Index: Is now diffable and appears to properly implement diffing against other items as well as the working tree

Diff.Diffable: added callback allowing superclasses to preprocess diff arguments
Diff.Diff: added eq, ne and hash methods, string methods would be nice

Author: Byron

TODO

@@ -58,6 +58,15 @@ Index
   creating several tree objects, so in the end it might be slower. 
   Hmm, probably its okay to use the command unless we go c(++)
 
+
+Head.reset
+----------
+* Should better be an instance method. Problem was that there is no class specifying
+  the HEAD  - in a way reset would always effect the active branch.
+  Probably it would be okay to have a special type called SymbolicReference
+  which represents items like HEAD. These could naturally carry the reset 
+  instance method.
+  
 Remote
 ------
 * 'push' method needs a test, a true test repository is required though, a fork 

lib/git/diff.py

@@ -18,13 +18,18 @@ class Diffable(object):
 	"""
 	__slots__ = tuple()
 
-	# subclasses provide additional arguments to the git-diff comamnd by supplynig 
-	# them in this tuple
-	_diff_args = tuple()
-	
-	# Temporary standin for Index type until we have a real index type
+	# standin indicating you want to diff against the index
 	class Index(object):
 		pass 
+		
+	def _process_diff_args(self, args):
+		"""
+		Returns
+			possibly altered version of the given args list.
+			Method is called right before git command execution.
+			Subclasses can use it to alter the behaviour of the superclass
+		"""
+		return args
 
 	def diff(self, other=Index, paths=None, create_patch=False, **kwargs):
 		"""
@@ -60,13 +65,13 @@ def diff(self, other=Index, paths=None, create_patch=False, **kwargs):
 			On a bare repository, 'other' needs to be provided as Index or as 
 			as Tree/Commit, or a git command error will occour
 		"""
-		args = list(self._diff_args[:])
+		args = list()
 		args.append( "--abbrev=40" )		# we need full shas
 		args.append( "--full-index" )		# get full index paths, not only filenames
 
 		if create_patch:
 			args.append("-p")
-			args.append("-M") # check for renames
+			args.append("-M") 				# check for renames
 		else:
 			args.append("--raw")
 
@@ -87,7 +92,7 @@ def diff(self, other=Index, paths=None, create_patch=False, **kwargs):
 		# END paths handling
 
 		kwargs['as_process'] = True
-		proc = self.repo.git.diff(*args, **kwargs)
+		proc = self.repo.git.diff(*self._process_diff_args(args), **kwargs)
 
 		diff_method = Diff._index_from_raw_format
 		if create_patch:
@@ -96,7 +101,7 @@ def diff(self, other=Index, paths=None, create_patch=False, **kwargs):
 
 		status = proc.wait()
 		if status != 0:
-			raise GitCommandError("git-diff", status, proc.stderr )
+			raise GitCommandError(("git diff",)+tuple(args), status, proc.stderr.read())
 
 		return index
 
@@ -207,6 +212,20 @@ def __init__(self, repo, a_path, b_path, a_blob_id, b_blob_id, a_mode,
 
 		self.diff = diff
 
+
+	def __eq__(self, other):
+		for name in self.__slots__:
+			if getattr(self, name) != getattr(other, name):
+				return False
+		# END for each name
+		return True
+		
+	def __ne__(self, other):
+		return not ( self == other )
+		
+	def __hash__(self):
+		return hash(tuple(getattr(self,n) for n in self.__slots__))
+
 	@property
 	def renamed(self):
 		"""

lib/git/errors.py

@@ -25,8 +25,8 @@ def __init__(self, command, status, stderr=None):
 		self.stderr = stderr
 		self.status = status
 		self.command = command
-
+		
 	def __str__(self):
-		return repr("'%s' returned exit status %d: %r" %
-					(' '.join(self.command), self.status, str(self.stderr)))
+		return ("'%s' returned exit status %i: %s" %
+					(' '.join(str(i) for i in self.command), self.status, self.stderr))
 

lib/git/index.py

@@ -14,9 +14,11 @@
 import tempfile
 import os
 import stat
-from git.objects import Blob, Tree
+import git.diff as diff
+
+from git.objects import Blob, Tree, Object
 from git.utils import SHA1Writer, LazyMixin, ConcurrentWriteOperation
-from git.diff import Diffable
+
 
 class _TemporaryFileSwap(object):
 	"""
@@ -140,7 +142,7 @@ def from_blob(cls, blob):
 		return IndexEntry((time, time, 0, 0, blob.mode, 0, 0, blob.size, blob.id, 0, blob.path))
 
 
-class Index(LazyMixin):
+class Index(LazyMixin, diff.Diffable):
 	"""
 	Implements an Index that can be manipulated using a native implementation in 
 	order to save git command function calls wherever possible.
@@ -154,7 +156,7 @@ class Index(LazyMixin):
 	The index contains an entries dict whose keys are tuples of type IndexEntry
 	to facilitate access.
 	"""
-	__slots__ = ( "repo", "version", "entries", "_extension_data" )
+	__slots__ = ( "repo", "version", "entries", "_extension_data", "_is_default_index" )
 	_VERSION = 2			# latest version we support
 	S_IFGITLINK	= 0160000
 
