Merge lp:~jelmer/brz/get-delta-revisions-files into lp:brz/3.1

[Jelmer Vernooij]

Add Repository.get_revision_deltas.

Add Repository.get_revision_deltas.

[Jelmer Vernooij (jelmer)]

review: Approve

=== modified file 'breezy/bzr/remote.py'
--- breezy/bzr/remote.py	2020-03-22 19:12:43 +0000
+++ breezy/bzr/remote.py	2020-06-28 18:07:07 +0000
@@ -2590,12 +2590,14 @@
             # file-id based, so filtering before or afterwards is
             # currently easier.
             if specific_fileids is None:
-                trees = dict((t.get_revision_id(), t) for
-                             t in self.revision_trees(required_trees))
+                trees = {
+                    t.get_revision_id(): t
+                    for t in self.revision_trees(required_trees)}
             else:
-                trees = dict((t.get_revision_id(), t) for
-                             t in self._filtered_revision_trees(required_trees,
-                                                                specific_fileids))
+                trees = {
+                    t.get_revision_id(): t
+                    for t in self._filtered_revision_trees(
+                        required_trees, specific_fileids)}
 
             # Calculate the deltas
             for revision in revisions:
@@ -2608,7 +2610,7 @@
     def get_revision_delta(self, revision_id):
         with self.lock_read():
             r = self.get_revision(revision_id)
-            return list(self.get_deltas_for_revisions([r]))[0]
+            return list(self.get_revision_deltas([r]))[0]
 
     def revision_trees(self, revision_ids):
         with self.lock_read():
 
=== modified file 'breezy/bzr/tests/per_repository_vf/test_repository.py'
--- breezy/bzr/tests/per_repository_vf/test_repository.py	2020-06-10 02:56:53 +0000
+++ breezy/bzr/tests/per_repository_vf/test_repository.py	2020-06-28 18:07:07 +0000
@@ -364,8 +364,19 @@
         revisions = [repository.get_revision(r) for r in
                      [b'rev1', b'rev2', b'rev3', b'rev4']]
         deltas1 = list(repository.get_deltas_for_revisions(revisions))
-        deltas2 = [repository.get_revision_delta(r.revision_id) for r in
-                   revisions]
+        deltas2 = [repository.get_revision_delta(r.revision_id)
+                   for r in revisions]
+        self.assertEqual(deltas1, deltas2)
+
+    def test_get_revision_deltas(self):
+        repository = self.controldir.open_repository()
+        repository.lock_read()
+        self.addCleanup(repository.unlock)
+        revisions = [repository.get_revision(r) for r in
+                     [b'rev1', b'rev2', b'rev3', b'rev4']]
+        deltas1 = list(repository.get_revision_deltas(revisions))
+        deltas2 = [repository.get_revision_delta(r.revision_id)
+                   for r in revisions]
         self.assertEqual(deltas1, deltas2)
 
     def test_all_revision_ids(self):
 
=== modified file 'breezy/git/tree.py'
--- breezy/git/tree.py	2020-06-27 14:48:19 +0000
+++ breezy/git/tree.py	2020-06-28 18:07:07 +0000
@@ -1029,10 +1029,11 @@
         ret = {}
         changes = self._iter_git_changes(specific_files=paths)[0]
         for (change_type, old, new) in changes:
-            oldpath = old[0]
-            newpath = new[0]
+            if old[0] is None:
+                continue
+            oldpath = decode_git_path(old[0])
             if oldpath in paths:
-                ret[oldpath] = newpath
+                ret[oldpath] = decode_git_path(new[0]) if new[0] else None
         for path in paths:
             if path not in ret:
                 if self.source.has_filename(path):
@@ -1049,10 +1050,11 @@
         ret = {}
         changes = self._iter_git_changes(specific_files=paths)[0]
         for (change_type, old, new) in changes:
-            oldpath = old[0]
-            newpath = new[0]
+            if new[0] is None:
+                continue
+            newpath = decode_git_path(new[0])
             if newpath in paths:
-                ret[newpath] = oldpath
+                ret[newpath] = decode_git_path(old[0]) if old[0] else None
         for path in paths:
             if path not in ret:
                 if self.target.has_filename(path):
 
