update domain-specific rules to make flickr replay work better!

pywb/cdx/cdxdomainspecific.py

@@ -95,7 +95,11 @@ class FuzzyQuery:
         if not matched_rule:
             return None
 
-        inx = url.find('?')
+        repl = '?'
+        if matched_rule.replace:
+            repl = matched_rule.replace
+
+        inx = url.rfind(repl)
         if inx > 0:
             url = url[:inx + 1]
 
@@ -104,6 +108,8 @@ class FuzzyQuery:
                   'filter': filter_,
                   'output': output}
 
+        print params
+
         return CDXQuery(**params)
 
 

pywb/rules.yaml

@@ -27,13 +27,20 @@ rules:
               replace: '/* {0} */'
 
 
-    # yahoo rules
+    # flickr rules
     #=================================================================
     - url_prefix: ['com,yimg,l)/g/combo', 'com,yahooapis,yui)/combo']
 
       fuzzy_lookup: '([^/]+(?:\.css|\.js))'
 
 
+    - url_prefix: 'com,staticflickr,'
+
+      fuzzy_lookup:
+        match: '([0-9]+_[a-z0-9]+).*?.jpg'
+        replace: '/'
+
+
     # testing rules -- not for valid domain
     #=================================================================
     # this rule block is a non-existent prefix merely for testing