java-topology/defects/spark/unit/SparkAnalyzerWindowAggTest.java

package unit;

import java.util.*;

/**
 * Unit test for spark-0001: seenWindowAggregates ArrayBuffer CWE-407.
 *
 * Defect:  Analyzer.scala uses ArrayBuffer[AggregateExpression] for
 *          seenWindowAggregates and calls .contains(agg) on it during
 *          window function extraction. O(W) per check, O(A×W) total
 *          where A = aggregate expressions, W = window aggregates in query.
 *
 * Fix:     Replace ArrayBuffer with mutable.LinkedHashSet — O(1) contains.
 *
 * This test uses a self-contained Java model of the two strategies:
 *   DefectiveWindowExtractor  — ArrayList.contains()
 *   FixedWindowExtractor      — LinkedHashSet.contains()
 *
 * Measurement: count element-level comparisons in the membership check.
 */
public class SparkAnalyzerWindowAggTest {

    // ── Minimal AggregateExpression stub ────────────────────────────────────

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

    // ── Result ──────────────────────────────────────────────────────────────

    public static class Result {
        final List<AggExpr> extracted;
        public final long comparisons;        // element-level equality checks
        Result(List<AggExpr> extracted, long comparisons) {
            this.extracted   = extracted;
            this.comparisons = comparisons;
        }
    }

    // ── DEFECTIVE: ArrayList.contains() ─────────────────────────────────────

    public static Result extractDefective(List<AggExpr> windowAggs, List<AggExpr> otherAggs) {
        List<AggExpr> seen = new ArrayList<>();
        long comparisons = 0;

        // Window aggregates are added first (equivalent to WindowExpression case)
        for (AggExpr agg : windowAggs) {
            seen.add(agg);
        }

        // Other aggregates check seen list — O(W) each
        List<AggExpr> extracted = new ArrayList<>();
        for (AggExpr agg : otherAggs) {
            // Simulate ArrayList.contains() — count each comparison
            boolean found = false;
            for (AggExpr s : seen) {
                comparisons++;
                if (s.equals(agg)) { found = true; break; }
            }
            if (!found) {
                extracted.add(agg);
                seen.add(agg);
            }
        }
        return new Result(extracted, comparisons);
    }

    // ── FIXED: LinkedHashSet.contains() ─────────────────────────────────────

    public static Result extractFixed(List<AggExpr> windowAggs, List<AggExpr> otherAggs) {
        LinkedHashSet<AggExpr> seen = new LinkedHashSet<>();
        long comparisons = 0;

        for (AggExpr agg : windowAggs) {
            seen.add(agg);
        }

        List<AggExpr> extracted = new ArrayList<>();
        for (AggExpr agg : otherAggs) {
            comparisons++;   // O(1) hash lookup — counts as 1
            if (!seen.contains(agg)) {
                extracted.add(agg);
                seen.add(agg);
            }
        }
        return new Result(extracted, comparisons);
    }

    // ── Build test fixtures ──────────────────────────────────────────────────

    public static List<AggExpr> windowAggs(int w) {
        List<AggExpr> list = new ArrayList<>();
        for (int i = 0; i < w; i++) list.add(new AggExpr("win_" + i));
        return list;
    }

    public static List<AggExpr> otherAggs(int a, List<AggExpr> existing) {
        // Mix of novel aggregates and ones already in existing (the common case)
        List<AggExpr> list = new ArrayList<>();
        for (int i = 0; i < a; i++) {
            list.add(i % 3 == 0 && i / 3 < existing.size()
                ? existing.get(i / 3)          // already seen
                : new AggExpr("agg_" + i));    // novel
        }
        return list;
    }

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

    static void testCorrectnessMatch() {
        List<AggExpr> wins = windowAggs(5);
        List<AggExpr> aggs = otherAggs(10, wins);
        Result def = extractDefective(wins, aggs);
        Result fix = extractFixed(wins, aggs);
        assert new HashSet<>(def.extracted).equals(new HashSet<>(fix.extracted))
            : "defective and fixed must extract identical aggregates";
        System.out.println("PASS testCorrectnessMatch");
    }

    static void testDefectiveGrowsQuadratically() {
        // W=A, grow together: comparisons should scale ~quadratically
        long prev = -1;
        double prevRatio = -1;
        for (int n : new int[]{10, 20, 40}) {
            List<AggExpr> wins = windowAggs(n);
            List<AggExpr> aggs = otherAggs(n, wins);
            long c = extractDefective(wins, aggs).comparisons;
            if (prev > 0) {
                double ratio = (double) c / prev;
                assert ratio > 2.5 : "defective comparisons should grow >2.5x when n doubles; got " + ratio;
            }
            prev = c;
        }
        System.out.println("PASS testDefectiveGrowsQuadratically");
    }

    static void testFixedGrowsLinearly() {
        // Fixed: exactly 1 comparison per otherAgg regardless of W
        for (int n : new int[]{10, 20, 40}) {
            List<AggExpr> wins = windowAggs(n);
            List<AggExpr> aggs = otherAggs(n, wins);
            long c = extractFixed(wins, aggs).comparisons;
            assert c == n : "fixed must make exactly A comparisons (one per otherAgg); got " + c + " for A=" + n;
        }
        System.out.println("PASS testFixedGrowsLinearly");
    }

    static void testRatioAtScaleIsLarge() {
        int w = 100, a = 100;
        List<AggExpr> wins = windowAggs(w);
        List<AggExpr> aggs = otherAggs(a, wins);
        long def_c = extractDefective(wins, aggs).comparisons;
        long fix_c = extractFixed(wins, aggs).comparisons;
        double ratio = (double) def_c / fix_c;
        assert ratio > 10 : "at W=A=100, defective should be >10x worse; ratio=" + ratio;
        System.out.printf("PASS testRatioAtScaleIsLarge  (defective=%d, fixed=%d, ratio=%.1fx)%n",
            def_c, fix_c, ratio);
    }

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