[TASK] Performance related changes.

if foo is not None is slightly faster than
if foo != None.
for foo in my_dict is faster than
for foo in my_dict.iterkeys()

hshetl/__init__.py

@@ -150,7 +150,7 @@ class NameResolver(object):
             @functools.wraps(fn)
             def decorated(*args, **kwargs):
                 m_args, _, _, defaults = inspect.getargspec(fn)
-                defaults = dict(zip(m_args[-len(defaults):], defaults)) if defaults != None else {}
+                defaults = dict(zip(m_args[-len(defaults):], defaults)) if defaults is not None else {}
                 if self.argument_name in kwargs:
                     self.argument_value = kwargs.get(self.argument_name)
                 elif self.argument_name in defaults:
@@ -211,7 +211,7 @@ class AbstractRepository(OrderedDict, object):
         '''Return if the given reference or object is known by the repository'''
         if isinstance(ref, self.cls):
             ref = getattr(ref, self.ref_attr)
-        return ref in self.iterkeys()
+        return ref in self
 
     def extend(self, items):
         '''Extend the repository with a list of entities'''

hshetl/entities.py

@@ -162,7 +162,7 @@ class Container(list):
         return mapped_record
 
     def is_regular_property(self, prop):
-        return prop in self.properties.iterkeys() or prop in self.mapping.iterkeys()
+        return prop in self.properties or prop in self.mapping
 
     def _map_property(self, prop, raw_value, reverse = False):
         '''Look up the name mapping and perform a type conversion.'''
@@ -333,7 +333,7 @@ class Record(object):
         self.identifier_dicts[container.name] = None
         self.operation[container.name] = OPERATION_UNKNOWN
         self.last_container = container
-        for prop in self.properties.iterkeys():
+        for prop in self.properties:
             self.container_properties[container.name][prop] = None
 
     def load(self, identifier_dictionary, properties, container, operation = OPERATION_UNKNOWN):
@@ -384,7 +384,7 @@ class Record(object):
 
     def dirty(self, source_container, target_container):
         '''Returns whether the entity is dirty or not'''
-        for prop in self.properties.iterkeys():
+        for prop in self.properties:
             if self.is_property_dirty(prop, source_container, target_container):
                 return True
         return False
@@ -399,7 +399,7 @@ class Record(object):
 
     def _set_container_identifier(self, identifier_dictionary, container):
         '''Set the container identifier. Ensures that this only happends once.'''
-        if container.name in self.identifier_dicts.iterkeys() and self.identifier_dicts[container.name] != None:
+        if container.name in self.identifier_dicts and self.identifier_dicts[container.name] is not None:
             msg = 'For one record each identifier is only allowed to be set once. Existing id: %s, new id: %s, join id: %s' % (self.identifier_dicts[container.name], identifier_dictionary, self.identifier)
             raise DuplicatedSystemIdException(msg)
         self.identifier_dicts[container.name] = identifier_dictionary

hshetl/extractors.py

@@ -49,7 +49,7 @@ class AbstractExtractor(object):
     def __init__(self, connector = None):
         '''Create a extractor.'''
         logging.debug('Initializing extractor: ' + self.__class__.__name__)
-        if connector != None and not self.can_execute(connector):
+        if connector is not None and not self.can_execute(connector):
             raise ExtractorException('The extractor {} can not be executed using this connector {}'.format(self, connector))
         self.connector = connector
         '''The connector used by this extractor.'''
@@ -72,7 +72,7 @@ class AbstractExtractor(object):
         '''
         if not self.can_execute(self.connector):
             raise ExtractorException('The extractor {} can not be executed using this connector {}'.format(self, self.connector))
-        logging.info('Extracting data from %s with %s' % (self.connector.name if self.connector != None else 'Fake', unicode(self)))
+        logging.info('Extracting data from %s with %s' % (self.connector.name if self.connector is not None else 'Fake', unicode(self)))
         result = self._execute()
         logging.info('Extraction resolved %d records' % len(result))
         return result
@@ -373,7 +373,7 @@ class FakerExtractor(AbstractExtractor):
         '''The properties that will be faked.'''
         self._faker = Factory.create(locale, providers = providers)
         '''The faker factory used to generate the data.'''
-        if seed != None:
+        if seed is not None:
             self._faker.seed(seed)
 
     def can_execute(self, connector):

