Initial set of improvementes merged into master, including a class hierarchy redesign and performance improvements

Merge commit 'origin/improvements'

* commit 'origin/improvements': (38 commits)
  test_performance: module containing benchmarks to get an idea of the achieved throughput
  Removed plenty of mocked tree tests as they cannot work anymore with persistent commands that require stdin AND binary data - not even an adapter would help here. These tests will have to be replaced.
  tree: now reads tress directly by parsing the binary data, allowing it to safe possibly hundreds of command calls
  Refs are now truly dynamic - this costs a little bit of (persistent command) work, but assures refs behave as expected
  persistent command signature changed to also return the hexsha from a possible input ref - the objects pointed to by refs are now baked on demand - perhaps it should change to always be re-retrieved using a property as it is relatively fast - this way refs can always be cached
  test_blob: removed many redundant tests that would fail now as the mock cannot handle the complexity of the command backend
  Implemented git command facility to keep persistent commands for fast object information retrieval
  test: Added time-consuming test which could also be a benchmark in fact - currently it cause hundreds of command invocations which is slow
  cmd: added option to return the process directly, allowing to read the output directly from the output stream
  added Iterable interface to Ref type
  renamed find_all to list_all, changed commit to use iterable interface in preparation for command changes
  Added base for all iteratable objects
  unified name of utils module, recently it was named util and utils in different packages
  tree: renamed content_from_string to _from_string to make it private. Removed tests that were testing that method
  tree: now behaves like a list with string indexing functionality - using a dict as cache is a problem as the tree is ordered, added blobs, trees and traverse method
  test_base: Improved basic object creation as well as set hash tests
  repo.active_branch now returns a Head object, not a string
  IndexObjects are now checking their slots to raise a proper error message in case someone tries to access an unset path or mode - this information cannot be retrieved afterwards as IndexObject information is kept in the object that pointed at them. To find this information, one would have to search all objects which is not feasible
  refs now take repo as first argument and derive from LazyMixin to allow them to dynamically retrieve their objects
  renamed from_string and list_from_string to _from_string and _list_from_string to indicate their new status as private method, adjusted all callers respectively
  ...

1 files changed, 151 insertions, 0 deletions

diff --git a/lib/git/objects/base.py b/lib/git/objects/base.py
new file mode 100644
index 00000000..07538ada
--- /dev/null
+++ b/lib/git/objects/base.py
@@ -0,0 +1,151 @@
+# base.py
+# Copyright (C) 2008, 2009 Michael Trier (mtrier@gmail.com) and contributors
+#
+# This module is part of GitPython and is released under
+# the BSD License: http://www.opensource.org/licenses/bsd-license.php
+import os
+from git.utils import LazyMixin
+	
+_assertion_msg_format = "Created object %r whose python type %r disagrees with the acutal git object type %r"
+
+class Object(LazyMixin):
+	"""
+	Implements an Object which may be Blobs, Trees, Commits and Tags
+	"""
+	TYPES = ("blob", "tree", "commit", "tag")
+	__slots__ = ("repo", "id", "size", "data" )
+	type = None			# to be set by subclass
+	
+	def __init__(self, repo, id):
+		"""
+		Initialize an object by identifying it by its id. All keyword arguments
+		will be set on demand if None.
+		
+		``repo``
+			repository this object is located in
+			
+		``id``
+			SHA1 or ref suitable for git-rev-parse
+		"""
+		super(Object,self).__init__()
+		self.repo = repo
+		self.id = id
+		
+	def _set_self_from_args_(self, args_dict):
+		"""
+		Initialize attributes on self from the given dict that was retrieved
+		from locals() in the calling method.
+		
+		Will only set an attribute on self if the corresponding value in args_dict
+		is not None
+		"""
+		for attr, val in args_dict.items():
+			if attr != "self" and val is not None:
+				setattr( self, attr, val )
+		# END set all non-None attributes
+	
+	def _set_cache_(self, attr):
+		"""
+		Retrieve object information
+		"""
+		if attr  == "size":
+			hexsha, typename, self.size = self.repo.git.get_object_header(self.id)
+			assert typename == self.type, _assertion_msg_format % (self.id, typename, self.type)
+		elif attr == "data":
+			hexsha, typename, self.size, self.data = self.repo.git.get_object_data(self.id)
+			assert typename == self.type, _assertion_msg_format % (self.id, typename, self.type)
+		else:
+			super(Object,self)._set_cache_(attr)
+		
+	def __eq__(self, other):
+		"""
+		Returns
+			True if the objects have the same SHA1
+		"""
+		return self.id == other.id
+		
+	def __ne__(self, other):
+		"""
+		Returns
+			True if the objects do not have the same SHA1
+		"""
+		return self.id != other.id
+		
+	def __hash__(self):
+		"""
+		Returns
+			Hash of our id allowing objects to be used in dicts and sets
+		"""
+		return hash(self.id)
+		
+	def __str__(self):
+		"""
+		Returns
+			string of our SHA1 as understood by all git commands
+		"""
+		return self.id
+		
+	def __repr__(self):
+		"""
+		Returns
+			string with pythonic representation of our object
+		"""
+		return '<git.%s "%s">' % (self.__class__.__name__, self.id)
+
+
+class IndexObject(Object):
+	"""
+	Base for all objects that can be part of the index file , namely Tree, Blob and
+	SubModule objects
+	"""
+	__slots__ = ("path", "mode") 
+	
+	def __init__(self, repo, id, mode=None, path=None):
+		"""
+		Initialize a newly instanced IndexObject
+		``repo``
+			is the Repo we are located in
+
+		``id`` : string
+			is the git object id as hex sha
+
+		``mode`` : int
+			is the file mode as int, use the stat module to evaluate the infomration
+
+		``path`` : str
+			is the path to the file in the file system, relative to the git repository root, i.e.
+			file.ext or folder/other.ext
+				
+		NOTE
+			Path may not be set of the index object has been created directly as it cannot
+			be retrieved without knowing the parent tree.
+		"""
+		super(IndexObject, self).__init__(repo, id)
+		self._set_self_from_args_(locals())
+		if isinstance(mode, basestring):
+			self.mode = self._mode_str_to_int(mode)
+	
+	def _set_cache_(self, attr):
+		if attr in IndexObject.__slots__:
+			# they cannot be retrieved lateron ( not without searching for them )
+			raise AttributeError( "path and mode attributes must have been set during %s object creation" % type(self).__name__ )
+		else:
+			super(IndexObject, self)._set_cache_(attr)
+	
+	@classmethod
+	def _mode_str_to_int(cls, modestr):
+		"""
+		``modestr``
+			string like 755 or 644 or 100644 - only the last 3 chars will be used
+			
+		Returns
+			String identifying a mode compatible to the mode methods ids of the 
+			stat module regarding the rwx permissions for user, group and other
+		"""
+		mode = 0
+		for iteration,char in enumerate(reversed(modestr[-3:])):
+			mode += int(char) << iteration*3
+		# END for each char
+		return mode
+	  
+