Huge refactoring of media and change get_metadata function creation

dozo/exiftool.py

@@ -176,7 +176,7 @@ class ExifTool:
         return self._exiftoolproc.process
 
     def setvalue(self, tag, value):
-        """Set tag to value(s); if value is None, will delete tag
+        """Set tag to value(s); if value is None, tag will not be set
 
         Args:
             tag: str; name of tag to set
@@ -191,7 +191,7 @@ class ExifTool:
         """
 
         if value is None:
-            value = ""
+            return False
         command = [f"-{tag}={value}"]
         if self.overwrite and not self._context_mgr:
             command.append("-overwrite_original")

dozo/filesystem.py

@@ -238,12 +238,12 @@ class FileSystem(object):
                 'title'):
             if metadata[item]:
                 part = metadata[item]
-        elif item in ('original_name'):
+        elif item == 'original_name':
             # First we check if we have metadata['original_name'].
             # We have to do this for backwards compatibility because
             #   we original did not store this back into EXIF.
             if metadata[item]:
-                part = os.path.splitext(metadata['original_name'])[0]
+                part = metadata['original_name']
         elif item in 'custom':
             # Fallback string
             part = mask[1:-1]

dozo/media/media.py

@@ -12,7 +12,7 @@ import logging
 # load modules
 from dateutil.parser import parse
 import re
-from dozo.exiftool import ExifToolCaching
+from dozo.exiftool import ExifTool, ExifToolCaching
 
 class Media():
 
@@ -34,184 +34,213 @@ class Media():
 
     extensions = PHOTO + AUDIO + VIDEO
 
-
     def __init__(self, sources=None, ignore_tags=set(), logger=logging.getLogger()):
         self.source = sources
-        self.reset_cache()
+        self.ignore_tags = ignore_tags
-        self.date_original = [
+        self.tags_keys = self.get_tags()
+        self.exif_metadata = None
+        self.metadata = None
+        self.logger = logger
+
+    def get_tags(self):
+        tags_keys = {}
+        tags_keys['date_original'] = [
             'EXIF:DateTimeOriginal',
             'H264:DateTimeOriginal',
             'QuickTime:ContentCreateDate'
         ]
-        self.date_created = [
+        tags_keys['date_created'] = [
             'EXIF:CreateDate',
             'QuickTime:CreationDate',
             'QuickTime:CreateDate',
             'QuickTime:CreationDate-und-US',
             'QuickTime:MediaCreateDate'
         ]
-        self.date_modified = ['File:FileModifyDate', 'QuickTime:ModifyDate']
+        tags_keys['date_modified'] = [
-        self.camera_make_keys = ['EXIF:Make', 'QuickTime:Make']
+            'File:FileModifyDate',
-        self.camera_model_keys = ['EXIF:Model', 'QuickTime:Model']
+            'QuickTime:ModifyDate'
-        self.album_keys = ['XMP-xmpDM:Album', 'XMP:Album']
+        ]
-        self.title_keys = ['XMP:Title', 'XMP:DisplayName']
+        tags_keys['camera_make'] = ['EXIF:Make', 'QuickTime:Make']
-        self.latitude_keys = [
+        tags_keys['camera_model'] = ['EXIF:Model', 'QuickTime:Model']
+        tags_keys['album'] = ['XMP-xmpDM:Album', 'XMP:Album']
+        tags_keys['title'] = ['XMP:Title', 'XMP:DisplayName']
+        tags_keys['latitude'] = [
             'EXIF:GPSLatitude',
             'XMP:GPSLatitude',
             # 'QuickTime:GPSLatitude',
             'Composite:GPSLatitude'
         ]
-        self.longitude_keys = [
+        tags_keys['longitude'] = [
             'EXIF:GPSLongitude',
             'XMP:GPSLongitude',
             # 'QuickTime:GPSLongitude',
             'Composite:GPSLongitude'
         ]
-        self.latitude_ref_key = 'EXIF:GPSLatitudeRef'
+        tags_keys['latitude_ref'] = ['EXIF:GPSLatitudeRef']
-        self.longitude_ref_key = 'EXIF:GPSLongitudeRef'
+        tags_keys['longitude_ref'] = ['EXIF:GPSLongitudeRef']
-        self.original_name_key = 'XMP:OriginalFileName'
+        tags_keys['original_name'] = ['XMP:OriginalFileName']
-        self.set_gps_ref = True
-        self.metadata = None
-        self.exif_metadata = None
-        self.ignore_tags = ignore_tags
-        self.logger = logger
 
