GitPlugin: optimiziation: make use of git's 1.5.6+ tree-ls -l featu…

…re for grabbing blob size instead of

individual `cat-file -s` calls; bump up minimum required git version to 1.5.6

tracext/git/PyGIT.py

@@ -146,7 +146,7 @@ def __rev_key(rev):
 
     @staticmethod
     def git_version(git_bin="git"):
-        GIT_VERSION_MIN_REQUIRED = (1,5,2)
+        GIT_VERSION_MIN_REQUIRED = (1,5,6)
         try:
             g = GitCore(git_bin=git_bin)
             output = g.version()
@@ -203,9 +203,7 @@ def __init__(self, git_dir, log, git_bin='git'):
         self.__commit_msg_cache = SizedDict(200)
         self.__commit_msg_lock = Lock()
 
-        # cache the last 2000 file sizes
-        self.__fs_obj_size_cache = SizedDict(2000)
-        self.__fs_obj_size_lock = Lock()
+
 
     def __del__(self):
         self.logger.debug("PyGIT.Storage instance %d destructed" % id(self))
@@ -483,18 +481,21 @@ def ls_tree(self, rev, path=""):
         if path.startswith('/'):
             path = path[1:]
 
-        if path:
-            tree = self.repo.ls_tree("-z", rev, "--", path)
-        else:
-            tree = self.repo.ls_tree("-z", rev)
+        tree = self.repo.ls_tree("-z", "-l", rev, "--", path).read().split('\0')
 
         def split_ls_tree_line(l):
-            "split according to '<mode> <type> <sha>\t<fname>'"
+            "split according to '<mode> <type> <sha> <size>\t<fname>'"
             meta,fname = l.split('\t')
-            _mode,_type,_sha = meta.split(' ')
-            return _mode,_type,_sha,fname
+            _mode,_type,_sha,_size = meta.split()
+
+            if _size == '-':
+                _size = None
+            else:
+                _size = int(_size)
 
-        return [split_ls_tree_line(e) for e in tree.read().split('\0') if e]
+            return _mode,_type,_sha,_size,fname
+
+        return [ split_ls_tree_line(e) for e in tree if e ]
 
     def read_commit(self, commit_id):
         if not commit_id:
@@ -539,13 +540,9 @@ def get_file(self, sha):
 
     def get_obj_size(self, sha):
         sha = str(sha)
+
         try:
-            with self.__fs_obj_size_lock:
-                if self.__fs_obj_size_cache.has_key(sha):
-                    obj_size = self.__fs_obj_size_cache[sha]
-                else:
-                    obj_size = int(self.repo.cat_file("-s", sha).read().strip())
-                    self.__fs_obj_size_cache[sha] = obj_size
+            obj_size = int(self.repo.cat_file("-s", sha).read().strip())
         except ValueError:
             raise GitErrorSha("object '%s' not found" % sha)
 
@@ -770,8 +767,6 @@ def check4loops(head):
         for sha in g.children_recursive(head):
             if sha in seen:
                 print "dupe detected :-/", sha, len(seen)
-                #print seen
-                #break
             seen.add(sha)
         return seen
 
@@ -802,10 +797,10 @@ def shortrev_test():
 
     # perform typical trac operations:
 
-    if 0:
+    if 1:
         print "--------------"
         rev = g.head()
-        for mode,type,sha,name in g.ls_tree(rev):
+        for mode,type,sha,_size,name in g.ls_tree(rev):
             [last_rev] = g.history(rev, name, limit=1)
             s = g.get_obj_size(sha) if type == "blob" else 0
             msg = g.read_commit(last_rev)

tracext/git/git_fs.py

@@ -290,7 +290,7 @@ def __init__(self, git, path, rev, log, ls_tree_info=None):
 			if not ls_tree_info:
 				raise NoSuchNode(path, rev)
 
-			(self.fs_perm, k, self.fs_sha, fn) = ls_tree_info
+			(self.fs_perm, k, self.fs_sha, self.fs_size, fn) = ls_tree_info
 
 			# fix-up to the last commit-rev that touched this node
 			rev = self.git.last_change(rev, p)
@@ -338,7 +338,7 @@ def get_entries(self):
 			return
 
 		for ent in self.git.ls_tree(self.rev, self.__git_path()):
-			yield GitNode(self.git, ent[3], self.rev, self.log, ent)
+			yield GitNode(self.git, ent[-1], self.rev, self.log, ent)
 
 	def get_content_type(self):
 		if self.isdir: