Merge branch 'fromtree'

Author: Byron

lib/git/index/base.py

@@ -59,14 +59,16 @@
 						)
 
 from fun import (
+					entry_key,
 					write_cache,
 					read_cache,
-					write_tree_from_cache,
-					entry_key
+					aggressive_tree_merge,
+					write_tree_from_cache
 				)
 
 from gitdb.base import IStream
 from gitdb.db import MemoryDB
+from itertools import izip
 
 __all__ = ( 'IndexFile', 'CheckoutError' )
 
@@ -133,7 +135,6 @@ def _set_cache_(self, attr):
 	def _index_path(self):
 		return join_path_native(self.repo.git_dir, "index")
 
-
 	@property
 	def path(self):
 		""" :return: Path to the index file we are representing """
@@ -240,6 +241,31 @@ def merge_tree(self, rhs, base=None):
 		self.repo.git.read_tree(args)
 		return self
 
+	@classmethod
+	def new(cls, repo, *tree_sha):
+		""" Merge the given treeish revisions into a new index which is returned.
+		This method behaves like git-read-tree --aggressive when doing the merge.
+
+		:param repo: The repository treeish are located in.
+
+		:param *tree_sha:
+			see ``from_tree``
+
+		:return:
+			New IndexFile instance. Its path will be undefined. 
+			If you intend to write such a merged Index, supply an alternate file_path 
+			to its 'write' method."""
+		base_entries = aggressive_tree_merge(repo.odb, [str(t) for t in tree_sha])
+		
+		inst = cls(repo)
+		# convert to entries dict
+		entries = dict(izip(((e.path, e.stage) for e in base_entries), 
+							(IndexEntry.from_base(e) for e in base_entries)))
+		
+		inst.entries = entries
+		return inst
+
+
 	@classmethod
 	def from_tree(cls, repo, *treeish, **kwargs):
 		"""
@@ -275,8 +301,7 @@ def from_tree(cls, repo, *treeish, **kwargs):
 
 			As the underlying git-read-tree command takes into account the current index,
 			it will be temporarily moved out of the way to assure there are no unsuspected
-			interferences.
-		"""
+			interferences."""
 		if len(treeish) == 0 or len(treeish) > 3:
 			raise ValueError("Please specify between 1 and 3 treeish, got %i" % len(treeish))
 

lib/git/index/fun.py

@@ -5,15 +5,19 @@
 from stat import S_IFDIR
 from cStringIO import StringIO
 
+from git.utils import IndexFileSHA1Writer
 from git.errors import UnmergedEntriesError
-from git.objects.fun import tree_to_stream
-from git.utils import (
-							IndexFileSHA1Writer, 
-						)
+from git.objects.fun import (
+								tree_to_stream,
+								traverse_tree_recursive,
+								traverse_trees_recursive
+							)
 
 from typ import (
+					BaseIndexEntry,
 					IndexEntry,
-					CE_NAMEMASK
+					CE_NAMEMASK, 
+					CE_STAGESHIFT
 				)
 
 from util import 	(
@@ -23,7 +27,6 @@
 
 from gitdb.base import IStream
 from gitdb.typ import str_tree_type
-from binascii import a2b_hex
 
 __all__ = ('write_cache', 'read_cache', 'write_tree_from_cache', 'entry_key' )
 
@@ -75,15 +78,16 @@ def write_cache(entries, stream, extension_data=None, ShaStreamCls=IndexFileSHA1
 def read_entry(stream):
 	"""Return: One entry of the given stream"""
 	beginoffset = stream.tell()
-	ctime = unpack(">8s", stream.read(8))[0]
-	mtime = unpack(">8s", stream.read(8))[0]
+	read = stream.read
+	ctime = unpack(">8s", read(8))[0]
+	mtime = unpack(">8s", read(8))[0]
 	(dev, ino, mode, uid, gid, size, sha, flags) = \
-		unpack(">LLLLLL20sH", stream.read(20 + 4 * 6 + 2))
+		unpack(">LLLLLL20sH", read(20 + 4 * 6 + 2))
 	path_size = flags & CE_NAMEMASK
-	path = stream.read(path_size)
+	path = read(path_size)
 
 	real_size = ((stream.tell() - beginoffset + 8) & ~7)
-	data = stream.read((beginoffset + real_size) - stream.tell())
+	data = read((beginoffset + real_size) - stream.tell())
 	return IndexEntry((mode, sha, flags, path, ctime, mtime, dev, ino, uid, gid, size))
 
