tracker

943 B

1. // tools/count-duplicate-ids.hl — mission 047 (hybriel#113): duplicate @ids per table. ONLY on a COPY (opening a table rewrites
2. // its files): TRACKER_STORAGE=$PWD/.scratch/realdata/storage/mpackdb ./bin/hybriel tools/count-duplicate-ids.hl
3. import { MPackDB } from 'hl:mpackdb'
4. import { env } from 'hl:proc'
5. dir = env('TRACKER_STORAGE')
6. for (t of ['episodes', 'follows', 'genres', 'persons', 'seasons', 'shows', 'users', 'watches']) {
7. let idx = t == 'users' ? ['!identity'] : ['!oldId']
8. let db = new MPackDB(file = dir + '/' + t + '.db', primaryKey = '@id', indexes = idx)
9. let seen = {}
10. let rows = 0
11. let dupIds = 0
12. let dupRows = 0
13. for (r of db.find(null, null)) {
14. rows = rows + 1
15. let k = '' + r.id
16. if (seen[k] == null) { seen[k] = 1 } else {
17. if (seen[k] == 1) { dupIds = dupIds + 1 }
18. seen[k] = seen[k] + 1
19. dupRows = dupRows + 1
20. }
21. }
22. print(t + ' rows=' + rows + ' duplicateIds=' + dupIds + ' extraRows=' + dupRows)
23. }