+        # Remove ignored tag from list
+        for tag_regex in self.ignore_tags:
+            ignored_tags = set()
+            for key, tags in tags_keys.items():
+                for n, tag in enumerate(tags):
+                    if re.match(tag_regex, tag):
+                        del(tags_keys[key][n])
 
-    def format_metadata(self, **kwargs):
+        return tags_keys
-        """Method to consistently return a populated metadata dictionary.
-
-        :returns: dict
-        """
-
-    def get_file_path(self):
-        """Get the full path to the video.
-
-        :returns: string
-        """
-        return self.source
-
-
-    def get_extension(self):
-        """Get the file extension as a lowercased string.
-
-        :returns: string or None for a non-video
-        """
-        if(not self.is_valid()):
-            return None
-
-        source = self.source
-        return os.path.splitext(source)[1][1:]
-
-
-    def get_metadata(self, update_cache=False, album_from_folder=False):
-        """Get a dictionary of metadata for any file.
-
-        All keys will be present and have a value of None if not obtained.
-
-        :returns: dict or None for non-text files
-        """
-        if(not self.is_valid()):
-            return None
-
-        if(isinstance(self.metadata, dict) and update_cache is False):
-            return self.metadata
-
-        source = self.source
-        folder = os.path.basename(os.path.dirname(source))
-        album = self.get_album()
-        if album_from_folder and (album is None or album == ''):
-            album = folder
-
-        self.metadata = {
-            'date_original': self.get_date_attribute(self.date_original),
-            'date_created': self.get_date_attribute(self.date_created),
-            'date_modified': self.get_date_attribute(self.date_modified),
-            'camera_make': self.get_camera_make(),
-            'camera_model': self.get_camera_model(),
-            'latitude': self.get_coordinate('latitude'),
-            'longitude': self.get_coordinate('longitude'),
-            'album': album,
-            'title': self.get_title(),
-            'mime_type': self.get_mimetype(),
-            'original_name': self.get_original_name(),
-            'base_name': os.path.basename(os.path.splitext(source)[0]),
-            'ext': self.get_extension(),
-            'directory_path': os.path.dirname(source)
-        }
-
-        return self.metadata
 
+    def _del_ignored_tags(self, exif_metadata):
+        for tag_regex in self.ignore_tags:
+            ignored_tags = set()
+            for tag in exif_metadata:
+                if re.search(tag_regex, tag) is not None:
+                    ignored_tags.add(tag)
+            for ignored_tag in ignored_tags:
+                del exif_metadata[ignored_tag]
 
     def get_mimetype(self):
         """Get the mimetype of the file.
 
-        :returns: str or None for unsupported files.
+        :returns: str or None
         """
-        if(not self.is_valid()):
+        mimetype = mimetypes.guess_type(self.source)
-            return None
-
-        source = self.source
-        mimetype = mimetypes.guess_type(source)
         if(mimetype is None):
             return None
 
         return mimetype[0]
 
+    def _get_key_values(self, key):
+        """Get the first value of a tag set
 