 def read_header(stream):
@@ -150,6 +154,7 @@ def write_tree_from_cache(entries, odb, sl, si=0):
 	:return: tuple(binsha, list(tree_entry, ...)) a tuple of a sha and a list of 
 		tree entries being a tuple of hexsha, mode, name"""
 	tree_items = list()
+	tree_items_append = tree_items.append
 	ci = sl.start
 	end = sl.stop
 	while ci < end:
@@ -161,7 +166,7 @@ def write_tree_from_cache(entries, odb, sl, si=0):
 		rbound = entry.path.find('/', si)
 		if rbound == -1:
 			# its not a tree
-			tree_items.append((entry.binsha, entry.mode, entry.path[si:]))
+			tree_items_append((entry.binsha, entry.mode, entry.path[si:]))
 		else:
 			# find common base range
 			base = entry.path[si:rbound]
@@ -178,7 +183,7 @@ def write_tree_from_cache(entries, odb, sl, si=0):
 			# enter recursion
 			# ci - 1 as we want to count our current item as well
 			sha, tree_entry_list = write_tree_from_cache(entries, odb, slice(ci-1, xi), rbound+1)
-			tree_items.append((sha, S_IFDIR, base))
+			tree_items_append((sha, S_IFDIR, base))
 
 			# skip ahead
 			ci = xi
@@ -193,5 +198,93 @@ def write_tree_from_cache(entries, odb, sl, si=0):
 	istream = odb.store(IStream(str_tree_type, len(sio.getvalue()), sio))
 	return (istream.binsha, tree_items)
 
+def _tree_entry_to_baseindexentry(tree_entry, stage):
+	return BaseIndexEntry((tree_entry[1], tree_entry[0], stage <<CE_STAGESHIFT, tree_entry[2]))
 
+def aggressive_tree_merge(odb, tree_shas):
+	"""
+	:return: list of BaseIndexEntries representing the aggressive merge of the given
+		trees. All valid entries are on stage 0, whereas the conflicting ones are left 
+		on stage 1, 2 or 3, whereas stage 1 corresponds to the common ancestor tree, 
+		2 to our tree and 3 to 'their' tree.
+	:param tree_shas: 1, 2 or 3 trees as identified by their shas
+		If 1 or two, the entries will effectively correspond to the last given tree
+		If 3 are given, a 3 way merge is performed"""
+	out = list()
+	out_append = out.append
 
+	# one and two way is the same for us, as we don't have to handle an existing
+	# index, instrea
+	if len(tree_shas) in (1,2):
+		for entry in traverse_tree_recursive(odb, tree_shas[-1], ''):
+			out_append(_tree_entry_to_baseindexentry(entry, 0))
+		# END for each entry
+		return out
+	# END handle single tree 
+	
+	if len(tree_shas) > 3:
+		raise ValueError("Cannot handle %i trees at once" % len(tree_shas))
+
+	# three trees
+	for base, ours, theirs in traverse_trees_recursive(odb, tree_shas, ''):
+		if base is not None:
+			# base version exists
+			if ours is not None:
+				# ours exists
+				if theirs is not None:
+					# it exists in all branches, if it was changed in both
+					# its a conflict, otherwise we take the changed version
+					# This should be the most common branch, so it comes first
+					if( base[0] != ours[0] and base[0] != theirs[0] and ours[0] != theirs[0] ) or \
+						( base[1] != ours[1] and base[1] != theirs[1] and ours[1] != theirs[1] ):
+						# changed by both
+						out_append(_tree_entry_to_baseindexentry(base, 1))
+						out_append(_tree_entry_to_baseindexentry(ours, 2))
+						out_append(_tree_entry_to_baseindexentry(theirs, 3))
+					elif base[0] != ours[0] or base[1] != ours[1]:
+						# only we changed it
+						out_append(_tree_entry_to_baseindexentry(ours, 0))
+					else:
+						# either nobody changed it, or they did. In either
+						# case, use theirs
+						out_append(_tree_entry_to_baseindexentry(theirs, 0))
+					# END handle modification 
+				else:
+					
+					if ours[0] != base[0] or ours[1] != base[1]:
+						# they deleted it, we changed it, conflict 
+						out_append(_tree_entry_to_baseindexentry(base, 1))
+						out_append(_tree_entry_to_baseindexentry(ours, 2))
+					# else:
+					#	we didn't change it, ignore
+					#	pass
+					# END handle our change
+				# END handle theirs
+			else:
+				if theirs is None:
+					# deleted in both, its fine - its out
+					pass
+				else:
+					if theirs[0] != base[0] or theirs[1] != base[1]:
+						# deleted in ours, changed theirs, conflict
+						out_append(_tree_entry_to_baseindexentry(base, 1))
+						out_append(_tree_entry_to_baseindexentry(theirs, 3))
+					# END theirs changed
+					#else:
+					# 	theirs didnt change
+					#	pass
+				# END handle theirs
+			# END handle ours
+		else:
+			# all three can't be None
+			if ours is None:
+				# added in their branch
+				out_append(_tree_entry_to_baseindexentry(theirs, 0))
+			elif theirs is None:
+				# added in our branch
+				out_append(_tree_entry_to_baseindexentry(ours, 0))
+			# END hanle heads
+		# END handle base exists
+	# END for each entries tuple
+
+	return out

lib/git/index/typ.py

@@ -56,6 +56,9 @@ class BaseIndexEntry(tuple):
 
 	def __str__(self):
 		return "%o %s %i\t%s" % (self.mode, self.hexsha, self.stage, self.path)
+		
+	def __repr__(self):
+		return "(%o, %s, %i, %s)" % (self.mode, self.hexsha, self.stage, self.path)
 
