test(node/url): add more node:test parallel cases (#16404)

src/bun.js/bindings/BunObject.cpp

@@ -604,23 +604,67 @@ JSC_DEFINE_HOST_FUNCTION(functionFileURLToPath, (JSC::JSGlobalObject * globalObj
             url = WTF::URL(arg0.toWTFString(globalObject));
             RETURN_IF_EXCEPTION(scope, {});
         } else {
-            throwTypeError(globalObject, scope, "Argument must be a URL"_s);
+            Bun::ERR::INVALID_ARG_TYPE(scope, globalObject, "url"_s, "string"_s, arg0);
             return {};
         }
     } else {
         url = domURL->href();
     }
 
+    /// cannot turn non-`file://` URLs into file paths
     if (UNLIKELY(!url.protocolIsFile())) {
-        throwTypeError(globalObject, scope, "Argument must be a file URL"_s);
+        Bun::ERR::INVALID_URL_SCHEME(scope, globalObject, "file"_s);
         return {};
     }
 
+// NOTE: On Windows, WTF::URL::fileSystemPath will handle UNC paths
+// (`file:\\server\share\etc` -> `\\server\share\etc`), so hostname check only
+// needs to happen on posix systems
+#if !OS(WINDOWS)
+    // file://host/path is illegal if `host` is not `localhost`.
+    // Should be `file:///` instead
+    if (UNLIKELY(url.host().length() > 0 && url.host() != "localhost"_s)) {
+
+#if OS(DARWIN)
+        Bun::ERR::INVALID_FILE_URL_HOST(scope, globalObject, "darwin"_s);
+        return {};
+#else
+        Bun::ERR::INVALID_FILE_URL_HOST(scope, globalObject, "linux"_s);
+        return {};
+#endif
+    }
+#endif
+
+    // ban url-encoded slashes. '/' on posix, '/' and '\' on windows.
+    StringView p = url.path();
+    if (p.length() > 3) {
+        for (int i = 0; i < p.length() - 2; i++) {
+            if (p[i] == '%') {
+                const char second = p[i + 1];
+                const uint8_t third = p[i + 2] | 0x20;
+#if OS(WINDOWS)
+                if (
+                    (second == '2' && third == 102) || // 2f 2F    '/'
+                    (second == '5' && third == 99) // 5c 5C    '\'
+                ) {
+                    Bun::ERR::INVALID_FILE_URL_PATH(scope, globalObject, "must not include encoded \\ or / characters"_s);
+                    return {};
+                }
+#else
+                if (second == '2' && third == 102) {
+                    Bun::ERR::INVALID_FILE_URL_PATH(scope, globalObject, "must not include encoded / characters"_s);
+                    return {};
+                }
+#endif
+            }
+        }
+    }
+
     auto fileSystemPath = url.fileSystemPath();
 
 #if OS(WINDOWS)
     if (!isAbsolutePath(fileSystemPath)) {
-        throwTypeError(globalObject, scope, "File URL path must be absolute"_s);
+        Bun::ERR::INVALID_FILE_URL_PATH(scope, globalObject, "must be an absolute path"_s);
         return {};
     }
 #endif

src/bun.js/bindings/ErrorCode.cpp

@@ -567,6 +567,32 @@ JSC::EncodedJSValue INVALID_ARG_VALUE(JSC::ThrowScope& throwScope, JSC::JSGlobal
     return {};
 }
 
+JSC::EncodedJSValue INVALID_URL_SCHEME(JSC::ThrowScope& throwScope, JSC::JSGlobalObject* globalObject, const WTF::String& expectedScheme)
+{
+    auto message = makeString("The URL must be of scheme "_s, expectedScheme);
+    throwScope.throwException(globalObject, createError(globalObject, ErrorCode::ERR_INVALID_URL_SCHEME, message));
+    return {};
+}
+JSC::EncodedJSValue INVALID_FILE_URL_HOST(JSC::ThrowScope& throwScope, JSC::JSGlobalObject* globalObject, const WTF::String& platform)
+{
+    auto message = makeString("File URL host must be \"localhost\" or empty on "_s, platform);
+    throwScope.throwException(globalObject, createError(globalObject, ErrorCode::ERR_INVALID_FILE_URL_HOST, message));
+    return {};
+}
+JSC::EncodedJSValue INVALID_FILE_URL_HOST(JSC::ThrowScope& throwScope, JSC::JSGlobalObject* globalObject, const ASCIILiteral platform)
+{
+    auto message = makeString("File URL host must be \"localhost\" or empty on "_s, platform);
+    throwScope.throwException(globalObject, createError(globalObject, ErrorCode::ERR_INVALID_FILE_URL_HOST, message));
+    return {};
+}
+/// `File URL path {suffix}`
+JSC::EncodedJSValue INVALID_FILE_URL_PATH(JSC::ThrowScope& throwScope, JSC::JSGlobalObject* globalObject, const ASCIILiteral suffix)
+{
+    auto message = makeString("File URL path "_s, suffix);
+    throwScope.throwException(globalObject, createError(globalObject, ErrorCode::ERR_INVALID_FILE_URL_PATH, message));
+    return {};
+}
+
 JSC::EncodedJSValue UNKNOWN_ENCODING(JSC::ThrowScope& throwScope, JSC::JSGlobalObject* globalObject, const WTF::StringView encoding)
 {
     auto message = makeString("Unknown encoding: "_s, encoding);
@@ -655,7 +681,7 @@ JSC::EncodedJSValue CRYPTO_JWK_UNSUPPORTED_CURVE(JSC::ThrowScope& throwScope, JS
     return {};
 }
 
