From c10df57e07719ddb6521babdb9aef56d7182a0f5 Mon Sep 17 00:00:00 2001
From: Ilya Kreymer <ikreymer@gmail.com>
Date: Mon, 24 Nov 2014 11:10:49 -0800
Subject: [PATCH] rules: add support for customizing matchType prefix, adding
 multiple filters

---
 pywb/cdx/cdxdomainspecific.py | 21 ++++++++++++++++++---
 pywb/rules.yaml               | 11 ++++++++---
 2 files changed, 26 insertions(+), 6 deletions(-)

diff --git a/pywb/cdx/cdxdomainspecific.py b/pywb/cdx/cdxdomainspecific.py
index ce15e024..c7c05545 100644
--- a/pywb/cdx/cdxdomainspecific.py
+++ b/pywb/cdx/cdxdomainspecific.py
@@ -2,6 +2,7 @@ import yaml
 import re
 import logging
 import pkg_resources
+import urlparse
 
 from pywb.utils.dsrules import BaseRule, RuleSet
 
@@ -89,7 +90,8 @@ class FuzzyQuery:
 
             groups = m.groups()
             for g in groups:
-                filter_.append(rule.filter.format(g))
+                for f in matched_rule.filter:
+                    filter_.append(f.format(g))
 
             break
 
@@ -104,9 +106,15 @@ class FuzzyQuery:
         if inx > 0:
             url = url[:inx + 1]
 
+
+        if matched_rule.match_type == 'domain':
+            host = urlparse.urlsplit(url).netloc
+            # remove the subdomain
+            url = host.split('.', 1)[1]
+
         params = query.params
         params.update({'url': url,
-                       'matchType': 'prefix',
+                       'matchType': matched_rule.match_type,
                        'filter': filter_})
 
         if 'reverse' in params:
@@ -115,12 +123,17 @@ class FuzzyQuery:
         if 'closest' in params:
             del params['closest']
 
+        if 'end_key' in params:
+            del params['end_key']
+
+        print(params)
         return params
 
 
 #=================================================================
 class CDXDomainSpecificRule(BaseRule):
-    DEFAULT_FILTER = '~urlkey:{0}'
+    DEFAULT_FILTER = ['~urlkey:{0}']
+    DEFAULT_MATCH_TYPE = 'prefix'
 
     def __init__(self, name, config):
         super(CDXDomainSpecificRule, self).__init__(name, config)
@@ -129,10 +142,12 @@ class CDXDomainSpecificRule(BaseRule):
             self.regex = self.make_regex(config)
             self.replace = None
             self.filter = self.DEFAULT_FILTER
+            self.match_type = self.DEFAULT_MATCH_TYPE
         else:
             self.regex = self.make_regex(config.get('match'))
             self.replace = config.get('replace')
             self.filter = config.get('filter', self.DEFAULT_FILTER)
+            self.match_type = config.get('type', self.DEFAULT_MATCH_TYPE)
 
     def unsurt(self):
         """
diff --git a/pywb/rules.yaml b/pywb/rules.yaml
index 8cdcb528..68b2759d 100644
--- a/pywb/rules.yaml
+++ b/pywb/rules.yaml
@@ -142,8 +142,13 @@ rules:
 
     - url_prefix: 'com,googlevideo,'
 
-    #fuzzy_lookup: 'com,googlevideo.*/videoplayback?.*(id=[^&]).*(mime=[^&]).*(signature=[^&])'
-      fuzzy_lookup: 'com,googlevideo.*/videoplayback?.*(id=[^&]+).*(itag=[^&]+).*(mime=[^&]+)'
+      fuzzy_lookup:
+          match: 'com,googlevideo.*/videoplayback.*(id=[^&]+).*(itag=[^&]+).*(mime=[^&]+)'
+          filter:
+              - '~urlkey:{0}'
+              - '!mimetype:text/plain'
+
+          type: 'domain'
 
 
     # testing rules -- not for valid domain
@@ -169,5 +174,5 @@ rules:
     - url_prefix: ''
       fuzzy_lookup:
         match: '(.*)[&?](?:_|uncache)=[\d]+[&]?'
-        filter: '=urlkey:{0}'
+        filter: ['=urlkey:{0}']
         replace: '?'