Format UK school postcodes

[grega]

Context
UK postcode data is used in our reporting tools in order to measure how effective we are at reaching deprived areas (using the IMD / Indices of Multiple Deprivation: https://www.gov.uk/government/collections/english-indices-of-deprivation). This is a key part our programme impact monitoring work.

The IMD lookup relies on a correctly formatted postcode as defined by the NSPL (National Statistics Postcode Lookup) data: 2-4 char outward code, a space, then a 3-char inward code.

Problem
Postcode validation (as in "is this an actual UK postcode that is definitely in the Royal Mail Postcode Address File / PAF") is suboptimal, as doing so with a regex is a pain and potentially error prone (with the onus to 'fix' a potentially valid postcode being put on the user). We're also currently discounting using an API to look up / validate postcodes, though this may be a valid approach going forwards.

Proposed approach
Whilst we don't want to take a heavy-handed approach to postcode validation, as outlined above, we do want to make sure that any postcode that meets the criteria of being "very likely to be in the PAF and NSPL data sets" is formatted in a way which makes lookups possible (in order to prevent cases such as CB11NT, CB 11NT):

• Take any >=5 char UK postcode and ensure it has one space in it before the final 3 chars / inward code.

Additional steps
Existing school data will need to have the same transform / corrections applied, something like:

UPDATE schools
SET postal_code = 
    CASE 
        WHEN LENGTH(postal_code) >= 5 AND POSITION(' ' IN postal_code) = 0
        THEN CONCAT(SUBSTRING(postal_code FROM 1 FOR LENGTH(postal_code) - 3), ' ', SUBSTRING(postal_code FROM LENGTH(postal_code) - 2))
        ELSE postal_code
    END
WHERE country_code = 'GB';

[sra405]

😍 NICEEEE!! Love a little code snatching from other apps 😅