@@ -168,10 +170,13 @@ def __init__(self, repo, stream = None):
 		self.repo = repo
 		self.version = self._VERSION
 		self._extension_data = ''
+		self._is_default_index = True
 		if stream is not None:
+			self._is_default_index = False
 			self._read_from_stream(stream)
 		# END read from stream immediatly
 
+	
 	def _set_cache_(self, attr):
 		if attr == "entries":
 			# read the current index
@@ -187,6 +192,18 @@ def _set_cache_(self, attr):
 	def _index_path(self):
 		return os.path.join(self.repo.path, "index")
 
+	
+	@property
+	def path(self):
+		"""
+		Returns 
+			Path to the index file we are representing or None if we are 
+			a loose index that was read from a stream.
+		"""
+		if self._is_default_index:
+			return self._index_path()
+		return None
+	
 	@classmethod
 	def _read_entry(cls, stream):
 		"""Return: One entry of the given stream"""
@@ -535,4 +552,51 @@ def write_tree(self):
 		# END write tree handling 
 		return Tree(self.repo, tree_sha, 0, '')
 
+		
+	def _process_diff_args(self, args):
+		try:
+			args.pop(args.index(self))
+		except IndexError:
+			pass
+		# END remove self
+		return args
+		
+	def diff(self, other=diff.Diffable.Index, paths=None, create_patch=False, **kwargs):
+		"""
+		Diff this index against the working copy or a Tree or Commit object
+		
+		For a documentation of the parameters and return values, see 
+		Diffable.diff
+		
+		Note
+			Will only work with indices that represent the default git index as 
+			they have not been initialized with a stream.
+		"""
+		if not self._is_default_index:
+			raise AssertionError( "Cannot diff custom indices as they do not represent the default git index" )
+		
+		# index against index is always empty
+		if other is self.Index:
+			return diff.DiffIndex()
+			
+		# index against anything but None is a reverse diff with the respective
+		# item. Handle existing -R flags properly. Transform strings to the object
+		# so that we can call diff on it
+		if isinstance(other, basestring):
+			other = Object.new(self.repo, other)
+		# END object conversion
+		
+		if isinstance(other, Object):
+			# invert the existing R flag
+			cur_val = kwargs.get('R', False)
+			kwargs['R'] = not cur_val
+			return other.diff(self.Index, paths, create_patch, **kwargs)
+		# END diff against other item handlin
+		
+		# if other is not None here, something is wrong 
+		if other is not None:
+			raise ValueError( "other must be None, Diffable.Index, a Tree or Commit, was %r" % other )
+		
+		# diff against working copy - can be handled by superclass natively
+		return super(Index, self).diff(other, paths, create_patch, **kwargs)
 

lib/git/objects/__init__.py

@@ -2,6 +2,7 @@
 Import all submodules main classes into the package space
 """
 import inspect
+from base import *
 from tag import *
 from blob import *
 from tree import *

test/git/test_diff.py

@@ -59,6 +59,14 @@ def test_diff_interface(self):
 								assertion_map.setdefault(key, 0)
 								assertion_map[key] = assertion_map[key]+len(list(diff_index.iter_change_type(ct)))  
 							# END for each changetype
+							
+							# check entries
+							diff_set = set()
+							diff_set.add(diff_index[0])
+							diff_set.add(diff_index[0])
+							assert len(diff_set) == 1
+							assert diff_index[0] == diff_index[0]
+							assert not (diff_index[0] != diff_index[0])
 						# END diff index checking 
 					# END for each patch option
 				# END for each path option
@@ -71,7 +79,4 @@ def test_diff_interface(self):
 		for key,value in assertion_map.items():
 			assert value, "Did not find diff for %s" % key
 		# END for each iteration type 
-		
-	def test_diff_index_working_tree(self):
-		self.fail("""Find a good way to diff an index against the working tree
-which is not possible with the current interface""")
+	

test/git/test_index.py

@@ -102,12 +102,45 @@ def test_from_tree(self):
 		assert num_blobs == len(three_way_index.entries)
 
 	@with_rw_repo('0.1.6')
-	def test_from_index(self, rw_repo):
+	def test_from_index_and_diff(self, rw_repo):
 		# default Index instance points to our index
 		index = Index(rw_repo)
+		assert index.path is not None
 		assert len(index.entries)
 
 		# write the file back
 		index.write()
 
 		# could sha it, or check stats
+		
+		# test diff
+		# resetting the head will leave the index in a different state, and the 
+		# diff will yield a few changes
+		cur_head_commit = rw_repo.head.commit
+		ref = rw_repo.head.reset(rw_repo, 'HEAD~6', index=True, working_tree=False)
+		
+		# diff against same index is 0
+		diff = index.diff()
+		assert len(diff) == 0
+		
+		# against HEAD as string, must be the same as it matches index
+		diff = index.diff('HEAD')
+		assert len(diff) == 0
+		
+		# against previous head, there must be a difference
+		diff = index.diff(cur_head_commit)
+		assert len(diff)
+		
+		# we reverse the result
+		adiff = index.diff(str(cur_head_commit), R=True)
+		odiff = index.diff(cur_head_commit, R=False)	# now its not reversed anymore
+		assert adiff != odiff
+		assert odiff == diff					# both unreversed diffs against HEAD
+		
+		# against working copy - its still at cur_commit
+		wdiff = index.diff(None)
+		assert wdiff != adiff
+		assert wdiff != odiff
+		
+		# against something unusual
+		self.failUnlessRaises(ValueError, index.diff, int)