1.0.14dev: fix IndexError raised from diff_tree() raised when git repository has a file started with a colon character (closes #12758)

git-svn-id: http://trac.edgewall.org/intertrac/log:/branches/1.0-stable@15759 af82e41b-90c4-0310-8c96-b1721e28e2e2

Author: jomae

tracopt/versioncontrol/git/PyGIT.py

@@ -1011,44 +1011,44 @@ def diff_tree(self, tree1, tree2, path='', find_renames=False):
         if find_renames:
             diff_tree_args.append('-M')
         diff_tree_args.extend([str(tree1) if tree1 else '--root',
-                               str(tree2),
-                               '--', path])
+                               str(tree2), '--', path])
+        result = self.repo.diff_tree(*diff_tree_args)
+        if not result:
+            return
 
-        lines = self.repo.diff_tree(*diff_tree_args).split('\0')
+        def iter_entry(result):
+            start = 0
+            while True:
+                idx = result.find('\0', start)
+                if idx == -1:
+                    return
+                yield result[start:idx]
+                start = idx + 1
+
+        iterate = iter_entry(result)
 
-        assert lines[-1] == ''
-        del lines[-1]
+        def next_entry():
+            return iterate.next()
 
-        if tree1 is None and lines:
+        if not tree1:
             # if only one tree-sha is given on commandline,
             # the first line is just the redundant tree-sha itself...
-            assert not lines[0].startswith(':')
-            del lines[0]
-
-        # FIXME: the following code is ugly, needs rewrite
-
-        chg = None
-
-        def __chg_tuple():
-            if len(chg) == 6:
-                chg.append(None)
-            else:
-                chg[6] = self._fs_to_unicode(chg[6])
-            chg[5] = self._fs_to_unicode(chg[5])
+            entry = next_entry()
+            assert not entry.startswith(':')
 
-            assert len(chg) == 7
-            return tuple(chg)
-
-        for line in lines:
-            if line.startswith(':'):
-                if chg:
-                    yield __chg_tuple()
-
-                chg = line[1:].split()
-                assert len(chg) == 5
-            else:
-                chg.append(line)
-
-        # handle left-over chg entry
-        if chg:
-            yield __chg_tuple()
+        while True:
+            try:
+                entry = next_entry()
+            except StopIteration:
+                return
+            assert entry.startswith(':')
+            values = entry[1:].split(' ')
+            assert len(values) == 5
+            old_mode, new_mode, old_sha, new_sha, change = values
+            change = change[:1]
+            old_path = self._fs_to_unicode(next_entry())
+            new_path = None
+            if change in ('R', 'C'):  # renamed or copied
+                new_path = self._fs_to_unicode(next_entry())
+            yield (old_mode, new_mode, old_sha, new_sha, change, old_path,
+                   new_path)

tracopt/versioncontrol/git/git_fs.py

@@ -872,7 +872,7 @@ def get_changes(self):
                    if mode2.startswith('04') or mode1.startswith('04') \
                    else Node.FILE
 
-            action = GitChangeset.action_map[action[0]]
+            action = GitChangeset.action_map[action]
 
             if action == Changeset.ADD:
                 p_path = p_rev = None

tracopt/versioncontrol/git/tests/git_fs.py

@@ -778,6 +778,70 @@ def test_iter_nodes(self):
             (root_commit, 'B/b2.txt'),
             ], [(node.created_rev, node.path) for node in nodes])
 
+    def test_colon_character_in_filename(self):
+        self._git_init(data=False)
+        self._git_fast_import(self._data_colon_character_in_filename)
+        self._add_repository('gitrepos')
+        repos = self._repomgr.get_repository('gitrepos')
+        repos.sync()
+        rev1 = '382e1e6b85ba20ce8a84af1a875eaa50b8e1e092'  # root commit
+        rev2 = 'd8001832aad079f85a39a54a388a8b15fe31093d'
+        ADD = Changeset.ADD
+        MOVE = Changeset.MOVE
+        FILE = Node.FILE
+
+        cset = repos.get_changeset(rev1)
+        self.assertEqual(set([('0100644',     FILE, ADD, None, None),
+                              ('0100644.txt', FILE, ADD, None, None),
+                              (':100644',     FILE, ADD, None, None),
+                              (':100644.txt', FILE, ADD, None, None),
+                              ('a100644',     FILE, ADD, None, None),
+                              ('a100644.txt', FILE, ADD, None, None)
+                             ]),
+                         set(cset.get_changes()))
+
+        cset = repos.get_changeset(rev2)
+        self.assertEqual(set([(':100666', FILE, MOVE, ':100644', rev1)]),
+                         set(cset.get_changes()))
+
+    _data_colon_character_in_filename = """\
+blob
+mark :1
+data 0
+
+blob
+mark :2
+data 16
+...............
+
+reset refs/heads/master
+commit refs/heads/master
+mark :3
+author Joe <[email protected]> 1491387182 +0000
+committer Joe <[email protected]> 1491387182 +0000
+data 9
+(#12758)
+M 100644 :1 0100644.txt
+M 100644 :1 0100644
+M 100644 :1 :100644.txt
+M 100644 :2 :100644
+M 100644 :1 a100644.txt
+M 100644 :1 a100644
+
+commit refs/heads/master
+mark :4
+author Joe <[email protected]> 1491387183 +0000
+committer Joe <[email protected]> 1491387183 +0000
+data 16
+(#12758) rename
+from :3
+D :100644
+M 100644 :2 :100666
+
+reset refs/heads/master
+from :4
+"""
+
     def _get_quickjump_names(self, repos):
         return list(name for type, name, path, rev
                          in repos.get_quickjump_entries('HEAD'))