hshetl/jobs.py

@@ -133,7 +133,7 @@ class Job(object):
             choice = raw_input()
             if len(choice) and choice[0].lower() in ('y', 'n'):
                 return choice == 'y'
-            elif default != None and not len(choice):
+            elif default is not None and not len(choice):
                 return default
             else:
                 sys.stdout.write('Please enter `y` or `n`.\n')
@@ -440,7 +440,7 @@ class ExtractionJob(EntityJob, ConnectorJob):
         if not isinstance(extractor, AbstractExtractor):
             raise ConfigurationException('The extractor needs to be an instance of hshetl.extractors.AbstractExtractor. Encoutered: {}'.format(str(type(extractor))))
         self.extractor = extractor
-        if self.connector != None and self.extractor.connector == None:
+        if self.connector is not None and self.extractor.connector == None:
             self.extractor.connector = self.connector
         '''The extractor used to fetch data.'''
         if name is None: self.name = self.entity.name + '_' + self.extractor.connector.name
@@ -498,14 +498,14 @@ class TransformerJob(EntityJob, ConnectorJob):
             raise ConfigurationException('The transformer needs to be an instance of hshetl.loaders.AbstractTransformer. Encoutered: {}'.format(str(type(transformer))))
         self.transformer = transformer
         '''The transformer used by the job.'''
-        if self.entity != None:
+        if self.entity is not None:
             self.source = NameResolver(self.entity, 'source', source)()
             self.target = NameResolver(self.entity, 'target', target)()
             if self.transformer.entity == None:
                 self.transformer.entity = self.entity
-        if self.source != None and self.transformer.source == None:
+        if self.source is not None and self.transformer.source == None:
             self.transformer.source = self.source
-        if self.target != None and self.transformer.target == None:
+        if self.target is not None and self.transformer.target == None:
             self.transformer.target = self.target
         self.transformer.name = self.name
 
@@ -645,7 +645,7 @@ class LoadJob(EntityJob, ConnectorJob):
             raise ConfigurationException('The loader needs to be an instance of hshetl.loaders.AbstractLoader. Encoutered: {}'.format(str(type(loader))))
         self.loader = loader
         '''The loader used to write the data in the database.'''
-        if self.connector != None and self.loader.connector == None:
+        if self.connector is not None and self.loader.connector == None:
             self.loader.connector = self.connector
         self.source = source
         '''The entity related container, that data will be written.'''

hshetl/loaders.py

@@ -50,7 +50,7 @@ class AbstractLoader(object):
     def __init__(self, connector = None, operations = ['insert', 'update', 'delete']):
         '''Initializes the loader.'''
         logging.debug('Initializing loader: ' + self.__class__.__name__)
-        if connector != None and not self.can_execute(connector):
+        if connector is not None and not self.can_execute(connector):
             raise LoaderException('The loader {} can not be executed using this connector {}'.format(self, connector))
         self.connector = connector
         '''The connector this loader uses to write the data.'''
@@ -237,7 +237,7 @@ class SqlAlchemyLoader(AbstractLoader):
         if identifiers == {}:
             raise ConfigurationException('No identifier given, cannot perform update or delete without identifier')
         for column, value in identifiers.iteritems():
-            if value != None:
+            if value is not None:
                 logging.debug('Adding where clause for column ' + str(column) + ' with value ' + repr(value))
                 statement = statement.where(self.table.c.get(column) == value)
             else:

hshetl/transformers.py

@@ -49,7 +49,7 @@ class AbstractTransformer(object):
 
     def can_execute(self):
         '''Whether the Transformer has everything set needed for its transform operation.'''
-        return self.entity != None
+        return self.entity is not None
 
     def execute(self):
         '''Execute the transformer.
@@ -117,13 +117,13 @@ class CompareTransformer(AbstractTransformer):
         '''The target container to work on.'''
         self.mapping = mapping
         '''If you do not want the properties to be mapped like in target system, provide a mapping here.'''
-        if self.entity != None:
+        if self.entity is not None:
             self.source = NameResolver(self.entity, 'source', source)()
             self.target = NameResolver(self.entity, 'target', target)()
 
     def can_execute(self):
         '''Whether the Transformer has everything set needed for its transform operation.'''
-        return self.entity != None and self.source != None and self.target != None
+        return self.entity is not None and self.source is not None and self.target is not None
 
     def _execute(self):
         '''Execute the compare operation.