Handle 'Only in <dir>: ' header in output from diff -ru d1 d2

cdiff.py

@@ -168,6 +168,9 @@ class Diff(object):
     def is_eof(self, line):
         return False
 
+    def is_only_in_dir(self, line):
+        return False
+
     def markup_traditional(self):
         """Returns a generator"""
         for line in self._headers:
@@ -260,13 +263,18 @@ class Diff(object):
 
             return ''.join(out)
 
-        # Setup line width and number width
+        # Set up line width
         if width <= 0:
             width = 80
-        (start, offset) = self._hunks[-1]._old_addr
-        max1 = start + offset - 1
-        (start, offset) = self._hunks[-1]._new_addr
-        max2 = start + offset - 1
+
+        # Set up number width, note last hunk might be empty
+        try:
+            (start, offset) = self._hunks[-1]._old_addr
+            max1 = start + offset - 1
+            (start, offset) = self._hunks[-1]._new_addr
+            max2 = start + offset - 1
+        except IndexError:
+            max1 = max2 = 0
         num_width = max(len(str(max1)), len(str(max2)))
 
         # Setup lineno and line format
@@ -412,6 +420,9 @@ class Udiff(Diff):
         # \ No newline at end of property
         return line.startswith(r'\ No newline at end of')
 
+    def is_only_in_dir(self, line):
+        return line.startswith('Only in ')
+
 
 class PatchStream(object):
 
@@ -492,7 +503,9 @@ class DiffParser(object):
             line = decode(line)
 
             if difflet.is_old_path(line):
-                # FIXME: '--- ' breaks here, need to probe next 3 lines
+                # FIXME: '--- ' breaks here, need to probe next 3 lines,
+                # reproducible with github raw patch
+                #
                 if diff._old_path and diff._new_path and len(diff._hunks) > 0:
                     # One diff constructed
                     yield diff
@@ -518,6 +531,16 @@ class DiffParser(object):
                 # ignore
                 pass
 
+            elif difflet.is_only_in_dir(line):
+                # 'Only in foo: ' is considered a separate diff, so yield
+                # current diff, then this line
+                #
+                if diff._old_path and diff._new_path and len(diff._hunks) > 0:
+                    # One diff constructed
+                    yield diff
+                yield Diff([line], '', '', [])
+                diff = Diff([], None, None, [])
+
             else:
                 # All other non-recognized lines are considered as headers or
                 # hunk headers respectively
@@ -527,12 +550,14 @@ class DiffParser(object):
         if headers:
             raise RuntimeError('dangling header(s):\n%s' % ''.join(headers))
 
-        # Validate and yield the last patch set
-        assert diff._old_path is not None
-        assert diff._new_path is not None
-        assert len(diff._hunks) > 0
-        assert len(diff._hunks[-1]._hunk_meta) > 0
-        yield diff
+        # Validate and yield the last patch set if it is not yielded yet
+        if diff._old_path:
+            assert diff._new_path is not None
+            assert len(diff._hunks) > 0
+            if diff._hunks:
+                assert len(diff._hunks[-1]._hunk_meta) > 0
+                assert len(diff._hunks[-1]._hunk_list) > 0
+            yield diff
 
 
 class DiffMarkup(object):

tests/README

@@ -1,6 +1,6 @@
 # To generate expected output, chdir to a subdir and use following command, then
 # review with `less -R`
 #
-../../cdiff.py -c always in.diff > out.normal 
-../../cdiff.py -c always -s in.diff > out.side-by-side
-../../cdiff.py -c always -s in.diff -w70 > out.w70
+../../cdiff.py -c always < in.diff > out.normal
+../../cdiff.py -c always -s < in.diff > out.side-by-side
+../../cdiff.py -c always -s < in.diff -w70 > out.w70

tests/dir/in.diff

@@ -0,0 +1,22 @@
+diff -ru a/README b/README
+--- a/README	2013-02-22 20:26:43.000000000 +0800
++++ b/README	2013-02-22 20:27:51.000000000 +0800
+@@ -1,6 +1,6 @@
+-# To generate expected output, chdir to a subdir and use following command, then
++# To generate expected output, cd to a subdir and use following command, then
+ # review with `less -R`
+-#
+ ../../cdiff.py -c always in.diff > out.normal 
+ ../../cdiff.py -c always -s in.diff > out.side-by-side
+ ../../cdiff.py -c always -s in.diff -w70 > out.w70
++# EOF
+Only in a: a1
+Only in b: b1
+diff -ru a/common/foo.txt b/common/foo.txt
+--- a/common/foo.txt	2013-02-22 20:28:32.000000000 +0800
++++ b/common/foo.txt	2013-02-22 20:30:18.000000000 +0800
+@@ -1 +1 @@
+-Hello
++world
+Only in b: date.txt
+Only in a: time.txt

tests/dir/out.normal

@@ -0,0 +1,23 @@
+diff -ru a/README b/README
+--- a/README	2013-02-22 20:26:43.000000000 +0800
++++ b/README	2013-02-22 20:27:51.000000000 +0800
+@@ -1,6 +1,6 @@
+-# To generate expected output, chdir to a subdir and use following command, then
++# To generate expected output, cd to a subdir and use following command, then
+ # review with `less -R`
+-#
+ ../../cdiff.py -c always in.diff > out.normal 
+ ../../cdiff.py -c always -s in.diff > out.side-by-side
+ ../../cdiff.py -c always -s in.diff -w70 > out.w70
++# EOF
+Only in a: a1
+Only in b: b1
+diff -ru a/common/foo.txt b/common/foo.txt
+--- a/common/foo.txt	2013-02-22 20:28:32.000000000 +0800
++++ b/common/foo.txt	2013-02-22 20:30:18.000000000 +0800
+@@ -1 +1 @@
+-Hello
++world
+Only in b: date.txt
+Only in a: time.txt
+

