fix(llm-security): v8 Phase 1 — Berry lockfile, nested-v1 recursion, per-occurrence strip attribution

Three TDD-first fixes surviving the B8 roadmap bucket (v8.0.0-plan.local.md
Phase 1, items 1-3; item 4 JAR hardening scoped out at review):

- supply-chain-recheck.mjs parseYarnLock: ported the hook's per-entry parser
  (pre-install-supply-chain.mjs) so Berry's `version: x` format (unquoted) is
  recognized alongside Classic's `version "x"` — Berry lockfiles previously
  yielded zero deps, silently missing pinned compromised packages.
- supply-chain-recheck.mjs parsePackageLock: lockfileVersion-1 fallback now
  recurses nested `dependencies`, mirroring the hook's walk() — a transitive,
  non-hoisted compromised copy below the top level was previously invisible.
- content-extractor.mjs stripInjection: attribution moved from a global
  `Set<label>` to `Set<label::lineIndex>`. The old check silenced the
  unstripped flag for ANY occurrence of a label once ANY occurrence had been
  line-redacted, so a second, cross-line-only encoded occurrence of the same
  label survived into sanitized output without being flagged.

Full suite 2019/2019 (one known-flaky timing test confirmed green in isolation).

scanners/content-extractor.mjs

@@ -137,7 +137,11 @@ function stripInjection(text, file) {
   // Runs first so that line indices still line up with the original text.
   const lines = text.split('\n');
   const normalizedLines = isDifferent ? lines.map(l => normalizeForScan(l)) : [];
-  const attributed = new Set();
+  // Keyed by `${label}::${lineIndex}`, not just label — a label redacted on
+  // one line must not silence the unstripped check for a DIFFERENT
+  // occurrence of the same label elsewhere (e.g. a second, cross-line-only
+  // encoding Pass 1 couldn't catch). Attribution is judged per occurrence.
+  const attributedLines = new Set();
 
   if (isDifferent) {
     for (const { pattern, label } of allPatterns) {
@@ -147,7 +151,7 @@ function stripInjection(text, file) {
         if (lines[i].includes(STRIP_MARKER)) continue;
         if (toGlobal(pattern).test(normalizedLines[i])) {
           lines[i] = `[${STRIP_MARKER}: ${label}]`;
-          attributed.add(label);
+          attributedLines.add(`${label}::${i}`);
         }
       }
     }
@@ -164,8 +168,10 @@ function stripInjection(text, file) {
         const finding = { file, line, label, severity };
         const before = sanitized;
         sanitized = sanitized.replace(match[0], `[${STRIP_MARKER}: ${label}]`);
-        // Neither a literal replace nor a line redaction removed this one.
-        if (sanitized === before && !attributed.has(label)) finding.unstripped = true;
+        // Neither a literal replace nor a line redaction removed THIS occurrence.
+        if (sanitized === before && !attributedLines.has(`${label}::${line - 1}`)) {
+          finding.unstripped = true;
+        }
         findings.push(finding);
       }
     }

scanners/supply-chain-recheck.mjs

@@ -84,21 +84,32 @@ async function parsePackageLock(filePath) {
       }
     }
 
-    // v1 fallback: dependencies object
+    // v1 fallback: dependencies is a nested tree — recurse so a non-hoisted
+    // nested copy of a package is still found, mirroring the hook's walk()
+    // (pre-install-supply-chain.mjs scanNpmLockfile).
     if (deps.length === 0 && lock.dependencies) {
-      for (const [name, info] of Object.entries(lock.dependencies)) {
-        if (info.version) {
-          deps.push({ name, version: info.version, ecosystem: 'npm' });
+      const walk = (dependencies) => {
+        for (const [name, info] of Object.entries(dependencies)) {
+          if (info && info.version) {
+            deps.push({ name, version: info.version, ecosystem: 'npm' });
+          }
+          if (info && info.dependencies) walk(info.dependencies);
         }
-      }
+      };
+      walk(lock.dependencies);
     }
   } catch { /* parse error — skip */ }
   return deps;
 }
 
