From f851d4b47368bfb83caa5085938020a6e75ff85f Mon Sep 17 00:00:00 2001 From: Ilya Kreymer Date: Fri, 13 Oct 2017 11:37:03 -0700 Subject: [PATCH] fuzzymatcher: fix fuzzymatcher to remove '~' from prefix match, per changes from #250 --- pywb/warcserver/index/fuzzymatcher.py | 2 +- pywb/warcserver/index/test/test_fuzzymatcher.py | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/pywb/warcserver/index/fuzzymatcher.py b/pywb/warcserver/index/fuzzymatcher.py index 8c0827bd..a38ca170 100644 --- a/pywb/warcserver/index/fuzzymatcher.py +++ b/pywb/warcserver/index/fuzzymatcher.py @@ -17,7 +17,7 @@ FuzzyRule = namedtuple('FuzzyRule', # ============================================================================ class FuzzyMatcher(object): - DEFAULT_FILTER = ['~urlkey:{0}'] + DEFAULT_FILTER = ['urlkey:{0}'] DEFAULT_MATCH_TYPE = 'prefix' DEFAULT_REPLACE_AFTER = '?' diff --git a/pywb/warcserver/index/test/test_fuzzymatcher.py b/pywb/warcserver/index/test/test_fuzzymatcher.py index 66e3e167..f1285692 100644 --- a/pywb/warcserver/index/test/test_fuzzymatcher.py +++ b/pywb/warcserver/index/test/test_fuzzymatcher.py @@ -36,7 +36,7 @@ class TestFuzzy(object): return params def get_expected(self, url, mime='text/html', filters=None): - filters = filters or ['~urlkey:'] + filters = filters or ['urlkey:'] exp = [{'filter': filters, 'is_fuzzy': True, 'urlkey': canonicalize(url), @@ -107,7 +107,7 @@ class TestFuzzy(object): actual_url = 'http://youtube.com/get_video_info?a=d&html5=true&___abc=125&video_id=ABCD&id=1234' params = self.get_params(url, actual_url) cdx_iter, errs = self.fuzzy(self.source, params) - filters = ['~urlkey:html5=true', '~urlkey:video_id=abcd'] + filters = ['urlkey:html5=true', 'urlkey:video_id=abcd'] assert list(cdx_iter) == self.get_expected(url=actual_url, filters=filters) def test_no_fuzzy_custom_rule_video_id_diff(self):