Merged be.faster-diff branch, fixing #bea/ed5#.

6 files changed, 454 insertions, 54 deletions

diff --git a/libbe/storage/vcs/arch.py b/libbe/storage/vcs/arch.py
index f9ae32b..f9b01fd 100644
--- a/libbe/storage/vcs/arch.py
+++ b/libbe/storage/vcs/arch.py
@@ -27,7 +27,7 @@ import os
 import re
 import shutil
 import sys
-import time
+import time # work around http://mercurial.selenic.com/bts/issue618
 
 import libbe
 import libbe.ui.util.user
@@ -68,6 +68,7 @@ class Arch(base.VCS):
         self.versioned = True
         self.interspersed_vcs_files = True
         self.paranoid = False
+        self.__updated = [] # work around http://mercurial.selenic.com/bts/issue618
 
     def _vcs_version(self):
         status,output,error = self._u_invoke_client('--version')
@@ -288,7 +289,7 @@ class Arch(base.VCS):
             shutil.rmtree(arch_ids)
 
     def _vcs_update(self, path):
-        pass
+        self.__updated.append(path) # work around http://mercurial.selenic.com/bts/issue618
 
     def _vcs_is_versioned(self, path):
         if '.arch-ids' in path:
@@ -300,16 +301,29 @@ class Arch(base.VCS):
             return base.VCS._vcs_get_file_contents(self, path, revision)
         else:
             status,output,error = \
-                self._invoke_client('file-find', path, revision)
-            relpath = output.rstrip('\n')
+                self._invoke_client(
+                'file-find', '--unescaped', path, revision)
+            relpath = output.rstrip('\n').splitlines()[-1]
+            print >> sys.stderr, 'getting', relpath
             return base.VCS._vcs_get_file_contents(self, relpath)
 
+    def _vcs_path(self, id, revision):
+        raise NotImplementedError
+
     def _vcs_commit(self, commitfile, allow_empty=False):
         if allow_empty == False:
             # arch applies empty commits without complaining, so check first
             status,output,error = self._u_invoke_client('changes',expect=(0,1))
             if status == 0:
-                raise base.EmptyCommit()
+                # work around http://mercurial.selenic.com/bts/issue618
+                time.sleep(1)
+                for path in self.__updated:
+                    os.utime(os.path.join(self.repo, path), None)
+                self.__updated = []
+                status,output,error = self._u_invoke_client('changes',expect=(0,1))
+                if status == 0:
+                # end work around
+                    raise base.EmptyCommit()
         summary,body = self._u_parse_commitfile(commitfile)
         args = ['commit', '--summary', summary]
         if body != None:
@@ -342,6 +356,9 @@ class Arch(base.VCS):
             return None
         return '%s--%s' % (self._archive_project_name(), log)
 
+    def _vcs_changed(self, revision):
+        raise NotImplementedError
+
 
 if libbe.TESTING == True:
     base.make_vcs_testcase_subclasses(Arch, sys.modules[__name__])
diff --git a/libbe/storage/vcs/base.py b/libbe/storage/vcs/base.py
index 8390cbc..15460b0 100644
--- a/libbe/storage/vcs/base.py
+++ b/libbe/storage/vcs/base.py
@@ -168,7 +168,7 @@ class CachedPathID (object):
     >>> c.path('qrs')
     Traceback (most recent call last):
       ...
-    InvalidID: 'qrs'
+    InvalidID: qrs in revision None
     >>> c.disconnect()
     >>> c.destroy()
     >>> dir.cleanup()
@@ -363,7 +363,7 @@ class VCS (libbe.storage.base.VersionedStorage):
     BugDir to search for an installed Storage backend and initialize
     it in the root directory.  This is a convenience option for
     supporting tests of versioning functionality
-    (e.g. .duplicate_bugdir).
+    (e.g. RevisionedBugDir).
 