 /**
- * Parse yarn.lock (v1 format).
- * Extracts package name and resolved version from each entry.
+ * Parse yarn.lock — both Classic (v1) and Berry (v2+) formats.
+ * Parses per entry: an entry starts at column 0 with one or more specs
+ * ending in ':' ("pkg@range" / Berry '"pkg@npm:range"'), followed by an
+ * indented body carrying `version "x"` (Classic) or `version: x` (Berry).
+ * The version is associated with ITS OWN entry (no unanchored substring
+ * matching across entries), mirroring the hook's parser
+ * (pre-install-supply-chain.mjs scanNpmLockfile).
  * @param {string} filePath - Absolute path to yarn.lock
  * @returns {Promise<{ name: string, version: string, ecosystem: string }[]>}
  */
@@ -106,28 +117,21 @@ async function parseYarnLock(filePath) {
   const deps = [];
   try {
     const raw = await readFile(filePath, 'utf8');
-    const lines = raw.split('\n');
-    let currentPkg = null;
-
-    for (const line of lines) {
-      // Package header: "pkg@^1.0.0", "pkg@1.0.0:" or "@scope/pkg@^1.0.0":
-      if (!line.startsWith(' ') && !line.startsWith('#') && line.includes('@')) {
-        const trimmed = line.replace(/[":]/g, '').trim();
-        if (trimmed.startsWith('@')) {
-          // Scoped: @scope/pkg@version
-          const rest = trimmed.slice(1);
-          const atIdx = rest.indexOf('@');
-          if (atIdx > 0) currentPkg = '@' + rest.slice(0, atIdx);
-        } else {
-          const atIdx = trimmed.indexOf('@');
-          if (atIdx > 0) currentPkg = trimmed.slice(0, atIdx);
-        }
-      }
-      // Version line: "  version "1.2.3""
-      const versionMatch = line.match(/^\s+version\s+"([^"]+)"/);
-      if (versionMatch && currentPkg) {
-        deps.push({ name: currentPkg, version: versionMatch[1], ecosystem: 'npm' });
-        currentPkg = null;
+    for (const entry of raw.split(/\n(?=\S)/)) {
+      const nl = entry.indexOf('\n');
+      if (nl === -1) continue;
+      const header = entry.slice(0, nl).trim();
+      if (!header.endsWith(':') || header.startsWith('#')) continue;
+      const vMatch = entry.match(/^\s+version:?\s+"?([^"\s]+)"?\s*$/m);
+      const entryVersion = vMatch ? vMatch[1] : null;
+      if (!entryVersion) continue;
+      const names = new Set(header.slice(0, -1).split(',').map(part => {
+        const spec = part.trim().replace(/^"|"$/g, '');
+        const at = spec.indexOf('@', spec.startsWith('@') ? 1 : 0);
+        return at > 0 ? spec.slice(0, at) : spec;
+      }));
+      for (const name of names) {
+        deps.push({ name, version: entryVersion, ecosystem: 'npm' });
       }
     }
   } catch { /* parse error — skip */ }

tests/fixtures/supply-chain/yarn-berry-compromised.lock

@@ -0,0 +1,20 @@
+# This file is generated by running "yarn install" inside your project.
+# yarn lockfile v6
+
+__metadata:
+  version: 8
+  cacheKey: 10
+
+"event-stream@npm:^3.3.6":
+  version: 3.3.6
+  resolution: "event-stream@npm:3.3.6"
+  checksum: 0123456789abcdef
+  languageName: node
+  linkType: hard
+
+"lodash@npm:^4.17.21":
+  version: 4.17.21
+  resolution: "lodash@npm:4.17.21"
+  checksum: fedcba9876543210
+  languageName: node
+  linkType: hard

