Avoid regenerating full blob in Blob.splitlines.

dulwich/objects.py

@@ -573,7 +573,27 @@ class Blob(ShaFile):
 
         This preserves the original line endings.
         """
-        return self.data.splitlines(True)
+        chunks = self.chunked
+        if not chunks:
+            return []
+        if len(chunks) == 1:
+            return chunks[0].splitlines(True)
+        remaining = None
+        ret = []
+        for chunk in chunks:
+            lines = chunk.splitlines(True)
+            if len(lines) > 1:
+                ret.append((remaining or "") + lines[0])
+                ret.extend(lines[1:-1])
+                remaining = lines[-1]
+            elif len(lines) == 1:
+                if remaining is None:
+                    remaining = lines.pop()
+                else:
+                    remaining += lines.pop()
+        if remaining is not None:
+            ret.append(remaining)
+        return ret
 
 
 def _parse_message(chunks):

dulwich/tests/test_objects.py

@@ -137,24 +137,20 @@ class BlobReadTests(TestCase):
         self.assertEqual([string], b.chunked)
 
     def test_splitlines(self):
-        b = Blob()
-        b.chunked = ['bl\na', 'blie']
-        self.assertEqual(['bl\n', 'ablie'], b.splitlines())
-
-    def test_splitlines_nonewline(self):
-        b = Blob()
-        b.chunked = ['bl\na', 'blie', 'bloe\n']
-        self.assertEqual(['bl\n', 'abliebloe\n'], b.splitlines())
-
-    def test_splitlines_firstempty(self):
-        b = Blob()
-        b.chunked = ['', 'bl\na', 'blie', 'bloe\n']
-        self.assertEqual(['bl\n', 'abliebloe\n'], b.splitlines())
-
-    def test_splitlines_nonewline(self):
-        b = Blob()
-        b.chunked = []
-        self.assertEqual([], b.splitlines())
+        for case in [
+            [],
+            ['foo\nbar\n'],
+            ['bl\na', 'blie'],
+            ['bl\na', 'blie', 'bloe\n'],
+            ['', 'bl\na', 'blie', 'bloe\n'],
+            ['', '', '', 'bla\n'],
+            ['', '', '', 'bla\n', ''],
+            ['bl', '', 'a\naaa'],
+            ['a\naaa', 'a'],
+            ]:
+            b = Blob()
+            b.chunked = case
+            self.assertEqual(b.data.splitlines(True), b.splitlines())
 
     def test_set_chunks(self):
         b = Blob()