete"

phodupe/__init__.py

@@ -0,0 +1,27 @@
+import sys
+from phodupe.dupe_finder import DupeFinder
+
+
+def main():
+
+    destination1 = sys.argv[1]
+    destination2 = sys.argv[2]
+    
+    dupe_files = DupeFinder.getDuplicateFileNames(destination1, destination2)
+
+    if len(dupe_files) is 0:
+        print('No dupe files found!')
+        exit()
+
+    user_input = input("{} duplicate file names found. Enter 'y' to delete or 'n' to abort:\n".format(len(dupe_files)))
+    
+    if user_input == 'y':
+        DupeFinder.deleteFiles(dupe_files, destination1, destination2)
+        print('Files deleted, exiting...')
+    else:
+        print('No files deleted, aborting...')
+        exit()
+
+
+if __name__ == "__main__":
+    main()

phodupe/__main__.py

@@ -0,0 +1,81 @@
+import os
+import pathlib
+import glob
+import itertools
+
+class DupeFinder:
+    """
+        Simple utility class for finding and manipulating duplicate files
+    """
+
+    @staticmethod
+    def getDuplicateFileNames(directory1, directory2):
+        """
+        Return a list of duplicate file name "stems" in both directories.
+        
+        e.g. : 
+        
+        directory1 : [a.png, b.png]
+        
+        directory2 : [a.png, b.png]
+        
+        returns: [a, b]
+
+        Parameters
+        ----------
+        directory1 : str
+            Path of the first directory
+        directory2 : str
+            Path of the second directory
+        
+        Returns
+        -------
+        list
+            A list of the duplicate file names between both directories. If no
+            duplicate files are found, an empty list is returned
+        """
+        directory1Path = pathlib.Path(directory1)
+        directory2Path = pathlib.Path(directory2)
+
+        directory1FilesNoExt = []
+        directory2FilesNoExt = []
+
+        for file in os.listdir(directory1Path):
+            purePath = pathlib.Path(os.path.join(directory1, file))
+            directory1FilesNoExt.append(purePath.stem)
+        for file in os.listdir(directory2Path):
+            purePath = pathlib.Path(os.path.join(directory2, file))
+            directory2FilesNoExt.append(purePath.stem)
+
+        dupeFiles = []
+
+        for fileName in directory1FilesNoExt:
+            if fileName in directory2FilesNoExt:
+                dupeFiles.append(fileName)
+        
+        return dupeFiles
+
+    @staticmethod
+    def deleteFiles(fileNames, directory1, directory2):
+        """
+        Deletes the files provided in both directories
+
+        Parameters
+        ----------
+        fileNames : list
+            List of string files names to delete
+        directory1 : str
+            Path of the first directory
+        directory2 : str
+            Path of the second directory
+        """
+        for file in fileNames:
+            
+            directory1Path = os.path.join(directory1, file)
+            directory2Path = os.path.join(directory2, file)
+            
+            directory1Glob = glob.glob('{}.*'.format(directory1Path))
+            directory2Glob = glob.glob('{}.*'.format(directory2Path))
+            
+            for filePath in itertools.chain(directory1Glob, directory2Glob):
+                os.remove(filePath)

phodupe/dupe_finder.py

@@ -0,0 +1,11 @@
+from setuptools import setup
+
+setup(
+    name='phodupe',
+    version='v0.0.1',
+    packages=['phodupe'],
+    url='https://github.com/NYPD/phodupe',
+    license='MIT',
+    author='NYPD',
+    description='Duplicate file finder'
+)