tests/scanners/content-extractor-strip.test.mjs

@@ -74,4 +74,37 @@ describe('content-extractor — stripInjection removes what it reports', () => {
     assert.match(sanitized, /keep-after/);
     assert.ok(!sanitized.includes(encoded));
   });
+
+  it('flags a second, cross-line occurrence of an already-stripped label as unstripped', () => {
+    // Two occurrences of the SAME label: the first is single-line encoded
+    // (caught + redacted by the per-line pass), the second is split across
+    // two lines so no individual line's decoded form matches — only the
+    // whole-text normalization does (the documented residual gap). A global
+    // `attributed.has(label)` check wrongly treats the second occurrence as
+    // handled because the FIRST one was; it must be judged on its own line
+    // span instead.
+    const singleLine = htmlEntities('ignore all previous');
+    const crossLineA = htmlEntities('ignore all');
+    const crossLineB = htmlEntities('previous');
+    const text = [
+      'keep-before',
+      singleLine,
+      'keep-middle',
+      crossLineA,
+      crossLineB,
+      'keep-after',
+      '',
+    ].join('\n');
+    const { sanitized, findings } = stripInjection(text, 'README.md');
+
+    assert.ok(!sanitized.includes(singleLine), 'the single-line occurrence must be stripped');
+    assert.ok(
+      sanitized.includes(crossLineA) && sanitized.includes(crossLineB),
+      'the cross-line occurrence is expected to survive stripping (documented residual gap)'
+    );
+    assert.ok(
+      findings.some(f => f.unstripped === true),
+      'a payload that survives into sanitized output must be flagged unstripped, not silently reported as handled'
+    );
+  });
 });

tests/scanners/supply-chain-recheck.test.mjs

@@ -117,6 +117,39 @@ describe('supply-chain-recheck: npm blocklist', () => {
     }
   });
 
+  it('detects compromised event-stream@3.3.6 nested under lockfileVersion 1 dependencies', async () => {
+    // lockfileVersion 1's `dependencies` is a nested tree — a transitive,
+    // non-hoisted copy sits under a parent's own `dependencies` object.
+    // parsePackageLock must recurse, mirroring the hook's walk()
+    // (pre-install-supply-chain.mjs #48).
+    if (existsSync(TEMP)) rmSync(TEMP, { recursive: true });
+    mkdirSync(TEMP, { recursive: true });
+    writeFileSync(join(TEMP, 'package-lock.json'), JSON.stringify({
+      name: 'sample',
+      lockfileVersion: 1,
+      dependencies: {
+        a: {
+          version: '1.0.0',
+          dependencies: {
+            'event-stream': { version: '3.3.6' },
+          },
+        },
+      },
+    }));
+    try {
+      const result = await scan(TEMP, { files: [] });
+      const compromised = result.findings.filter(
+        f => f.title.includes('Compromised') && f.title.includes('event-stream')
+      );
+      assert.ok(
+        compromised.length >= 1,
+        `Expected compromised finding for nested event-stream, got ${result.findings.map(f => f.title).join('; ')}`
+      );
+    } finally {
+      cleanupTemp();
+    }
+  });
+
   it('does not flag clean packages in package-lock.json', async () => {
     setupTemp({ 'package-lock.json': 'package-lock-clean.json' });
     try {
@@ -140,6 +173,26 @@ describe('supply-chain-recheck: npm blocklist', () => {
       cleanupTemp();
     }
   });
+
+  it('detects compromised event-stream@3.3.6 in a Berry yarn.lock (version: x format)', async () => {
+    // Berry (yarn v2+) writes `version: x` with no quotes, unlike Classic's
+    // `version "x"`. parseYarnLock must recognize both, mirroring the hook's
+    // parser (pre-install-supply-chain.mjs #17).
+    setupTemp({ 'yarn.lock': 'yarn-berry-compromised.lock' });
+    try {
+      const result = await scan(TEMP, { files: [] });
+      const compromised = result.findings.filter(
+        f => f.title.includes('Compromised') && f.title.includes('event-stream')
+      );
+      assert.ok(
+        compromised.length >= 1,
+        `Expected compromised finding for event-stream, got ${result.findings.map(f => f.title).join('; ')}`
+      );
+      assert.equal(compromised[0].severity, 'critical');
+    } finally {
+      cleanupTemp();
+    }
+  });
 });
 
 // ============================================================================