     Disable encoding manipulation
     =============================
@@ -597,6 +597,14 @@ os.listdir(self.get_path("bugs")):
         """
         return None
 
+    def _vcs_changed(self, revision):
+        """
+        Return a tuple of lists of ids
+          (new, modified, removed)
+        from the specified revision to the current situation.
+        """
+        return ([], [], [])
+
     def version(self):
         # Cache version string for efficiency.
         if not hasattr(self, '_version'):
@@ -737,6 +745,27 @@ os.listdir(self.get_path("bugs")):
             if p.startswith(path):
                 self._cached_path_id.remove_id(id)
 
+    def _ancestors(self, id=None, revision=None):
+        if revision == None:
+            id_to_path = self._cached_path_id.path
+        else:
+            id_to_path = lambda id : self._vcs_path(id, revision)
+        if id==None:
+            path = self.be_dir
+        else:
+            path = id_to_path(id)
+        ancestors = []
+        while True:
+            if not path.startswith(self.repo + os.path.sep):
+                break
+            path = os.path.dirname(path)
+            try:
+                id = self._u_path_to_id(path)
+                ancestors.append(id)
+            except (SpacerCollision, InvalidPath):
+                pass    
+        return ancestors
+
     def _children(self, id=None, revision=None):
         if revision == None:
             id_to_path = self._cached_path_id.path
@@ -786,12 +815,14 @@ os.listdir(self.get_path("bugs")):
         try:
             contents = self._vcs_get_file_contents(relpath, revision)
         except InvalidID, e:
-            if InvalidID == None:
-                e.id = InvalidID
+            if e.id == None:
+                e.id = id
+            if e.revision == None:
+                e.revision = revision
             raise
         if contents in [libbe.storage.base.InvalidDirectory,
                         libbe.util.InvalidObject]:
-            raise InvalidID(id)
+            raise InvalidID(id, revision)
         elif len(contents) == 0:
             return None
         return contents
@@ -839,6 +870,20 @@ os.listdir(self.get_path("bugs")):
             raise libbe.storage.base.InvalidRevision(index)
         return revid
 
+    def changed(self, revision):
+        new,mod,rem = self._vcs_changed(revision)
+        def paths_to_ids(paths):
+            for p in paths:
+                try:
+                    id = self._u_path_to_id(p)
+                    yield id
+                except (SpacerCollision, InvalidPath):
+                    pass
+        new_id = list(paths_to_ids(new))
+        mod_id = list(paths_to_ids(mod))
+        rem_id = list(paths_to_ids(rem))
+        return (new_id, mod_id, rem_id)
+
     def _u_any_in_string(self, list, string):
         """
         Return True if any of the strings in list are in string.
@@ -883,6 +928,33 @@ os.listdir(self.get_path("bugs")):
             return None
         return ret
 
+    def _u_find_id_from_manifest(self, id, manifest, revision=None):
+        """
+        Search for the relative path to id using manifest, a list of all files.
+        
+        Returns None if the id is not found.
+        """
+        be_dir = self._cached_path_id._spacer_dirs[0]
+        be_dir_sep = self._cached_path_id._spacer_dirs[0] + os.path.sep
+        files = [f for f in manifest if f.startswith(be_dir_sep)]
+        for file in files:
+            if not file.startswith(be_dir+os.path.sep):
+                continue
+            parts = file.split(os.path.sep)
+            dir = parts.pop(0) # don't add the first spacer dir
+            for part in parts[:-1]:
+                dir = os.path.join(dir, part)
+                if not dir in files:
+                    files.append(dir)
+        for file in files:
+            try:
+                p_id = self._u_path_to_id(file)
+                if p_id == id:
+                    return file
+            except (SpacerCollision, InvalidPath):
+                pass
+        raise InvalidID(id, revision=revision)
+
     def _u_find_id(self, id, revision):
         """
         Search for the relative path to id as of revision.
@@ -1023,8 +1095,7 @@ if libbe.TESTING == True:
 
     class VCS_installed_TestCase (VCSTestCase):
         def test_installed(self):
-            """
-            See if the VCS is installed.
+            """See if the VCS is installed.
             """
             self.failUnless(self.s.installed() == True,
                             '%(name)s VCS not found' % vars(self.Class))
@@ -1032,8 +1103,7 @@ if libbe.TESTING == True:
 