-    def is_valid(self):
+        :returns: str or None if no exif tag
-        # Disable extension check
+        """
+        if self.exif_metadata is None:
+            return None
+
+        for tag in self.tags_keys[key]:
+            if tag in self.exif_metadata:
+                yield self.exif_metadata[tag]
+
+    def get_value(self, tag):
+        """Get given value from EXIF.
+
+        :returns: str or None
+        """
+        exiftool_attributes = self.get_exiftool_attributes()
+        if exiftool_attributes is None:
+            return None
+        if(tag not in exiftool_attributes):
+            return None
+
+        return exiftool_attributes[tag]
+
+    def get_date_format(self, value):
+        """Formate date attribute.
+        :returns: datetime object or None
+        """
+        # We need to parse a string to datetime format.
+        # EXIF DateTimeOriginal and EXIF DateTime are both stored
+        #   in %Y:%m:%d %H:%M:%S format
+        if value is None:
+            return None
+
+        try:
+            # correct nasty formated date
+            regex = re.compile(r'(\d{4}):(\d{2}):(\d{2})')
+            if(re.match(regex , value) is not None):  # noqa
+                value = re.sub(regex , r'\g<1>-\g<2>-\g<3>', value)
+            return parse(value)
+        except BaseException  or dateutil.parser._parser.ParserError as e:
+            self.logger.error(e)
+            return None
+
+    def get_coordinates(self, key, value):
+        """Get latitude or longitude value
+
+        :param str key: Type of coordinate to get. Either "latitude" or
+            "longitude".
+        :returns: float or None
+        """
+        if value is None:
+            return None
+
+        if isinstance(value, str) and len(value) == 0:
+            # If exiftool GPS output is empty, the data returned will be a str
+            # with 0 length.
+            # https://github.com/jmathai/elodie/issues/354
+            return None
+
+        # Cast coordinate to a float due to a bug in exiftool's
+        #   -json output format.
+        # https://github.com/jmathai/elodie/issues/171
+        # http://u88.n24.queensu.ca/exiftool/forum/index.php/topic,7952.0.html  # noqa
+        this_coordinate = float(value)
+
+        direction_multiplier = 1.0
+        #   when self.set_gps_ref != True
+        if key == 'latitude':
+            if 'EXIF:GPSLatitudeRef' in self.exif_metadata:
+                if self.exif_metadata['EXIF:GPSLatitudeRef'] == 'S':
+                    direction_multiplier = -1.0
+        elif key == 'longitude':
+            if 'EXIF:GPSLongitudeRef' in self.exif_metadata:
+                if self.exif_metadata['EXIF:GPSLongitudeRef'] == 'W':
+                    direction_multiplier = -1.0
+        return this_coordinate * direction_multiplier
+
+        return None
+
+    def get_metadata(self):
+        """Get a dictionary of metadata from exif.
+        All keys will be present and have a value of None if not obtained.
+
+        :returns: dict
+        """
+        # Get metadata from exiftool.
+        self.exif_metadata = ExifToolCaching(self.source, logger=self.logger).asdict()
+
+        # TODO to be removed
+        self.metadata = {}
+        # Retrieve selected metadata to dict
+        if not self.exif_metadata:
+            return self.metadata
+
+        for key in self.tags_keys:
+            formated_data = None
+            for value in self._get_key_values(key):
+                if 'date' in key:
+                    formated_data = self.get_date_format(value)
+                elif key in ('latitude', 'longitude'):
+                    formated_data = self.get_coordinates(key, value)
+                else:
+                    if value is not None and value != '':
+                        formated_data = value
+                    else:
+                        formated_data = None
+                if formated_data:
+                    # Use this data and break
+                    break
+
+            self.metadata[key] = formated_data
+
+        self.metadata['base_name']  = os.path.basename(os.path.splitext(self.source)[0])
+        self.metadata['ext'] = os.path.splitext(self.source)[1][1:]
+        self.metadata['directory_path'] = os.path.dirname(self.source)
+
+        return self.metadata
+
+    def has_exif_data(self):
+        """Check if file has metadata, date original"""
+        if not self.metadata:
+            return False
+
+        if 'date_original' in self.metadata:
+            if self.metadata['date_original'] != None:
                 return True
 
-
-    def set_album_from_folder(self, path):
-        """Set the album attribute based on the leaf folder name
-
-        :returns: bool
-        """
-        metadata = self.get_metadata()
-
-        # If this file has an album already set we do not overwrite EXIF
-        if(not isinstance(metadata, dict) or metadata['album'] is not None):
         return False
 
