Fix #43: os path join raise unicode decode error

In python2, UnicodeDecodeError raised when dir/file name is non-ASCII
using getcwdu (six provide getcwd to be compatible with Python3)
to get current dir and passing it to listdir which also return unicode
and decoding folder-relative args as utf-8

Arrange the structure of unittest, and add a testcase for classifying by
date which will raise the Error mentioned before

Author: yittg

classifier/classifier.py

@@ -1,8 +1,11 @@
 #!/usr/bin/env python
-import os
-import sys
 import argparse
 import arrow
+import os
+import six
+import sys
+
+from six.moves import getcwd
 
 """
 All format lists were taken from wikipedia, not all of them were added due to extensions
@@ -28,7 +31,7 @@ def moveto(file, from_folder, to_folder):
 def classify(formats, output):
     print("Scanning Files")
 
-    directory = os.getcwd()
+    directory = getcwd()
 
     for file in os.listdir(directory):
         filename, file_ext = os.path.splitext(file)
@@ -46,7 +49,7 @@ def classify(formats, output):
 def classify_by_date(date_format, output_dir):
     print("Scanning Files")
 
-    directory = os.getcwd()
+    directory = getcwd()
     files = [x for x in os.listdir(directory) if not x.startswith('.')]
     creation_dates = map(lambda x: (x, arrow.get(os.path.getctime(x))), files)
 
@@ -57,6 +60,18 @@ def classify_by_date(date_format, output_dir):
     print("Done!")
 
 
+def _format_text_arg(arg):
+    if not isinstance(arg, six.text_type):
+        arg = arg.decode('utf-8')
+    return arg
+
+
+def _format_arg(arg):
+    if isinstance(arg, six.string_types):
+        arg = _format_text_arg(arg)
+    return arg
+
+
 def main():
     description = "Organize files in your directory instantly,by classifying them into different folders"
     parser = argparse.ArgumentParser(description=description)
@@ -92,14 +107,17 @@ def main():
         sys.exit()
 
     if args.specific_folder and args.specific_types:
-        formats = {args.specific_folder: args.specific_types}
+        specific_folder = _format_arg(args.specific_folder)
+        formats = {specific_folder: args.specific_types}
 
     if args.output is None:
-        args.output = os.getcwd()
+        output = getcwd()
+    else:
+        output = _format_arg(args.output)
 
     if args.date:
-        classify_by_date('DD-MM-YYYY', args.output)
+        classify_by_date('DD-MM-YYYY', output)
     else:
-        classify(formats, args.output)
+        classify(formats, output)
 
     sys.exit()

setup.py

@@ -15,6 +15,7 @@
         """,
         install_requires=[
             'arrow',
+            'six',
         ],
         zip_safe=False
 )

tests/test_classifier.py

@@ -1,28 +1,59 @@
 # -*- coding: utf-8 -*-
 
+import arrow
 import os
+import shutil
 import unittest
 
 from classifier import classifier as clf
+from six.moves import getcwd
 
 
 class ClassifierTest(unittest.TestCase):
 
     __location = os.path.realpath(
-        os.path.join(os.getcwd(), os.path.dirname(__file__)))
+        os.path.join(getcwd(), os.path.dirname(__file__), '.unittest'))
 
-    __tmp_file_name = '.classifier_test'
+    __tmp_files = [u'test_file', u'test_file_中文']
+    __tmp_dirs = [u'test_dir', u'test_dir_中文']
 
-    def test_moveto(self):
+    def setUp(self):
+        if not os.path.exists(self.__location):
+            os.mkdir(self.__location)
         os.chdir(self.__location)
-        tmp_file = open(self.__tmp_file_name, 'w')
+        for file_ in self.__tmp_files:
+            open(file_, 'w').close()
+        for dir_ in self.__tmp_dirs:
+            if not os.path.exists(dir_):
+                os.mkdir(dir_)
+        super(ClassifierTest, self).setUp()
+
+    def tearDown(self):
+        shutil.rmtree(self.__location)
+        super(ClassifierTest, self).tearDown()
 
-        parent_dir = os.path.abspath(os.path.join(self.__location, os.pardir))
-        clf.moveto(self.__tmp_file_name, self.__location, parent_dir)
+    def test_moveto(self):
+        target_dir = os.path.abspath(os.path.join(self.__location, 'moveto'))
+        for file_ in self.__tmp_files:
+            clf.moveto(file_, self.__location, target_dir)
+
+        for file_ in self.__tmp_files:
+            final_file_path = os.path.join(target_dir, file_)
+            self.assertTrue(os.path.exists(final_file_path))
+
+    def test_classify_bydate(self):
+        date_format = 'DD-MM-YYYY'
+        target_files = []
+        for file_ in self.__tmp_files:
+            target_dir = arrow.get(os.path.getctime(file_)).format(date_format)
+            final_file_path = os.path.join(target_dir, file_)
+            target_files.append(final_file_path)
+        clf.classify_by_date(date_format, self.__location)
+        for file_ in target_files:
+            self.assertTrue(os.path.exists(final_file_path))
+        for dir_ in self.__tmp_dirs:
+            self.assertTrue(os.path.exists(dir_))
 
-        final_file_path = os.path.join(parent_dir, self.__tmp_file_name)
-        self.assertTrue(os.path.exists(final_file_path))
-        os.remove(final_file_path)
 
 if __name__ == '__main__':
     unittest.main()