     class VCS_detection_TestCase (VCSTestCase):
         def test_detection(self):
-            """
-            See if the VCS detects its installed repository
+            """See if the VCS detects its installed repository
             """
             if self.s.installed():
                 self.s.disconnect()
@@ -1043,8 +1113,7 @@ if libbe.TESTING == True:
                 self.s.connect()
 
         def test_no_detection(self):
-            """
-            See if the VCS detects its installed repository
+            """See if the VCS detects its installed repository
             """
             if self.s.installed() and self.Class.name != 'None':
                 self.s.disconnect()
diff --git a/libbe/storage/vcs/bzr.py b/libbe/storage/vcs/bzr.py
index 4d72fd0..e1cd2e5 100644
--- a/libbe/storage/vcs/bzr.py
+++ b/libbe/storage/vcs/bzr.py
@@ -36,13 +36,13 @@ import os.path
 import re
 import shutil
 import StringIO
+import sys
 
 import libbe
 import base
 
 if libbe.TESTING == True:
     import doctest
-    import sys
     import unittest
 
 
@@ -134,7 +134,9 @@ class Bzr(base.VCS):
         return cmd.outf.getvalue()        
 
     def _vcs_path(self, id, revision):
-        return self._u_find_id(id, revision)
+        manifest = self._vcs_listdir(
+            self.repo, revision=revision, recursive=True)
+        return self._u_find_id_from_manifest(id, manifest, revision=revision)
 
     def _vcs_isdir(self, path, revision):
         try:
@@ -145,13 +147,13 @@ class Bzr(base.VCS):
             raise
         return True
 
-    def _vcs_listdir(self, path, revision):
+    def _vcs_listdir(self, path, revision, recursive=False):
         path = os.path.join(self.repo, path)
         revision = self._parse_revision_string(revision)
         cmd = bzrlib.builtins.cmd_ls()
         cmd.outf = StringIO.StringIO()
         try:
-            cmd.run(revision=revision, path=path)
+            cmd.run(revision=revision, path=path, recursive=recursive)
         except bzrlib.errors.BzrCommandError, e:
             if 'not present in revision' in str(e):
                 raise base.InvalidPath(path, root=self.repo, revision=revision)
@@ -188,6 +190,87 @@ class Bzr(base.VCS):
             return str(index) # bzr commit 0 is the empty tree.
         return str(current_revision+index+1)
 
+    def _diff(self, revision):
+        revision = self._parse_revision_string(revision)
+        cmd = bzrlib.builtins.cmd_diff()
+        cmd.outf = StringIO.StringIO()
+        # for some reason, cmd_diff uses sys.stdout not self.outf for output.
+        stdout = sys.stdout
+        sys.stdout = cmd.outf
+        try:
+            status = cmd.run(revision=revision, file_list=[self.repo])
+        finally:
+            sys.stdout = stdout
+        assert status in [0,1], "Invalid status %d" % status
+        return cmd.outf.getvalue()
+
+    def _parse_diff(self, diff_text):
+        """
+        Example diff text:
+                
+        === modified file 'dir/changed'
+        --- dir/changed	2010-01-16 01:54:53 +0000
+        +++ dir/changed	2010-01-16 01:54:54 +0000
+        @@ -1,3 +1,3 @@
+         hi
+        -there
+        +everyone and
+         joe
+        
+        === removed file 'dir/deleted'
+        --- dir/deleted	2010-01-16 01:54:53 +0000
+        +++ dir/deleted	1970-01-01 00:00:00 +0000
+        @@ -1,3 +0,0 @@
+        -in
+        -the
+        -beginning
+        
+        === removed file 'dir/moved'
+        --- dir/moved	2010-01-16 01:54:53 +0000
+        +++ dir/moved	1970-01-01 00:00:00 +0000
+        @@ -1,4 +0,0 @@
+        -the
+        -ants
+        -go
+        -marching
+        
+        === added file 'dir/moved2'
+        --- dir/moved2	1970-01-01 00:00:00 +0000
+        +++ dir/moved2	2010-01-16 01:54:34 +0000
+        @@ -0,0 +1,4 @@
+        +the
+        +ants
+        +go
+        +marching
+        
+        === added file 'dir/new'
+        --- dir/new	1970-01-01 00:00:00 +0000
+        +++ dir/new	2010-01-16 01:54:54 +0000
+        @@ -0,0 +1,2 @@
+        +hello
+        +world
+        
+        """
+        new = []
+        modified = []
+        removed = []
+        for line in diff_text.splitlines():
+            if not line.startswith('=== '):
+                continue
+            fields = line.split()
+            action = fields[1]
+            file = fields[-1].strip("'")
+            if action == 'added':
+                new.append(file)
+            elif action == 'modified':
+                modified.append(file)
+            elif action == 'removed':
+                removed.append(file)
+        return (new,modified,removed)
+
+    def _vcs_changed(self, revision):
+        return self._parse_diff(self._diff(revision))
+
 
