@blamejs/blamejs-shop 0.4.31 → 0.4.32

package/lib/vendor/blamejs/test/layer-0-primitives/safe-url-canonicalize.test.js

@@ -0,0 +1,309 @@
+"use strict";
+/**
+ * b.safeUrl.canonicalize — URL/host canonicalizer for safe comparison and
+ * SSRF-allowlist defense.
+ *
+ * Every obfuscated spelling of one destination must reduce to one string so a
+ * host allowlist / dedup key / SSRF pre-check compares equal. The adversarial
+ * cases below are the exact shapes an attacker reaches for to slip a naive
+ * `===` allowlist: IPv4 in octal / hex / decimal-int / shorthand, IPv4-mapped
+ * and zero-compressed IPv6, IDN homographs, trailing-dot and default-port and
+ * case variations, and over-encoded path bytes.
+ *
+ * Source-discipline: NO attack characters are typed as literals. The one
+ * mixed-script case is expressed as its already-encoded punycode A-label
+ * (`xn--ggle-55da.com`, same fixture the homograph suite uses) so this file
+ * stays pure ASCII.
+ */
+
+var b     = require("../..");
+var check = require("../helpers/check").check;
+
+// ---- Equivalence classes: obfuscated loopback collapses to one string ----
+
+function testIpv4LoopbackEquivalenceClass() {
+  // Decimal-dotted, octal, hex-word, decimal-int, and shorthand spellings of
+  // 127.0.0.1 — in the OLD world (raw string compare, no canonicalizer) every
+  // pair below compared UNEQUAL, so an allowlist of "127.0.0.1" let them all
+  // through. canonicalize collapses them to one form.
+  var forms = [
+    "http://127.0.0.1/",
+    "http://0177.0.0.1/",      // octal first octet
+    "http://0x7f.1/",          // hex octet + shorthand
+    "http://0x7f000001/",      // single hex dword
+    "http://2130706433/",      // single decimal dword
+    "http://127.1/",           // a.d shorthand
+  ];
+  var expected = "http://127.0.0.1/";
+  for (var i = 0; i < forms.length; i += 1) {
+    var got = b.safeUrl.canonicalize(forms[i]);
+    check("ipv4 loopback form '" + forms[i] + "' canonicalizes to " + expected,
+          got === expected);
+    // The pairwise equality the allowlist actually relies on.
+    check("ipv4 loopback '" + forms[i] + "' === canonical '127.0.0.1' form",
+          b.safeUrl.canonicalize(forms[i]) === b.safeUrl.canonicalize(forms[0]));
+    // Prove the test is meaningful: the RAW strings were NOT equal.
+    if (forms[i] !== forms[0]) {
+      check("raw '" + forms[i] + "' !== raw canonical (old-world bypass shape)",
+            forms[i] !== forms[0]);
+    }
+  }
+}
+
+function testIpv6MappedEquivalenceClass() {
+  // IPv4-mapped IPv6 in dotted, hex, and fully-expanded spellings — all the
+  // same 16 bytes, all collapse to one bracketed RFC 5952 form.
+  var forms = [
+    "http://[::ffff:127.0.0.1]/",
+    "http://[::ffff:7f00:1]/",
+    "http://[0:0:0:0:0:ffff:7f00:1]/",
+    "http://[0:0:0:0:0:FFFF:7F00:1]/",   // mixed-case hex
+  ];
+  var first = b.safeUrl.canonicalize(forms[0]);
+  check("ipv4-mapped IPv6 canonical is bracketed RFC 5952",
+        first === "http://[::ffff:7f00:1]/");
+  for (var i = 1; i < forms.length; i += 1) {
+    check("ipv4-mapped form '" + forms[i] + "' === first canonical",
+          b.safeUrl.canonicalize(forms[i]) === first);
+    check("raw '" + forms[i] + "' !== raw '" + forms[0] + "' (old-world unequal)",
+          forms[i] !== forms[0]);
+  }
+}
+
+function testIpv6ZeroCompressionEquivalenceClass() {
+  // ::1 in compressed, fully-expanded, and zero-padded spellings.
