ClickHouse/tests/performance/ngram_distance.xml

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

    <substitutions>
        <substitution>
            <name>small_table</name>
                <values>
                    <!-- the same for now -->
                    <value>hits_10m_single</value>
                </values>
        </substitution>
    </substitutions>

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

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