db-subclass-to-dto/tools/ExtractMethod/Tooling/SignatureBuilder.cs

using Microsoft.CodeAnalysis;
using Microsoft.CodeAnalysis.CSharp.Syntax;

namespace ExtractMethod.Tooling;

/// <summary>The suggested signature line of the extraction (parent spec:
/// "ends with a suggested signature: <c>int Extract(int i, Order order)</c>").</summary>
/// <param name="ReturnType">Return type, <c>void</c> when nothing must flow
/// out, or a tuple type when several declared-inside variables must.</param>
/// <param name="MethodName">v1 always suggests <c>Extract</c>.</param>
/// <param name="Params">The coherent parameter list (ref-ness preserved from
/// the classification).</param>
public sealed record SignatureSuggestion(
    string ReturnType,
    string MethodName,
    IReadOnlyList<ParamSuggestion> Params);

/// <summary>
/// Turns the raw em 02 buckets into ONE coherent signature line. The buckets
/// over-report on purpose (parent spec: a variable can sit in several
/// buckets); building a signature is where the overlaps are resolved — the
/// "promotion" em 02's comment promises.
///
/// RULES (v1):
/// - A variable DECLARED inside the selection cannot be a parameter of the
///   extracted method (it does not exist before the selection): drop it from
///   the params. If it is also a return candidate, it becomes the RETURN —
///   e.g. `total` (declared, reassigned, read after) in Summarize lands in
///   params[ref] ∧ returns raw, and promotes to the plain return value.
/// - A return candidate declared OUTSIDE the selection that the extraction
///   already receives by-ref (read inside ∧ written inside) keeps its ref
///   parameter: the write-back already flows the value out, and also
///   returning it would hand the caller the same value twice.
/// - A return candidate declared OUTSIDE that is NOT a param (written
///   inside, never read inside) becomes a return slot too: the caller
///   reassigns it from the extracted method's result.
/// - Trailing composite returns (`return a + b;`) are not nameable in v1 —
///   no return slot; the report's note suggests extracting a local first.
///
/// Determinism: parameters keep the classifier's ordinal-by-name order and
/// return slots are ordered by name, so the line is stable for tests.
/// </summary>
public static class SignatureBuilder
{
    public static SignatureSuggestion Build(SemanticModel model, SelectionReport selection, ExtractionSuggestion suggestion)
    {
        var declaredInside = DeclaredInsideNames(model, selection);
        var paramNames = suggestion.Params.Select(p => p.Name).ToHashSet(StringComparer.Ordinal);

        var parameters = suggestion.Params
            .Where(p => !declaredInside.Contains(p.Name))
            .ToList();

        var returnSlots = suggestion.Returns
            .Where(r => declaredInside.Contains(r.Name) || !paramNames.Contains(r.Name))
            .OrderBy(r => r.Name, StringComparer.Ordinal)
            .ToList();

        var returnType = returnSlots.Count switch
        {
            0 => "void",
            1 => returnSlots[0].Type,
            // Branch-insensitive over-approximation can over-report returns
            // (decision #4); the tuple keeps every candidate flowing rather
            // than guessing which is real.
            _ => $"({string.Join(", ", returnSlots.Select(r => r.Type))})",
        };

        return new SignatureSuggestion(returnType, "Extract", parameters);
    }

    /// <summary>
    /// Names of the locals DECLARED inside the selection. v1 covers the
    /// declaration shapes the fixture uses — local declarations, for-loop
    /// headers (`for (int i = ...)`) and foreach headers; other declaring
    /// shapes (pattern declarations, using/fixed statements, local functions)
    /// are not selected by the v1 resolver and would need their own cases.
    ///
    /// Names, not symbols, are matched against the bucket entries — sound
    /// here because C# forbids a local from shadowing another local or a
    /// parameter of the same method, so names are unique per method.
    /// </summary>
    public static HashSet<string> DeclaredInsideNames(SemanticModel model, SelectionReport selection)
    {
        var names = new HashSet<string>(StringComparer.Ordinal);

        foreach (var statement in selection.Statements)
        {
            foreach (var node in statement.DescendantNodesAndSelf())
            {
                switch (node)
                {
                    case LocalDeclarationStatementSyntax localDeclaration:
                        foreach (var variable in localDeclaration.Declaration.Variables)
                        {
                            if (model.GetDeclaredSymbol(variable) is ILocalSymbol)
                            {
                                names.Add(variable.Identifier.ValueText);
                            }
                        }

                        break;

                    case ForStatementSyntax forStatement when forStatement.Declaration is not null:
                        foreach (var variable in forStatement.Declaration.Variables)
                        {
                            if (model.GetDeclaredSymbol(variable) is ILocalSymbol)
                            {
                                names.Add(variable.Identifier.ValueText);
                            }
                        }

                        break;

                    case ForEachStatementSyntax forEach:
                        if (model.GetDeclaredSymbol(forEach) is ILocalSymbol)
                        {
                            names.Add(forEach.Identifier.ValueText);
                        }

                        break;
                }
            }
        }

        return names;
    }
}