+  var forms = [
+    "http://[::1]/",
+    "http://[0:0:0:0:0:0:0:1]/",
+    "http://[0000:0000:0000:0000:0000:0000:0000:0001]/",
+  ];
+  var first = b.safeUrl.canonicalize(forms[0]);
+  check("IPv6 loopback canonical is [::1]", first === "http://[::1]/");
+  for (var i = 1; i < forms.length; i += 1) {
+    check("IPv6 loopback form '" + forms[i] + "' === [::1] canonical",
+          b.safeUrl.canonicalize(forms[i]) === first);
+    check("raw '" + forms[i] + "' !== raw '[::1]' (old-world unequal)",
+          forms[i] !== forms[0]);
+  }
+  // A documentation address with an INTERIOR zero run picks the longest run
+  // (RFC 5952 §4.2.3) and stays lower-hex.
+  check("interior zero-run compresses correctly",
+        b.safeUrl.canonicalize("https://[2001:0DB8:0:0:0:0:0:1]/") ===
+        "https://[2001:db8::1]/");
+}
+
+// ---- IP-literal cross-check against the SSRF classifier's byte parser ----
+
+function testIpCanonicalAgreesWithClassifier() {
+  // The canonical host string and the SSRF verdict are derived from the SAME
+  // bytes — so the canonical loopback string classifies as loopback and the
+  // canonical mapped form re-classifies its embedded v4 as loopback too.
+  check("canonical 127.0.0.1 classifies loopback",
+        b.ssrfGuard.classify("127.0.0.1") === "loopback");
+  // canonicalizeHost is the IP-byte + case layer: it byte-canonicalizes
+  // net.isIP-recognized literals (the numeric-base decode of 0x.../octal/dword
+  // is the WHATWG-URL-parser layer that safeUrl.canonicalize runs first).
+  check("ssrfGuard.canonicalizeHost passes an already-canonical dotted-quad",
+        b.ssrfGuard.canonicalizeHost("127.0.0.1") === "127.0.0.1");
+  check("ssrfGuard.canonicalizeHost collapses expanded v6 to ::1",
+        b.ssrfGuard.canonicalizeHost("[0:0:0:0:0:0:0:1]") === "::1");
+  check("ssrfGuard.canonicalizeHost compresses + lower-hexes a mixed-case v6",
+        b.ssrfGuard.canonicalizeHost("2001:0DB8:0:0:0:0:0:1") === "2001:db8::1");
+  check("ssrfGuard.canonicalizeHost lowercases + strips trailing dot",
+        b.ssrfGuard.canonicalizeHost("Example.COM.") === "example.com");
+}
+
+// ---- IDN: A-label emission + homograph refusal ----
+
+function testIdnEmittedAsPunycode() {
+  // A pure-Cyrillic+Latin-tld host parses (single script per label) and is
+  // emitted as its xn-- A-label, never the Unicode form.
+  var got = b.safeUrl.canonicalize("https://xn--80akhbyknj4f.com/");
+  check("IDN host emitted as xn-- A-label",
+        got === "https://xn--80akhbyknj4f.com/");
+}
+
+function testConfusableHostThrows() {
+  // xn--ggle-55da.com decodes to a label mixing Cyrillic + Latin (the
+  // gооgle.com homograph). canonicalize must THROW, never silently pass it.
+  var threw = false;
+  var code = null;
+  try { b.safeUrl.canonicalize("https://xn--ggle-55da.com/"); }
+  catch (e) { threw = true; code = e.code; }
+  check("confusable / mixed-script host throws", threw === true);
+  check("confusable host throw carries safe-url/idn-homograph",
+        code === "safe-url/idn-homograph");
+}
+
+function testMixedScriptOptInCanonicalizes() {
+  // Operators with a legitimate non-Latin host opt in; the A-label is then
+  // canonicalized rather than refused.
+  var got = b.safeUrl.canonicalize("https://xn--ggle-55da.com/", {
+    allowMixedScript: true,
+  });
+  check("allowMixedScript:true canonicalizes the mixed-script host",
+        got === "https://xn--ggle-55da.com/");
+}
+
+// ---- scheme / host case, default port, trailing dot, path ----
+
+function testEndToEndNormalization() {
+  // The headline end-to-end case from the spec: case-folded host, stripped
+  // default port, resolved `..`, all in one.