=== modified file 'breezy/log.py'
--- breezy/log.py	2020-06-21 02:15:25 +0000
+++ breezy/log.py	2020-06-28 18:07:07 +0000
@@ -982,7 +982,7 @@
         revisions = [rev[1] for rev in revs]
         new_revs = []
         if delta_type == 'full' and not check_fileids:
-            deltas = repository.get_deltas_for_revisions(revisions)
+            deltas = repository.get_revision_deltas(revisions)
             for rev, delta in zip(revs, deltas):
                 new_revs.append((rev[0], rev[1], delta))
         else:
 
=== modified file 'breezy/plugins/stats/cmds.py'
--- breezy/plugins/stats/cmds.py	2019-12-23 00:34:25 +0000
+++ breezy/plugins/stats/cmds.py	2020-06-28 18:07:07 +0000
@@ -315,7 +315,7 @@
     with ui.ui_factory.nested_progress_bar() as pb:
         with repository.lock_read():
             i = 0
-            for delta in repository.get_deltas_for_revisions(revs):
+            for delta in repository.get_revision_deltas(revs):
                 pb.update("classifying commits", i, len(revs))
                 for c in classify_delta(delta):
                     if c not in ret:
@@ -364,7 +364,7 @@
                     if ps is not None and r != NULL_REVISION]
         revs = repository.get_revisions(ancestry)
         with ui.ui_factory.nested_progress_bar() as pb:
-            iterator = zip(revs, repository.get_deltas_for_revisions(revs))
+            iterator = zip(revs, repository.get_revision_deltas(revs))
             for i, (rev, delta) in enumerate(iterator):
                 pb.update("analysing revisions", i, len(revs))
                 # Don't count merges
 
=== modified file 'breezy/repository.py'
--- breezy/repository.py	2020-01-20 00:03:54 +0000
+++ breezy/repository.py	2020-06-28 18:07:07 +0000
@@ -913,6 +913,44 @@
             r = self.get_revision(revision_id)
             return list(self.get_deltas_for_revisions([r]))[0]
 
+    def get_revision_deltas(self, revisions, specific_files=None):
+        """Produce a generator of revision deltas.
+
+        Note that the input is a sequence of REVISIONS, not revision ids.
+        Trees will be held in memory until the generator exits.
+        Each delta is relative to the revision's lefthand predecessor.
+
+        specific_files should exist in the first revision.
+
+        :param specific_files: if not None, the result is filtered
+          so that only those files, their parents and their
+          children are included.
+        """
+        from .tree import InterTree
+        # Get the revision-ids of interest
+        required_trees = set()
+        for revision in revisions:
+            required_trees.add(revision.revision_id)
+            required_trees.update(revision.parent_ids[:1])
+
+        trees = {
+            t.get_revision_id(): t
+            for t in self.revision_trees(required_trees)}
+
+        # Calculate the deltas
+        for revision in revisions:
+            if not revision.parent_ids:
+                old_tree = self.revision_tree(_mod_revision.NULL_REVISION)
+            else:
+                old_tree = trees[revision.parent_ids[0]]
+            intertree = InterTree.get(old_tree, trees[revision.revision_id])
+            yield intertree.compare(specific_files=specific_files)
+            if specific_files is not None:
+                specific_files = [
+                    p for p in intertree.find_source_paths(
+                        specific_files).values()
+                    if p is not None]
+
     def store_revision_signature(self, gpg_strategy, plaintext, revision_id):
         raise NotImplementedError(self.store_revision_signature)
 
 
=== modified file 'breezy/tests/per_repository/test_repository.py'
--- breezy/tests/per_repository/test_repository.py	2019-09-21 16:12:01 +0000
+++ breezy/tests/per_repository/test_repository.py	2020-06-28 18:07:07 +0000
@@ -997,10 +997,10 @@
         self.assertEqual(repo.control_url, repo.control_transport.base)
 
 
-class TestDeltaRevisionFiltered(per_repository.TestCaseWithRepository):
+class TestDeltaRevisionFileIdFiltered(per_repository.TestCaseWithRepository):
 
