mirror of
https://github.com/GreptimeTeam/greptimedb.git
synced 2026-01-13 16:52:56 +00:00
dd63068df6
* feat: add `matches_term` function Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * merge & fix Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * address comments Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * fix & skip char after boundary mismatch Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> --------- Signed-off-by: Zhenchi <zhongzc_arch@outlook.com>
315 lines
9.3 KiB
Plaintext
315 lines
9.3 KiB
Plaintext
-- Test basic term matching
|
|
-- Expect: true
|
|
SELECT matches_term('cat!', 'cat') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| true |
|
|
+--------+
|
|
|
|
-- Test phrase matching with spaces
|
|
-- Expect: true
|
|
SELECT matches_term('warning:hello world!', 'hello world') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| true |
|
|
+--------+
|
|
|
|
-- Test numbers in term
|
|
SELECT matches_term('v1.0!', 'v1.0') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| true |
|
|
+--------+
|
|
|
|
-- Test case sensitivity
|
|
-- Expect: true
|
|
SELECT matches_term('Cat', 'Cat') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| true |
|
|
+--------+
|
|
|
|
-- Expect: false
|
|
SELECT matches_term('cat', 'Cat') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| false |
|
|
+--------+
|
|
|
|
-- Test empty string handling
|
|
-- Expect: true
|
|
SELECT matches_term('', '') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| true |
|
|
+--------+
|
|
|
|
-- Expect: false
|
|
SELECT matches_term('any', '') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| false |
|
|
+--------+
|
|
|
|
-- Expect: false
|
|
SELECT matches_term('', 'any') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| false |
|
|
+--------+
|
|
|
|
-- Test partial matches (should fail)
|
|
-- Expect: false
|
|
SELECT matches_term('category', 'cat') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| false |
|
|
+--------+
|
|
|
|
-- Expect: false
|
|
SELECT matches_term('rebooted', 'boot') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| false |
|
|
+--------+
|
|
|
|
-- Test adjacent alphanumeric characters
|
|
SELECT matches_term('cat5', 'cat') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| false |
|
|
+--------+
|
|
|
|
SELECT matches_term('dogcat', 'dog') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| false |
|
|
+--------+
|
|
|
|
-- Test leading non-alphanumeric
|
|
-- Expect: true
|
|
SELECT matches_term('dog/cat', '/cat') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| true |
|
|
+--------+
|
|
|
|
-- Expect: true
|
|
SELECT matches_term('dog/cat', 'dog/') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| true |
|
|
+--------+
|
|
|
|
-- Expect: true
|
|
SELECT matches_term('dog/cat', 'dog/cat') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| true |
|
|
+--------+
|
|
|
|
-- Test unicode characters
|
|
-- Expect: true
|
|
SELECT matches_term('café>', 'café') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| true |
|
|
+--------+
|
|
|
|
-- Expect: true
|
|
SELECT matches_term('русский!', 'русский') as result;
|
|
|
|
+--------+
|
|
| result |
|
|
+--------+
|
|
| true |
|
|
+--------+
|
|
|
|
-- Test complete word matching
|
|
CREATE TABLE logs (
|
|
`id` TIMESTAMP TIME INDEX,
|
|
`log_message` STRING
|
|
);
|
|
|
|
Affected Rows: 0
|
|
|
|
INSERT INTO logs VALUES
|
|
(1, 'An error occurred!'),
|
|
(2, 'Critical error: system failure'),
|
|
(3, 'error-prone'),
|
|
(4, 'errors'),
|
|
(5, 'error123'),
|
|
(6, 'errorLogs'),
|
|
(7, 'Version v1.0 released'),
|
|
(8, 'v1.0!'),
|
|
(9, 'v1.0a'),
|
|
(10, 'v1.0beta'),
|
|
(11, 'GET /app/start'),
|
|
(12, 'Command: /start-prosess'),
|
|
(13, 'Command: /start'),
|
|
(14, 'start'),
|
|
(15, 'start/stop'),
|
|
(16, 'Alert: system failure detected'),
|
|
(17, 'system failure!'),
|
|
(18, 'system-failure'),
|
|
(19, 'system failure2023'),
|
|
(20, 'critical error: system failure'),
|
|
(21, 'critical failure detected'),
|
|
(22, 'critical issue'),
|
|
(23, 'failure imminent'),
|
|
(24, 'Warning: high temperature'),
|
|
(25, 'WARNING: system overload'),
|
|
(26, 'warned'),
|
|
(27, 'warnings');
|
|
|
|
Affected Rows: 27
|
|
|
|
-- Test complete word matching for 'error'
|
|
-- Expect:
|
|
-- 1|An error occurred!|true
|
|
-- 2|Critical error: system failure|true
|
|
-- 3|error-prone|true
|
|
-- 4|errors|false
|
|
-- 5|error123|false
|
|
-- 6|errorLogs|false
|
|
SELECT `id`, `log_message`, matches_term(`log_message`, 'error') as `matches_error` FROM logs WHERE `id` <= 6 ORDER BY `id`;
|
|
|
|
+-------------------------+--------------------------------+---------------+
|
|
| id | log_message | matches_error |
|
|
+-------------------------+--------------------------------+---------------+
|
|
| 1970-01-01T00:00:00.001 | An error occurred! | true |
|
|
| 1970-01-01T00:00:00.002 | Critical error: system failure | true |
|
|
| 1970-01-01T00:00:00.003 | error-prone | true |
|
|
| 1970-01-01T00:00:00.004 | errors | false |