tests/dir/out.side-by-side

@@ -0,0 +1,21 @@
+diff -ru a/README b/README
+--- a/README	2013-02-22 20:26:43.000000000 +0800
++++ b/README	2013-02-22 20:27:51.000000000 +0800
+@@ -1,6 +1,6 @@
+1 # To generate expected output, chdir to a subdir and use following command, then 1 # To generate expected output, cd to a subdir and use following command, then
+2 # review with `less -R`                                                          2 # review with `less -R`
+3 #   
+4 ../../cdiff.py -c always in.diff > out.normal                                    3 ../../cdiff.py -c always in.diff > out.normal 
+5 ../../cdiff.py -c always -s in.diff > out.side-by-side                           4 ../../cdiff.py -c always -s in.diff > out.side-by-side
+6 ../../cdiff.py -c always -s in.diff -w70 > out.w70                               5 ../../cdiff.py -c always -s in.diff -w70 > out.w70
+                                                                                   6 # EOF
+Only in a: a1
+Only in b: b1
+diff -ru a/common/foo.txt b/common/foo.txt
+--- a/common/foo.txt	2013-02-22 20:28:32.000000000 +0800
++++ b/common/foo.txt	2013-02-22 20:30:18.000000000 +0800
+@@ -1 +1 @@
+1 Hello                                                                            1 world
+Only in b: date.txt
+Only in a: time.txt
+

tests/dir/out.w70

@@ -0,0 +1,21 @@
+diff -ru a/README b/README
+--- a/README	2013-02-22 20:26:43.000000000 +0800
++++ b/README	2013-02-22 20:27:51.000000000 +0800
+@@ -1,6 +1,6 @@
+1 # To generate expected output, chdir to a subdir and use following co> 1 # To generate expected output, cd to a subdir and use following comma>
+2 # review with `less -R`                                                2 # review with `less -R`
+3 #   
+4 ../../cdiff.py -c always in.diff > out.normal                          3 ../../cdiff.py -c always in.diff > out.normal 
+5 ../../cdiff.py -c always -s in.diff > out.side-by-side                 4 ../../cdiff.py -c always -s in.diff > out.side-by-side
+6 ../../cdiff.py -c always -s in.diff -w70 > out.w70                     5 ../../cdiff.py -c always -s in.diff -w70 > out.w70
+                                                                         6 # EOF
+Only in a: a1
+Only in b: b1
+diff -ru a/common/foo.txt b/common/foo.txt
+--- a/common/foo.txt	2013-02-22 20:28:32.000000000 +0800
++++ b/common/foo.txt	2013-02-22 20:30:18.000000000 +0800
+@@ -1 +1 @@
+1 Hello                                                                  1 world
+Only in b: date.txt
+Only in a: time.txt
+

tests/test_cdiff.py

@@ -318,7 +318,7 @@ spam
         parser = cdiff.DiffParser(stream)
         self.assertRaises(AssertionError, list, parser._parse())
 
-    def test_parse_missing_hunk(self):
+    def test_parse_missing_hunk_meta(self):
         patch = r"""\
 --- a
 +++ b
@@ -334,6 +334,69 @@ spam
         parser = cdiff.DiffParser(stream)
         self.assertRaises(AssertionError, list, parser._parse())
 
+    def test_parse_missing_hunk_list(self):
+        patch = r"""\
+--- a
++++ b
+@@ -1,2 +1,2 @@
+-foo
++bar
+ common
+--- c
++++ d
+@@ -1,2 +1,2 @@
+"""
+        items = patch.splitlines(True)
+        stream = cdiff.PatchStream(Sequential(items))
+        parser = cdiff.DiffParser(stream)
+        self.assertRaises(AssertionError, list, parser._parse())
+
+    def test_parse_only_in_dir(self):
+        patch = r"""\
+--- a
++++ b
+@@ -1,2 +1,2 @@
+-foo
++bar
+ common
+Only in foo: foo
+--- c
++++ d
+@@ -1,2 +1,2 @@
+-foo
++bar
+ common
+"""
+        items = patch.splitlines(True)
+        stream = cdiff.PatchStream(Sequential(items))
+        parser = cdiff.DiffParser(stream)
+
+        out = list(parser._parse())
+        self.assertEqual(len(out), 3)
+        self.assertEqual(len(out[1]._hunks), 0)
+        self.assertEqual(out[1]._headers, ['Only in foo: foo\n'])
+        self.assertEqual(len(out[2]._hunks), 1)
+        self.assertEqual(len(out[2]._hunks[0]._hunk_list), 3)
+
+    def test_parse_only_in_dir_at_last(self):
+        patch = r"""\
+--- a
++++ b
+@@ -1,2 +1,2 @@
+-foo
++bar
+ common
+Only in foo: foo
+"""
+        items = patch.splitlines(True)
+        stream = cdiff.PatchStream(Sequential(items))
+        parser = cdiff.DiffParser(stream)
+
+        out = list(parser._parse())
+        self.assertEqual(len(out), 2)
+        self.assertEqual(len(out[1]._hunks), 0)
+        self.assertEqual(out[1]._headers, ['Only in foo: foo\n'])
+
     def test_parse_svn_prop(self):
         patch = r"""\
 --- a