Two customer systems use different identifiers for the same customer. How would you integrate them without merging the wrong records?

Practice task

Implement resolve_customer_ids(records, approved_pairs) for this fictional customer integration. Return one result for each input record, in the original order, preserving its fields and adding status and target_id. Do not modify the input data.

Input contract

• The entity is a customer account. All inputs have already been schema-validated, and the shown identifier fields are nonempty strings. Input records do not contain status or target_id; those fields belong to the result.
• An identifier belongs to its tenant and source system. Reusing the same text in another tenant or source does not establish the same identity.
• The target system is billing, one fixed namespace. Each target_id refers to an account within the same tenant. A deployment with multiple target systems would need to include the target namespace too.
• approved_pairs is an approved mapping feed. It can contain repeated entries or conflicting approvals that need resolution. This exercise does not authenticate the caller, approve mappings or prove that a business identity is true.
• Permitted statuses are matched, ambiguous and unresolved. Return a target only when the approved input supports one unambiguous target for the scoped source record. Do not choose between conflicting targets. An identical repeated pair does not create a new candidate.
• Approved many-to-one relationships are allowed; do not impose target uniqueness that the customer has not requested.

Fictional inputs

records = [
    {"tenant": "north", "source": "crm", "source_id": "C7"},
    {"tenant": "south", "source": "crm", "source_id": "C7"},
    {"tenant": "north", "source": "crm", "source_id": "C8"},
    {"tenant": "north", "source": "erp", "source_id": "C7"},
]
approved_pairs = [
    {"tenant": "north", "source": "crm", "source_id": "C7", "target_id": "B12"},
    {"tenant": "north", "source": "crm", "source_id": "C8", "target_id": "B15"},
    {"tenant": "north", "source": "crm", "source_id": "C8", "target_id": "B16"},
]

Explain how you would test tenant separation, source-system separation, conflicting approvals, an identical repeated pair and an approved many-to-one relationship. State what must be checked outside this function before a mapping may authorize a cross-system write.