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

Author: Byron

lib/git/cmd.py

@@ -272,7 +272,7 @@ def _parse_object_header(self, header_line):
 			<hex_sha> type_string size_as_int
 			
 		Returns
-			(type_string, size_as_int)
+			(hex_sha, type_string, size_as_int)
 			
 		Raises
 			ValueError if the header contains indication for an error due to incorrect 
@@ -282,7 +282,7 @@ def _parse_object_header(self, header_line):
 		if len(tokens) != 3:
 			raise ValueError( "SHA named %s could not be resolved" % tokens[0] )
 
-		return (tokens[1], int(tokens[2]))
+		return (tokens[0], tokens[1], int(tokens[2]))
 
 	def __prepare_ref(self, ref):
 		# required for command to separate refs on stdin
@@ -318,7 +318,7 @@ def get_object_header(self, ref):
 			once and reuses the command in subsequent calls. 
 		
 		Return:
-			(type_string, size_as_int)
+			(hexsha, type_string, size_as_int)
 		"""
 		cmd = self.__get_persistent_cmd("cat_file_header", "cat_file", batch_check=True)
 		return self.__get_object_header(cmd, ref)
@@ -328,11 +328,11 @@ def get_object_data(self, ref):
 		As get_object_header, but returns object data as well
 		
 		Return:
-			(type_string, size_as_int,data_string)
+			(hexsha, type_string, size_as_int,data_string)
 		"""
 		cmd = self.__get_persistent_cmd("cat_file_all", "cat_file", batch=True)
-		typename, size = self.__get_object_header(cmd, ref)
+		hexsha, typename, size = self.__get_object_header(cmd, ref)
 		data = cmd.stdout.read(size)
 		cmd.stdout.read(1)		# finishing newlines
 
-		return (typename, size, data)
+		return (hexsha, typename, size, data)

lib/git/objects/base.py

@@ -48,10 +48,10 @@ def _set_cache_(self, attr):
 		Retrieve object information
 		"""
 		if attr  == "size":
-			typename, self.size = self.repo.git.get_object_header(self.id)
+			hexsha, typename, self.size = self.repo.git.get_object_header(self.id)
 			assert typename == self.type, "Created object whose python type %r disagrees with the acutal git object type %r" % (typename, self.type)
 		elif attr == "data":
-			typename, self.size, self.data = self.repo.git.get_object_data(self.id)
+			hexsha, typename, self.size, self.data = self.repo.git.get_object_data(self.id)
 			assert typename == self.type, "Created object whose python type %r disagrees with the acutal git object type %r" % (typename, self.type)
 		else:
 			super(Object,self)._set_cache_(attr)

lib/git/refs.py

@@ -38,10 +38,11 @@ def _set_cache_(self, attr):
 		if attr == "object":
 			# have to be dynamic here as we may be a tag which can point to anything
 			# it uses our path to stay dynamic
-			typename, size = self.repo.git.get_object_header(self.path)
-			# explicitly do not set the size as it may change if the our ref path points 
-			# at some other place when the head changes for instance ... 
-			self.object = get_object_type_by_name(typename)(self.repo, self.path)
+			hexsha, typename, size = self.repo.git.get_object_header(self.path)
+			# pin-point our object to a specific sha, even though it might not 
+			# reflect the our cached object anymore in case our rev now points 
+			# to a different commit
+			self.object = get_object_type_by_name(typename)(self.repo, hexsha)
 		else:
 			super(Ref, self)._set_cache_(attr)
 
@@ -128,7 +129,9 @@ def _from_string(cls, repo, line):
 		full_path, hexsha, type_name, object_size = line.split("\x00")
 
 		# No, we keep the object dynamic by allowing it to be retrieved by
-		# our path on demand - due to perstent commands it is fast
+		# our path on demand - due to perstent commands it is fast.
+		# This reduces the risk that the object does not match 
+		# the changed ref anymore in case it changes in the meanwhile
 		return cls(repo, full_path)
 
 		# obj = get_object_type_by_name(type_name)(repo, hexsha)

test/git/test_git.py

@@ -84,6 +84,6 @@ def test_persistent_cat_file_command(self):
 
 
 		# same can be achived using the respective command functions
-		typename, size =  self.git.get_object_header(hexsha)
-		typename_two, size_two, data = self.git.get_object_data(hexsha)
+		hexsha, typename, size =  self.git.get_object_header(hexsha)
+		hexsha, typename_two, size_two, data = self.git.get_object_data(hexsha)
 		assert typename == typename_two and size == size_two