+  check("https://Example.COM:443/a/../b -> https://example.com/b",
+        b.safeUrl.canonicalize("https://Example.COM:443/a/../b") ===
+        "https://example.com/b");
+  check("http default :80 stripped + host lowercased",
+        b.safeUrl.canonicalize("http://Example.COM:80/") ===
+        "http://example.com/");
+  check("trailing-dot host removed",
+        b.safeUrl.canonicalize("https://example.com./path") ===
+        "https://example.com/path");
+  check("non-default port preserved",
+        b.safeUrl.canonicalize("https://example.com:8443/") ===
+        "https://example.com:8443/");
+}
+
+function testPathPercentNormalization() {
+  // RFC 3986 §6.2.2: escapes of unreserved chars are decoded, hex digits
+  // uppercased — in the PATH only.
+  check("path %7E decodes to ~ and %2D to -",
+        b.safeUrl.canonicalize("https://example.com/%7Euser/a%2Db") ===
+        "https://example.com/~user/a-b");
+  // A reserved char (%2F = '/') is NOT decoded (would change structure); its
+  // hex is uppercased.
+  check("reserved %2f path escape uppercased, NOT decoded",
+        b.safeUrl.canonicalize("https://example.com/a%2fb") ===
+        "https://example.com/a%2Fb");
+  // Query semantics are conservative: a %2D in the QUERY is left verbatim
+  // (lowercase, undecoded) so value semantics are never altered.
+  check("query percent-escape left byte-for-byte",
+        b.safeUrl.canonicalize("https://example.com/?x=%2d") ===
+        "https://example.com/?x=%2d");
+}
+
+// ---- throw paths: disallowed scheme + unparseable + scheme allowlist ----
+
+function testDisallowedSchemeThrows() {
+  var threw = false;
+  var code = null;
+  try { b.safeUrl.canonicalize("ftp://example.com/file"); }
+  catch (e) { threw = true; code = e.code; }
+  check("ftp scheme throws", threw === true);
+  check("disallowed-scheme code is safe-url/protocol-disallowed",
+        code === "safe-url/protocol-disallowed");
+}
+
+function testUnparseableThrows() {
+  var threw = false;
+  var code = null;
+  try { b.safeUrl.canonicalize("this is not a url"); }
+  catch (e) { threw = true; code = e.code; }
+  check("unparseable input throws", threw === true);
+  check("unparseable code is safe-url/malformed", code === "safe-url/malformed");
+}
+
+function testMissingInputThrows() {
+  var threw = false;
+  var code = null;
+  try { b.safeUrl.canonicalize(""); }
+  catch (e) { threw = true; code = e.code; }
+  check("empty input throws safe-url/missing", threw === true && code === "safe-url/missing");
+}
+
+function testCustomSchemeAllowlist() {
+  // Narrow the allowlist: https-only refuses an http URL.
+  var threw = false;
+  var code = null;
+  try {
+    b.safeUrl.canonicalize("http://example.com/", {
+      allowedSchemes: b.safeUrl.ALLOW_HTTP_TLS,
+    });
+  } catch (e) { threw = true; code = e.code; }
+  check("custom allowedSchemes (https-only) refuses http",
+        threw === true && code === "safe-url/protocol-disallowed");
+  // ws canonicalizes under the default ALLOW_ANY.
+  check("ws scheme canonicalizes under default allowlist",
+        b.safeUrl.canonicalize("ws://Example.COM:80/s") === "ws://example.com/s");
+}
+
+function testIdempotence() {
+  // canonicalize(canonicalize(x)) === canonicalize(x) for the whole adversarial
+  // set — a canonical form must be a fixed point.
+  var inputs = [
+    "http://0177.0.0.1/",
+    "http://[0:0:0:0:0:ffff:7f00:1]/",
+    "https://Example.COM:443/a/../b",
+    "https://example.com/%7Euser",
+    "https://xn--80akhbyknj4f.com/",
+  ];
+  for (var i = 0; i < inputs.length; i += 1) {
+    var once = b.safeUrl.canonicalize(inputs[i]);
+    var twice = b.safeUrl.canonicalize(once);
+    check("canonicalize is idempotent for '" + inputs[i] + "'", once === twice);
+  }
+}
+
+function testUncanonicalizableCodeIsRegistered() {
+  // The post-parse fallback code is reachable as a documented SafeUrlError
+  // code — constructing it directly proves the class accepts it.
