Merge pull request #155 from jmathai/regenerate-db-gh-154

Add command to regenerate the hash database

Readme.md

@@ -52,7 +52,7 @@ You'll notice that the photo was organized into an *Unknown Location* folder. Th
 
 ### Usage Instructions
 
-You can view these instructions on the command line by typing `./elodie.py import --help` or `./elodie.py update --help`.
+You can view these instructions on the command line by typing `./elodie.py import --help`, `./elodie.py update --help` or `./elodie.py generate-db --help`.
 
 ```
 Usage: elodie.py import [OPTIONS] [PATHS]...
@@ -88,6 +88,17 @@ Options:
   --help           Show this message and exit.
 ```
 
+```
+Usage: elodie.py generate-db [OPTIONS]
+
+  Regenerate the hash.json database which contains all of the sha1
+  signatures of media files.
+
+Options:
+  --source DIRECTORY  Source of your photo library.  [required]
+  --help              Show this message and exit.
+```
+
 Now you're ready to learn more about Elodie.
 
 <p align="center"><img src ="creative/logo@300x.png" /></p>

elodie.py

@@ -20,7 +20,7 @@ from elodie import geolocation
 from elodie import log
 from elodie.filesystem import FileSystem
 from elodie.localstorage import Db
-from elodie.media.base import Base
+from elodie.media.base import Base, get_all_subclasses
 from elodie.media.media import Media
 from elodie.media.text import Text
 from elodie.media.audio import Audio
@@ -29,7 +29,6 @@ from elodie.media.video import Video
 from elodie.result import Result
 
 
-DB = Db()
 FILESYSTEM = FileSystem()
 RESULT = Result()
 
@@ -107,6 +106,42 @@ def _import(destination, source, file, album_from_folder, trash, paths, allow_du
     RESULT.write()
 
 
+@click.command('generate-db')
+@click.option('--source', type=click.Path(file_okay=False),
+              required=True, help='Source of your photo library.')
+def _generate_db(source):
+    """Regenerate the hash.json database which contains all of the sha1 signatures of media files.
+    """
+    source = os.path.abspath(os.path.expanduser(source))
+
+    extensions = set()
+    all_files = set()
+    valid_files = set()
+
+    if not os.path.isdir(source):
+        log.error('Source is not a valid directory %s' % source)
+        sys.exit(1)
+        
+    subclasses = get_all_subclasses(Base)
+    for cls in subclasses:
+        extensions.update(cls.extensions)
+
+    all_files.update(FILESYSTEM.get_all_files(source, None))
+
+    db = Db()
+    db.backup_hash_db()
+    db.reset_hash_db()
+
+    for current_file in all_files:
+        if os.path.splitext(current_file)[1][1:].lower() not in extensions:
+            log.info('Skipping invalid file %s' % current_file)
+            continue
+
+        db.add_hash(db.checksum(current_file), current_file)
+    
+    db.update_hash_db()
+
+
 def update_location(media, file_path, location_name):
     """Update location exif metadata of media.
     """
@@ -237,6 +272,7 @@ def main():
 
 main.add_command(_import)
 main.add_command(_update)
+main.add_command(_generate_db)
 
 
 if __name__ == '__main__':

elodie/localstorage.py

@@ -6,10 +6,13 @@ from builtins import object
 
 import hashlib
 import json
-from math import radians, cos, sqrt
 import os
 import sys
 
+from math import radians, cos, sqrt
+from shutil import copyfile
+from time import strftime
+
 from elodie import constants
 
 
@@ -66,6 +69,38 @@ class Db(object):
         if(write is True):
             self.update_hash_db()
 
