Simplify code to determine if files need to be changed and prep to move data transformation into PSQL

This commit is contained in:
2024-04-04 21:49:52 +01:00
parent 99523430c6
commit bcfa0c2d89
5 changed files with 97 additions and 38 deletions
+24
View File
@@ -0,0 +1,24 @@
with timelineItems as (select jsonb_array_elements(raw_files.json -> 'timelineItems') as timelineItem
from raw_files),
places as (select distinct on (md5(timelineItem ->> 'place' :: text)) timelineItem -> 'place' as place,
timelineItem -> 'place' ->> 'placeId' as placeId,
(timelineItem -> 'place' ->> 'lastSaved') :: timestamptz as lastSaved
from timelineItems
where timelineItem ->> 'place' is not null),
places_with_max_last_saved as (select place -> 'placeId' as placeId,
max((place ->> 'lastSaved') :: timestamptz) as latest_last_saved
from places
group by place -> 'placeId'),
latest_places as (select places.*
from places_with_max_last_saved
inner join places on places.place -> 'placeId' = places_with_max_last_saved.placeId and
places.lastSaved =
places_with_max_last_saved.latest_last_saved)
insert
into public.place (place_id, json, last_saved, server_last_updated)
select (placeId :: uuid) as place_id, place as json, lastSaved as last_saved, now() as server_last_updated
from latest_places
on conflict (place_id) do update set json = excluded.json,
last_saved = excluded.last_saved,
server_last_updated = excluded.server_last_updated
where excluded.last_saved > public.place.last_saved;
+28
View File
@@ -0,0 +1,28 @@
with timelineItems as (select jsonb_array_elements(raw_files.json -> 'timelineItems') as timelineItem
from raw_files),
max_last_saved as (select timelineItem ->> 'itemId' as itemId,
max((timelineItem ->> 'lastSaved') :: timestamptz) as latest_last_saved
from timelineItems
group by timelineItem ->> 'itemId'),
unique_timline_items as (select distinct on (max_last_saved.itemId) *
from max_last_saved
inner join timelineItems
on timelineItems.timelineItem ->> 'itemId' = max_last_saved.itemId
and (timelineItems.timelineItem ->> 'lastSaved') :: timestamptz =
max_last_saved.latest_last_saved)
insert
into public.timeline_item (item_id, json, place_id, end_date, last_saved, server_last_updated)
select unique_timline_items.itemId :: uuid as item_id,
unique_timline_items.timelineItem as json,
(unique_timline_items.timelineItem -> 'place' ->> 'placeId') :: uuid as place_id,
(unique_timline_items.timelineItem ->> 'endDate') :: timestamptz as end_date,
unique_timline_items.latest_last_saved :: timestamptz as last_saved,
now() as server_last_updated
from unique_timline_items
on conflict (item_id) do update set json = excluded.json,
place_id = excluded.place_id,
end_date = excluded.end_date,
last_saved = excluded.last_saved,
server_last_updated = excluded.server_last_updated
where excluded.last_saved > public.timeline_item.last_saved
returning item_id;