ClickHouse/dbms/tests/performance/ngram_distance.xml

<test>

    <tags>
        <tag>search</tag>
    </tags>

    <preconditions>
        <table_exists>hits_10m_single</table_exists>
    </preconditions>

    <type>loop</type>

    <settings>
        <max_memory_usage>20000000000</max_memory_usage>
    </settings>

    <stop_conditions>
        <all_of>
            <iterations>5</iterations>
            <min_time_not_changing_for_ms>10000</min_time_not_changing_for_ms>
        </all_of>
        <any_of>
            <iterations>50</iterations>
            <total_time_ms>60000</total_time_ms>
        </any_of>
    </stop_conditions>

    <substitutions>
        <substitution>
           <name>small_table</name>
           <values>
               <value>(select * from hits_10m_single limit 2000000)</value>
           </values>
       </substitution>
    </substitutions>


    <query>SELECT DISTINCT URL,Title, ngramDistance(Title, URL) AS distance FROM {small_table} FORMAT Null</query>
    <query>SELECT DISTINCT SearchPhrase,Title, ngramDistance(Title, SearchPhrase) AS distance FROM {small_table} FORMAT Null</query>
    <query>SELECT DISTINCT Title, ngramDistance(Title, 'what is love') AS distance FROM hits_10m_single FORMAT Null</query>
    <query>SELECT DISTINCT Title, ngramDistance(Title, 'baby dont hurt me') AS distance FROM hits_10m_single FORMAT Null</query>
    <query>SELECT DISTINCT Title, ngramDistance(Title, 'no more') AS distance FROM hits_10m_single FORMAT Null</query>
    <query>SELECT DISTINCT Title, ngramDistanceCaseInsensitive(Title, 'wHAt Is lovE') AS distance FROM hits_10m_single FORMAT Null</query>
    <query>SELECT DISTINCT Title, ngramDistanceCaseInsensitive(Title, 'BABY DonT hUrT me') AS distance FROM hits_10m_single FORMAT Null</query>
    <query>SELECT DISTINCT Title, ngramDistanceCaseInsensitive(Title, 'nO MOrE') AS distance FROM hits_10m_single FORMAT Null</query>
    <query>SELECT DISTINCT URL,Title, ngramDistanceUTF8(Title, URL) AS distance FROM {small_table} FORMAT Null</query>
    <query>SELECT DISTINCT SearchPhrase,Title, ngramDistanceUTF8(Title, SearchPhrase) AS distance FROM {small_table} FORMAT Null</query>
    <query>SELECT DISTINCT Title, ngramDistanceUTF8(Title, 'метрика') AS distance FROM hits_10m_single FORMAT Null</query>
    <query>SELECT DISTINCT URL, ngramDistanceUTF8(URL, 'как дела') AS distance FROM {small_table} FORMAT Null</query>
    <query>SELECT DISTINCT URL, ngramDistanceUTF8(URL, 'чем занимаешься') AS distance FROM {small_table} FORMAT Null</query>

    <query>SELECT DISTINCT Title, ngramDistanceCaseInsensitiveUTF8(Title, 'Метрика') AS distance FROM hits_10m_single FORMAT Null</query>
    <query>SELECT DISTINCT URL, ngramDistanceCaseInsensitiveUTF8(URL, 'как дЕлА') AS distance FROM {small_table} FORMAT Null</query>
    <query>SELECT DISTINCT URL, ngramDistanceCaseInsensitiveUTF8(URL, 'Чем зАнимаешЬся') AS distance FROM {small_table} FORMAT Null</query>

</test>