keep newlines and replace comments with white space (#42)

* keep newlines and replace comments with white space

If strip_space is False, keep newlines and replace comments with white space so that the JSON parser reports the correct line and column numbers on parsing errors.

This fixes issue #5 for the python version.

* incorporate _space_template() into the main template()

remove the tiny (redundant) white space tests.
Check that the JSON with white space preserved can be further minimized and produces the same result when fed to the JSON parser.

Author: PoByBolek

json_minify/__init__.py

@@ -35,6 +35,10 @@ def json_minify(string, strip_space=True):
                 # replace white space as defined in standard
                 tmp = re.sub('[ \t\n\r]+', '', tmp)
             new_str.append(tmp)
+        elif not strip_space:
+            # Replace comments with white space so that the JSON parser reports
+            # the correct column numbers on parsing errors.
+            new_str.append(' ' * (match.start() - index))
 
         index = match.end()
         val = match.group()
@@ -53,10 +57,18 @@ def json_minify(string, strip_space=True):
                 in_single = True
         elif val == '*/' and in_multi and not (in_string or in_single):
             in_multi = False
+            if not strip_space:
+                new_str.append(' ' * len(val))
         elif val in '\r\n' and not (in_multi or in_string) and in_single:
             in_single = False
         elif not ((in_multi or in_single) or (val in ' \r\n\t' and strip_space)):  # noqa
             new_str.append(val)
 
+        if not strip_space:
+            if val in '\r\n':
+                new_str.append(val)
+            elif in_multi or in_single:
+                new_str.append(' ' * len(val))
+
     new_str.append(string[index:])
     return ''.join(new_str)

json_minify/test_json_minify.py

@@ -31,12 +31,32 @@ class JsonMinifyTestCase(unittest.TestCase):
                from such a file with the JSON.minify() project hosted
                here on github at http://github.com/getify/JSON.minify
             */''',
+            '''
+                                               \x20
+            {
+                "foo": "bar",                  \x20
+                "bar": [
+                    "baz", "bum"
+                ],
+                                                     \x20
+                                            \x20
+                "something": 10,
+                "else": 20
+            }
+
+                                                                    \x20
+                                                                    \x20
+                                                                    \x20
+              ''',
             '{"foo":"bar","bar":["baz","bum"],"something":10,"else":20}'
         ],
         [
             '''
             {"/*":"*/","//":"",/*"//"*/"/*/"://
             "//"}''',
+            '''
+            {"/*":"*/","//":"",        "/*/": \x20
+            "//"}''',
             '{"/*":"*/","//":"","/*/":"//"}'
         ],
         [
@@ -51,6 +71,19 @@ class JsonMinifyTestCase(unittest.TestCase):
                 ,    "b\"az":/*
             something else */"blah"
 
+            }
+            ''',
+            '''
+             \x20
+                    \x20
+                                 {
+
+            "foo"
+            :
+                "bar/*"           \x20
+                ,    "b\\\"az": \x20
+                             "blah"
+
             }
             ''',
             r'{"foo":"bar/*","b\"az":"blah"}'
@@ -61,16 +94,30 @@ class JsonMinifyTestCase(unittest.TestCase):
             "baz\\\\": /* yay */ "fo\\\\\"*/o"
             }
             ''',
+            r'''
+            {"foo": "ba\"r//", "bar\\": "b\\\"a/*z",
+            "baz\\\\":           "fo\\\\\"*/o"
+            }
+            ''',
             r'{"foo":"ba\"r//","bar\\":"b\\\"a/*z","baz\\\\":"fo\\\\\"*/o"}'  # noqa
         ]
     ]
 
     def template(self, index):
-        in_string, expected = self.tests[index - 1]
+        in_string, expected_whitespace, expected = self.tests[index - 1]
         in_dict = json.loads(json_minify(in_string))
         expected_dict = json.loads(textwrap.dedent(expected))
         self.assertEqual(in_dict, expected_dict)
 
+        with_whitespace = json_minify(in_string, strip_space=False)
+        with_whitespace_dict = json.loads(with_whitespace)
+        self.assertEqual(expected_whitespace, with_whitespace)
+        self.assertEqual(with_whitespace_dict, expected_dict)
+
+        no_whitespace = json_minify(with_whitespace)
+        no_whitespace_dict = json.loads(no_whitespace)
+        self.assertEqual(no_whitespace_dict, expected_dict)
+
     def test_1(self):
         self.template(1)
 
@@ -83,5 +130,6 @@ def test_3(self):
     def test_4(self):
         self.template(4)
 
+
 if __name__ == '__main__':
     unittest.main()