-        folder = os.path.basename(metadata['directory_path'])
-        # If folder is empty we skip
-        if(len(folder) == 0):
-            return False
-
-        status = self.set_album(folder, path)
-        if status == False:
-            return False
-        return True
-
-
-    def set_metadata_basename(self, new_basename):
-        """Update the basename attribute in the metadata dict for this instance.
-
-        This is used for when we update the EXIF title of a media file. Since
-        that determines the name of a file if we update the title of a file
-        more than once it appends to the file name.
-
-        i.e. 2015-12-31_00-00-00-my-first-title-my-second-title.jpg
-
-        :param str new_basename: New basename of file (with the old title
-            removed).
-        """
-        self.get_metadata()
-        self.metadata['base_name'] = new_basename
-
-
-    def set_metadata(self, **kwargs):
-        """Method to manually update attributes in metadata.
-
-        :params dict kwargs: Named parameters to update.
-        """
-        metadata = self.get_metadata()
-        for key in kwargs:
-            if(key in metadata):
-                self.metadata[key] = kwargs[key]
-
-
     @classmethod
     def get_class_by_file(cls, _file, classes, ignore_tags=set(), logger=logging.getLogger()):
         """Static method to get a media object by file.
@@ -233,246 +262,7 @@ class Media():
         else:
             return Media(_file, ignore_tags=ignore_tags, logger=logger)
 
-
+    def set_date_taken(self, date_key, time):
-    @classmethod
-    def get_valid_extensions(cls):
-        """Static method to access static extensions variable.
-
-        :returns: tuple(str)
-        """
-        return cls.extensions
-
-
-    def get_album(self):
-        """Get album from EXIF
-
-        :returns: None or string
-        """
-        if(not self.is_valid()):
-            return None
-
-        exiftool_attributes = self.get_exiftool_attributes()
-        if exiftool_attributes is None:
-            return None
-
-        for album_key in self.album_keys:
-            if album_key in exiftool_attributes:
-                return exiftool_attributes[album_key]
-
-        return None
-
-
-    def get_coordinate(self, type='latitude'):
-        """Get latitude or longitude of media from EXIF
-
-        :param str type: Type of coordinate to get. Either "latitude" or
-            "longitude".
-        :returns: float or None if not present in EXIF or a non-photo file
-        """
-
-        exif = self.get_exiftool_attributes()
-        if not exif:
-            return None
-
-        # The lat/lon _keys array has an order of precedence.
-        # The first key is writable and we will give the writable
-        #   key precence when reading.
-        direction_multiplier = 1.0
-        for key in self.latitude_keys + self.longitude_keys:
-            if key not in exif:
-                continue
-            if isinstance(exif[key], six.string_types) and len(exif[key]) == 0:
-                # If exiftool GPS output is empty, the data returned will be a str
-                # with 0 length.
-                # https://github.com/jmathai/elodie/issues/354
-                continue
-
-            # Cast coordinate to a float due to a bug in exiftool's
-            #   -json output format.
-            # https://github.com/jmathai/elodie/issues/171
-            # http://u88.n24.queensu.ca/exiftool/forum/index.php/topic,7952.0.html  # noqa
-            this_coordinate = float(exif[key])
-
-            # TODO: verify that we need to check ref key
-            #   when self.set_gps_ref != True
-            if type == 'latitude' and key in self.latitude_keys:
-                if self.latitude_ref_key in exif and \
-                        exif[self.latitude_ref_key] == 'S':
-                    direction_multiplier = -1.0
-                return this_coordinate * direction_multiplier
-            elif type == 'longitude' and key in self.longitude_keys:
-                if self.longitude_ref_key in exif and \
-                        exif[self.longitude_ref_key] == 'W':
-                    direction_multiplier = -1.0
-                return this_coordinate * direction_multiplier
-
-        return None
-
-
-    def get_exiftool_attributes(self):
-        """Get attributes for the media object from exiftool.
-
-        :returns: dict, or False if exiftool was not available.
-        """
-        source = self.source
-
-        #Cache exif metadata results and use if already exists for media
-        if(self.exif_metadata is None):
-            self.exif_metadata = ExifToolCaching(source, logger=self.logger).asdict()
-            for tag_regex in self.ignore_tags:
-                ignored_tags = set()
-                for tag in self.exif_metadata:
-                    if re.search(tag_regex, tag) is not None:
-                        ignored_tags.add(tag)
-                for ignored_tag in ignored_tags:
-                    del self.exif_metadata[ignored_tag]
-
-
-        if not self.exif_metadata:
-            return False
-
-        return self.exif_metadata
-
-
-    def get_date_attribute(self, tag):
-        """Get a date attribute.
-        :returns: time object or None
-        """
-        exif = self.get_exiftool_attributes()
-        if not exif:
-            return None
-        # We need to parse a string from EXIF into a timestamp.
-        # EXIF DateTimeOriginal and EXIF DateTime are both stored
-        #   in %Y:%m:%d %H:%M:%S format
-        # we split on a space and then r':|-' -> convert to int -> .timetuple()
-        #   the conversion in the local timezone
-        # EXIF DateTime is already stored as a timestamp
-        # Sourced from https://github.com/photo/frontend/blob/master/src/libraries/models/Photo.php#L500  # noqa
-        for key in tag:
-            try:
-                if(key in exif):
-                    # correct nasty formated date
-                    regex = re.compile(r'(\d{4}):(\d{2}):(\d{2})')
-                    if(re.match(regex , exif[key]) is not None):  # noqa
-                        exif[key] = re.sub(regex , r'\g<1>-\g<2>-\g<3>', exif[key])
-                    return parse(exif[key])
-                    # if(re.match('\d{4}(-|:)\d{2}(-|:)\d{2}', exif[key]) is not None):  # noqa
-                    #     dt, tm = exif[key].split(' ')
-                    #     dt_list = compile(r'-|:').split(dt)
-                    #     dt_list = dt_list + compile(r'-|:').split(tm)
-                    #     dt_list = map(int, dt_list)
-                    #     return datetime(*dt_list)
-            except BaseException  or dateutil.parser._parser.ParserError as e:
-                self.logger.error(e)
-                return None
-
-        return None
-
-
-    def get_camera_make(self):
-        """Get the camera make 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
-
-        for camera_make_key in self.camera_make_keys:
-            if camera_make_key in exiftool_attributes:
-                return exiftool_attributes[camera_make_key]
-
-        return None
-
-
-    def get_camera_model(self):
-        """Get the camera make 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
-
-        for camera_model_key in self.camera_model_keys:
-            if camera_model_key in exiftool_attributes:
-                return exiftool_attributes[camera_model_key]
-
-        return None
-
-
-    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
-
-        :returns: str or None if no title is set or not a valid media type
-        """
-        if(not self.is_valid()):
-            return None
-
-        exiftool_attributes = self.get_exiftool_attributes()
-
-        if exiftool_attributes is None:
-            return None
-
-        for title_key in self.title_keys:
-            if title_key in exiftool_attributes:
-                return exiftool_attributes[title_key]
-
-        return None
-
-
-    def reset_cache(self):
-        """Resets any internal cache
-        """
-        self.exiftool_attributes = None
-        self.exif_metadata = None
-
-
-    def set_album(self, name, path):
-        """Set album EXIF tag if not already set.
-
-        :returns: True, False, None
-        """
-        if self.get_album() is not None:
-            return None
-
-        tags = {}
-        for key in self.album_keys:
-            tags[key] = name
-        status = self.__set_tags(tags, path)
-        self.reset_cache()
-
-        return status
-
-
-    def set_date_original(self, time, path):
         """Set the date/time a photo was taken.
 
