Java 8 lambdas nested Map

Question

I am trying to use Java-8 lambdas to solve the following problem:

Given a List<Transaction>, for each Category.minorCategory I require the sum of Transaction.amount per Category.minorCategory and a Map of Transaction.accountNumber with the sum of Transaction.amount per Transaction.accountNumber. I have this working, as per the code below. I now have a requirement to group by Category.majorCategory, essentially returning a Map<String, Map<String, MinorCategorySummary>> keyed on Category.majorCategory.

I have everything working up until the stage of grouping by Category.majorCategory but struggle to see the solution; the paradigm shift of programming with lambdas is proving a steep learning curve.

TransactionBreakdown is where the action happens and where I'd like to return a Map<String, Map<String, MinorCategorySummary>>.

public class Transaction {

    private final String accountNumber;
    private final BigDecimal amount;
    private final Category category;

}

public class Category {

    private final String majorCategory;
    private final String minorCategory;

}

public class MinorCategorySummary {

    private final BigDecimal sumOfAmountPerMinorCategory;
    private final Map<String, BigDecimal> accountNumberSumOfAmountMap;
    private final Category category;

}

public class TransactionBreakdown {

    Function<Entry<String, List<Transaction>>, MinorCategorySummary> transactionSummariser = new TransactionSummariser();

    public Map<Object, MinorCategorySummary> getTransactionSummaries(List<Transaction> transactions) {
        return transactions
                .stream()
                .collect(groupingBy(t -> t.getCategory().getMinorCategory()))
                .entrySet()
                .stream()
                .collect(
                        toMap(Entry::getKey,
                                transactionSummariser));
    }

}

public class TransactionSummariser implements Function<Entry<String, List<Transaction>>, MinorCategorySummary> {

    @Override
    public MinorCategorySummary apply(
            Entry<String, List<Transaction>> entry) {
         return new MinorCategorySummary(
                    entry.getValue()
                            .stream()
                            .map(Transaction::getAmount)
                            .collect(reducing(BigDecimal.ZERO, BigDecimal::add)),
                    entry.getValue()
                            .stream()
                            .collect(
                                    groupingBy(Transaction::getAccountNumber,
                                            mapping(Transaction::getAmount,
                                                    reducing(BigDecimal.ZERO, BigDecimal::add)))), 
                                                    entry.getValue().get(0).getCategory());
    }

}

Answer 1

Your class design seems odd to me. Why put category into the summary class only to then have the category as a map key? It would make more sense to have a summary class without category in it:

public class TransactionSummary {

    private final BigDecimal amount;
    private final Map<String, BigDecimal> acctToTotal;

    TransactionSummary(Map<String, BigDecimal> acctToTotal) {
        this.acctToTotal = Collections.unmodifiableMap(acctToTotal);
        this.amount = acctToTotal.values().stream()
                .reduce(BigDecimal.ZERO, BigDecimal::add);
    }

    public static Collector<Transaction, ?, TransactionSummary> collector() {
        // this can be a static constant
        return collectingAndThen(
            toMap(Transaction::getAccountNumber,Transaction::getAmount,BigDecimal::add),
            TransactionSummary::new
        );
    }

    // getters
}

Now your two problems are solved clearly and with no redundancy:

Map<String, TransactionSummary> minorSummary = transactions.stream()
        .collect(groupingBy(
                t -> t.getCategory().getMinorCategory(),
                TransactionSummary.collector()
        ));

Map<String, Map<String, TransactionSummary>> majorMinorSummary = transactions.stream()
        .collect(groupingBy(
                t -> t.getCategory().getMajorCategory(),
                groupingBy(
                        t -> t.getCategory().getMinorCategory(),
                        TransactionSummary.collector()
                )
        ));