Canonicalize the request path before tripwire matching (#57)

packages/webdecoy/src/rules/tripwire-rule.ts

@@ -41,9 +41,77 @@ export const DEFAULT_TRIPWIRE_PATHS: readonly string[] = [
   '/.vscode/sftp.json',
 ];
 
-/** Strip query string and fragment for path matching. */
-function normalizePath(path: string): string {
-  return path.split('?')[0].split('#')[0];
+/**
+ * Canonicalize a request path before matching a hidden-path tripwire.
+ *
+ * A scanner probing for `/.env` or `/.git/config` can dress the path up —
+ * `//.env`, `/x/..%2f.git/config`, `/%2Eenv` — and a webserver still resolves it
+ * to the file it was hunting for. Matching the raw string would miss the scan,
+ * which is the one thing a tripwire exists to catch. So the query and fragment
+ * are dropped and the path is reduced the way a server would: percent-decode
+ * ASCII bytes, collapse duplicate slashes, and resolve `.` and `..` segments,
+ * keeping a single trailing slash.
+ */
+export function canonicalizeTripwirePath(path: string): string {
+  const noQuery = path.split('?')[0].split('#')[0];
+  return cleanSlashesAndDots(decodePercentAscii(noQuery));
+}
+
+function decodePercentAscii(s: string): string {
+  // Bounded, so a double-encoded separator (`%252F` -> `%2F` -> `/`) resolves
+  // without unbounded work. ASCII bytes only: high bytes stay encoded rather
+  // than risk throwing (decodeURIComponent) on invalid UTF-8.
+  for (let i = 0; i < 4; i++) {
+    const next = decodePercentAsciiOnce(s);
+    if (next === s) break;
+    s = next;
+  }
+  return s;
+}
+
+function decodePercentAsciiOnce(s: string): string {
+  if (!s.includes('%')) return s;
+  let out = '';
+  for (let i = 0; i < s.length; i++) {
+    if (s[i] === '%' && i + 2 < s.length) {
+      const hi = unhexNibble(s.charCodeAt(i + 1));
+      const lo = unhexNibble(s.charCodeAt(i + 2));
+      if (hi >= 0 && lo >= 0) {
+        const v = (hi << 4) | lo;
+        if (v < 0x80) {
+          out += String.fromCharCode(v);
+          i += 2;
+          continue;
+        }
+      }
+    }
+    out += s[i];
+  }
+  return out;
+}
+
+function unhexNibble(c: number): number {
+  if (c >= 48 && c <= 57) return c - 48; // 0-9
+  if (c >= 97 && c <= 102) return c - 97 + 10; // a-f
+  if (c >= 65 && c <= 70) return c - 65 + 10; // A-F
+  return -1;
+}
+
+function cleanSlashesAndDots(s: string): string {
+  if (s === '') return s;
+  const hadTrailing = s.length > 1 && s[s.length - 1] === '/';
+  const stack: string[] = [];
+  for (const p of s.split('/')) {
+    if (p === '' || p === '.') continue;
+    if (p === '..') {
+      if (stack.length > 0) stack.pop();
+      continue;
+    }
+    stack.push(p);
+  }
+  let res = '/' + stack.join('/');
+  if (hadTrailing && res !== '/') res += '/';
+  return res;
 }
 
 export class TripwireRule implements Rule {
@@ -57,15 +125,17 @@ export class TripwireRule implements Rule {
   constructor(config: TripwireConfig = {}) {
     const paths = [...(config.paths ?? [])];
     if (config.includeDefaults ?? true) paths.push(...DEFAULT_TRIPWIRE_PATHS);
-    this.exact = new Set(paths);
-    this.prefixes = config.prefixes ?? [];
+    // Canonicalize the configured paths too, so both sides of the match are
+    // reduced the same way and a decoy written as `/foo/../bar` still lines up.
+    this.exact = new Set(paths.map(canonicalizeTripwirePath));
+    this.prefixes = (config.prefixes ?? []).map(canonicalizeTripwirePath);
     this.patterns = config.patterns ?? [];
     this.action = config.action ?? 'DENY';
     this.dryRun = config.dryRun ?? false;
   }
 
   evaluate(context: RuleContext): RuleResult {
-    const path = normalizePath(context.path);
+    const path = canonicalizeTripwirePath(context.path);
     const hit =
       this.exact.has(path) ||
       this.prefixes.some((prefix) => path.startsWith(prefix)) ||

packages/webdecoy/src/rules/tripwire.test.ts

@@ -36,6 +36,28 @@ describe('TripwireRule', () => {
     expect(rule.evaluate(ctx('/.git/config#x')).action).toBe('DENY');
   });
 
+  it('catches a scan that obfuscates the decoy path (#57)', () => {
+    const rule = new TripwireRule();
+    // A server resolves each of these to a decoy path; matching the raw string
+    // would miss the scan.
+    for (const p of [
+      '//.env', // duplicate leading slash
+      '/%2Eenv', // percent-encoded dot
+      '/%252Eenv', // double-encoded dot
+      '/static/..%2f.git/config', // traversal with an encoded slash
+      '/x/../.env', // plain traversal
+    ]) {
+      expect(rule.evaluate(ctx(p)).action).toBe('DENY');
+    }
+  });
+
+  it('does not trip on a legitimate path that merely resembles a decoy (#57)', () => {
+    const rule = new TripwireRule();
+    for (const p of ['/environment', '/env/config', '/assets/env.js']) {
+      expect(rule.evaluate(ctx(p)).action).toBe('ALLOW');
+    }
+  });
+
   it('respects includeDefaults: false', () => {
     const rule = new TripwireRule({ paths: ['/trap'], includeDefaults: false });
     expect(rule.evaluate(ctx('/.env')).action).toBe('ALLOW');