jsonp rewriter: expand jsonp matching: (#336)

- treat as jsonp if url query contains 'callback=jsonp',
- fuzzy match query containing 'callback=jsonp'
- tests: add test for additional jsonp matching

pywb/rewrite/content_rewriter.py

@@ -313,6 +313,11 @@ class StreamingRewriter(object):
 class RewriteInfo(object):
     TAG_REGEX = re.compile(b'^\s*\<')
 
+    JSONP_CONTAINS = ['callback=jQuery',
+                      'callback=jsonp',
+                      '.json?'
+                     ]
+
     def __init__(self, record, content_rewriter, url_rewriter, cookie_rewriter=None):
         self.record = record
 
@@ -347,12 +352,14 @@ class RewriteInfo(object):
         orig_text_type = self.rewrite_types.get(mime)
 
         text_type = self._resolve_text_type(orig_text_type)
+        url = self.url_rewriter.wburl.url
 
         if text_type in ('guess-text', 'guess-bin'):
             text_type = None
 
         if text_type == 'js':
-            if 'callback=jQuery' in self.url_rewriter.wburl.url or '.json?' in self.url_rewriter.wburl.url:
+            # determine if url contains strings that indicate jsonp
+            if any(jsonp_string in url for jsonp_string in self.JSONP_CONTAINS):
                 text_type = 'json'
 
         if (text_type and orig_text_type != text_type) or text_type == 'html':

pywb/rewrite/test/test_content_rewriter.py

@@ -250,6 +250,19 @@ class TestContentRewriter(object):
         exp = 'jQuery_DEF({"foo": "bar"});'
         assert b''.join(gen).decode('utf-8') == exp
 
+    def test_rewrite_js_as_json_generic_jsonp(self):
+        headers = {'Content-Type': 'application/json'}
+        content = '/**/ jsonpCallbackABCDEF({"foo": "bar"});'
+
+        headers, gen, is_rw = self.rewrite_record(headers, content, ts='201701js_',
+                                                  url='http://example.com/path/file?callback=jsonpCallback12345')
+
+        # content-type unchanged
+        assert ('Content-Type', 'application/json') in headers.headers
+
+        exp = 'jsonpCallback12345({"foo": "bar"});'
+        assert b''.join(gen).decode('utf-8') == exp
+
     def test_rewrite_js_not_json(self):
         # callback not set
         headers = {}

pywb/rules.yaml

@@ -34,6 +34,10 @@ default_filters:
         - match: '[?&]utm_[^=]+=[^&]+(?=&|$)'
           replace: ''
 
+        # additional callback=jsonpCallbackXYZ
+        - match: '[?&](callback=jsonp)[^&]+(?=&|$)'
+          replace: '\1'
+
         # remove jquery callback dynamic timestamp
         - match: '[?&]((?:\w+)=jquery)[\d]+_[\d]+'
           replace: '\1'