fix repeatable migrations

This commit is contained in:
Dusan Vojacek
2026-04-19 20:15:46 +02:00
parent 0c93f493a4
commit 22bca9cd9e
73 changed files with 22 additions and 15 deletions

View File

@@ -0,0 +1,209 @@
-- =============================================================
-- R__031_fn_ote_import.sql
-- OTE CZ import parser a import funkce
-- Repeatable migration při změně funkce stačí upravit tento soubor
-- =============================================================
-- Parser: raw jsonb → 96 (nebo 92/100 při DST) cenových řádků po 15 min
CREATE OR REPLACE FUNCTION ems.fn_ote_parse_15m_price_json(
in_payload jsonb,
in_czk_per_eur numeric DEFAULT 25.000
)
RETURNS TABLE (
interval_start timestamptz,
interval_end timestamptz,
raw_price_czk_kwh numeric(10,6)
)
LANGUAGE plpgsql
AS $$
DECLARE
v_date_text text;
v_market_date date;
v_anchor timestamptz;
v_dl jsonb;
v_npts int;
BEGIN
IF in_payload IS NULL THEN
RAISE EXCEPTION 'in_payload must not be null';
END IF;
IF in_czk_per_eur IS NULL OR in_czk_per_eur <= 0 THEN
RAISE EXCEPTION 'in_czk_per_eur must be > 0, got: %', in_czk_per_eur;
END IF;
-- Datum z graph.title ve formátu "... DD.MM.YYYY"
v_date_text := substring(
in_payload #>> '{graph,title}'
FROM '([0-9]{2}\.[0-9]{2}\.[0-9]{4})'
);
IF v_date_text IS NULL THEN
RAISE EXCEPTION 'cannot parse date from graph.title: %',
in_payload #>> '{graph,title}';
END IF;
v_market_date := to_date(v_date_text, 'DD.MM.YYYY');
-- Začátek obchodního dne v Praze jako absolutní okamžik; pak +15 min v UTC.
-- Starý vzor (timestamp + interval) AT TIME ZONE při přechodu na letní čas
-- slučoval různé sloty → duplicitní interval_start a chyba ON CONFLICT.
v_anchor := v_market_date::timestamp AT TIME ZONE 'Europe/Prague';
-- OTE mění strukturu: novější data = 15min série (tooltip flash_chart_01_y_15m_*),
-- starší / jiná odpověď = jen 24 hodinových bodů s tooltip "Cena".
SELECT t.dl INTO v_dl
FROM jsonb_array_elements(in_payload #> '{data,dataLine}') AS t(dl)
WHERE dl ->> 'tooltip' = 'flash_chart_01_y_15m_price_tooltip'
AND jsonb_array_length(dl -> 'point') >= 90
LIMIT 1;
IF v_dl IS NULL THEN
SELECT t.dl INTO v_dl
FROM jsonb_array_elements(in_payload #> '{data,dataLine}') AS t(dl)
WHERE dl ->> 'tooltip' = 'flash_chart_01_y_60m_price_tooltip'
AND jsonb_array_length(dl -> 'point') >= 90
LIMIT 1;
END IF;
IF v_dl IS NULL THEN
SELECT t.dl INTO v_dl
FROM jsonb_array_elements(in_payload #> '{data,dataLine}') AS t(dl)
WHERE dl ->> 'tooltip' = 'Cena'
AND jsonb_array_length(dl -> 'point') >= 90
LIMIT 1;
END IF;
IF v_dl IS NULL THEN
SELECT t.dl INTO v_dl
FROM jsonb_array_elements(in_payload #> '{data,dataLine}') AS t(dl)
WHERE dl ->> 'tooltip' = 'Cena'
AND jsonb_array_length(dl -> 'point') BETWEEN 2 AND 28
LIMIT 1;
END IF;
IF v_dl IS NULL THEN
SELECT t.dl INTO v_dl
FROM jsonb_array_elements(in_payload #> '{data,dataLine}') AS t(dl)
WHERE dl ->> 'tooltip' = 'flash_chart_01_y_15m_price_tooltip'
AND jsonb_array_length(dl -> 'point') BETWEEN 2 AND 28
LIMIT 1;
END IF;
IF v_dl IS NULL THEN
RAISE EXCEPTION
'OTE price dataLine not found (očekáváno 15min flash_* nebo hodinová Cena); '
'dostupné tooltips: %',
(SELECT jsonb_agg(dl ->> 'tooltip' ORDER BY dl ->> 'tooltip')
FROM jsonb_array_elements(in_payload #> '{data,dataLine}') dl);
END IF;
v_npts := jsonb_array_length(v_dl -> 'point');
IF v_npts >= 90 THEN
