mirror of
https://github.com/hanzoai/datastore.git
synced 2026-09-26 13:55:07 +00:00
Elmi Ahmadov
164952e52d
Currently, `hasPhrase` does not utilize the text index in any way. This PR adds the missing support for it. However it does not store the positional information yet, intead the `HINT` mode would be used. The query internally would be converted from: ```sql WHERE hasPhrase(text_column, 'token1 token2') ``` to: ```sql WHERE hasAllTokens(text_column, 'token1 token2) AND hasPhrase(text_column, 'token1 token2') ```. The added `hasAllTokens` condition would filter out rows do not contains all tokens. The proper positional information for the phrase search would be added in the follow-up PRs.
60 lines
4.1 KiB
XML
60 lines
4.1 KiB
XML
<test>
|
|
<settings>
|
|
<use_skip_indexes_on_data_read>1</use_skip_indexes_on_data_read>
|
|
<query_plan_direct_read_from_text_index>1</query_plan_direct_read_from_text_index>
|
|
<force_data_skipping_indices>idx_sp</force_data_skipping_indices>
|
|
</settings>
|
|
|
|
<fill_query>ALTER TABLE hits_10m_single ADD INDEX idx_sp(SearchPhrase) TYPE text(tokenizer = 'splitByNonAlpha')</fill_query>
|
|
<fill_query>ALTER TABLE hits_10m_single MATERIALIZE INDEX idx_sp</fill_query>
|
|
|
|
<!-- The terms are ordered in the decreasing frequency -->
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasToken(SearchPhrase, 'video')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasToken(SearchPhrase, 'yandex')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasToken(SearchPhrase, 'google')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasToken(SearchPhrase, 'world')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasToken(SearchPhrase, 'market')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasToken(SearchPhrase, 'mail')</query>
|
|
|
|
<query>SELECT count() FROM hits_10m_single WHERE NOT hasToken(SearchPhrase, 'mail')</query>
|
|
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasToken(SearchPhrase, 'mail') OR hasToken(SearchPhrase, 'market')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasToken(SearchPhrase, 'mail') AND hasToken(SearchPhrase, 'market')</query>
|
|
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasToken(SearchPhrase, 'acompletelynonexistentwordxyz')</query>
|
|
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasAllTokens(SearchPhrase, ['the'])</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasAllTokens(SearchPhrase, ['the', 'video'])</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasAllTokens(SearchPhrase, ['the', 'video', 'world', 'market'])</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE NOT hasAllTokens(SearchPhrase, ['the', 'video', 'world', 'market'])</query>
|
|
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasAnyTokens(SearchPhrase, ['mail'])</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasAnyTokens(SearchPhrase, ['the', 'video', 'world', 'market'])</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE NOT hasAnyTokens(SearchPhrase, ['the', 'video', 'world', 'market'])</query>
|
|
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasToken(SearchPhrase, 'video') AND WatchID >= 10 </query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasAllTokens(SearchPhrase, ['video']) AND WatchID >= 10 </query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasAnyTokens(SearchPhrase, ['video']) AND WatchID >= 10 </query>
|
|
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasToken(SearchPhrase, '')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasAnyTokens(SearchPhrase, [''])</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasAllTokens(SearchPhrase, [''])</query>
|
|
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasAnyTokens(SearchPhrase, [])</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasAllTokens(SearchPhrase, [])</query>
|
|
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasPhrase(SearchPhrase, '2014 фото')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasPhrase(SearchPhrase, '2014 года')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasPhrase(SearchPhrase, 'ru -> eng')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasPhrase(SearchPhrase, 'google.com')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasPhrase(SearchPhrase, 'yandex.ru')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasPhrase(SearchPhrase, 'video world')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasPhrase(SearchPhrase, 'puma oxygen')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasPhrase(SearchPhrase, 'master video')</query>
|
|
<query>SELECT count() FROM hits_10m_single WHERE hasPhrase(SearchPhrase, 'grand world')</query>
|
|
|
|
<query>SELECT token, cardinality FROM mergeTreeTextIndex(currentDatabase(), hits_10m_single, idx_sp) FORMAT Null</query>
|
|
|
|
<drop_query>ALTER TABLE hits_10m_single DROP INDEX idx_sp</drop_query>
|
|
</test>
|