 	@property
 	def mode(self):

lib/git/objects/fun.py

@@ -2,6 +2,9 @@
 
 __all__ = ('tree_to_stream', 'tree_entries_from_data')
 
+from stat import S_ISDIR
+				
+
 def tree_to_stream(entries, write):
 	"""Write the give list of entries into a stream using its write method
 	:param entries: **sorted** list of tuples with (binsha, mode, name)
@@ -64,3 +67,118 @@ def tree_entries_from_data(data):
 		out.append((sha, mode, name))
 	# END for each byte in data stream
 	return out
+	
+	
+def _find_by_name(tree_data, name, is_dir, start_at):
+	"""return data entry matching the given name and tree mode
+	or None.
+	Before the item is returned, the respective data item is set 
+	None in the tree_data list to mark it done"""
+	try:
+		item = tree_data[start_at]
+		if item and  item[2] == name and S_ISDIR(item[1]) == is_dir:
+			tree_data[start_at] = None
+			return item
+	except IndexError:
+		pass
+	# END exception handling
+	for index, item in enumerate(tree_data):
+		if item and item[2] == name and S_ISDIR(item[1]) == is_dir:
+			tree_data[index] = None
+			return item
+		# END if item matches
+	# END for each item
+	return None
+
+def _to_full_path(item, path_prefix):
+	"""Rebuild entry with given path prefix"""
+	if not item:
+		return item
+	return (item[0], item[1], path_prefix+item[2])
+	
+def traverse_trees_recursive(odb, tree_shas, path_prefix):
+	"""
+	:return: list with entries according to the given tree-shas. 
+		The result is encoded in a list
+		of n tuple|None per blob/commit, (n == len(tree_shas)), where 
+		* [0] == 20 byte sha
+		* [1] == mode as int
+		* [2] == path relative to working tree root
+		The entry tuple is None if the respective blob/commit did not 
+		exist in the given tree.
+	:param tree_shas: iterable of shas pointing to trees. All trees must 
+		be on the same level. A tree-sha may be None in which case None
+	:param path_prefix: a prefix to be added to the returned paths on this level, 
+		set it '' for the first iteration
+	:note: The ordering of the returned items will be partially lost"""
+	trees_data = list()
+	nt = len(tree_shas)
+	for tree_sha in tree_shas:
+		if tree_sha is None:
+			data = list()
+		else:
+			data = tree_entries_from_data(odb.stream(tree_sha).read())
+		# END handle muted trees
+		trees_data.append(data)
+	# END for each sha to get data for
+	
+	out = list()
+	out_append = out.append
+	
+	# find all matching entries and recursively process them together if the match
+	# is a tree. If the match is a non-tree item, put it into the result.
+	# Processed items will be set None
+	for ti, tree_data in enumerate(trees_data):
+		for ii, item in enumerate(tree_data):
+			if not item:
+				continue
+			# END skip already done items
+			entries = [ None for n in range(nt) ]
+			entries[ti] = item
+			sha, mode, name = item							# its faster to unpack
+			is_dir = S_ISDIR(mode)							# type mode bits
+			
+			# find this item in all other tree data items
+			# wrap around, but stop one before our current index, hence 
+			# ti+nt, not ti+1+nt
+			for tio in range(ti+1, ti+nt):
+				tio = tio % nt
+				entries[tio] = _find_by_name(trees_data[tio], name, is_dir, ii)
+			# END for each other item data
+			
+			# if we are a directory, enter recursion
+			if is_dir:
+				out.extend(traverse_trees_recursive(odb, [((ei and ei[0]) or None) for ei in entries], path_prefix+name+'/'))
+			else:
+				out_append(tuple(_to_full_path(e, path_prefix) for e in entries))
+			# END handle recursion
+			
+			# finally mark it done
+			tree_data[ii] = None
+		# END for each item
+		
+		# we are done with one tree, set all its data empty
+		del(tree_data[:])
+	# END for each tree_data chunk
+	return out
+	
+def traverse_tree_recursive(odb, tree_sha, path_prefix):
+	"""
+	:return: list of entries of the tree pointed to by tree_sha. An entry
+		has the following format:
+		* [0] 20 byte sha
+		* [1] mode as int
+		* [2] path relative to the repository
+	:param path_prefix: prefix to prepend to the front of all returned paths"""
+	entries = list()
+	data = tree_entries_from_data(odb.stream(tree_sha).read())
+	
+	# unpacking/packing is faster than accessing individual items
+	for sha, mode, name in data:
+		if S_ISDIR(mode):
+			entries.extend(traverse_tree_recursive(odb, sha, path_prefix+name+'/'))
+		else:
+			entries.append((sha, mode, path_prefix+name))
+	# END for each item
+	
+	return entries