         :param datetime time: datetime object of when the photo was taken
@@ -481,110 +271,47 @@ class Media():
         if(time is None):
             return False
 
-        tags = {}
         formatted_time = time.strftime('%Y:%m:%d %H:%M:%S')
-        for key in self.date_original:
+        status = self.set_value('date_original', formatted_time)
-            tags[key] = formatted_time
-
-        status = self.__set_tags(tags, path)
         if status == False:
             # exif attribute date_original d'ont exist
-            for key in self.date_created:
+            status = self.set_value('date_created', formatted_time)
-                tags[key] = formatted_time
-
-            status = self.__set_tags(tags, path)
-        self.reset_cache()
-        return status
-
-
-    def set_location(self, latitude, longitude, path):
-        if(not self.is_valid()):
-            return None
-
-        # The lat/lon _keys array has an order of precedence.
-        # The first key is writable and we will give the writable
-        #   key precence when reading.
-        # TODO check
-        # tags = {
-        #     self.latitude_keys[0]: latitude,
-        #     self.longitude_keys[0]: longitude,
-        # }
-        tags = {}
-        for key in self.latitude_keys:
-            tags[key] = latitude
-        for key in self.longitude_keys:
-            tags[key] = longitude
-
-        # If self.set_gps_ref == True then it means we are writing an EXIF
-        #   GPS tag which requires us to set the reference key.
-        # That's because the lat/lon are absolute values.
-        # TODO set_gps_ref = False for Video ?
-        if self.set_gps_ref:
-            if latitude < 0:
-                tags[self.latitude_ref_key] = 'S'
-
-            if longitude < 0:
-                tags[self.longitude_ref_key] = 'W'
-
-        status = self.__set_tags(tags, path)
-        self.reset_cache()
 
