|
| 1 | +# Time: O(n * l^2 + (n * l) * log(n * l)) |
| 2 | +# Space: O(n * l^2) |
| 3 | + |
| 4 | +# recursive cte |
| 5 | +WITH RECURSIVE splitted_word_cte AS ( |
| 6 | + SELECT content_id, |
| 7 | + content_text, |
| 8 | + '^' AS delim, |
| 9 | + CASE |
| 10 | + WHEN INSTR(content_text, ' ') = 0 AND INSTR(content_text, '-') = 0 THEN |
| 11 | + content_text |
| 12 | + WHEN INSTR(content_text, ' ') != 0 AND INSTR(content_text, '-') = 0 THEN |
| 13 | + SUBSTRING(content_text, 1, INSTR(content_text, ' ') - 1) |
| 14 | + WHEN INSTR(content_text, ' ') = 0 AND INSTR(content_text, '-') != 0 THEN |
| 15 | + SUBSTRING(content_text, 1, INSTR(content_text, '-') - 1) |
| 16 | + WHEN INSTR(content_text, ' ') != 0 AND INSTR(content_text, '-') != 0 THEN |
| 17 | + SUBSTRING(content_text, 1, INSTR(content_text, IF(INSTR(content_text, ' ') < INSTR(content_text, '-'), ' ', '-')) - 1) |
| 18 | + END AS word, |
| 19 | + CASE |
| 20 | + WHEN INSTR(content_text, ' ') = 0 AND INSTR(content_text, '-') = 0 THEN |
| 21 | + '' |
| 22 | + WHEN INSTR(content_text, ' ') != 0 AND INSTR(content_text, '-') = 0 THEN |
| 23 | + SUBSTRING(content_text, INSTR(content_text, ' '), 255) |
| 24 | + WHEN INSTR(content_text, ' ') = 0 AND INSTR(content_text, '-') != 0 THEN |
| 25 | + SUBSTRING(content_text, INSTR(content_text, '-'), 255) |
| 26 | + WHEN INSTR(content_text, ' ') != 0 AND INSTR(content_text, '-') != 0 THEN |
| 27 | + SUBSTRING(content_text, INSTR(content_text, IF(INSTR(content_text, ' ') < INSTR(content_text, '-'), ' ', '-')), 255) |
| 28 | + END AS remain, |
| 29 | + 1 AS idx |
| 30 | + FROM user_content |
| 31 | + UNION ALL |
| 32 | + SELECT |
| 33 | + content_id, |
| 34 | + content_text, |
| 35 | + SUBSTR(remain, 1, 1) AS delim, |
| 36 | + CASE |
| 37 | + WHEN INSTR(SUBSTRING(remain, 2, 255), ' ') = 0 AND INSTR(SUBSTRING(remain, 2, 255), '-') = 0 THEN |
| 38 | + SUBSTRING(remain, 2, 255) |
| 39 | + WHEN INSTR(SUBSTRING(remain, 2, 255), ' ') != 0 AND INSTR(SUBSTRING(remain, 2, 255), '-') = 0 THEN |
| 40 | + SUBSTRING(SUBSTRING(remain, 2, 255), 1, INSTR(SUBSTRING(remain, 2, 255), ' ') - 1) |
| 41 | + WHEN INSTR(SUBSTRING(remain, 2, 255), ' ') = 0 AND INSTR(SUBSTRING(remain, 2, 255), '-') != 0 THEN |
| 42 | + SUBSTRING(SUBSTRING(remain, 2, 255), 1, INSTR(SUBSTRING(remain, 2, 255), '-') - 1) |
| 43 | + WHEN INSTR(SUBSTRING(remain, 2, 255), ' ') != 0 AND INSTR(SUBSTRING(remain, 2, 255), '-') != 0 THEN |
| 44 | + SUBSTRING(SUBSTRING(remain, 2, 255), 1, INSTR(SUBSTRING(remain, 2, 255), IF(INSTR(SUBSTRING(remain, 2, 255), ' ') < INSTR(SUBSTRING(remain, 2, 255), '-'), ' ', '-')) - 1) |
| 45 | + END AS word, |
| 46 | + CASE |
| 47 | + WHEN INSTR(SUBSTRING(remain, 2, 255), ' ') = 0 AND INSTR(SUBSTRING(remain, 2, 255), '-') = 0 THEN |
| 48 | + '' |
| 49 | + WHEN INSTR(SUBSTRING(remain, 2, 255), ' ') != 0 AND INSTR(SUBSTRING(remain, 2, 255), '-') = 0 THEN |
| 50 | + SUBSTRING(SUBSTRING(remain, 2, 255), INSTR(SUBSTRING(remain, 2, 255), ' '), 255) |
| 51 | + WHEN INSTR(SUBSTRING(remain, 2, 255), ' ') = 0 AND INSTR(SUBSTRING(remain, 2, 255), '-') != 0 THEN |
| 52 | + SUBSTRING(SUBSTRING(remain, 2, 255), INSTR(SUBSTRING(remain, 2, 255), '-'), 255) |
| 53 | + WHEN INSTR(SUBSTRING(remain, 2, 255), ' ') != 0 AND INSTR(SUBSTRING(remain, 2, 255), '-') != 0 THEN |
| 54 | + SUBSTRING(SUBSTRING(remain, 2, 255), INSTR(SUBSTRING(remain, 2, 255), IF(INSTR(SUBSTRING(remain, 2, 255), ' ') < INSTR(SUBSTRING(remain, 2, 255), '-'), ' ', '-')), 255) |
| 55 | + END AS remain, |
| 56 | + idx + 1 |
| 57 | + FROM splitted_word_cte |
| 58 | + WHERE remain != '' |
| 59 | +), |
| 60 | +formatted_word_cte AS ( |
| 61 | + SELECT content_id, |
| 62 | + content_text, |
| 63 | + idx, |
| 64 | + word, |
| 65 | + CONCAT(IF(delim = '^', '', delim), UPPER(SUBSTRING(word, 1, 1)), LOWER(SUBSTRING(word, 2, 255))) AS formatted_word |
| 66 | + FROM splitted_word_cte |
| 67 | +) |
| 68 | + |
| 69 | +SELECT content_id, |
| 70 | + content_text AS original_text, |
| 71 | + GROUP_CONCAT(formatted_word ORDER BY idx SEPARATOR '') AS converted_text |
| 72 | +FROM formatted_word_cte |
| 73 | +GROUP BY 1, 2 |
| 74 | +ORDER BY 1; |
0 commit comments