     def setUp(self):
-        super(TestDeltaRevisionFiltered, self).setUp()
+        super(TestDeltaRevisionFileIdFiltered, self).setUp()
         self.tree_a = self.make_branch_and_tree('a')
         self.build_tree(['a/foo', 'a/bar/', 'a/bar/b1', 'a/bar/b2', 'a/baz'])
         self.tree_a.add(['foo', 'bar', 'bar/b1', 'bar/b2', 'baz'])
@@ -1067,3 +1067,91 @@
         self.assertEqual([
             ('bar/b3', 'file'),
             ], [(c.path[1], c.kind[1]) for c in delta.added])
+
+
+class TestDeltaRevisionFilesFiltered(per_repository.TestCaseWithRepository):
+
+    def setUp(self):
+        super(TestDeltaRevisionFilesFiltered, self).setUp()
+        self.tree_a = self.make_branch_and_tree('a')
+        self.build_tree(
+            ['a/foo', 'a/bar/', 'a/bar/b1', 'a/bar/b2', 'a/baz', 'a/oldname'])
+        self.tree_a.add(['foo', 'bar', 'bar/b1', 'bar/b2', 'baz', 'oldname'])
+        self.rev1 = self.tree_a.commit('rev1')
+        self.build_tree(['a/bar/b3'])
+        self.tree_a.add('bar/b3')
+        self.tree_a.rename_one('oldname', 'newname')
+        self.rev2 = self.tree_a.commit('rev2')
+        self.repository = self.tree_a.branch.repository
+        self.addCleanup(self.repository.lock_read().unlock)
+
+    def test_multiple_files(self):
+        # Test multiple files
+        delta = list(self.repository.get_revision_deltas(
+            [self.repository.get_revision(self.rev1)], specific_files=[
+                'foo', 'baz']))[0]
+        self.assertIsInstance(delta, _mod_delta.TreeDelta)
+        self.assertEqual([
+            ('baz', 'file'),
+            ('foo', 'file'),
+            ], [(c.path[1], c.kind[1]) for c in delta.added])
+
+    def test_directory(self):
+        # Test a directory
+        delta = list(self.repository.get_revision_deltas(
+            [self.repository.get_revision(self.rev1)],
+            specific_files=['bar']))[0]
+        self.assertIsInstance(delta, _mod_delta.TreeDelta)
+        self.assertEqual([
+            ('bar', 'directory'),
+            ('bar/b1', 'file'),
+            ('bar/b2', 'file'),
+            ], [(c.path[1], c.kind[1]) for c in delta.added])
+
+    def test_unrelated(self):
+        # Try another revision
+        delta = list(self.repository.get_revision_deltas(
+            [self.repository.get_revision(self.rev2)],
+            specific_files=['foo']))[0]
+        self.assertIsInstance(delta, _mod_delta.TreeDelta)
+        self.assertEqual([], delta.added)
+
+    def test_renamed(self):
+        # Try another revision
+        self.assertTrue(
+            self.repository.revision_tree(self.rev2).has_filename('newname'))
+        self.assertTrue(
+            self.repository.revision_tree(self.rev1).has_filename('oldname'))
+        revs = [
+            self.repository.get_revision(self.rev2),
+            self.repository.get_revision(self.rev1)]
+        delta2, delta1 = list(self.repository.get_revision_deltas(
+            revs, specific_files=['newname']))
+        self.assertIsInstance(delta1, _mod_delta.TreeDelta)
+        self.assertEqual([('oldname', 'newname')], [c.path for c in delta2.renamed])
+        self.assertIsInstance(delta2, _mod_delta.TreeDelta)
+        self.assertEqual(['oldname'], [c.path[1] for c in delta1.added])
+
+    def test_file_in_directory(self):
+        # Test a file in a directory, both of which were added
+        delta = list(self.repository.get_revision_deltas(
+            [self.repository.get_revision(self.rev1)],
+            specific_files=['bar/b2']))[0]
+        self.assertIsInstance(delta, _mod_delta.TreeDelta)
+        self.assertEqual([
+            ('bar', 'directory'),
+            ('bar/b2', 'file'),
+            ], [(c.path[1], c.kind[1]) for c in delta.added])
+
+    def test_file_in_unchanged_directory(self):
+        delta = list(self.repository.get_revision_deltas(
+            [self.repository.get_revision(self.rev2)],
+            specific_files=['bar/b3']))[0]
+        self.assertIsInstance(delta, _mod_delta.TreeDelta)
+        if [(c.path[1], c.kind[1]) for c in delta.added] == [
+                ('bar', 'directory'), ('bar/b3', 'file')]:
+            self.knownFailure("bzr incorrectly reports 'bar' as added - "
+                              "bug 878217")
+        self.assertEqual([
+            ('bar/b3', 'file'),
+            ], [(c.path[1], c.kind[1]) for c in delta.added])