Handle 'Binary files ... differ'

cdiff.py

@@ -169,6 +169,11 @@ def is_eof(self, line):
     def is_only_in_dir(self, line):
         return False
 
+    def is_binary_differ(self, line):
+        return False
+
+    # Followings are not suppose to override
+    #
     def markup_traditional(self):
         """Returns a generator"""
         for line in self._headers:
@@ -421,6 +426,9 @@ def is_eof(self, line):
     def is_only_in_dir(self, line):
         return line.startswith('Only in ')
 
+    def is_binary_differ(self, line):
+        return re.match('^Binary files .* differ$', line.rstrip())
+
 
 class PatchStream(object):
 
@@ -529,14 +537,16 @@ def _parse(self):
                 # ignore
                 pass
 
-            elif difflet.is_only_in_dir(line):
-                # 'Only in foo: ' is considered a separate diff, so yield
-                # current diff, then this line
+            elif difflet.is_only_in_dir(line) or difflet.is_binary_differ(line):
+                # 'Only in foo:' and 'Binary files ... differ' are considered as
+                # separate diffs, so yield current diff, then this line
                 #
                 if diff._old_path and diff._new_path and len(diff._hunks) > 0:
-                    # One diff constructed
+                    # Current diff is comppletely constructed
                     yield diff
-                yield Diff([line], '', '', [])
+                headers.append(line)
+                yield Diff(headers, '', '', [])
+                headers = []
                 diff = Diff([], None, None, [])
 
             else:

tests/diff-ru-bin/in.diff

@@ -0,0 +1,20 @@
+diff -ru a/README b/README
+--- a/README	2013-02-23 19:51:41.000000000 +0800
++++ b/README	2013-02-23 19:52:06.000000000 +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
+Binary files a/foo.pdf and b/foo.pdf differ
+diff -ru a/foo.txt b/foo.txt
+--- a/foo.txt	2013-02-23 19:55:03.000000000 +0800
++++ b/foo.txt	2013-02-23 19:55:10.000000000 +0800
+@@ -1,2 +1,2 @@
+-Hello
++hella
+ world

tests/diff-ru-bin/out.normal

@@ -0,0 +1,21 @@
+diff -ru a/README b/README
+--- a/README	2013-02-23 19:51:41.000000000 +0800
++++ b/README	2013-02-23 19:52:06.000000000 +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
+Binary files a/foo.pdf and b/foo.pdf differ
+diff -ru a/foo.txt b/foo.txt
+--- a/foo.txt	2013-02-23 19:55:03.000000000 +0800
++++ b/foo.txt	2013-02-23 19:55:10.000000000 +0800
+@@ -1,2 +1,2 @@
+-Hello
++hella
+ world
+

tests/diff-ru-bin/out.side-by-side

@@ -0,0 +1,18 @@
+diff -ru a/README b/README
+--- a/README	2013-02-23 19:51:41.000000000 +0800
++++ b/README	2013-02-23 19:52:06.000000000 +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
+Binary files a/foo.pdf and b/foo.pdf differ
+diff -ru a/foo.txt b/foo.txt
+--- a/foo.txt	2013-02-23 19:55:03.000000000 +0800
++++ b/foo.txt	2013-02-23 19:55:10.000000000 +0800
+@@ -1,2 +1,2 @@
+1 Hello                                                                            1 hella
+2 world                                                                            2 world

tests/diff-ru-bin/out.w70

@@ -0,0 +1,18 @@
+diff -ru a/README b/README
+--- a/README	2013-02-23 19:51:41.000000000 +0800
++++ b/README	2013-02-23 19:52:06.000000000 +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
+Binary files a/foo.pdf and b/foo.pdf differ
+diff -ru a/foo.txt b/foo.txt
+--- a/foo.txt	2013-02-23 19:55:03.000000000 +0800
++++ b/foo.txt	2013-02-23 19:55:10.000000000 +0800
+@@ -1,2 +1,2 @@
+1 Hello                                                                  1 hella
+2 world                                                                  2 world

tests/git-bin/in.diff

@@ -0,0 +1,26 @@
+diff --git a/2pdf.sh b/2pdf.sh
+index 529d8a3..ad71911 100755
+--- a/2pdf.sh
++++ b/2pdf.sh
+@@ -13,3 +13,5 @@ INPUT=${1:-foo.html}
+ OUTPUT=${INPUT%.html}.pdf
+
+ wkhtmltopdf --page-size A4 $INPUT $OUTPUT  # very very long comments ends here
++
++# EOF
+diff --git a/example.pdf b/example.pdf
+index 1eacfd8..3696851 100644
+Binary files a/example.pdf and b/example.pdf differ
+diff --git a/foo.html b/foo.html
+index d2fd3fb..13afa6e 100644
+--- a/foo.html
++++ b/foo.html
+@@ -1,4 +1,4 @@
+-<!doctype html>
++<!DOCTYPE html>
+ <html>
+
+ <head>
+diff --git a/foo.pdf b/foo.pdf
+index 0e90017..3c3b90d 100644
+Binary files a/foo.pdf and b/foo.pdf differ

