java-topology/defects/hive/unit/HiveGenMRSeenOpsTest.java

package unit;

import java.util.*;

/**
 * Unit tests for hive-0001 and hive-0002: GenMRProcContext seenOps CWE-407.
 *
 * hive-0001: taskToSeenOps maps Task → List<Operator>. isSeenOp() calls
 *            ArrayList.contains() — O(n) per lookup. Called O(T) times during
 *            MapReduce plan generation. Total O(T²) per task.
 *
 * hive-0002: seenFileSinkOps is List<FileSinkOperator>. contains() in
 *            GenMRFileSink1 is O(n) on each file sink encountered.
 *
 * Fix:       List → HashSet for both. O(1) contains().
 *
 * Measurement: count element-level comparisons in contains().
 */
public class HiveGenMRSeenOpsTest {

    // ── Minimal stub types ───────────────────────────────────────────────────

    static class Task {
        final String id;
        Task(String id) { this.id = id; }
        @Override public boolean equals(Object o) {
            return o instanceof Task && ((Task) o).id.equals(id);
        }
        @Override public int hashCode() { return id.hashCode(); }
        @Override public String toString() { return "Task(" + id + ")"; }
    }

    static class Operator {
        final String id;
        Operator(String id) { this.id = id; }
        @Override public boolean equals(Object o) {
            return o instanceof Operator && ((Operator) o).id.equals(id);
        }
        @Override public int hashCode() { return id.hashCode(); }
        @Override public String toString() { return "Op(" + id + ")"; }
    }

    // ── DEFECTIVE: List<Operator> per task ───────────────────────────────────

    static class DefectiveSeenOps {
        final Map<Task, List<Operator>> taskToSeenOps = new HashMap<>();
        long comparisons = 0;

        boolean isSeenOp(Task task, Operator op) {
            List<Operator> seen = taskToSeenOps.get(task);
            if (seen == null) return false;
            for (Operator s : seen) {      // linear scan
                comparisons++;
                if (s.equals(op)) return true;
            }
            return false;
        }

        void addSeenOp(Task task, Operator op) {
            taskToSeenOps.computeIfAbsent(task, k -> new ArrayList<>()).add(op);
        }
    }

    // ── FIXED: Set<Operator> per task ────────────────────────────────────────

    static class FixedSeenOps {
        final Map<Task, Set<Operator>> taskToSeenOps = new HashMap<>();
        long comparisons = 0;

        boolean isSeenOp(Task task, Operator op) {
            Set<Operator> seen = taskToSeenOps.get(task);
            if (seen == null) return false;
            comparisons++;     // O(1) hash lookup
            return seen.contains(op);
        }

        void addSeenOp(Task task, Operator op) {
            taskToSeenOps.computeIfAbsent(task, k -> new HashSet<>()).add(op);
        }
    }

    // ── Simulate MapReduce plan generation ───────────────────────────────────
    // For T table-scan operators per task: each is added then checked T times.

    public static long simulateDefective(int T) {
        Task task = new Task("mapTask");
        DefectiveSeenOps ctx = new DefectiveSeenOps();
        List<Operator> ops = new ArrayList<>();
        for (int i = 0; i < T; i++) ops.add(new Operator("op_" + i));

        for (Operator op : ops) {
            // Check if seen (not yet) then add — models isSeenOp + addSeenOp
            if (!ctx.isSeenOp(task, op)) ctx.addSeenOp(task, op);
        }
        // Second pass: re-check all (models merged task path through same operators)
        for (Operator op : ops) {
            ctx.isSeenOp(task, op);
        }
        return ctx.comparisons;
    }

    public static long simulateFixed(int T) {
        Task task = new Task("mapTask");
        FixedSeenOps ctx = new FixedSeenOps();
        List<Operator> ops = new ArrayList<>();
        for (int i = 0; i < T; i++) ops.add(new Operator("op_" + i));

        for (Operator op : ops) {
            if (!ctx.isSeenOp(task, op)) ctx.addSeenOp(task, op);
        }
        for (Operator op : ops) {
            ctx.isSeenOp(task, op);
        }
        return ctx.comparisons;
    }

    // ── Tests ────────────────────────────────────────────────────────────────

    static void testCorrectnessMatch() {
        Task task = new Task("t1");
        DefectiveSeenOps def = new DefectiveSeenOps();
        FixedSeenOps fix = new FixedSeenOps();
        List<Operator> ops = new ArrayList<>();
        for (int i = 0; i < 10; i++) ops.add(new Operator("op_" + i));

        for (Operator op : ops) {
            boolean d = def.isSeenOp(task, op); def.addSeenOp(task, op);
            boolean f = fix.isSeenOp(task, op); fix.addSeenOp(task, op);
            assert d == f : "isSeenOp result mismatch before add";
        }
        for (Operator op : ops) {
            assert def.isSeenOp(task, op) == fix.isSeenOp(task, op)
                : "isSeenOp result mismatch after add";
        }
        System.out.println("PASS testCorrectnessMatch");
    }

    static void testDefectiveGrowsQuadratically() {
        long prev = -1;
        for (int T : new int[]{20, 40, 80}) {
            long c = simulateDefective(T);
            if (prev > 0) {
                double ratio = (double) c / prev;
                assert ratio > 2.5
                    : "defective comparisons should grow >2.5x when T doubles; got " + ratio;
            }
            prev = c;
        }
        System.out.println("PASS testDefectiveGrowsQuadratically");
    }

    static void testFixedGrowsLinearly() {
        long prev = -1;
        for (int T : new int[]{20, 40, 80}) {
            long c = simulateFixed(T);
            if (prev > 0) {
                double ratio = (double) c / prev;
                // Should be very close to 2x (linear) — allow 2.0±0.2
                assert ratio < 2.25
                    : "fixed comparisons should grow ≈2x when T doubles; got " + ratio;
            }
            prev = c;
        }
        System.out.println("PASS testFixedGrowsLinearly");
    }

    static void testRatioAtScale() {
        int T = 100;
        long def_c = simulateDefective(T);
        long fix_c = simulateFixed(T);
        double ratio = (double) def_c / fix_c;
        assert ratio > 20
            : "at T=100, defective should be >20x slower; ratio=" + ratio;
        System.out.printf("PASS testRatioAtScale  (defective=%d, fixed=%d, ratio=%.1fx)%n",
            def_c, fix_c, ratio);
    }

    public static void main(String[] args) {
        testCorrectnessMatch();
        testDefectiveGrowsQuadratically();
        testFixedGrowsLinearly();
        testRatioAtScale();
        System.out.println("All hive-0001/0002 tests passed.");
    }
}