Suppress unified_diff for files deemed to be "binary".

dulwich/patch.py

@@ -31,6 +31,9 @@
     S_ISGITLINK,
     )
 
+FIRST_FEW_BYTES = 8000
+
+
 def write_commit_patch(f, commit, contents, progress, version=None):
     """Write a individual file patch.
 
@@ -103,6 +106,11 @@ def unified_diff(a, b, fromfile='', tofile='', n=3):
                     yield '+' + line
 
 
+def is_binary(content):
+    """See if the first few bytes contains any null characters."""
+    return '\0' in content[:FIRST_FEW_BYTES]
+
+
 def write_object_diff(f, store, (old_path, old_mode, old_id),
                                 (new_path, new_mode, new_id)):
     """Write the diff for an object.
@@ -119,13 +127,21 @@ def shortid(hexsha):
             return "0" * 7
         else:
             return hexsha[:7]
-    def lines(mode, hexsha):
+
+    def content(mode, hexsha):
         if hexsha is None:
-            return []
+            return ''
         elif S_ISGITLINK(mode):
-            return ["Submodule commit " + hexsha + "\n"]
+            return "Submodule commit " + hexsha + "\n"
         else:
-            return store[hexsha].data.splitlines(True)
+            return store[hexsha].data
+
+    def lines(content):
+        if not content:
+            return []
+        else:
+            return content.splitlines(True)
+
     if old_path is None:
         old_path = "/dev/null"
     else:
@@ -146,10 +162,13 @@ def lines(mode, hexsha):
     if new_mode is not None:
         f.write(" %o" % new_mode)
     f.write("\n")
-    old_contents = lines(old_mode, old_id)
-    new_contents = lines(new_mode, new_id)
-    f.writelines(unified_diff(old_contents, new_contents,
-        old_path, new_path))
+    old_content = content(old_mode, old_id)
+    new_content = content(new_mode, new_id)
+    if is_binary(old_content) or is_binary(new_content):
+        f.write("Binary files %s and %s differ\n" % (old_path, new_path))
+    else:
+        f.writelines(unified_diff(lines(old_content), lines(new_content),
+            old_path, new_path))
 
 
 def write_blob_diff(f, (old_path, old_mode, old_blob),

dulwich/tests/test_patch.py

@@ -369,6 +369,57 @@ def test_object_diff_remove_blob(self):
             '-same'
             ], f.getvalue().splitlines())
 
+    def test_object_diff_bin_blob(self):
+        f = StringIO()
+        # Prepare two slightly different PNG headers
+        b1 = Blob.from_string(
+            "\x89\x50\x4e\x47\x0d\x0a\x1a\x0a\x00\x00\x00\x0d\x49\x48\x44\x52"
+            "\x00\x00\x01\xd5\x00\x00\x00\x9f\x08\x04\x00\x00\x00\x05\x04\x8b")
+        b2 = Blob.from_string(
+            "\x89\x50\x4e\x47\x0d\x0a\x1a\x0a\x00\x00\x00\x0d\x49\x48\x44\x52"
+            "\x00\x00\x01\xd5\x00\x00\x00\x9f\x08\x03\x00\x00\x00\x98\xd3\xb3")
+        store = MemoryObjectStore()
+        store.add_objects([(b1, None), (b2, None)])
+        write_object_diff(f, store, ('foo.png', 0644, b1.id),
+                                    ('bar.png', 0644, b2.id))
+        self.assertEqual([
+            'diff --git a/foo.png b/bar.png',
+            'index f73e47d..06364b7 644',
+            'Binary files a/foo.png and b/bar.png differ'
+            ], f.getvalue().splitlines())
+
+    def test_object_diff_add_bin_blob(self):
+        f = StringIO()
+        b2 = Blob.from_string(
+            '\x89\x50\x4e\x47\x0d\x0a\x1a\x0a\x00\x00\x00\x0d\x49\x48\x44\x52'
+            '\x00\x00\x01\xd5\x00\x00\x00\x9f\x08\x03\x00\x00\x00\x98\xd3\xb3')
+        store = MemoryObjectStore()
+        store.add_object(b2)
+        write_object_diff(f, store, (None, None, None),
+                                    ('bar.png', 0644, b2.id))
+        self.assertEqual([
+            'diff --git /dev/null b/bar.png',
+            'new mode 644',
+            'index 0000000..06364b7 644',
+            'Binary files /dev/null and b/bar.png differ'
+            ], f.getvalue().splitlines())
+
+    def test_object_diff_remove_bin_blob(self):
+        f = StringIO()
+        b1 = Blob.from_string(
+            '\x89\x50\x4e\x47\x0d\x0a\x1a\x0a\x00\x00\x00\x0d\x49\x48\x44\x52'
+            '\x00\x00\x01\xd5\x00\x00\x00\x9f\x08\x04\x00\x00\x00\x05\x04\x8b')
+        store = MemoryObjectStore()
+        store.add_object(b1)
+        write_object_diff(f, store, ('foo.png', 0644, b1.id),
+                                    (None, None, None))
+        self.assertEqual([
+            'diff --git a/foo.png /dev/null',
+            'deleted mode 644',
+            'index f73e47d..0000000',
+            'Binary files a/foo.png and /dev/null differ'
+            ], f.getvalue().splitlines())
+
     def test_object_diff_kind_change(self):
         f = StringIO()
         b1 = Blob.from_string("new\nsame\n")