Alternatives à une auto FULL OUTER JOIN pour correspondance obtaint entre deux entrées

Question

dans mon modèle db j'ai deux entités: Droit et Entitlement_Data. Chaque droit est identifié par un ID incrémenter (d'identité), et peut avoir plusieurs entrées de Entitlement_Data, chacune avec un autre type - Wich peut être soit 0 (hebdomadaire) ou 1 (mois) (EntitlementID et type sont PK pour Entitlement_Data, EntitlementID étant FK droit de déposer). Pour chaque droit, je dois:

1. Obtenir la première valeur non NULL à partir soit mensuel (type = 1) ou hebdomadaire (type = 0) Entitlement_Data, pour un ensemble de Entitlement_Data donné les attributs (SharesPaid, LocalTaxRate, etc ...) - la première partie de la SELECT;
2. Obtenir une indication sur Match / décalage entre les entrées mensuelles et hebdomadaires Entitlement_Data (SharesPaid_Match, etc ...) -. La dernière partie du SELECT

Ceci est la requête que je utilise atm:

SELECT 
    COALESCE(MD.EntitlementID, WD.EntitlementID) AS EntitlementID,
    COALESCE(MD.LocalTaxRate, WD.LocalTaxRate) AS LocalTaxRate,
    COALESCE(MD.SharesPaid, WD.SharesPaid) AS SharesPaid,
    COALESCE(MD.LocalTaxAmount, WD.LocalTaxAmount) AS LocalTaxAmount,
    COALESCE(MD.LocalTaxEquivalent, WD.LocalTaxEquivalent) AS LocalTaxEquivalent,
    COALESCE(MD.NetReceived, WD.NetReceived) AS NetReceived,
    COALESCE(MD.LocalTaxCurrency, WD.LocalTaxCurrency) AS LocalTaxCurrency,
    COALESCE(MD.Currency, WD.Currency) AS Currency,
    COALESCE(MD.ReleaseDate, WD.ReleaseDate) AS ReleaseDate,
    (
        CASE
            WHEN MD.LocalTaxEquivalent IS NULL OR WD.LocalTaxEquivalent IS NULL THEN NULL
            WHEN MD.LocalTaxEquivalent <> WD.LocalTaxEquivalent THEN 0
            ELSE 1
        END
    ) AS LocalTaxEquivalent_Match,
    (
        CASE
            WHEN MD.NetReceived IS NULL OR WD.NetReceived IS NULL THEN NULL
            WHEN MD.NetReceived <> WD.NetReceived THEN 0
            ELSE 1
        END
    ) AS NetReceived_Match,
    (
        CASE
            WHEN MD.SharesPaid IS NULL OR WD.SharesPaid IS NULL THEN NULL
            WHEN MD.SharesPaid <> WD.SharesPaid THEN 0
            ELSE 1
        END
    ) AS SharesPaid_Match,
    (
        CASE
            WHEN MD.Currency IS NULL OR WD.Currency IS NULL THEN NULL
            WHEN MD.Currency <> WD.Currency THEN 0
            ELSE 1
        END
    ) AS Currency_Match,
    (
        CASE
            WHEN MD.DividendRate IS NULL OR WD.DividendRate IS NULL THEN NULL
            WHEN MD.DividendRate <> WD.DividendRate THEN 0
            ELSE 1
        END
    ) AS DividendRate_Match,
    (
        CASE
            WHEN MD.LocalTaxRate IS NULL OR WD.LocalTaxRate IS NULL THEN NULL
            WHEN MD.LocalTaxRate <> WD.LocalTaxRate THEN 0
            ELSE 1
        END
    ) AS LocalTaxRate_Match
FROM 
    Entitlement_Data MD
    FULL OUTER JOIN Entitlement_Data WD
        ON MD.EntitlementID = WD.EntitlementID
WHERE
    WD.Type = 0 AND MD.Type = 1

Y at-il un moyen d'éviter FULL OUTER JOIN et obtenir les informations nécessaires? J'évalué l'utilisation d'un PAR GROUPE mais je ne semble pas obtenir les résultats dont j'ai besoin. Merci.

Answer 1

Pourquoi ne pas left join à "table principale" qui est Entitlement comme dans le code suivant:

select
    COALESCE(MD.EntitlementID, WD.EntitlementID) AS EntitlementID,
    COALESCE(MD.LocalTaxRate, WD.LocalTaxRate) AS LocalTaxRate,
    COALESCE(MD.SharesPaid, WD.SharesPaid) AS SharesPaid,
    COALESCE(MD.LocalTaxAmount, WD.LocalTaxAmount) AS LocalTaxAmount,
    COALESCE(MD.LocalTaxEquivalent, WD.LocalTaxEquivalent) AS LocalTaxEquivalent,
    COALESCE(MD.NetReceived, WD.NetReceived) AS NetReceived,
    COALESCE(MD.LocalTaxCurrency, WD.LocalTaxCurrency) AS LocalTaxCurrency,
    COALESCE(MD.Currency, WD.Currency) AS Currency,
    COALESCE(MD.ReleaseDate, WD.ReleaseDate) AS ReleaseDate,
    (
        CASE
            WHEN MD.LocalTaxEquivalent IS NULL OR WD.LocalTaxEquivalent IS NULL THEN NULL
            WHEN MD.LocalTaxEquivalent <> WD.LocalTaxEquivalent THEN 0
            ELSE 1
        END
    ) AS LocalTaxEquivalent_Match,
    (
        CASE
            WHEN MD.NetReceived IS NULL OR WD.NetReceived IS NULL THEN NULL
            WHEN MD.NetReceived <> WD.NetReceived THEN 0
            ELSE 1
        END
    ) AS NetReceived_Match,
    (
        CASE
            WHEN MD.SharesPaid IS NULL OR WD.SharesPaid IS NULL THEN NULL
            WHEN MD.SharesPaid <> WD.SharesPaid THEN 0
            ELSE 1
        END
    ) AS SharesPaid_Match,
    (
        CASE
            WHEN MD.Currency IS NULL OR WD.Currency IS NULL THEN NULL
            WHEN MD.Currency <> WD.Currency THEN 0
            ELSE 1
        END
    ) AS Currency_Match,
    (
        CASE
            WHEN MD.DividendRate IS NULL OR WD.DividendRate IS NULL THEN NULL
            WHEN MD.DividendRate <> WD.DividendRate THEN 0
            ELSE 1
        END
    ) AS DividendRate_Match,
    (
        CASE
            WHEN MD.LocalTaxRate IS NULL OR WD.LocalTaxRate IS NULL THEN NULL
            WHEN MD.LocalTaxRate <> WD.LocalTaxRate THEN 0
            ELSE 1
        END
    ) AS LocalTaxRate_Match
from Entitlement e
left join Entitlement_Data wd on e.id = wd.entitlementID and wd.type = 0
left join Entitlement_Data md on e.id = md.entitlementID and md.type = 1

Vous pourriez limiter aux lignes Entitlement qui ont enfin une ligne dans la table de Entitlement_Data (sans se soucier si est faible ou mensuelle) en utilisant:

where wd.type is not null or md.type is not null