Add original file name methods to text and media class #186 (#187)

elodie/compatability.py

@@ -3,6 +3,7 @@ import shutil
 
 from elodie import constants
 
+
 def _decode(string, encoding='utf8'):
     """Return a utf8 encoded unicode string.
 
@@ -22,6 +23,7 @@ def _decode(string, encoding='utf8'):
 
     return string
 
+
 def _copyfile(src, dst):
     # Python 3 hangs using open/write method
     if (constants.python_version == 3):
@@ -45,7 +47,11 @@ def _copyfile(src, dst):
         for x in iter(lambda: os.read(fin, BUFFER_SIZE), ""):
             os.write(fout, x)
     finally:
-        try: os.close(fin)
-        except: pass
-        try: os.close(fout)
-        except: pass
+        try:
+            os.close(fin)
+        except:
+            pass
+        try:
+            os.close(fout)
+        except:
+            pass

elodie/filesystem.py

@@ -291,6 +291,7 @@ class FileSystem(object):
             print('%s is not a valid media file. Skipping...' % _file)
             return
 
+        media.set_original_name()
         metadata = media.get_metadata()
 
         directory_name = self.get_folder_path(metadata)
@@ -334,9 +335,8 @@ class FileSystem(object):
             # Do not use copy2(), will have an issue when copying to a
             # network/mounted drive using copy and manual
             # set_date_from_filename gets the job done
-            # Do not use copy2(), will have an issue when copying to a network/mounted drive
             # shutil.copy seems slow, changing to streaming according to
-            # http://stackoverflow.com/questions/22078621/python-how-to-copy-files-fast
+            # http://stackoverflow.com/questions/22078621/python-how-to-copy-files-fast  # noqa
             compatability._copyfile(_file, dest_path)
             self.set_utime(media)
 

elodie/media/base.py

@@ -90,6 +90,7 @@ class Base(object):
             'album': self.get_album(),
             'title': self.get_title(),
             'mime_type': self.get_mimetype(),
+            'original_name': self.get_original_name(),
             'base_name': os.path.splitext(os.path.basename(source))[0],
             'extension': self.get_extension(),
             'directory_path': os.path.dirname(source)
@@ -100,7 +101,7 @@ class Base(object):
     def get_mimetype(self):
         """Get the mimetype of the file.
 
-        :returns: str or None for a non-video
+        :returns: str or None for unsupported files.
         """
         if(not self.is_valid()):
             return None
@@ -112,6 +113,15 @@ class Base(object):
 
         return mimetype[0]
 
+    def get_original_name(self):
+        """Get the original name of the file from before it was imported.
+        Does not include the extension.
+        Overridden by Media class for files with EXIF.
+
+        :returns: str or None for unsupported files.
+        """
+        return None
+
     def get_title(self):
         """Base method for getting the title of a file
 
@@ -185,6 +195,12 @@ class Base(object):
             if(key in metadata):
                 self.metadata[key] = kwargs[key]
 
+    def set_original_name(self):
+        """Stores the original file name into EXIF/metadata.
+        :returns: bool
+        """
+        return False
+
     @classmethod
     def get_class_by_file(cls, _file, classes):
         """Static method to get a media object by file.

elodie/media/media.py

@@ -10,6 +10,8 @@ are used to represent the actual files.
 """
 from __future__ import print_function
 
+import os
+
 # load modules
 from elodie import constants
 from elodie.dependencies import get_exiftool
@@ -46,6 +48,7 @@ class Media(Base):
         self.longitude_keys = ['EXIF:GPSLongitude']
         self.latitude_ref_key = 'EXIF:GPSLatitudeRef'
         self.longitude_ref_key = 'EXIF:GPSLongitudeRef'
+        self.original_name_key = 'XMP:OriginalFileName'
         self.set_gps_ref = True
         self.exiftool_addedargs = [
             '-overwrite_original',
@@ -129,6 +132,24 @@ class Media(Base):
 
         return metadata
 
+    def get_original_name(self):
+        """Get the original name stored in EXIF.
+
+        :returns: str
+        """
+        if(not self.is_valid()):
+            return None
+
+        exiftool_attributes = self.get_exiftool_attributes()
+
+        if exiftool_attributes is None:
+            return None
+
+        if(self.original_name_key not in exiftool_attributes):
+            return None
+
+        return exiftool_attributes[self.original_name_key]
+
     def get_title(self):
         """Get the title for a photo of video
 
@@ -213,6 +234,25 @@ class Media(Base):
 
         return status
 
+    def set_original_name(self):
+        """Sets the original name EXIF tag if not already set.
+
+        :returns: True, False, None
+        """
+        if(not self.is_valid()):
+            return None
+
+        # If EXIF original name tag is set then we return.
+        if self.get_original_name() is not None:
+            return None
+
+        source = self.source
+        name = os.path.basename(source)
+        tags = {self.original_name_key: name}
+        status = self.__set_tags(tags)
+        self.reset_cache()
+        return status
+
     def set_title(self, title):
         """Set title for a photo.
 