 if libbe.TESTING == True:
     base.make_vcs_testcase_subclasses(Bzr, sys.modules[__name__])
diff --git a/libbe/storage/vcs/darcs.py b/libbe/storage/vcs/darcs.py
index 92dc1a7..c6892b4 100644
--- a/libbe/storage/vcs/darcs.py
+++ b/libbe/storage/vcs/darcs.py
@@ -157,24 +157,14 @@ class Darcs(base.VCS):
             return output
         # Darcs versions < 2.0.0pre2 lack the 'show contents' command
 
-        status,output,error = self._u_invoke_client( \
-            'diff', '--unified', '--from-patch', revision, path,
-            unicode_output=False)
-        major_patch = output
-        status,output,error = self._u_invoke_client( \
-            'diff', '--unified', '--patch', revision, path,
-            unicode_output=False)
-        target_patch = output
+        patch = self._diff(revision, path=path, unicode_output=False)
 
         # '--output -' to be supported in GNU patch > 2.5.9
         # but that hasn't been released as of June 30th, 2009.
 
         # Rewrite path to status before the patch we want
         args=['patch', '--reverse', path]
-        status,output,error = self._u_invoke(args, stdin=major_patch)
-        # Now apply the patch we want
-        args=['patch', path]
-        status,output,error = self._u_invoke(args, stdin=target_patch)
+        status,output,error = self._u_invoke(args, stdin=patch)
 
         if os.path.exists(os.path.join(self.repo, path)) == True:
             contents = base.VCS._vcs_get_file_contents(self, path)
@@ -182,11 +172,8 @@ class Darcs(base.VCS):
             contents = ''
 
         # Now restore path to it's current incarnation
-        args=['patch', '--reverse', path]
-        status,output,error = self._u_invoke(args, stdin=target_patch)
         args=['patch', path]
-        status,output,error = self._u_invoke(args, stdin=major_patch)
-        current_contents = base.VCS._vcs_get_file_contents(self, path)
+        status,output,error = self._u_invoke(args, stdin=patch)
         return contents
 
     def _vcs_path(self, id, revision):
@@ -257,7 +244,10 @@ class Darcs(base.VCS):
             revision = match.groups()[0]
         return revision
 
-    def _vcs_revision_id(self, index):
+    def _revisions(self):
+        """
+        Return a list of revisions in the repository.
+        """
         status,output,error = self._u_invoke_client('changes', '--xml')
         revisions = []
         xml_str = output.encode('unicode_escape').replace(r'\n', '\n')
@@ -270,6 +260,10 @@ class Darcs(base.VCS):
                     text = unescape(unicode(child.text).decode('unicode_escape').strip())
                     revisions.append(text)
         revisions.reverse()
+        return revisions
+
+    def _vcs_revision_id(self, index):
+        revisions = self._revisions()
         try:
             if index > 0:
                 return revisions[index-1]
@@ -280,6 +274,105 @@ class Darcs(base.VCS):
         except IndexError:
             return None
 