|
|
| 1970-01-01T00:00:00.005 | error123 | false |
|
|
| 1970-01-01T00:00:00.006 | errorLogs | false |
|
|
+-------------------------+--------------------------------+---------------+
|
|
|
|
-- Test complete word matching for 'v1.0'
|
|
-- Expect:
|
|
-- 7|Version v1.0 released|true
|
|
-- 8|v1.0!|true
|
|
-- 9|v1.0a|false
|
|
-- 10|v1.0beta|false
|
|
SELECT `id`, `log_message`, matches_term(`log_message`, 'v1.0') as `matches_version` FROM logs WHERE `id` BETWEEN 7 AND 10 ORDER BY `id`;
|
|
|
|
+-------------------------+-----------------------+-----------------+
|
|
| id | log_message | matches_version |
|
|
+-------------------------+-----------------------+-----------------+
|
|
| 1970-01-01T00:00:00.007 | Version v1.0 released | true |
|
|
| 1970-01-01T00:00:00.008 | v1.0! | true |
|
|
| 1970-01-01T00:00:00.009 | v1.0a | false |
|
|
| 1970-01-01T00:00:00.010 | v1.0beta | false |
|
|
+-------------------------+-----------------------+-----------------+
|
|
|
|
-- Test complete word matching for '/start'
|
|
-- Expect:
|
|
-- 11|GET /app/start|true
|
|
-- 12|Command: /start-prosess|true
|
|
-- 13|Command: /start|true
|
|
-- 14|start|false
|
|
-- 15|start/stop|false
|
|
SELECT `id`, `log_message`, matches_term(`log_message`, '/start') as `matches_start` FROM logs WHERE `id` BETWEEN 11 AND 15 ORDER BY `id`;
|
|
|
|
+-------------------------+-------------------------+---------------+
|
|
| id | log_message | matches_start |
|
|
+-------------------------+-------------------------+---------------+
|
|
| 1970-01-01T00:00:00.011 | GET /app/start | true |
|
|
| 1970-01-01T00:00:00.012 | Command: /start-prosess | true |
|
|
| 1970-01-01T00:00:00.013 | Command: /start | true |
|
|
| 1970-01-01T00:00:00.014 | start | false |
|
|
| 1970-01-01T00:00:00.015 | start/stop | false |
|
|
+-------------------------+-------------------------+---------------+
|
|
|
|
-- Test phrase matching for 'system failure'
|
|
-- Expect:
|
|
-- 16|Alert: system failure detected|true
|
|
-- 17|system failure!|true
|
|
-- 18|system-failure|false
|
|
-- 19|system failure2023|false
|
|
SELECT `id`, `log_message`, matches_term(`log_message`, 'system failure') as `matches_phrase` FROM logs WHERE `id` BETWEEN 16 AND 19 ORDER BY `id`;
|
|
|
|
+-------------------------+--------------------------------+----------------+
|
|
| id | log_message | matches_phrase |
|
|
+-------------------------+--------------------------------+----------------+
|
|
| 1970-01-01T00:00:00.016 | Alert: system failure detected | true |
|
|
| 1970-01-01T00:00:00.017 | system failure! | true |
|
|
| 1970-01-01T00:00:00.018 | system-failure | false |
|
|
| 1970-01-01T00:00:00.019 | system failure2023 | false |
|
|
+-------------------------+--------------------------------+----------------+
|
|
|
|
-- Test multi-word matching using AND
|
|
-- Expect:
|
|
-- 20|critical error: system failure|true|true|true
|
|
-- 21|critical failure detected|true|true|true
|
|
-- 22|critical issue|true|false|false
|
|
-- 23|failure imminent|false|true|false
|
|
SELECT `id`, `log_message`,
|
|
matches_term(`log_message`, 'critical') as `matches_critical`,
|
|
matches_term(`log_message`, 'failure') as `matches_failure`,
|
|
matches_term(`log_message`, 'critical') AND matches_term(`log_message`, 'failure') as `matches_both`
|
|
FROM logs WHERE `id` BETWEEN 20 AND 23 ORDER BY `id`;
|
|
|
|
+-------------------------+--------------------------------+------------------+-----------------+--------------+
|
|
| id | log_message | matches_critical | matches_failure | matches_both |
|
|
+-------------------------+--------------------------------+------------------+-----------------+--------------+
|
|
| 1970-01-01T00:00:00.020 | critical error: system failure | true | true | true |
|
|
| 1970-01-01T00:00:00.021 | critical failure detected | true | true | true |
|
|
| 1970-01-01T00:00:00.022 | critical issue | true | false | false |
|
|
| 1970-01-01T00:00:00.023 | failure imminent | false | true | false |
|
|
+-------------------------+--------------------------------+------------------+-----------------+--------------+
|
|
|
|
-- Test case-insensitive matching using lower()
|
|
-- Expect:
|
|
-- 24|Warning: high temperature|true
|
|
-- 25|WARNING: system overload|true
|
|
-- 26|warned|false
|
|
-- 27|warnings|false
|
|
SELECT `id`, `log_message`, matches_term(lower(`log_message`), 'warning') as `matches_warning` FROM logs WHERE `id` >= 24 ORDER BY `id`;
|
|
|
|
+-------------------------+---------------------------+-----------------+
|
|
| id | log_message | matches_warning |
|
|
+-------------------------+---------------------------+-----------------+
|
|
| 1970-01-01T00:00:00.024 | Warning: high temperature | true |
|
|
| 1970-01-01T00:00:00.025 | WARNING: system overload | true |
|
|
| 1970-01-01T00:00:00.026 | warned | false |
|
|
| 1970-01-01T00:00:00.027 | warnings | false |
|
|
+-------------------------+---------------------------+-----------------+
|
|
|
|
DROP TABLE logs;
|
|
|
|
Affected Rows: 0
|
|
|