         return status
 
+    def set_coordinates(self, latitude, longitude):
+        status = []
+        if self.metadata['latitude_ref']:
+            latitude = abs(latitude)
+            if latitude > 0:
+                status.append(self.set_value('latitude_ref', 'N'))
+            else:
+                status.append(self.set_value('latitude_ref', 'S'))
 
-    def set_original_name(self, path, name=None):
+        status.append(self.set_value('latitude', latitude))
-        """Sets the original name EXIF tag if not already set.
 
-        :returns: True, False, None
+        if  self.metadata['longitude_ref']:
-        """
+            longitude = abs(longitude)
-        # If EXIF original name tag is set then we return.
+            if longitude > 0:
-        if self.get_original_name() is not None:
+                status.append(self.set_value('latitude_ref', 'E'))
-            return None
+            else:
+                status.append(self.set_value('longitude_ref', 'W'))
 
-        if name == None:
+        status.append(self.set_value('longitude', longitude))
-            name = os.path.basename(self.source)
 
-        tags = {self.original_name_key: name}
+        if all(status):
-        status = self.__set_tags(tags, path)
+            return True
-        self.reset_cache()
+        else:
+            return False
 
-        return status
+    def set_album_from_folder(self, path):
+        """Set the album attribute based on the leaf folder name
 
-
-    def set_title(self, title, path):
-        """Set title for a photo.
-
-        :param str title: Title of the photo.
         :returns: bool
         """
-        if(not self.is_valid()):
+        folder = os.path.basename(os.path.dirname(self.source))
-            return None
 
-        if(title is None):
+        return set_value(self, 'album', folder)
-            return None
-
-        tags = {}
-        for key in self.title_keys:
-            tags[key] = title
-        status = self.__set_tags(tags, path)
-        self.reset_cache()
-
-        return status
-
-
-    def __set_tags(self, tags, path):
-        if(not self.is_valid()):
-            return None
-
-        status = ''
-        for tag, value in tags.items():
-            status = ExifToolCaching(path, self.logger).setvalue(tag, value)
-        if status.decode().find('unchanged') != -1 or status == '':
-            return False
-        if status.decode().find('error') != -1:
-            return False
-
-        return True
 
 
 def get_all_subclasses(cls=None):

tests/test_filesystem.py

@@ -56,24 +56,15 @@ class TestFilesystem:
                 '{%Y-%m-%b}'
                 ]
 
-        media = Media()
-        exif_tags = {
-                'album': media.album_keys,
-                'camera_make': media.camera_make_keys,
-                'camera_model': media.camera_model_keys,
-                # 'date_original': media.date_original,
-                # 'date_created': media.date_created,
-                # 'date_modified': media.date_modified,
-                'latitude': media.latitude_keys,
-                'longitude': media.longitude_keys,
-                'original_name': [media.original_name_key],
-                'title': media.title_keys
-                }
-
         subdirs = Path('a', 'b', 'c', 'd')
 
         for file_path in self.file_paths:
             media = Media(str(file_path))
+            exif_tags = {}
+            for key in ('album', 'camera_make', 'camera_model', 'latitude',
+                    'longitude', 'original_name', 'title'):
+                exif_tags[key] = media.tags_keys[key]
+
             exif_data = ExifToolCaching(str(file_path)).asdict()
             metadata = media.get_metadata()
             for item, regex in items.items():
@@ -127,25 +118,22 @@ class TestFilesystem:
             metadata = media.get_metadata()
             date_taken = filesystem.get_date_taken(metadata)
 