+    def _diff(self, revision, path=None, unicode_output=True):
+        revisions = self._revisions()
+        i = revisions.index(revision)
+        args = ['diff', '--unified']
+        if i+1 < len(revisions):
+            next_rev = revisions[i+1]
+            args.extend(['--from-patch', next_rev])
+        if path != None:
+            args.append(path)
+        kwargs = {'unicode_output':unicode_output}
+        status,output,error = self._u_invoke_client(
+            *args, **kwargs)
+        return output
+
+    def _parse_diff(self, diff_text):
+        """
+        Example diff text:
+
+        Mon Jan 18 15:19:30 EST 2010  None <None@invalid.com>
+          * Final state
+        diff -rN --unified old-BEtestgQtDuD/.be/dir/bugs/modified new-BEtestgQtDuD/.be/dir/bugs/modified
+        --- old-BEtestgQtDuD/.be/dir/bugs/modified      2010-01-18 15:19:30.000000000 -0500
+        +++ new-BEtestgQtDuD/.be/dir/bugs/modified      2010-01-18 15:19:30.000000000 -0500
+        @@ -1 +1 @@
+        -some value to be modified
+        \ No newline at end of file
+        +a new value
+        \ No newline at end of file
+        diff -rN --unified old-BEtestgQtDuD/.be/dir/bugs/moved new-BEtestgQtDuD/.be/dir/bugs/moved
+        --- old-BEtestgQtDuD/.be/dir/bugs/moved 2010-01-18 15:19:30.000000000 -0500
+        +++ new-BEtestgQtDuD/.be/dir/bugs/moved 1969-12-31 19:00:00.000000000 -0500
+        @@ -1 +0,0 @@
+        -this entry will be moved
+        \ No newline at end of file
+        diff -rN --unified old-BEtestgQtDuD/.be/dir/bugs/moved2 new-BEtestgQtDuD/.be/dir/bugs/moved2
+        --- old-BEtestgQtDuD/.be/dir/bugs/moved2        1969-12-31 19:00:00.000000000 -0500
+        +++ new-BEtestgQtDuD/.be/dir/bugs/moved2        2010-01-18 15:19:30.000000000 -0500
+        @@ -0,0 +1 @@
+        +this entry will be moved
+        \ No newline at end of file
+        diff -rN --unified old-BEtestgQtDuD/.be/dir/bugs/new new-BEtestgQtDuD/.be/dir/bugs/new
+        --- old-BEtestgQtDuD/.be/dir/bugs/new   1969-12-31 19:00:00.000000000 -0500
+        +++ new-BEtestgQtDuD/.be/dir/bugs/new   2010-01-18 15:19:30.000000000 -0500
+        @@ -0,0 +1 @@
+        +this entry is new
+        \ No newline at end of file
+        diff -rN --unified old-BEtestgQtDuD/.be/dir/bugs/removed new-BEtestgQtDuD/.be/dir/bugs/removed
+        --- old-BEtestgQtDuD/.be/dir/bugs/removed       2010-01-18 15:19:30.000000000 -0500
+        +++ new-BEtestgQtDuD/.be/dir/bugs/removed       1969-12-31 19:00:00.000000000 -0500
+        @@ -1 +0,0 @@
+        -this entry will be deleted
+        \ No newline at end of file
+        
+        """
+        new = []
+        modified = []
+        removed = []
+        lines = diff_text.splitlines()
+        repodir = os.path.basename(self.repo) + os.path.sep
+        i = 0
+        while i < len(lines):
+            line = lines[i]; i += 1
+            if not line.startswith('diff '):
+                continue
+            file_a,file_b = line.split()[-2:]
+            assert file_a.startswith('old-'), \
+                'missformed file_a %s' % file_a
+            assert file_b.startswith('new-'), \
+                'missformed file_a %s' % file_b
+            file = file_a[4:]
+            assert file_b[4:] == file, \
+                'diff file missmatch %s != %s' % (file_a, file_b)
+            assert file.startswith(repodir), \
+                'missformed file_a %s' % file_a
+            file = file[len(repodir):]
+            lines_added = 0
+            lines_removed = 0
+            line = lines[i]; i += 1
+            assert line.startswith('--- old-'), \
+                'missformed "---" line %s' % line
+            time_a = line.split('\t')[1]
+            line = lines[i]; i += 1
+            assert line.startswith('+++ new-'), \
+                'missformed "+++" line %s' % line
+            time_b = line.split('\t')[1]
+            zero_time = time.strftime('%Y-%m-%d %H:%M:%S.000000000 ',
+                                      time.localtime(0))
+            # note that zero_time is missing the trailing timezone offset
+            if time_a.startswith(zero_time):
+                new.append(file)
+            elif time_b.startswith(zero_time):
+                removed.append(file)
+            else:
+                modified.append(file)
+        return (new,modified,removed)
+
+    def _vcs_changed(self, revision):
+        return self._parse_diff(self._diff(revision))
+
 
