java-topology/defects/octave/unit/OctaveVecdimTest.java

package unit;

import java.util.ArrayList;
import java.util.Collections;

/**
 * Unit tests modelling CWE-407 defect in GNU Octave:
 *
 *   octave-0001 — data.cc and numeric/max.cc: vecdim is already sorted by
 *                 std::sort, but the subsequent loop uses std::find (O(|vecdim|))
 *                 instead of std::binary_search (O(log |vecdim|)).
 *
 *                 Modelled as: outer loop over ndims=64 dimensions, inner
 *                 ArrayList.contains() (unsorted scan, O(|vecdim|)) vs
 *                 Collections.binarySearch() on a sorted list (O(log |vecdim|)).
 *
 * Pure Java stdlib, instrumented operation counts.
 */
public class OctaveVecdimTest {

    /**
     * Simulate defective path: ndims iterations, std::find scan per iteration.
     * vecdim has n entries drawn from [0, ndims).
     */
    static long vecdimDefective(int ndims, int vecdimSize) {
        ArrayList<Integer> vecdim = new ArrayList<>();
        // Populate sorted vecdim with every other dim (representative subset)
        for (int i = 0; i < vecdimSize; i++) {
            vecdim.add(i * (ndims / vecdimSize));
        }
        // Mimics std::sort already done — list is sorted
        Collections.sort(vecdim);

        long ops = 0;
        for (int i = 0; i < ndims; i++) {
            // O(|vecdim|) — mirrors std::find(vecdim.begin(), vecdim.end(), i)
            boolean found = false;
            for (int j = 0; j < vecdim.size(); j++) {
                ops++;
                if (vecdim.get(j).equals(i)) { found = true; break; }
                // std::find does not exploit sort order; keep scanning past hits
                // would continue, but we break on first match (same as std::find)
            }
            // remaining iterations for elements not in vecdim scan to the end
            if (!found) {
                // already counted full scan above — no extra ops needed
            }
        }
        return ops;
    }

    /**
     * Simulate defective path with full scan (no early exit) to model worst
     * case where dims not in vecdim cause a full O(|vecdim|) scan.
     */
    static long vecdimDefectiveWorstCase(int ndims, int vecdimSize) {
        ArrayList<Integer> vecdim = new ArrayList<>();
        for (int i = 0; i < vecdimSize; i++) {
            vecdim.add(i * (ndims / vecdimSize));
        }
        Collections.sort(vecdim);

        long ops = 0;
        for (int i = 0; i < ndims; i++) {
            // Full O(|vecdim|) scan regardless (std::find scans all when not found)
            for (int j = 0; j < vecdim.size(); j++) {
                ops++;
                if (vecdim.get(j).equals(i)) break;
            }
        }
        return ops;
    }

    /**
     * Simulate fixed path: ndims iterations, Collections.binarySearch per iteration.
     * Each binarySearch is O(log |vecdim|).
     */
    static long vecdimFixed(int ndims, int vecdimSize) {
        ArrayList<Integer> vecdim = new ArrayList<>();
        for (int i = 0; i < vecdimSize; i++) {
            vecdim.add(i * (ndims / vecdimSize));
        }
        Collections.sort(vecdim);

        long ops = 0;
        for (int i = 0; i < ndims; i++) {
            // O(log |vecdim|) — mirrors std::binary_search(vecdim.begin(), vecdim.end(), i)
            int lo = 0, hi = vecdim.size() - 1;
            while (lo <= hi) {
                ops++;
                int mid = (lo + hi) >>> 1;
                int cmp = vecdim.get(mid).compareTo(i);
                if (cmp < 0)       lo = mid + 1;
                else if (cmp > 0)  hi = mid - 1;
                else               break;
            }
        }
        return ops;
    }

    // -----------------------------------------------------------------------
    // Test methods
    // -----------------------------------------------------------------------

    /**
     * Defective op count must grow as ndims × |vecdim|.
     * Confirm: opsLarge > opsSmall × 3 when scaling ndims from 16 to 64.
     */
    static void testDefectiveCountGrows() {
        long opsSmall = vecdimDefectiveWorstCase(16, 8);
        long opsFull  = vecdimDefectiveWorstCase(64, 32);
        assert opsFull > opsSmall * 3
            : "octave-0001: expected super-linear growth, opsSmall=" + opsSmall
              + " opsFull=" + opsFull;
        System.out.printf("PASS testDefectiveCountGrows: opsSmall=%d opsFull=%d ratio=%.1fx%n",
            opsSmall, opsFull, (double) opsFull / opsSmall);
    }

    /**
     * Fixed op count must be O(ndims × log|vecdim|).
     * At ndims=64, vecdimSize=32: ops <= ndims * ceil(log2(32)+1) = 64*6 = 384.
     */
    static void testFixedCountLogBound() {
        int ndims = 64, vecdimSize = 32;
        long ops = vecdimFixed(ndims, vecdimSize);
        long bound = (long) ndims * (long) Math.ceil(Math.log(vecdimSize) / Math.log(2) + 1);
        assert ops <= bound
            : "octave-0001: fixed ops=" + ops + " exceeded log bound=" + bound;
        System.out.printf("PASS testFixedCountLogBound: ops=%d bound=%d%n", ops, bound);
    }

    /**
     * Speedup ratio defective/fixed must exceed 5× at ndims=64, vecdimSize=32.
     */
    static void testSpeedupRatio() {
        int ndims = 64, vecdimSize = 32;
        long defOps = vecdimDefectiveWorstCase(ndims, vecdimSize);
        long fixOps = vecdimFixed(ndims, vecdimSize);
        double ratio = (double) defOps / fixOps;
        assert ratio > 5.0
            : "octave-0001: speedup ratio " + ratio + " not > 5x";
        System.out.printf("PASS testSpeedupRatio: defective=%d fixed=%d ratio=%.1fx%n",
            defOps, fixOps, ratio);
    }

    /**
     * Correctness: defective and fixed paths agree on which dims are excluded.
     * Both should exclude the same set of dims not present in vecdim.
     */
    static void testCorrectnessAgreement() {
        int ndims = 64, vecdimSize = 32;
        ArrayList<Integer> vecdim = new ArrayList<>();
        for (int i = 0; i < vecdimSize; i++) {
            vecdim.add(i * (ndims / vecdimSize));
        }
        Collections.sort(vecdim);

        ArrayList<Integer> excludedByFind   = new ArrayList<>();
        ArrayList<Integer> excludedByBsearch = new ArrayList<>();

        for (int i = 0; i < ndims; i++) {
            // std::find equivalent
            if (!vecdim.contains(i)) excludedByFind.add(i);
            // std::binary_search equivalent
            if (Collections.binarySearch(vecdim, i) < 0) excludedByBsearch.add(i);
        }

        assert excludedByFind.equals(excludedByBsearch)
            : "octave-0001: correctness mismatch: find=" + excludedByFind
              + " bsearch=" + excludedByBsearch;
        System.out.printf("PASS testCorrectnessAgreement: %d dims excluded (identical)%n",
            excludedByFind.size());
    }

    // -----------------------------------------------------------------------
    // Main
    // -----------------------------------------------------------------------

    public static void main(String[] args) {
        System.out.println("=== OctaveVecdimTest ===");
        testDefectiveCountGrows();
        testFixedCountLogBound();
        testSpeedupRatio();
        testCorrectnessAgreement();
        System.out.println("All tests passed.");
    }
}