tests/git-bin/out.normal

@@ -0,0 +1,27 @@
+diff --git a/2pdf.sh b/2pdf.sh
+index 529d8a3..ad71911 100755
+--- a/2pdf.sh
++++ b/2pdf.sh
+@@ -13,3 +13,5 @@ INPUT=${1:-foo.html}
+ OUTPUT=${INPUT%.html}.pdf
+ 
+ wkhtmltopdf --page-size A4 $INPUT $OUTPUT  # very very long comments ends here
++
++# EOF
+diff --git a/example.pdf b/example.pdf
+index 1eacfd8..3696851 100644
+Binary files a/example.pdf and b/example.pdf differ
+diff --git a/foo.html b/foo.html
+index d2fd3fb..13afa6e 100644
+--- a/foo.html
++++ b/foo.html
+@@ -1,4 +1,4 @@
+-<!doctype html>
++<!DOCTYPE html>
+ <html>
+ 
+ <head>
+diff --git a/foo.pdf b/foo.pdf
+index 0e90017..3c3b90d 100644
+Binary files a/foo.pdf and b/foo.pdf differ
+

tests/git-bin/out.side-by-side

@@ -0,0 +1,26 @@
+diff --git a/2pdf.sh b/2pdf.sh
+index 529d8a3..ad71911 100755
+--- a/2pdf.sh
++++ b/2pdf.sh
+@@ -13,3 +13,5 @@ INPUT=${1:-foo.html}
+13 OUTPUT=${INPUT%.html}.pdf                                                        13 OUTPUT=${INPUT%.html}.pdf
+14                                                                                  14 
+15 wkhtmltopdf --page-size A4 $INPUT $OUTPUT  # very very long comments ends here   15 wkhtmltopdf --page-size A4 $INPUT $OUTPUT  # very very long comments ends here
+                                                                                    16 
+                                                                                    17 # EOF
+diff --git a/example.pdf b/example.pdf
+index 1eacfd8..3696851 100644
+Binary files a/example.pdf and b/example.pdf differ
+diff --git a/foo.html b/foo.html
+index d2fd3fb..13afa6e 100644
+--- a/foo.html
++++ b/foo.html
+@@ -1,4 +1,4 @@
+1 <!doctype html>                                                                  1 <!DOCTYPE html>
+2 <html>                                                                           2 <html>
+3                                                                                  3 
+4 <head>                                                                           4 <head>
+diff --git a/foo.pdf b/foo.pdf
+index 0e90017..3c3b90d 100644
+Binary files a/foo.pdf and b/foo.pdf differ
+

tests/git-bin/out.w70

@@ -0,0 +1,26 @@
+diff --git a/2pdf.sh b/2pdf.sh
+index 529d8a3..ad71911 100755
+--- a/2pdf.sh
++++ b/2pdf.sh
+@@ -13,3 +13,5 @@ INPUT=${1:-foo.html}
+13 OUTPUT=${INPUT%.html}.pdf                                              13 OUTPUT=${INPUT%.html}.pdf
+14                                                                        14 
+15 wkhtmltopdf --page-size A4 $INPUT $OUTPUT  # very very long comments > 15 wkhtmltopdf --page-size A4 $INPUT $OUTPUT  # very very long comments >
+                                                                          16 
+                                                                          17 # EOF
+diff --git a/example.pdf b/example.pdf
+index 1eacfd8..3696851 100644
+Binary files a/example.pdf and b/example.pdf differ
+diff --git a/foo.html b/foo.html
+index d2fd3fb..13afa6e 100644
+--- a/foo.html
++++ b/foo.html
+@@ -1,4 +1,4 @@
+1 <!doctype html>                                                        1 <!DOCTYPE html>
+2 <html>                                                                 2 <html>
+3                                                                        3 
+4 <head>                                                                 4 <head>
+diff --git a/foo.pdf b/foo.pdf
+index 0e90017..3c3b90d 100644
+Binary files a/foo.pdf and b/foo.pdf differ
+

tests/test_cdiff.py

@@ -397,6 +397,72 @@ def test_parse_only_in_dir_at_last(self):
         self.assertEqual(len(out[1]._hunks), 0)
         self.assertEqual(out[1]._headers, ['Only in foo: foo\n'])
 
+    def test_parse_binary_differ_diff_ru(self):
+        patch = r"""\
+--- a
++++ b
+@@ -1,2 +1,2 @@
+-foo
++bar
+ common
+Binary files a/1.pdf and b/1.pdf differ
+--- 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]._old_path, '')
+        self.assertEqual(out[1]._new_path, '')
+        self.assertEqual(len(out[1]._headers), 1)
+        self.assertTrue(out[1]._headers[0].startswith('Binary files'))
+        self.assertEqual(len(out[2]._hunks), 1)
+        self.assertEqual(len(out[2]._hunks[0]._hunk_list), 3)
+
+    def test_parse_binary_differ_git(self):
+        patch = r"""\
+diff --git a/foo b/foo
+index 529d8a3..ad71911 100755
+--- a/foo
++++ b/foo
+@@ -1,2 +1,2 @@
+-foo
++bar
+ common
+diff --git a/example.pdf b/example.pdf
+index 1eacfd8..3696851 100644
+Binary files a/example.pdf and b/example.pdf differ
+diff --git a/bar b/bar
+index 529e8a3..ad71921 100755
+--- a/bar
++++ b/bar
+@@ -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]._old_path, '')
+        self.assertEqual(out[1]._new_path, '')
+        self.assertEqual(len(out[1]._headers), 3)
+        self.assertTrue(out[1]._headers[2].startswith('Binary files'))
+        self.assertEqual(len(out[2]._hunks), 1)
+        self.assertEqual(len(out[2]._hunks[0]._hunk_list), 3)
+
     def test_parse_svn_prop(self):
         patch = r"""\
 --- a