-}
+} // namespace ERR
 
 static JSC::JSValue ERR_INVALID_ARG_TYPE(JSC::ThrowScope& scope, JSC::JSGlobalObject* globalObject, JSValue arg0, JSValue arg1, JSValue arg2)
 {

src/bun.js/bindings/ErrorCode.h

@@ -1,3 +1,4 @@
+// To add a new error code, put it in ErrorCode.ts
 #pragma once
 
 #include "ZigGlobalObject.h"
@@ -86,6 +87,15 @@ JSC::EncodedJSValue ASSERTION(JSC::ThrowScope& throwScope, JSC::JSGlobalObject*
 JSC::EncodedJSValue CRYPTO_INVALID_CURVE(JSC::ThrowScope& throwScope, JSC::JSGlobalObject* globalObject);
 JSC::EncodedJSValue CRYPTO_JWK_UNSUPPORTED_CURVE(JSC::ThrowScope&, JSC::JSGlobalObject*, const WTF::String&);
 
+// URL
+
+/// `URL must be of scheme {expectedScheme}`
+JSC::EncodedJSValue INVALID_URL_SCHEME(JSC::ThrowScope& throwScope, JSC::JSGlobalObject* globalObject, const WTF::String& expectedScheme);
+/// `File URL host must be "localhost" or empty on {platform}`
+JSC::EncodedJSValue INVALID_FILE_URL_HOST(JSC::ThrowScope& throwScope, JSC::JSGlobalObject* globalObject, const WTF::String& platform);
+/// `File URL path {suffix}`
+JSC::EncodedJSValue INVALID_FILE_URL_PATH(JSC::ThrowScope& throwScope, JSC::JSGlobalObject* globalObject, const ASCIILiteral suffix);
+
 }
 
 void throwBoringSSLError(JSC::VM& vm, JSC::ThrowScope& scope, JSGlobalObject* globalObject, int errorCode);

src/bun.js/bindings/ErrorCode.ts

@@ -2,11 +2,11 @@
 type ErrorCodeMapping = Array<
   [
     /** error.code  */
-    string,
+    code: string,
     /** Constructor **/
-    typeof TypeError | typeof RangeError | typeof Error | typeof SyntaxError,
+    ctor: typeof TypeError | typeof RangeError | typeof Error | typeof SyntaxError,
     /** error.name. Defaults to `Constructor.name` (that is, mapping[1].name  */
-    string?,
+    name?: string,
     (typeof TypeError | typeof RangeError | typeof Error | typeof SyntaxError)?,
     (typeof TypeError | typeof RangeError | typeof Error | typeof SyntaxError)?,
   ]
@@ -43,7 +43,6 @@ const errors: ErrorCodeMapping = [
   ["ERR_INVALID_STATE", Error, undefined, TypeError, RangeError],
   ["ERR_INVALID_THIS", TypeError],
   ["ERR_INVALID_URI", URIError],
-  ["ERR_INVALID_URL", TypeError],
   ["ERR_IPC_CHANNEL_CLOSED", Error],
   ["ERR_IPC_DISCONNECTED", Error],
   ["ERR_IPC_ONE_PIPE", Error],
@@ -64,16 +63,16 @@ const errors: ErrorCodeMapping = [
   ["ERR_SOCKET_DGRAM_IS_CONNECTED", Error],
   ["ERR_SOCKET_DGRAM_NOT_CONNECTED", Error],
   ["ERR_SOCKET_DGRAM_NOT_RUNNING", Error],
+  ["ERR_STREAM_PREMATURE_CLOSE", Error],
   ["ERR_STREAM_ALREADY_FINISHED", Error],
   ["ERR_STREAM_CANNOT_PIPE", Error],
   ["ERR_STREAM_DESTROYED", Error],
   ["ERR_STREAM_NULL_VALUES", TypeError],
-  ["ERR_STREAM_PREMATURE_CLOSE", Error],
+  ["ERR_STREAM_WRITE_AFTER_END", Error],
   ["ERR_STREAM_PUSH_AFTER_EOF", Error],
   ["ERR_STREAM_RELEASE_LOCK", Error, "AbortError"],
   ["ERR_STREAM_UNABLE_TO_PIPE", Error],
   ["ERR_STREAM_UNSHIFT_AFTER_END_EVENT", Error],
-  ["ERR_STREAM_WRITE_AFTER_END", Error],
   ["ERR_STRING_TOO_LONG", Error],
   ["ERR_UNAVAILABLE_DURING_EXIT", Error],
   ["ERR_UNCAUGHT_EXCEPTION_CAPTURE_ALREADY_SET", Error],
@@ -176,5 +175,11 @@ const errors: ErrorCodeMapping = [
   ["ERR_S3_INVALID_ENDPOINT", Error],
   ["ERR_S3_INVALID_SIGNATURE", Error],
   ["ERR_S3_INVALID_SESSION_TOKEN", Error],
-] as const;
+
+  // URL
+  ["ERR_INVALID_URL", TypeError],
+  ["ERR_INVALID_URL_SCHEME", TypeError],
+  ["ERR_INVALID_FILE_URL_HOST", TypeError],
+  ["ERR_INVALID_FILE_URL_PATH", TypeError],
+];
 export default errors;

