ClickHouse/tests/performance/url_hits.xml

<test>
    <preconditions>
        <table_exists>hits_100m_single</table_exists>
        <table_exists>hits_10m_single</table_exists>
        <table_exists>test.hits</table_exists>
    </preconditions>


    <substitutions>
        <substitution>
           <name>func_slow</name>
           <values>
               <value>URLHierarchy</value>
               <value>URLPathHierarchy</value>
           </values>
       </substitution>

        <substitution>
           <name>func_fast</name>
           <values>
               <value>protocol</value>
               <value>domain</value>
               <value>domainWithoutWWW</value>
               <value>topLevelDomain</value>
               <value>firstSignificantSubdomain</value>
               <value>cutToFirstSignificantSubdomain</value>
               <value>path</value>
               <value>pathFull</value>
               <value>queryString</value>
               <value>fragment</value>
               <value>queryStringAndFragment</value>
               <value>extractURLParameters</value>
               <value>extractURLParameterNames</value>
               <value>decodeURLComponent</value>
               <value>cutWWW</value>
               <value>cutQueryString</value>
               <value>cutQueryStringAndFragment</value>
           </values>
       </substitution>
    </substitutions>

    <query>SELECT count() FROM hits_100m_single WHERE NOT ignore({func_fast}(URL))</query>
    <query>SELECT count() FROM hits_10m_single  WHERE NOT ignore({func_slow}(URL))</query>

    <!-- firstSignificantSubdomain/firstSignificantSubdomainCustom -->
    <query>SELECT count() FROM test.hits WHERE NOT ignore(firstSignificantSubdomain(URL)) SETTINGS max_threads=1</query>
    <query>SELECT count() FROM test.hits WHERE NOT ignore(firstSignificantSubdomainCustom(URL, 'public_suffix_list')) SETTINGS max_threads=1</query>
    <!-- cutToFirstSignificantSubdomain/cutToFirstSignificantSubdomainCustom -->
    <query>SELECT count() FROM test.hits WHERE NOT ignore(cutToFirstSignificantSubdomain(URL)) SETTINGS max_threads=1</query>
    <query>SELECT count() FROM test.hits WHERE NOT ignore(cutToFirstSignificantSubdomainCustom(URL, 'public_suffix_list')) SETTINGS max_threads=1</query>
</test>