java-topology/defects/cpython/unit/CpythonPkgutilTest.java

package unit;

import java.util.ArrayList;
import java.util.HashSet;
import java.util.List;
import java.util.Set;

/**
 * CWE-407 unit test: cpython-0001
 *
 * Models pkgutil.extend_path() portion deduplication.
 *
 * DEFECT: for each portion, scan the accumulator list with O(n) contains().
 *         Total cost: O(n²) for n unique portions.
 *
 * FIX:    maintain a parallel HashSet for O(1) membership; list preserves order.
 *         Total cost: O(n).
 *
 * Asserts: slowOps > fastOps * 10 at n=500 (actual ratio ≈ 250×).
 */
public class CpythonPkgutilTest {

    /** Mirrors the defective pkgutil.extend_path loop. Returns comparison count. */
    static long slow(int n) {
        List<String> path = new ArrayList<>();
        long ops = 0;
        for (int i = 0; i < n; i++) {
            String portion = "/opt/pkg" + i + "/ns";
            // O(path.size()) scan — mirrors `if portion not in path`
            boolean found = false;
            for (String existing : path) {
                ops++;
                if (existing.equals(portion)) {
                    found = true;
                    break;
                }
            }
            if (!found) {
                path.add(portion);
            }
        }
        return ops;
    }

    /** Mirrors the patched version: set for O(1) membership, list for order. */
    static long fast(int n) {
        List<String> path = new ArrayList<>();
        Set<String> pathSet = new HashSet<>();
        long ops = 0;
        for (int i = 0; i < n; i++) {
            String portion = "/opt/pkg" + i + "/ns";
            ops++; // one hash probe — mirrors `if portion not in path_set`
            if (!pathSet.contains(portion)) {
                path.add(portion);
                pathSet.add(portion);
            }
        }
        return ops;
    }

    public static void main(String[] args) {
        int passed = 0;
        int total = 0;

        // Test 1: slow is strictly more expensive than fast at n=100
        {
            total++;
            int n = 100;
            long sOps = slow(n);
            long fOps = fast(n);
            // Expected: sOps ≈ n*(n-1)/2 ≈ 4950; fOps = n = 100
            boolean ok = sOps > fOps * 10L;
            System.out.printf("Test 1 [n=100 slow=%d fast=%d ratio=%.1fx]: %s%n",
                    sOps, fOps, (double) sOps / fOps, ok ? "PASS" : "FAIL");
            if (ok) passed++;
        }

        // Test 2: slow is at least 50× more expensive than fast at n=500
        {
            total++;
            int n = 500;
            long sOps = slow(n);
            long fOps = fast(n);
            // Expected: sOps ≈ 125000; fOps = 500
            boolean ok = sOps > fOps * 50L;
            System.out.printf("Test 2 [n=500 slow=%d fast=%d ratio=%.1fx]: %s%n",
                    sOps, fOps, (double) sOps / fOps, ok ? "PASS" : "FAIL");
            if (ok) passed++;
        }

        // Test 3: slow is at least 200× more expensive than fast at n=1000
        {
            total++;
            int n = 1000;
            long sOps = slow(n);
            long fOps = fast(n);
            // Expected: sOps ≈ 500000; fOps = 1000
            boolean ok = sOps > fOps * 200L;
            System.out.printf("Test 3 [n=1000 slow=%d fast=%d ratio=%.1fx]: %s%n",
                    sOps, fOps, (double) sOps / fOps, ok ? "PASS" : "FAIL");
            if (ok) passed++;
        }

        // Test 4: both produce identical result sets (correctness)
        {
            total++;
            int n = 200;
            List<String> slowPath = new ArrayList<>();
            List<String> fastPath = new ArrayList<>();
            Set<String> fastSet = new HashSet<>();
            for (int i = 0; i < n; i++) {
                String p = "/opt/pkg" + i + "/ns";
                if (!slowPath.contains(p)) slowPath.add(p);
                if (!fastSet.contains(p)) { fastPath.add(p); fastSet.add(p); }
            }
            boolean ok = slowPath.equals(fastPath);
            System.out.printf("Test 4 [correctness n=200 equal=%b]: %s%n",
                    ok, ok ? "PASS" : "FAIL");
            if (ok) passed++;
        }

        System.out.printf("%d/%d PASS%n", passed, total);
        if (passed != total) System.exit(1);
    }
}