html_rewriter: fixed attribute 'srcset' rewriting (webrecorder#712)

Co-authored-by: Yasar Kunduz <[email protected]>

pywb/rewrite/html_rewriter.py

@@ -177,7 +177,7 @@ def _rewrite_srcset(self, value, mod=''):
             return ''
 
         values = (url.strip() for url in re.split(self.SRCSET_REGEX, value) if url)
-        values = [self._rewrite_url(v.strip()) for v in values]
+        values = [self._rewrite_url(v.split(' ')[0].strip()) + (' ' + ' '.join(v.split(' ')[1:])).rstrip() for v in values if v]
         return ', '.join(values)
 
     def _rewrite_meta_refresh(self, meta_refresh):

pywb/rewrite/test/test_html_rewriter.py

@@ -185,6 +185,10 @@
 >>> parse('<img srcset="//example.com/1x,1x 2w, //example1.com/foo 2x, http://example.com/bar,bar 4x">')
 <img srcset="/web/20131226101010///example.com/1x,1x 2w, /web/20131226101010///example1.com/foo 2x, /web/20131226101010/http://example.com/bar,bar 4x">
 
+# complex srcset attrib
+>>> parse('<img srcset="http://test.com/yaşar-kunduz.jpg 320w, http://test.com/yaşar-konçalves-273x300.jpg 273w">')
+<img srcset="/web/20131226101010/http://test.com/ya%C5%9Far-kunduz.jpg 320w, /web/20131226101010/http://test.com/ya%C5%9Far-konc%CC%A7alves-273x300.jpg 273w">
+
 # empty srcset attrib
 >>> parse('<img srcset="">')
 <img srcset="">