Merge lp:~ian-clatworthy/bzr/faster-diff into lp:~bzr/bzr/trunk-old

[Ian Clatworthy]

[Ian Clatworthy (ian-clatworthy)]

This small patch makes diff faster on large trees. On OOo with a single added file, the time improves from 2.0 seconds to 1.2 seconds.

[Martin Pool (mbp)]

review: Approve

[Ian Clatworthy (ian-clatworthy)]

review: Needs Resubmitting

[Ian Clatworthy (ian-clatworthy)]

review: Approve

=== modified file 'NEWS'
--- NEWS	2009-05-28 17:14:38 +0000
+++ NEWS	2009-05-28 18:36:14 +0000
@@ -23,6 +23,8 @@
 
 * ``bzr branch --notree`` is now faster. (Ian Clatworthy)
 
+* ``bzr diff`` is now faster on large trees. (Ian Clatworthy)
+
 Bug Fixes
 *********
 

=== modified file 'bzrlib/diff.py'
--- bzrlib/diff.py	2009-03-28 02:10:54 +0000
+++ bzrlib/diff.py	2009-05-28 18:36:15 +0000
@@ -620,9 +620,11 @@
             return self.CANNOT_DIFF
         from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
         to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
-        return self.diff_text(from_file_id, to_file_id, from_label, to_label)
+        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
+            old_path, new_path)
 
-    def diff_text(self, from_file_id, to_file_id, from_label, to_label):
+    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
+        from_path=None, to_path=None):
         """Diff the content of given files in two trees
 
         :param from_file_id: The id of the file in the from tree.  If None,
@@ -630,15 +632,17 @@
         :param to_file_id: The id of the file in the to tree.  This may refer
             to a different file from from_file_id.  If None,
             the file is not present in the to tree.
+        :param from_path: The path in the from tree or None if unknown.
+        :param to_path: The path in the to tree or None if unknown.
         """
-        def _get_text(tree, file_id):
+        def _get_text(tree, file_id, path):
             if file_id is not None:
-                return tree.get_file(file_id).readlines()
+                return tree.get_file(file_id, path).readlines()
             else:
                 return []
         try:
-            from_text = _get_text(self.old_tree, from_file_id)
-            to_text = _get_text(self.new_tree, to_file_id)
+            from_text = _get_text(self.old_tree, from_file_id, from_path)
+            to_text = _get_text(self.new_tree, to_file_id, to_path)
             self.text_differ(from_label, from_text, to_label, to_text,
                              self.to_file)
         except errors.BinaryFile:
@@ -882,7 +886,7 @@
                 self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
                                    newpath_encoded, prop_str))
             if changed_content:
-                self.diff(file_id, oldpath, newpath)
+                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
                 has_changes = 1
             if renamed:
                 has_changes = 1
@@ -903,7 +907,10 @@
             new_kind = self.new_tree.kind(file_id)
         except (errors.NoSuchId, errors.NoSuchFile):
             new_kind = None
-
+        self._diff(file_id, old_path, new_path, old_kind, new_kind)
+
+
+    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
         result = DiffPath._diff_many(self.differs, file_id, old_path,
                                        new_path, old_kind, new_kind)
         if result is DiffPath.CANNOT_DIFF:

=== modified file 'bzrlib/workingtree_4.py'
--- bzrlib/workingtree_4.py	2009-05-23 04:55:52 +0000
+++ bzrlib/workingtree_4.py	2009-05-28 18:36:15 +0000
@@ -1572,6 +1572,9 @@
         self._repo_supports_tree_reference = getattr(
             repository._format, "supports_tree_reference",
             False)
+        # Simple cache of last dirstate lookup
+        self._last_entry_lookup = None
+        self._last_entry = None
 
     def __repr__(self):
         return "<%s of %s in %s>" % \
@@ -1644,8 +1647,13 @@
             raise errors.BzrError('must supply file_id or path')
         if path is not None:
             path = path.encode('utf8')
-        parent_index = self._get_parent_index()
-        return self._dirstate._get_entry(parent_index, fileid_utf8=file_id, path_utf8=path)
+        cache_key = file_id or ("path:%s" % path)
+        if self._last_entry_lookup != cache_key:
+            parent_index = self._get_parent_index()
+            self._last_entry = self._dirstate._get_entry(parent_index,
+                fileid_utf8=file_id, path_utf8=path)
+            self._last_entry_lookup = cache_key
+        return self._last_entry
 
     def _generate_inventory(self):
         """Create and set self.inventory from the dirstate object.