+    # Location database
+    # Currently quite simple just a list of long/lat pairs with a name
+    # If it gets many entries a lookup might take too long and a better
+    # structure might be needed. Some speed up ideas:
+    # - Sort it and inter-half method can be used
+    # - Use integer part of long or lat as key to get a lower search list
+    # - Cache a small number of lookups, photos are likely to be taken in
+    #   clusters around a spot during import.
+    def add_location(self, latitude, longitude, place, write=False):
+        """Add a location to the database.
+
+        :param float latitude: Latitude of the location.
+        :param float longitude: Longitude of the location.
+        :param str place: Name for the location.
+        :param bool write: If true, write the location db to disk.
+        """
+        data = {}
+        data['lat'] = latitude
+        data['long'] = longitude
+        data['name'] = place
+        self.location_db.append(data)
+        if(write is True):
+            self.update_location_db()
+
+    def backup_hash_db(self):
+        """Backs up the hash db."""
+        if os.path.isfile(constants.hash_db):
+            mask = strftime('%Y-%m-%d_%H-%M-%S')
+            backup_file_name = '%s-%s' % (constants.hash_db, mask)
+            copyfile(constants.hash_db, backup_file_name)
+            return backup_file_name
+
     def check_hash(self, key):
         """Check whether a hash is present for the given key.
 
@@ -74,21 +109,6 @@ class Db(object):
         """
         return key in self.hash_db
 
-    def get_hash(self, key):
-        """Get the hash value for a given key.
-
-        :param str key:
-        :returns: str or None
-        """
-        if(self.check_hash(key) is True):
-            return self.hash_db[key]
-        return None
-
-    def update_hash_db(self):
-        """Write the hash db to disk."""
-        with open(constants.hash_db, 'w') as f:
-            json.dump(self.hash_db, f)
-
     def checksum(self, file_path, blocksize=65536):
         """Create a hash value for the given file.
 
@@ -109,30 +129,15 @@ class Db(object):
             return hasher.hexdigest()
         return None
 
-    # Location database
-    # Currently quite simple just a list of long/lat pairs with a name
-    # If it gets many entries a lookup might take too long and a better
-    # structure might be needed. Some speed up ideas:
-    # - Sort it and inter-half method can be used
-    # - Use integer part of long or lat as key to get a lower search list
-    # - Cache a small number of lookups, photos are likely to be taken in
-    #   clusters around a spot during import.
+    def get_hash(self, key):
+        """Get the hash value for a given key.
 
-    def add_location(self, latitude, longitude, place, write=False):
-        """Add a location to the database.
-
-        :param float latitude: Latitude of the location.
-        :param float longitude: Longitude of the location.
-        :param str place: Name for the location.
-        :param bool write: If true, write the location db to disk.
+        :param str key:
+        :returns: str or None
         """
-        data = {}
-        data['lat'] = latitude
-        data['long'] = longitude
-        data['name'] = place
-        self.location_db.append(data)
-        if(write is True):
-            self.update_location_db()
+        if(self.check_hash(key) is True):
+            return self.hash_db[key]
+        return None
 
     def get_location_name(self, latitude, longitude, threshold_m):
         """Find a name for a location in the database.
@@ -178,6 +183,14 @@ class Db(object):
 
         return None
 
+    def reset_hash_db(self):
+        self.hash_db = {}
+
+    def update_hash_db(self):
+        """Write the hash db to disk."""
+        with open(constants.hash_db, 'w') as f:
+            json.dump(self.hash_db, f)
+
     def update_location_db(self):
         """Write the location db to disk."""
         with open(constants.location_db, 'w') as f:

elodie/media/base.py

@@ -187,6 +187,8 @@ class Base(object):
 
     @classmethod
     def get_class_by_file(cls, _file, classes):
+        """Static method to get a media object by file.
+        """
         if not isinstance(_file, basestring) or not os.path.isfile(_file):
             return None
 
@@ -206,3 +208,21 @@ class Base(object):
         :returns: tuple(str)
         """
         return cls.extensions
+
+
+def get_all_subclasses(cls=None):
+    """Module method to get all subclasses of Base.
+    """
+    subclasses = set()
+
+    this_class = Base
+    if cls is not None:
+        this_class = cls
+
+    subclasses.add(this_class)
+
+    this_class_subclasses = this_class.__subclasses__()
+    for child_class in this_class_subclasses:
+        subclasses.update(get_all_subclasses(child_class))
+
+    return subclasses

elodie/media/text.py

@@ -25,7 +25,7 @@ class Text(Base):
     __name__ = 'Text'
 
     #: Valid extensions for text files.
-    extensions = ('txt')
+    extensions = ('txt',)
 
     def __init__(self, source=None):
         super(Text, self).__init__(source)

elodie/tests/elodie_test.py

@@ -4,7 +4,9 @@ import os
 import sys
 import shutil
 
+from click.testing import CliRunner
 from nose.plugins.skip import SkipTest
+from nose.tools import assert_raises
 
 sys.path.insert(0, os.path.abspath(os.path.dirname(os.path.dirname(os.path.dirname(os.path.realpath(__file__))))))
 sys.path.insert(0, os.path.abspath(os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(os.path.realpath(__file__)))))))
@@ -13,6 +15,7 @@ import helper
 elodie = load_source('elodie', os.path.abspath('{}/../../elodie.py'.format(os.path.dirname(os.path.realpath(__file__)))))
 
 from elodie import constants
+from elodie.localstorage import Db
 from elodie.media.audio import Audio
 from elodie.media.photo import Photo
 from elodie.media.text import Text
@@ -334,6 +337,48 @@ def test_update_time_on_video():
     assert metadata['date_taken'] != metadata_processed['date_taken']
     assert metadata_processed['date_taken'] == helper.time_convert((2000, 1, 1, 12, 0, 0, 5, 1, 0)), metadata_processed['date_taken']
 
+def test_regenerate_db_invalid_source():
+    runner = CliRunner()
+    result = runner.invoke(elodie._generate_db, ['--source', '/invalid/path'])
+    assert result.exit_code == 1, result.exit_code
+
+def test_regenerate_valid_source():
+    temporary_folder, folder = helper.create_working_folder()
+
+    origin = '%s/valid.txt' % folder
+    shutil.copyfile(helper.get_file('valid.txt'), origin)
+
+    reset_hash_db()
+    runner = CliRunner()
+    result = runner.invoke(elodie._generate_db, ['--source', folder])
+    db = Db()
+    restore_hash_db()
+
+    shutil.rmtree(folder)
+
+    assert result.exit_code == 0, result.exit_code
+    assert 'bde2dc0b839a5d20b0b4c1f57605f84e0e2a4562aaebc1c362de6cb7cc02eeb3' in db.hash_db, db.hash_db
+
+def test_regenerate_valid_source_with_invalid_files():
+    temporary_folder, folder = helper.create_working_folder()
+
+    origin_valid = '%s/valid.txt' % folder
+    shutil.copyfile(helper.get_file('valid.txt'), origin_valid)
+    origin_invalid = '%s/invalid.invalid' % folder
+    shutil.copyfile(helper.get_file('invalid.invalid'), origin_invalid)
+
+    reset_hash_db()
+    runner = CliRunner()
+    result = runner.invoke(elodie._generate_db, ['--source', folder])
+    db = Db()
+    restore_hash_db()
+
+    shutil.rmtree(folder)
+
+    assert result.exit_code == 0, result.exit_code
+    assert 'bde2dc0b839a5d20b0b4c1f57605f84e0e2a4562aaebc1c362de6cb7cc02eeb3' in db.hash_db, db.hash_db
+    assert 'e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855' not in db.hash_db, db.hash_db
+
 def reset_hash_db():
     hash_db = constants.hash_db
     if os.path.isfile(hash_db):

elodie/tests/localstorage_test.py

@@ -62,6 +62,14 @@ def test_add_hash_explicit_write():
     # Instnatiate new db class to confirm random_key exists
     db2 = Db()
     assert db2.check_hash(random_key) == True
+
+def test_backup_hash_db():
+    db = Db()
+    backup_file_name = db.backup_hash_db()
+    file_exists = os.path.isfile(backup_file_name)
+    os.remove(backup_file_name)
+    
+    assert file_exists, backup_file_name
     
 def test_check_hash_exists():
     db = Db()

elodie/tests/media/base_test.py

@@ -14,7 +14,7 @@ sys.path.insert(0, os.path.abspath(os.path.dirname(os.path.dirname(os.path.dirna
 sys.path.insert(0, os.path.abspath(os.path.dirname(os.path.dirname(os.path.realpath(__file__)))))
 
 import helper
-from elodie.media.base import Base
+from elodie.media.base import Base, get_all_subclasses
 from elodie.media.media import Media
 from elodie.media.audio import Audio
 from elodie.media.text import Text
@@ -106,3 +106,8 @@ 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