LEFT JOIN 右表中的缺失值

Question

I am trying to split a table into two referenced tables and am having issues with the population of one of the new tables. Say the old table has columns A, B, C, X, Y, Z and the first new table to have A, B, C, ID and the second to be ID, X, Y, Z .

Populating the second table is trivial:

INSERT INTO newTable2 (`X`,`Y`,`Z`)
SELECT DISTINCT `X`,`Y`,`Z`
FROM oldTable;

I can check newTable2 after this has been run and see that is populated properly. ID is populated by the table definition and there are no null values. I seem to have issues with populating the first table. I attempted to use this script:

INSERT INTO newTable1
SELECT oldTable.`A`
     , oldTable.`B`
     , oldTable.`C`
     , newTable2.`ID`
FROM oldTable
LEFT JOIN newTable2
ON newTable2.`X` = oldTable.`X`
    AND newTable2.`Y` = oldTable.`Y`
    AND newTable2.`Z` = oldTable.`Z`;

But when I check the resulting table I get null values for ID for most rows. Due to how it's populated newTable2 should have a row and ID for every row of oldTable and every row with a null that I have checked manually has had a value that was simply not found.

I am running MySql 5.7 and all columns except ID are of the type varchar .

Answer 1

Your JOIN condition does not handle NULL values. You have to handle them if you want to get all your original values back.

Use something like this for each columns which can contain NULL values.

(newTable2.`X` = oldTable.`X` OR (newTable2.`X` IS NULL AND oldTable.`X` IS NULL))

Answer 2

As @Pred mentioned you should handle the NULL case in your Join statements. I would use the null-safe <=> operator to avoid the OR statement:

Something like the following:

 INSERT INTO newTable1
    SELECT oldTable.`A`
         , oldTable.`B`
         , oldTable.`C`
         , newTable2.`ID`
    FROM oldTable
    LEFT JOIN newTable2
    ON newTable2.`X` <=> oldTable.`X`
        AND newTable2.`Y` <=> oldTable.`Y`
        AND newTable2.`Z` <=> oldTable.`Z`;