+  var e = new b.safeUrl.SafeUrlError("safe-url/uncanonicalizable", "x");
+  check("safe-url/uncanonicalizable is a valid SafeUrlError code",
+        e.code === "safe-url/uncanonicalizable" && e.isSafeUrlError === true);
+}
+
+// ---- Credentials are never carried into the canonical form ----
+
+function testUserinfoDroppedFromCanonicalForm() {
+  // The canonical string is built to be compared, used as a dedup / cache key,
+  // or logged — it must never carry user:pass credentials, and the creds are
+  // not part of the target identity for an allowlist / SSRF decision. parse()
+  // refuses userinfo by default; even opted-in, the canonical output omits it.
+  var user = "alice";
+  var token = "s3cr" + "et-pw-9f3a";          // split so no secret-shaped literal sits in source
+  var withCreds = "https://" + user + ":" + token + "@host.example.com/p";
+
+  var deniedCode = null;
+  try { b.safeUrl.canonicalize(withCreds); }
+  catch (e) { deniedCode = e && e.code; }
+  check("canonicalize refuses userinfo by default",
+        deniedCode === "safe-url/userinfo-disallowed");
+
+  var canon = b.safeUrl.canonicalize(withCreds, { allowUserinfo: true });
+  check("canonicalize drops the userinfo delimiter from the canonical form",
+        canon.indexOf("@") === -1);
+  check("canonicalize does not carry the password into the canonical form",
+        canon.indexOf(token) === -1);
+  check("canonicalize does not carry the username into the canonical form",
+        canon.indexOf("//" + user) === -1);
+  check("the credential-stripped canonical form is the bare target",
+        canon === "https://host.example.com/p");
+  check("URLs differing only in credentials canonicalize equal",
+        b.safeUrl.canonicalize("https://x:y@host.example.com/p", { allowUserinfo: true }) === canon);
+}
+
+async function run() {
+  testUserinfoDroppedFromCanonicalForm();
+  testIpv4LoopbackEquivalenceClass();
+  testIpv6MappedEquivalenceClass();
+  testIpv6ZeroCompressionEquivalenceClass();
+  testIpCanonicalAgreesWithClassifier();
+  testIdnEmittedAsPunycode();
+  testConfusableHostThrows();
+  testMixedScriptOptInCanonicalizes();
+  testEndToEndNormalization();
+  testPathPercentNormalization();
+  testDisallowedSchemeThrows();
+  testUnparseableThrows();
+  testMissingInputThrows();
+  testCustomSchemeAllowlist();
+  testIdempotence();
+  testUncanonicalizableCodeIsRegistered();
+}
+
+module.exports = { run: run };
+
+if (require.main === module) {
+  run().then(function () { console.log("OK"); })
+       .catch(function (e) { console.error(e.stack || e); process.exit(1); });
+}

package/lib/vendor/blamejs/test/layer-0-primitives/safe-xml.test.js

@@ -0,0 +1,143 @@
+"use strict";
+/**
+ * b.parsers.xml.parse — security-focused XML parser.
+ *
+ * Covers the structural defaults (XXE / DOCTYPE / billion-laughs / depth
+ * + element + attribute caps) and the prototype-pollution posture: the
+ * result tree and every nested object it contains carry a null prototype,
+ * and element / attribute names equal to __proto__ / constructor /
+ * prototype are rejected with xml/forbidden-name (CWE-1321 / OWASP
+ * prototype-pollution).
+ *
+ * Run standalone: `node test/layer-0-primitives/safe-xml.test.js`
+ */
+
+var helpers = require("../helpers");
+var b       = helpers.b;
+var check   = helpers.check;
+
+var xml = b.parsers.xml;
+
+// ---- Baseline parse + shape ----
+
+function testParsesAttributedElement() {
+  var result = xml.parse('<root id="x"><child>text</child></root>');
+  check("xml.parse returns the root key",
+        result && result.root && result.root["@attrs"] && result.root["@attrs"].id === "x");
+  check("xml.parse groups child element",
+        result.root.child === "text");
+}
+
+function testRejectsDoctype() {
+  var threw = null;
+  try { xml.parse('<!DOCTYPE x SYSTEM "file:///etc/passwd"><root/>'); }
+  catch (e) { threw = e; }
+  check("xml.parse rejects DOCTYPE (XXE defense)",
+        threw && threw.code === "xml/doctype");
+}
+
+function testRejectsCustomEntity() {
+  var threw = null;
+  try { xml.parse("<root>&xxe;</root>"); }
+  catch (e) { threw = e; }
+  check("xml.parse rejects custom entity (XXE defense)",
+        threw && threw.code === "xml/external-entity");
+}
+
+// ---- Prototype-pollution posture (CWE-1321) ----
+
+function testResultTreeHasNullPrototype() {
+  // A document that exercises every keyed accumulator: attributes (attrs),
+  // grouped child elements (grouped / obj), and the element-name wrapper
+  // (out). Each must carry a null prototype so a consumer reading an
+  // absent key sees undefined, never an inherited Object member.