 if libbe.TESTING == True:
     base.make_vcs_testcase_subclasses(Darcs, sys.modules[__name__])
diff --git a/libbe/storage/vcs/git.py b/libbe/storage/vcs/git.py
index 5382c7e..6d3aa87 100644
--- a/libbe/storage/vcs/git.py
+++ b/libbe/storage/vcs/git.py
@@ -175,6 +175,87 @@ class Git(base.VCS):
         except IndexError:
             return None
 
+    def _diff(self, revision):
+        status,output,error = self._u_invoke_client('diff', revision)
+        return output
+
+    def _parse_diff(self, diff_text):
+        """
+        Example diff text:
+        
+        diff --git a/dir/changed b/dir/changed
+        index 6c3ea8c..2f2f7c7 100644
+        --- a/dir/changed
+        +++ b/dir/changed
+        @@ -1,3 +1,3 @@
+         hi
+        -there
+        +everyone and
+         joe
+        diff --git a/dir/deleted b/dir/deleted
+        deleted file mode 100644
+        index 225ec04..0000000
+        --- a/dir/deleted
+        +++ /dev/null
+        @@ -1,3 +0,0 @@
+        -in
+        -the
+        -beginning
+        diff --git a/dir/moved b/dir/moved
+        deleted file mode 100644
+        index 5ef102f..0000000
+        --- a/dir/moved
+        +++ /dev/null
+        @@ -1,4 +0,0 @@
+        -the
+        -ants
+        -go
+        -marching
+        diff --git a/dir/moved2 b/dir/moved2
+        new file mode 100644
+        index 0000000..5ef102f
+        --- /dev/null
+        +++ b/dir/moved2
+        @@ -0,0 +1,4 @@
+        +the
+        +ants
+        +go
+        +marching
+        diff --git a/dir/new b/dir/new
+        new file mode 100644
+        index 0000000..94954ab
+        --- /dev/null
+        +++ b/dir/new
+        @@ -0,0 +1,2 @@
+        +hello
+        +world
+        """
+        new = []
+        modified = []
+        removed = []
+        lines = diff_text.splitlines()
+        for i,line in enumerate(lines):
+            if not line.startswith('diff '):
+                continue
+            file_a,file_b = line.split()[-2:]
+            assert file_a.startswith('a/'), \
+                'missformed file_a %s' % file_a
+            assert file_b.startswith('b/'), \
+                'missformed file_a %s' % file_b
+            file = file_a[2:]
+            assert file_b[2:] == file, \
+                'diff file missmatch %s != %s' % (file_a, file_b)
+            if lines[i+1].startswith('new '):
+                new.append(file)
+            elif lines[i+1].startswith('index '):
+                modified.append(file)
+            elif lines[i+1].startswith('deleted '):
+                removed.append(file)
+        return (new,modified,removed)
+
+    def _vcs_changed(self, revision):
+        return self._parse_diff(self._diff(revision))
+
 
 if libbe.TESTING == True:
     base.make_vcs_testcase_subclasses(Git, sys.modules[__name__])
diff --git a/libbe/storage/vcs/hg.py b/libbe/storage/vcs/hg.py
index 824f687..076943a 100644
--- a/libbe/storage/vcs/hg.py
+++ b/libbe/storage/vcs/hg.py
@@ -112,23 +112,9 @@ class Hg(base.VCS):
             return self._u_invoke_client('cat', '-r', revision, path)
 