-- x = 1..N = 15min bloky (96 běžně; 92 při zkráceném dni při přechodu na letní čas)
RETURN QUERY
SELECT s.interval_start, s.interval_end, s.raw_price_czk_kwh
FROM (
SELECT
(v_anchor + ((block_no - 1) * INTERVAL '15 minutes')) AS interval_start,
(v_anchor + (block_no * INTERVAL '15 minutes')) AS interval_end,
ROUND(
(price_eur_mwh * in_czk_per_eur / 1000.0)::numeric, 6
)::numeric(10,6) AS raw_price_czk_kwh,
block_no
FROM (
SELECT
(p ->> 'x')::int AS block_no,
(p ->> 'y')::numeric AS price_eur_mwh
FROM jsonb_array_elements(v_dl -> 'point') AS p
) pts
) AS s
ORDER BY s.block_no;
ELSIF v_npts BETWEEN 2 AND 28 THEN
-- x = 1..H = hodiny (ne nutně celý den); každou hodinovou cenu → 4× 15min slot
RETURN QUERY
SELECT s.interval_start, s.interval_end, s.raw_price_czk_kwh
FROM (
SELECT
(v_anchor + (((hour_no - 1) * 4 + qix) * INTERVAL '15 minutes'))
AS interval_start,
(v_anchor + (((hour_no - 1) * 4 + qix) * INTERVAL '15 minutes')
+ INTERVAL '15 minutes')
AS interval_end,
ROUND(
(price_eur_mwh * in_czk_per_eur / 1000.0)::numeric, 6
)::numeric(10,6) AS raw_price_czk_kwh,
(hour_no - 1) * 4 + qix + 1 AS sort_key
FROM (
SELECT (p ->> 'x')::int AS hour_no, (p ->> 'y')::numeric AS price_eur_mwh
FROM jsonb_array_elements(v_dl -> 'point') AS p
) h
CROSS JOIN (VALUES (0), (1), (2), (3)) AS q(qix)
) AS s
ORDER BY s.sort_key;
ELSE
RAISE EXCEPTION
'OTE price series: unexpected point count % (tooltip %)',
v_npts,
v_dl ->> 'tooltip';
END IF;
IF NOT FOUND THEN
RAISE EXCEPTION 'OTE price series had no points after parse';
END IF;
END;
$$;
COMMENT ON FUNCTION ems.fn_ote_parse_15m_price_json(jsonb, numeric) IS
'Parsuje raw JSON z OTE @@chart-data?time_resolution=PT15M.
Datum z graph.title (DD.MM.YYYY). Sloty: kotva půlnoci Europe/Prague jako timestamptz + násobky 15 min (bez DST duplicit).
Výběr série: flash_15m ≥90, flash_60m ≥90, Cena ≥90, Cena 228 h, flash_15m 228 h.
EUR/MWh → Kč/kWh. Typicky 96 řádků, 92 při zkráceném dni (letní čas).
Testování přímo v DB:
SELECT * FROM ems.fn_ote_parse_15m_price_json(pg_read_file(''/tmp/ote.json'')::jsonb, 25.0) LIMIT 5;';
CREATE OR REPLACE FUNCTION ems.fn_ote_import_from_json(
in_payload jsonb,
in_czk_per_eur numeric DEFAULT 25.000
)
RETURNS integer
LANGUAGE plpgsql
AS $$
DECLARE
v_rowcount integer;
BEGIN
INSERT INTO ems.market_interval_price (
market_source,
interval_start,
interval_end,
buy_raw_price_czk_kwh,
sell_raw_price_czk_kwh,
currency,
imported_at
)
SELECT DISTINCT ON (p.interval_start)
'OTE_CZ',
p.interval_start,
p.interval_end,
p.raw_price_czk_kwh,
p.raw_price_czk_kwh, -- spot trh: buy = sell
'CZK',
now()
FROM ems.fn_ote_parse_15m_price_json(in_payload, in_czk_per_eur) AS p
ORDER BY p.interval_start, p.interval_end DESC
ON CONFLICT (market_source, interval_start) DO UPDATE SET
interval_end = EXCLUDED.interval_end,
buy_raw_price_czk_kwh = EXCLUDED.buy_raw_price_czk_kwh,
sell_raw_price_czk_kwh = EXCLUDED.sell_raw_price_czk_kwh,
imported_at = EXCLUDED.imported_at;
GET DIAGNOSTICS v_rowcount = ROW_COUNT;
RETURN v_rowcount;
END;
$$;
COMMENT ON FUNCTION ems.fn_ote_import_from_json(jsonb, numeric) IS
'Uloží výstup fn_ote_parse_15m_price_json do ems.market_interval_price.
Python předá raw jsonb z HTTP response + kurz EUR/CZK.
Vrátí počet upsertnutých řádků (očekáváno 96).
Testování přímo v DB:
SELECT ems.fn_ote_import_from_json(
pg_read_file(''/tmp/ote.json'')::jsonb, 25.0
);';