+  var result = xml.parse(
+    '<root attr="v"><a>1</a><a>2</a><b><c>deep</c></b></root>'
+  );
+
+  check("result wrapper has null prototype",
+        Object.getPrototypeOf(result) === null);
+  check("element object has null prototype",
+        Object.getPrototypeOf(result.root) === null);
+  check("@attrs object has null prototype",
+        Object.getPrototypeOf(result.root["@attrs"]) === null);
+  check("nested element object has null prototype",
+        Object.getPrototypeOf(result.root.b) === null);
+  check("repeated child became an array",
+        Array.isArray(result.root.a) && result.root.a.length === 2);
+
+  // No inherited Object.prototype members leak through any accumulator —
+  // a plain {} would surface a function here.
+  check("no inherited toString on result",
+        result.toString === undefined);
+  check("no inherited hasOwnProperty on element object",
+        result.root.hasOwnProperty === undefined);
+  check("no inherited constructor on @attrs object",
+        result.root["@attrs"].constructor === undefined);
+}
+
+function testRejectsProtoAttributeName() {
+  var threw = null;
+  try { xml.parse('<root __proto__="polluted"/>'); }
+  catch (e) { threw = e; }
+  check("xml.parse rejects __proto__ attribute name",
+        threw && threw.code === "xml/forbidden-name");
+  // The global Object.prototype was not mutated either way.
+  check("Object.prototype not polluted by __proto__ attribute",
+        ({}).polluted === undefined);
+}
+
+function testRejectsProtoElementName() {
+  var threw = null;
+  try { xml.parse("<__proto__><x>y</x></__proto__>"); }
+  catch (e) { threw = e; }
+  check("xml.parse rejects __proto__ element name",
+        threw && threw.code === "xml/forbidden-name");
+}
+
+function testRejectsConstructorElementName() {
+  var threw = null;
+  try { xml.parse("<root><constructor>x</constructor></root>"); }
+  catch (e) { threw = e; }
+  check("xml.parse rejects constructor child element name",
+        threw && threw.code === "xml/forbidden-name");
+}
+
+function testRejectsPrototypeElementName() {
+  var threw = null;
+  try { xml.parse("<prototype/>"); }
+  catch (e) { threw = e; }
+  check("xml.parse rejects prototype element name",
+        threw && threw.code === "xml/forbidden-name");
+}
+
+function testConstructorAttributeNoFalseDuplicate() {
+  // With a plain-object accumulator, attrs["constructor"] would read the
+  // inherited Object constructor (not undefined) and a single occurrence
+  // would falsely trip the duplicate-attribute guard. The forbidden-name
+  // rejection now fires first; assert the code is forbidden-name, not
+  // duplicate-attr, so the diagnostic points at the real cause.
+  var threw = null;
+  try { xml.parse('<root constructor="once"/>'); }
+  catch (e) { threw = e; }
+  check("constructor attribute → forbidden-name (not false duplicate-attr)",
+        threw && threw.code === "xml/forbidden-name");
+}
+
+(function run() {
+  try {
+    testParsesAttributedElement();
+    testRejectsDoctype();
+    testRejectsCustomEntity();
+    testResultTreeHasNullPrototype();
+    testRejectsProtoAttributeName();
+    testRejectsProtoElementName();
+    testRejectsConstructorElementName();
+    testRejectsPrototypeElementName();
+    testConstructorAttributeNoFalseDuplicate();
+  } catch (e) {
+    console.error(e);
+    process.exit(1);
+  }
+  console.log("OK — safe-xml tests");
+})();