src/bun.js/bindings/PathInlines.h

@@ -14,6 +14,11 @@
 #define PLATFORM_SEP POSIX_PATH_SEP
 #endif
 
+#define IS_LETTER(byte) \
+    ((byte >= 'a' && byte <= 'z') || (byte >= 'A' && byte <= 'Z'))
+#define IS_SLASH(byte) \
+    (byte == '/' || byte == '\\')
+
 ALWAYS_INLINE bool isAbsolutePath(WTF::String input)
 {
 #if OS(WINDOWS)
@@ -22,23 +27,23 @@ ALWAYS_INLINE bool isAbsolutePath(WTF::String input)
         if (len < 1)
             return false;
         const auto bytes = input.span8().data();
-        if (bytes[0] == '/' || bytes[0] == '\\')
+        if (IS_SLASH(bytes[0]))
             return true;
         if (len < 2)
             return false;
-        if (bytes[1] == ':' && (bytes[2] == '/' || bytes[2] == '\\'))
+        if (IS_LETTER(bytes[0]) && bytes[1] == ':' && IS_SLASH(bytes[2]))
             return true;
         return false;
     } else {
         auto len = input.length();
         if (len < 1)
             return false;
         const auto bytes = input.span16().data();
-        if (bytes[0] == '/' || bytes[0] == '\\')
+        if (IS_SLASH(bytes[0]))
             return true;
         if (len < 2)
             return false;
-        if (bytes[1] == ':' && (bytes[2] == '/' || bytes[2] == '\\'))
+        if (IS_LETTER(bytes[0]) && bytes[1] == ':' && IS_SLASH(bytes[2]))
             return true;
         return false;
     }
@@ -47,6 +52,9 @@ ALWAYS_INLINE bool isAbsolutePath(WTF::String input)
 #endif
 }
 
+#undef IS_LETTER
+#undef IS_SLASH
+
 extern "C" BunString ResolvePath__joinAbsStringBufCurrentPlatformBunString(JSC::JSGlobalObject*, BunString);
 
 /// CWD is determined by the global object's current cwd.

src/bun.js/bindings/ZigGlobalObject.cpp

@@ -1,4 +1,3 @@
-
 #include "root.h"
 
 #include "ZigGlobalObject.h"

src/bun.js/bindings/bindings.cpp

@@ -6346,3 +6346,29 @@ extern "C" EncodedJSValue Bun__JSObject__getCodePropertyVMInquiry(JSC::JSGlobalO
 
     return JSValue::encode(slot.getPureResult());
 }
+
+using StackCodeType = JSC::StackVisitor::Frame::CodeType;
+CPP_DECL bool Bun__util__isInsideNodeModules(JSC::JSGlobalObject* globalObject, JSC::CallFrame* callFrame)
+{
+    JSC::VM& vm = globalObject->vm();
+    bool inNodeModules = false;
+    JSC::StackVisitor::visit(callFrame, vm, [&](JSC::StackVisitor& visitor) -> WTF::IterationStatus {
+        if (Zig::isImplementationVisibilityPrivate(visitor) || visitor->isNativeCalleeFrame()) {
+            return WTF::IterationStatus::Continue;
+        }
+
+        if (visitor->hasLineAndColumnInfo()) {
+            String sourceURL = Zig::sourceURL(visitor);
+            if (sourceURL.startsWith("node:"_s) || sourceURL.startsWith("bun:"_s))
+                return WTF::IterationStatus::Continue;
+            if (sourceURL.contains("node_modules"_s))
+                inNodeModules = true;
+
+            return WTF::IterationStatus::Done;
+        }
+
+        return WTF::IterationStatus::Continue;
+    });
+
+    return inNodeModules;
+}

src/bun.js/node/node_util_binding.zig

@@ -132,7 +132,20 @@ pub fn extractedSplitNewLinesFastPathStringsOnly(globalThis: *JSC.JSGlobalObject
     };
 }
 