     def _vcs_path(self, id, revision):
-        output = self._u_invoke_client('manifest', '--rev', revision)
-        be_dir = self._cached_path_id._spacer_dirs[0]
-        be_dir_sep = self._cached_path_id._spacer_dirs[0] + os.path.sep
-        files = [f for f in output.splitlines() if f.startswith(be_dir_sep)]
-        for file in files:
-            if not file.startswith(be_dir+os.path.sep):
-                continue
-            parts = file.split(os.path.sep)
-            dir = parts.pop(0) # don't add the first spacer dir
-            for part in parts[:-1]:
-                dir = os.path.join(dir, part)
-                if not dir in files:
-                    files.append(dir)
-        for file in files:
-            if self._u_path_to_id(file) == id:
-                return file
-        raise base.InvalidId(id, revision=revision)
+        manifest = self._u_invoke_client(
+            'manifest', '--rev', revision).splitlines()
+        return self._u_find_id_from_manifest(id, manifest, revision=revision)
 
     def _vcs_isdir(self, path, revision):
         output = self._u_invoke_client('manifest', '--rev', revision)
@@ -172,6 +158,77 @@ class Hg(base.VCS):
             return None # before initial commit.
         return id
 
+    def _diff(self, revision):
+        return self._u_invoke_client(
+            'diff', '-r', revision, '--git')
+
+    def _parse_diff(self, diff_text):
+        """
+        Example diff text:
+                
+        diff --git a/.be/dir/bugs/modified b/.be/dir/bugs/modified
+        --- a/.be/dir/bugs/modified
+        +++ b/.be/dir/bugs/modified
+        @@ -1,1 +1,1 @@ some value to be modified
+        -some value to be modified
+        \ No newline at end of file
+        +a new value
+        \ No newline at end of file
+        diff --git a/.be/dir/bugs/moved b/.be/dir/bugs/moved
+        deleted file mode 100644
+        --- a/.be/dir/bugs/moved
+        +++ /dev/null
+        @@ -1,1 +0,0 @@
+        -this entry will be moved
+        \ No newline at end of file
+        diff --git a/.be/dir/bugs/moved2 b/.be/dir/bugs/moved2
+        new file mode 100644
+        --- /dev/null
+        +++ b/.be/dir/bugs/moved2
+        @@ -0,0 +1,1 @@
+        +this entry will be moved
+        \ No newline at end of file
+        diff --git a/.be/dir/bugs/new b/.be/dir/bugs/new
+        new file mode 100644
+        --- /dev/null
+        +++ b/.be/dir/bugs/new
+        @@ -0,0 +1,1 @@
+        +this entry is new
+        \ No newline at end of file
+        diff --git a/.be/dir/bugs/removed b/.be/dir/bugs/removed
+        deleted file mode 100644
+        --- a/.be/dir/bugs/removed
+        +++ /dev/null
+        @@ -1,1 +0,0 @@
+        -this entry will be deleted
+        \ No newline at end of file
+        """
+        new = []
+        modified = []
+        removed = []
+        lines = diff_text.splitlines()
+        for i,line in enumerate(lines):
+            if not line.startswith('diff '):
+                continue
+            file_a,file_b = line.split()[-2:]
+            assert file_a.startswith('a/'), \
+                'missformed file_a %s' % file_a
+            assert file_b.startswith('b/'), \
+                'missformed file_a %s' % file_b
+            file = file_a[2:]
+            assert file_b[2:] == file, \
+                'diff file missmatch %s != %s' % (file_a, file_b)
+            if lines[i+1].startswith('new '):
+                new.append(file)
+            elif lines[i+1].startswith('deleted '):
+                removed.append(file)
+            else:
+                modified.append(file)
+        return (new,modified,removed)
+
+    def _vcs_changed(self, revision):
+        return self._parse_diff(self._diff(revision))
+
 
 if libbe.TESTING == True:
     base.make_vcs_testcase_subclasses(Hg, sys.modules[__name__])