elodie/media/text.py

@@ -72,6 +72,16 @@ class Text(Base):
         self.parse_metadata_line()
         return super(Text, self).get_metadata()
 
+    def get_original_name(self):
+        self.parse_metadata_line()
+
+        # We return the value if found in metadata
+        if(isinstance(self.metadata_line, dict) and
+                'original_name' in self.metadata_line):
+            return self.metadata_line['original_name']
+
+        return super(Text, self).get_original_name()
+
     def get_title(self):
         self.parse_metadata_line()
 
@@ -92,11 +102,6 @@ class Text(Base):
         self.reset_cache()
         return status
 
-    def set_location(self, latitude, longitude):
-        status = self.write_metadata(latitude=latitude, longitude=longitude)
-        self.reset_cache()
-        return status
-
     def set_date_taken(self, passed_in_time):
         if(time is None):
             return False
@@ -106,6 +111,29 @@ class Text(Base):
         self.reset_cache()
         return status
 
+    def set_original_name(self):
+        """Sets the original name if not already set.
+
+        :returns: True, False, None
+        """
+        if(not self.is_valid()):
+            return None
+
+        # If EXIF original name tag is set then we return.
+        if self.get_original_name() is not None:
+            return None
+
+        source = self.source
+        name = os.path.basename(source)
+        status = self.write_metadata(original_name=name)
+        self.reset_cache()
+        return status
+
+    def set_location(self, latitude, longitude):
+        status = self.write_metadata(latitude=latitude, longitude=longitude)
+        self.reset_cache()
+        return status
+
     def parse_metadata_line(self):
         if isinstance(self.metadata_line, dict):
             return self.metadata_line

elodie/tests/files/with-original-name.txt

@@ -0,0 +1,3 @@
+{"date_taken":1460027726.0,"latitude":"51.521435","longitude":"0.162714","title":"sample title","original_name":"originalname.txt"}
+
+This file has a valid header.

elodie/tests/helper.py

@@ -36,11 +36,16 @@ def create_working_folder():
 
 def download_file(name, destination):
     try:
+        url_to_file = 'https://s3.amazonaws.com/jmathai/github/elodie/{}'.format(name)
+        # urlretrieve works differently for python 2 and 3
+        if constants.python_version < 3:
             final_name = '{}/{}{}'.format(destination, random_string(10), os.path.splitext(name)[1])
             urllib.urlretrieve(
-            'https://s3.amazonaws.com/jmathai/github/elodie/{}'.format(name),
+                url_to_file,
                 final_name
             )
+        else:
+            final_name, headers = urllib.request.urlretrieve(url_to_file)
         return final_name
     except Exception as e:
         return False

elodie/tests/media/base_test.py

@@ -24,6 +24,10 @@ from elodie.media.video import Video
 os.environ['TZ'] = 'GMT'
 
 
+def test_get_all_subclasses():
+    subclasses = get_all_subclasses(Base)
+    expected = {Media, Base, Text, Photo, Video, Audio}
+    assert subclasses == expected, subclasses
 
 def test_get_class_by_file_without_extension():
     base_file = helper.get_file('withoutextension')
@@ -32,6 +36,32 @@ def test_get_class_by_file_without_extension():
     
     assert cls is None, cls
 
+def test_get_original_name():
+    temporary_folder, folder = helper.create_working_folder()
+
+    origin = '%s/%s' % (folder, 'with-original-name.jpg')
+    file = helper.get_file('with-original-name.jpg')
+    
+    shutil.copyfile(file, origin)
+
+    media = Media.get_class_by_file(origin, [Photo])
+    original_name = media.get_original_name()
+
+    assert original_name == 'originalfilename.jpg', original_name
+
+def test_get_original_name_invalid_file():
+    temporary_folder, folder = helper.create_working_folder()
+
+    origin = '%s/%s' % (folder, 'invalid.jpg')
+    file = helper.get_file('invalid.jpg')
+    
+    shutil.copyfile(file, origin)
+
+    media = Media.get_class_by_file(origin, [Photo])
+    original_name = media.get_original_name()
+
+    assert original_name is None, original_name
+
 def test_set_album_from_folder_invalid_file():
     temporary_folder, folder = helper.create_working_folder()
 