-fn split(comptime encoding: bun.strings.EncodingNonAscii, globalThis: *JSC.JSGlobalObject, allocator: Allocator, str: *const bun.String) bun.JSError!JSC.JSValue {
+extern fn Bun__util__isInsideNodeModules(globalObject: *JSC.JSGlobalObject, callFrame: *JSC.CallFrame) bool;
+/// Walks the call stack from bottom to top, returning `true` when it finds a
+/// frame within a `node_modules` directory.
+pub fn isInsideNodeModules(globalObject: *JSC.JSGlobalObject, callframe: *JSC.CallFrame) bun.JSError!JSC.JSValue {
+    const res = Bun__util__isInsideNodeModules(globalObject, callframe);
+    return JSC.JSValue.jsBoolean(res);
+}
+
+fn split(
+    comptime encoding: bun.strings.EncodingNonAscii,
+    globalThis: *JSC.JSGlobalObject,
+    allocator: Allocator,
+    str: *const bun.String,
+) bun.JSError!JSC.JSValue {
     var fallback = std.heap.stackFallback(1024, allocator);
     const alloc = fallback.get();
     const Char = switch (encoding) {

src/codegen/generate-node-errors.ts

@@ -25,6 +25,7 @@ let zig = ``;
 enumHeader = `
 // clang-format off
 // Generated by: src/codegen/generate-node-errors.ts
+// Input:        src/bun.js/bindings/ErrorCode.ts
 #pragma once
 
 #include <cstdint>

src/js/builtins.d.ts

@@ -73,11 +73,15 @@ declare function $getByIdDirectPrivate<T = any>(obj: any, key: string): T;
 declare function $getByValWithThis(target: any, receiver: any, propertyKey: string): void;
 /** gets the prototype of an object */
 declare function $getPrototypeOf(value: any): any;
-/** gets an internal property on a promise
+/**
+ * Gets an internal property on a promise
  *
  *  You can pass
- *  - $promiseFieldFlags - get a number with flags
- *  - $promiseFieldReactionsOrResult - get the result (like Bun.peek)
+ *  - {@link $promiseFieldFlags} - get a number with flags
+ *  - {@link $promiseFieldReactionsOrResult} - get the result (like {@link Bun.peek})
+ *
+ * @param promise the promise to get the field from
+ * @param key an internal field id.
  */
 declare function $getPromiseInternalField<K extends PromiseFieldType, V>(
   promise: Promise<V>,
@@ -99,6 +103,19 @@ declare function $getMapIteratorInternalField(): TODO;
 declare function $getSetIteratorInternalField(): TODO;
 declare function $getProxyInternalField(): TODO;
 declare function $idWithProfile(): TODO;
+/**
+ * True for object-like `JSCell`s. That is, this is roughly equivalent to this
+ * JS code:
+ * ```js
+ * typeof obj === "object" && obj !== null
+ * ```
+ *
+ * @param obj The object to check
+ * @returns `true` if `obj` is an object-like `JSCell`
+ *
+ * @see [JSCell.h](https://github.com/oven-sh/WebKit/blob/main/Source/JavaScriptCore/runtime/JSCell.h)
+ * @see [JIT implementation](https://github.com/oven-sh/WebKit/blob/433f7598bf3537a295d0af5ffd83b9a307abec4e/Source/JavaScriptCore/jit/JITOpcodes.cpp#L311)
+ */
 declare function $isObject(obj: unknown): obj is object;
 declare function $isArray(obj: unknown): obj is any[];
 declare function $isCallable(fn: unknown): fn is CallableFunction;
@@ -559,6 +576,11 @@ declare interface Function {
   path: string;
 }
 
+interface String {
+  $charCodeAt: String["charCodeAt"];
+  // add others as needed
+}
+
 declare var $Buffer: {
   new (a: any, b?: any, c?: any): Buffer;
 };

src/js/internal/util.ts

@@ -0,0 +1,5 @@
+const isInsideNodeModules: () => boolean = $newZigFunction("node_util_binding.zig", "isInsideNodeModules", 0);
+
+export default {
+  isInsideNodeModules,
+};