-            dates = {}
+            date_filename = None
-            for key, date in ('original', media.date_original), ('created',
+            for tag in media.tags_keys['original_name']:
-                    media.date_created), ('modified', media.date_modified):
+                if tag in exif_data:
-                dates[key] = media.get_date_attribute(date)
+                    date_filename = filesystem.get_date_from_string(exif_data[tag])
-
+                break
-            if media.original_name_key in exif_data:
+            if not date_filename:
-                date_filename = filesystem.get_date_from_string(
-                        exif_data[media.original_name_key])
-            else:
                 date_filename = filesystem.get_date_from_string(file_path.name)
 
-            if dates['original']:
+            if media.metadata['date_original']:
-                assert date_taken == dates['original']
+                assert date_taken == media.metadata['date_original']
             elif date_filename:
                 assert date_taken == date_filename
-            elif dates['created']:
+            elif media.metadata['date_created']:
-                assert date_taken == dates['created']
+                assert date_taken == media.metadata['date_created']
-            elif dates['modified']:
+            elif media.metadata['date_modified']:
-                assert date_taken == dates['modified']
+                assert date_taken == media.metadata['date_modified']
 
     def test_sort_files(self, tmp_path):
         db = Db(tmp_path)

tests/test_media.py

@@ -1,6 +1,7 @@
-
+from datetime import datetime
 import pytest
 from pathlib import Path
+import re
 import shutil
 import tempfile
 
@@ -10,28 +11,60 @@ from dozo.media.media import Media
 from dozo.media.audio import Audio
 from dozo.media.photo import Photo
 from dozo.media.video import Video
-from dozo.exiftool import ExifToolCaching
+from dozo.exiftool import ExifTool, ExifToolCaching
 
 DOZO_PATH = Path(__file__).parent.parent
+CACHING = True
 
 class TestMetadata:
 
     def setup_class(cls):
         cls.src_paths, cls.file_paths = copy_sample_files()
+        cls.ignore_tags = ('EXIF:CreateDate', 'File:FileModifyDate',
+                'File:FileAccessDate', 'EXIF:Make', 'Composite:LightValue')
 
-    def test_get_exiftool_attribute(self, tmp_path):
+    def get_media(self):
         for file_path in self.file_paths:
-            exif_data = ExifToolCaching(str(file_path)).asdict()
+            self.exif_data = ExifTool(str(file_path)).asdict()
-            ignore_tags = ('File:FileModifyDate', 'File:FileAccessDate')
+            yield Media(str(file_path), self.ignore_tags)
-            exif_data_filtered = {}
-            for key in exif_data:
-                if key not in ignore_tags:
-                    exif_data_filtered[key] = exif_data[key]
-            media = Media(str(file_path), ignore_tags)
-            exif = media.get_exiftool_attributes()
-            # Ensure returned value is a dictionary
-            assert isinstance(exif, dict)
-            for tag in ignore_tags:
-                assert tag not in exif
-            assert exif == exif_data_filtered
 
+    def test_get_metadata(self):
+        for media in self.get_media():
+            result = media.get_metadata()
+            assert result
+            assert isinstance(media.metadata, dict), media.metadata
+            #check if all tags key are present
+            for tags_key, tags in media.tags_keys.items():
+                assert tags_key in media.metadata
+                for tag in tags:
+                    for tag_regex in self.ignore_tags:
+                        assert not re.match(tag_regex, tag)
+            # Check for valid type
+            for key, value in media.metadata.items():
+                if value or value == '':
+                    if 'date' in key:
+                        assert isinstance(value, datetime)
+                    elif key in ('latitude', 'longitude'):
+                        assert isinstance(value, float)
+                    else:
+                        assert isinstance(value, str)
+                else:
+                    assert value is None
+            # Check if has_exif_data() is True if 'date_original' key is
+            # present, else check if it's false
+            has_exif_data = False
+            for tag in media.tags_keys['date_original']:
+                if tag in media.exif_metadata:
+                    if media.get_date_format(media.exif_metadata[tag]):
+                        has_exif_data = True
+                        assert media.has_exif_data()
+                        break
+            if has_exif_data == False:
+                assert not media.has_exif_data()
+
+        # Will be changed to get_metadata
+        # check if metatadata type are correct
+
+        # if(isinstance(self.metadata, dict) and update_cache is False):
+            # return self.metadata
+        # Album for folder implemented other place