refs https://github.com/TryGhost/Team/issues/1076
refs 70229e4fd3 (diff-b67ecda91b5bd79c598e5c5a9ec2ccf28dbfab6a924b21352273865e07cd7ceaR57)
- The "products" column has not been doing any logic anything since at least 5.20.0 (see refed commit). The concept of columns in the export file was mostly there for analytical/data filtering reasons - so the user could analyze their exports. CSV was never a good suite for relational data that "products" (or now tiers) represent
- The "tiers" column will still be present in the exported CSV file, but there is not going to be any logic attached to it.
- The only columns that can effect the "tiers" state of the member are: "complimentary_plan" (assign default tier to the member) and "stripe_customer_id" (pulls in subscription/tier data from Stripe)
closes https://github.com/TryGhost/Toolbox/issues/430
- The members importer used to import all fields present in the uploaded CSV if the headers match, even if they're not mapped in the UI. This behavior has lead to have misleading consequences and "hidden" features. For example, if the field was present but intentionally left as "Not imported" in the UI the field would still get imported.
- Having a strict list of supported import fields also allows for manageable long-term maintenance of the CSV Import API and detect/communicate changes when they happen.
- The list of the current default field mapping is:
email: 'email',
name: 'name',
note: 'note',
subscribed_to_emails: 'subscribed',
created_at: 'created_at',
complimentary_plan: 'complimentary_plan',
stripe_customer_id: 'stripe_customer_id',
labels: 'labels',
products: 'products'
refs https://github.com/TryGhost/Toolbox/issues/430
- To be able to introduce strict mapping rules (exclude unknown fields) we need to control the CSV header mapping on the importer level. This change moves the configuration up from CSV parser to the importer
- Also adds tests covering correct inserts for specially treated "subscribed_to_emails" field
refs https://github.com/TryGhost/Team/issues/916
- These tests are nowhere close to perfect of even good. They barely pass "good enough" watermark... it's a start for new tests to be added easier in the future once the features are developed around CSV improrter
