Brazilian Receivables Part 3 — Partial Contracts

Brazilian card transactions are registered as daily receivables. You are given two valid CSV strings.
transactionsCsv has header customer_id,merchant_id,payout_date,card_type,amount.
contractsCsv has header contract_id,merchant_id,payout_date,card_type,amount.
Aggregate transactions by (merchant_id, card_type, payout_date). Each contract buys only its stated amount from the matching merchant receivable. Create a contract-ID receivable for that amount and subtract it from the merchant receivable. Keep the merchant row when its remaining amount is positive; remove it when the remaining amount is zero.
Return CSV rows beginning with id,card_type,payout_date,amount. Sort the data rows by Unicode code points in ID, then card type, then payout date. When all three strings are equal, sort by amount numerically in ascending order. Write each output amount in base-10 notation without leading zeros.

Function
allocatePartialContracts(transactionsCsv: String, contractsCsv: String) → String[]

Examples
Example 1
transactionsCsv = "customer_id,merchant_id,payout_date,card_type,amount\ncust1,merchantA,2022-01-07,Visa,500"
contractsCsv = "contract_id,merchant_id,payout_date,card_type,amount\ncontract1,merchantA,2022-01-07,Visa,200"
return = ["id,card_type,payout_date,amount","contract1,Visa,2022-01-07,200","merchantA,Visa,2022-01-07,300"]
The contract buys 200 of the 500 receivable, leaving 300 under the merchant ID.
Example 2
transactionsCsv = "customer_id,merchant_id,payout_date,card_type,amount\nc1,m1,2024-01-01,Visa,300\nc2,m1,2024-01-01,Visa,250\nc3,m2,2024-01-02,MasterCard,700"
contractsCsv = "contract_id,merchant_id,payout_date,card_type,amount\nk1,m1,2024-01-01,Visa,550"
return = ["id,card_type,payout_date,amount","k1,Visa,2024-01-01,550","m2,MasterCard,2024-01-02,700"]
A full-amount contract removes the matching merchant receivable. The unrelated merchant remains.
Example 3
transactionsCsv = "customer_id,merchant_id,payout_date,card_type,amount\nc1,zeta,2024-02-01,Visa,100\nc2,alpha,2024-02-02,Amex,80"
contractsCsv = "contract_id,merchant_id,payout_date,card_type,amount"
return = ["id,card_type,payout_date,amount","alpha,Amex,2024-02-02,80","zeta,Visa,2024-02-01,100"]
With no contracts, both aggregates remain and are sorted by output ID.

Constraints
Each CSV contains its exact header followed by between 0 and 100000 data rows.
Fields contain no commas, line breaks, or surrounding whitespace.
Amounts are positive integers at most 10^9; every aggregate fits a signed 64-bit integer.
Each contract maps to exactly one aggregated merchant receivable, its amount does not exceed that receivable, and no two contracts map to the same receivable.

The trick is a hash map keyed on (merchant_id, card_type, payout_date) that sums amounts in one pass over transactions. Then walk the contracts once. Each contract looks up its key in O(1), emits a row under the contract ID for its amount, and subtracts that amount from the merchant total. Because every contract maps to exactly one receivable and no two share one, there's no splitting logic to worry about. After that, emit merchant rows only where the remainder is positive. The pitfalls are all in the edges. Sort by plain code point comparison, not locale-aware compare. Tie-break on amount numerically, not as a string. Handle a contracts CSV that has only a header. Use 64-bit sums. Always output the header row first. If you freeze on any of this live, StealthCoder can hand you the working skeleton while you check the details.

Memorize the pattern. If you can't, run StealthCoder. The proctor sees the IDE. They don't see what's behind it.