Support completely empty messages in git objects

As it turns out, a git object can be terminated by the last header, without empty lines. In that case, dulwich currently "eats" the last header line, which is obviously wrong. When such a case happens, return an object with a None message instead of an empty line. We make sure that serialization/deserialization still works as expected.
dysosmus · Feb 17, 2016 · f8e4295 · f8e4295
1 parent a5d6568
commit f8e4295
Show file tree

Hide file tree

Showing 2 changed files with 52 additions and 4 deletions.
diff --git a/dulwich/objects.py b/dulwich/objects.py
@@ -602,17 +602,39 @@ def _parse_message(chunks):
     f = BytesIO(b''.join(chunks))
     k = None
     v = ""
+    eof = False
+
+    # Parse the headers
+    #
+    # Headers can contain newlines. The next line is indented with a space.
+    # We store the latest key as 'k', and the accumulated value as 'v'.
     for l in f:
         if l.startswith(b' '):
+            # Indented continuation of the previous line
             v += l[1:]
         else:
             if k is not None:
+                # We parsed a new header, return its value
                 yield (k, v.rstrip(b'\n'))
             if l == b'\n':
                 # Empty line indicates end of headers
                 break
             (k, v) = l.split(b' ', 1)
-    yield (None, f.read())
+
+    else:
+        # We reached end of file before the headers ended. We still need to
+        # return the previous header, then we need to return a None field for
+        # the text.
+        eof = True
+        if k is not None:
+            yield (k, v.rstrip(b'\n'))
+        yield (None, None)
+
+    if not eof:
+        # We didn't reach the end of file while parsing headers. We can return
+        # the rest of the file as a message.
+        yield (None, f.read())
+
     f.close()
 
 
@@ -679,8 +701,9 @@ def _serialize(self):
                 chunks.append(git_line(
                     _TAGGER_HEADER, self._tagger, str(self._tag_time).encode('ascii'),
                     format_timezone(self._tag_timezone, self._tag_timezone_neg_utc)))
-        chunks.append(b'\n') # To close headers
-        chunks.append(self._message)
+        if self._message is not None:
+            chunks.append(b'\n') # To close headers
+            chunks.append(self._message)
         return chunks
 
     def _deserialize(self, chunks):

diff --git a/dulwich/tests/test_objects.py b/dulwich/tests/test_objects.py
@@ -818,6 +818,20 @@ def test_serialize_simple(self):
                           b'\n'
                           b'Tag 0.1'), x.as_raw_string())
 
+    def test_serialize_none_message(self):
+        x = make_object(Tag,
+                        tagger=b'Jelmer Vernooij <jelmer@samba.org>',
+                        name=b'0.1',
+                        message=None,
+                        object=(Blob, b'd80c186a03f423a81b39df39dc87fd269736ca86'),
+                        tag_time=423423423,
+                        tag_timezone=0)
+        self.assertEqual((b'object d80c186a03f423a81b39df39dc87fd269736ca86\n'
+                          b'type blob\n'
+                          b'tag 0.1\n'
+                          b'tagger Jelmer Vernooij <jelmer@samba.org> '
+                          b'423423423 +0000\n'), x.as_raw_string())
+
 
 default_tagger = (b'Linus Torvalds <torvalds@woody.linux-foundation.org> '
                   b'1183319674 -0700')
@@ -849,8 +863,8 @@ def make_tag_lines(self,
             lines.append(b'tag ' + name)
         if tagger is not None:
             lines.append(b'tagger ' + tagger)
-        lines.append(b'')
         if message is not None:
+            lines.append(b'')
             lines.append(message)
         return lines
 
@@ -877,6 +891,17 @@ def test_parse_no_tagger(self):
         self.assertEqual(None, x.tagger)
         self.assertEqual(b'v2.6.22-rc7', x.name)
 
+    def test_parse_no_message(self):
+        x = Tag()
+        x.set_raw_string(self.make_tag_text(message=None))
+        self.assertEqual(None, x.message)
+        self.assertEqual(
+            b'Linus Torvalds <torvalds@woody.linux-foundation.org>', x.tagger)
+        self.assertEqual(datetime.datetime.utcfromtimestamp(x.tag_time),
+                          datetime.datetime(2007, 7, 1, 19, 54, 34))
+        self.assertEqual(-25200, x.tag_timezone)
+        self.assertEqual(b'v2.6.22-rc7', x.name)
+
     def test_check(self):
         self.assertCheckSucceeds(Tag, self.make_tag_text())
         self.assertCheckFails(Tag, self.make_tag_text(object_sha=None))