diff --git a/tests/performance/nlp.xml b/tests/performance/nlp.xml new file mode 100644 index 00000000000..adc889b10ca --- /dev/null +++ b/tests/performance/nlp.xml @@ -0,0 +1,16 @@ + + + hits_100m_single + + + CREATE TABLE hits_100m_words (words Array(String), UserID UInt64) ENGINE Memory + CREATE TABLE hits_100m_words_ws (words Array(String), UserID UInt64) ENGINE Memory + + INSERT INTO hits_100m_words SELECT splitByNonAlpha(SearchPhrase) AS words, UserID FROM hits_100m_single WHERE length(words) > 0 + INSERT INTO hits_100m_words_ws SELECT splitByWhitespace(SearchPhrase) AS words, UserID FROM hits_100m_single WHERE length(words) > 0 + + SELECT arrayMap(x -> stem('ru', x), words) FROM hits_100m_words FORMAT Null + + DROP TABLE IF EXISTS hits_100m_words + DROP TABLE IF EXISTS hits_100m_words_ws + \ No newline at end of file