Use JSON values in query string when canonicalizing JSON request bodies

pywb/warcserver/inputrequest.py

@@ -240,6 +240,8 @@ def handle_binary(query):
             query = '__wb_post_data=' + query
             return query
 
+        # TODO: Forms keeps JSON values but doesn't account for multiple values
+        # with same name
         if mime.startswith('application/x-www-form-urlencoded'):
             try:
                 query = to_native_str(query.decode('utf-8'))
@@ -270,9 +272,11 @@ def handle_binary(query):
                     values.append((item.name, item.value))
 
                 query = urlencode(values, True)
+                # TODO: Convert Pythonic values to JSON values
 
         elif mime.startswith('application/x-amf'):
             query = self.amf_parse(query, environ)
+            # TODO: Convert Pythonic values to JSON values?
 
         elif mime.startswith('application/json'):
             try:
@@ -328,7 +332,14 @@ def _parser(json_obj, name=""):
                     _parser(v, name)
 
             elif name:
-                data[get_key(name)] = str(json_obj)
+                if isinstance(json_obj, bool) and json_obj:
+                    data[get_key(name)] = "true"
+                elif isinstance(json_obj, bool):
+                    data[get_key(name)] = "false"
+                elif json_obj is None:
+                    data[get_key(name)] = "null"
+                else:
+                    data[get_key(name)] = str(json_obj)
 
         _parser(json.loads(string))
         return urlencode(data)

pywb/warcserver/test/test_inputreq.py

@@ -82,44 +82,44 @@ def test_post_req(self):
 class TestPostQueryExtract(object):
     @classmethod
     def setup_class(cls):
-        cls.post_data = b'foo=bar&dir=%2Fbaz'
+        cls.post_data = b'foo=bar&dir=%2Fbaz&do=true&re=false&re=null'
         cls.binary_post_data = b'\x816l`L\xa04P\x0e\xe0r\x02\xb5\x89\x19\x00fP\xdb\x0e\xb0\x02,'
 
     def test_post_extract_1(self):
         mq = MethodQueryCanonicalizer('POST', 'application/x-www-form-urlencoded',
                                 len(self.post_data), BytesIO(self.post_data))
 
-        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&foo=bar&dir=/baz'
+        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&foo=bar&dir=/baz&do=true&re=false&re=null'
 
-        assert mq.append_query('http://example.com/?123=ABC') == 'http://example.com/?123=ABC&__wb_method=POST&foo=bar&dir=/baz'
+        assert mq.append_query('http://example.com/?123=ABC') == 'http://example.com/?123=ABC&__wb_method=POST&foo=bar&dir=/baz&do=true&re=false&re=null'
 
     def test_post_extract_json(self):
-        post_data = b'{"a": "b", "c": {"a": 2}, "d": "e"}'
+        post_data = b'{"a": "b", "c": {"a": 2}, "d": "e", "f": true, "g": [false, null]}'
         mq = MethodQueryCanonicalizer('POST', 'application/json',
                                 len(post_data), BytesIO(post_data))
 
-        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&a=b&a.2_=2&d=e'
+        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&a=b&a.2_=2&d=e&f=true&g=false&g.2_=null'
 
 
     def test_put_extract_method(self):
         mq = MethodQueryCanonicalizer('PUT', 'application/x-www-form-urlencoded',
                                 len(self.post_data), BytesIO(self.post_data))
 
-        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=PUT&foo=bar&dir=/baz'
+        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=PUT&foo=bar&dir=/baz&do=true&re=false&re=null'
 
     def test_post_extract_non_form_data_1(self):
         mq = MethodQueryCanonicalizer('POST', 'application/octet-stream',
                                 len(self.post_data), BytesIO(self.post_data))
 
         #base64 encoded data
-        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&__wb_post_data=Zm9vPWJhciZkaXI9JTJGYmF6'
+        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&__wb_post_data=Zm9vPWJhciZkaXI9JTJGYmF6JmRvPXRydWUmcmU9ZmFsc2UmcmU9bnVsbA=='
 
     def test_post_extract_non_form_data_2(self):
         mq = MethodQueryCanonicalizer('POST', 'text/plain',
                                 len(self.post_data), BytesIO(self.post_data))
 
         #base64 encoded data
-        assert mq.append_query('http://example.com/pathbar?id=123') == 'http://example.com/pathbar?id=123&__wb_method=POST&__wb_post_data=Zm9vPWJhciZkaXI9JTJGYmF6'
+        assert mq.append_query('http://example.com/pathbar?id=123') == 'http://example.com/pathbar?id=123&__wb_method=POST&__wb_post_data=Zm9vPWJhciZkaXI9JTJGYmF6JmRvPXRydWUmcmU9ZmFsc2UmcmU9bnVsbA=='
 
     def test_post_extract_length_invalid_ignore(self):
         mq = MethodQueryCanonicalizer('POST', 'application/x-www-form-urlencoded',
@@ -136,13 +136,13 @@ def test_post_extract_length_too_short(self):
         mq = MethodQueryCanonicalizer('POST', 'application/x-www-form-urlencoded',
                                 len(self.post_data) - 4, BytesIO(self.post_data))
 
-        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&foo=bar&dir=%2'
+        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&foo=bar&dir=%2&do=true&re=false&re='
 
     def test_post_extract_length_too_long(self):
         mq = MethodQueryCanonicalizer('POST', 'application/x-www-form-urlencoded',
                                 len(self.post_data) + 4, BytesIO(self.post_data))
 
-        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&foo=bar&dir=/baz'
+        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&foo=bar&dir=/baz&do=true&re=false&re=null'
 
     def test_post_extract_malformed_form_data(self):
         mq = MethodQueryCanonicalizer('POST', 'application/x-www-form-urlencoded',
@@ -155,7 +155,7 @@ def test_post_extract_no_boundary_in_multipart_form_mimetype(self):
         mq = MethodQueryCanonicalizer('POST', 'multipart/form-data',
                                 len(self.post_data), BytesIO(self.post_data))
 
-        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&__wb_post_data=Zm9vPWJhciZkaXI9JTJGYmF6'
+        assert mq.append_query('http://example.com/') == 'http://example.com/?__wb_method=POST&__wb_post_data=Zm9vPWJhciZkaXI9JTJGYmF6JmRvPXRydWUmcmU9ZmFsc2UmcmU9bnVsbA=='
 
 
     def test_options(self):