@@ -106,8 +136,3 @@ def test_set_metadata_basename():
     new_metadata = photo.get_metadata()
 
     assert new_metadata['base_name'] == new_basename, new_metadata['base_name']
-
-def test_get_all_subclasses():
-    subclasses = get_all_subclasses(Base)
-    expected = {Media, Base, Text, Photo, Video, Audio}
-    assert subclasses == expected, subclasses

elodie/tests/media/media_test.py

@@ -61,6 +61,88 @@ def test_get_class_by_file_invalid_type():
                                     [Photo, Video, Audio])
     assert media is None
 
+def test_get_original_name():
+    temporary_folder, folder = helper.create_working_folder()
+
+    origin = '%s/%s' % (folder, 'with-original-name.jpg')
+    file = helper.get_file('with-original-name.jpg')
+    
+    shutil.copyfile(file, origin)
+
+    media = Media.get_class_by_file(origin, [Photo])
+    original_name = media.get_original_name()
+
+    assert original_name == 'originalfilename.jpg', original_name
+
+def test_get_original_name_invalid_file():
+    temporary_folder, folder = helper.create_working_folder()
+
+    origin = '%s/%s' % (folder, 'invalid.jpg')
+    file = helper.get_file('invalid.jpg')
+    
+    shutil.copyfile(file, origin)
+
+    media = Media.get_class_by_file(origin, [Photo])
+    original_name = media.get_original_name()
+
+    assert original_name is None, original_name
+    
+def test_set_original_name_when_exists():
+    temporary_folder, folder = helper.create_working_folder()
+
+    origin = '%s/%s' % (folder, 'with-original-name.jpg')
+    file = helper.get_file('with-original-name.jpg')
+    
+    shutil.copyfile(file, origin)
+
+    media = Media.get_class_by_file(origin, [Photo])
+    result = media.set_original_name()
+
+    assert result is None, result
+
+def test_set_original_name_when_does_not_exist():
+    temporary_folder, folder = helper.create_working_folder()
+
+    origin = '%s/%s' % (folder, 'plain.jpg')
+    file = helper.get_file('plain.jpg')
+    
+    shutil.copyfile(file, origin)
+
+    media = Media.get_class_by_file(origin, [Photo])
+    metadata_before = media.get_metadata()
+    result = media.set_original_name()
+    metadata_after = media.get_metadata()
+
+    assert metadata_before['original_name'] is None, metadata_before
+    assert metadata_after['original_name'] == 'plain.jpg', metadata_after
+    assert result is True, result
+
+def test_set_original_name():
+    files = ['plain.jpg', 'audio.m4a', 'photo.nef', 'video.mov']
+
+    for file in files:
+        ext = os.path.splitext(file)[1]
+
+        temporary_folder, folder = helper.create_working_folder()
+
+        random_file_name = '%s%s' % (helper.random_string(10), ext)
+        origin = '%s/%s' % (folder, random_file_name)
+        file_path = helper.get_file(file)
+        if file_path is False:
+            file_path = helper.download_file(file, folder)
+
+        shutil.copyfile(file_path, origin)
+
+        media = Media.get_class_by_file(origin, [Audio, Media, Photo, Video])
+        metadata = media.get_metadata()
+        media.set_original_name()
+        metadata_updated = media.get_metadata()
+
+        shutil.rmtree(folder)
+
+        assert metadata['original_name'] is None, metadata['original_name']
+        assert metadata_updated['original_name'] == random_file_name, metadata_updated['original_name']
+
 def is_valid():
     media = Media()
 

elodie/tests/media/text_test.py

@@ -29,6 +29,18 @@ def test_text_extensions():
 
     assert extensions == valid_extensions, valid_extensions
 
+def test_get_original_name():
+    media = Text(helper.get_file('with-original-name.txt'))
+    original_name = media.get_original_name()
+
+    assert original_name == 'originalname.txt', original_name
+
+def test_get_original_name_when_does_not_exist():
+    media = Text(helper.get_file('valid.txt'))
+    original_name = media.get_original_name()
+
+    assert original_name is None, original_name
+
 def test_get_title():
     text = Text(helper.get_file('valid.txt'))
     text.get_metadata()
@@ -268,3 +280,20 @@ def test_set_location_without_header():
     shutil.rmtree(folder)
 
     assert helper.isclose(metadata['latitude'], 11.1111111111), metadata['latitude']
+
+def test_set_original_name():
+    temporary_folder, folder = helper.create_working_folder()
+
+    random_file_name = '%s.txt' % helper.random_string(10)
+    origin = '%s/%s' % (folder, random_file_name)
+    shutil.copyfile(helper.get_file('valid.txt'), origin)
+
+    text = Text(origin)
+    metadata = text.get_metadata()
+    text.set_original_name()
+    metadata_updated = text.get_metadata()
+
+    shutil.rmtree(folder)
+
+    assert metadata['original_name'] is None, metadata['original_name']
+    assert metadata_updated['original_name'] == random_file_name, metadata_updated['original_name']