Merge pull request #1023 from cloudflare/maizatskyi/2023-08-15-lazy-c…

…f-parsing

lazy cf property parsing

src/workerd/api/cf-property.c++

@@ -0,0 +1,131 @@
+// Copyright (c) 2017-2022 Cloudflare, Inc.
+// Licensed under the Apache 2.0 license found in the LICENSE file or at:
+//     https://opensource.org/licenses/Apache-2.0
+
+#include "cf-property.h"
+#include <workerd/io/features.h>
+
+namespace workerd::api {
+
+static constexpr auto kDefaultBotManagementValue = R"DATA({
+  "corporateProxy": false,
+  "verifiedBot": false,
+  "jsDetection": { "passed": false },
+  "staticResource": false,
+  "detectionIds": {},
+  "score": 99
+})DATA";
+
+static void handleDefaultBotManagement(jsg::Lock& js, v8::Local<v8::Object> handle) {
+  // When the cfBotManagementNoOp compatibility flag is set, we'll check the
+  // request cf blob to see if it contains a botManagement field. If it does
+  // *not* we will add it using the following default fields.
+  // Note that if the botManagement team changes any of the fields they provide,
+  // this default value may need to be changed also.
+  auto context = js.v8Context();
+  if (!js.v8Has(handle, "botManagement"_kj)) {
+    auto sym = v8::Private::ForApi(js.v8Isolate,
+        jsg::v8StrIntern(js.v8Isolate, "botManagement"_kj));
+    // For performance reasons, we only want to construct the default values
+    // once per isolate so we cache the constructed value using an internal
+    // private field on the global scope. Whenever we need to use it again we
+    // pull the exact same value.
+    auto defaultBm = jsg::check(context->Global()->GetPrivate(context, sym));
+    if (defaultBm->IsUndefined()) {
+      auto bm = js.parseJson(kj::StringPtr(kDefaultBotManagementValue));
+      KJ_DASSERT(bm.getHandle(js)->IsObject());
+      js.recursivelyFreeze(bm);
+      defaultBm = bm.getHandle(js);
+      jsg::check(context->Global()->SetPrivate(context, sym, defaultBm));
+    }
+    js.v8Set(handle, "botManagement"_kj, defaultBm);
+  }
+}
+
+jsg::Optional<v8::Local<v8::Object>> CfProperty::get(jsg::Lock& js) {
+  return getRef(js).map([&js](jsg::V8Ref<v8::Object>&& ref) mutable {
+    return ref.getHandle(js);
+  });
+}
+
+jsg::Optional<jsg::V8Ref<v8::Object>> CfProperty::getRef(jsg::Lock& js) {
+  KJ_IF_MAYBE(cf, value) {
+    KJ_SWITCH_ONEOF(*cf) {
+      KJ_CASE_ONEOF(parsed, jsg::V8Ref<v8::Object>) {
+        return parsed.addRef(js);
+      }
+      KJ_CASE_ONEOF(unparsed, kj::String) {
+        KJ_DBG(unparsed.asPtr());
+        auto parsed = js.parseJson(unparsed);
+        auto handle = parsed.getHandle(js);
+        KJ_ASSERT(handle->IsObject());
+
+        auto objectHandle = handle.As<v8::Object>();
+        if (!FeatureFlags::get(js).getNoCfBotManagementDefault()) {
+          handleDefaultBotManagement(js, objectHandle);
+        }
+
+        // For the inbound request, we make the `cf` blob immutable.
+        js.recursivelyFreeze(parsed);
+
+        // replace unparsed string with a parsed v8 object
+        auto parsedObject = parsed.cast<v8::Object>(js);
+        this->value = parsedObject.addRef(js);
+        return kj::mv(parsedObject);
+      }
+    }
+  }
+
+  return nullptr;
+}
+
+
+kj::Maybe<kj::String> CfProperty::serialize(jsg::Lock& js) {
+  KJ_IF_MAYBE(cf, value) {
+    KJ_SWITCH_ONEOF(*cf) {
+      KJ_CASE_ONEOF(parsed, jsg::V8Ref<v8::Object>) {
+        return js.serializeJson(parsed);
+      }
+      KJ_CASE_ONEOF(unparsed, kj::String) {
+        if (!FeatureFlags::get(js).getNoCfBotManagementDefault()) {
+          // we mess up with the value on this code path,
+          // need to parse it, fix it and serialize back
+          return js.serializeJson(KJ_ASSERT_NONNULL(getRef(js)));
+        }
+
+        return kj::str(unparsed);
+      }
+    }
+  }
+
+  return nullptr;
+}
+
+CfProperty CfProperty::deepClone(jsg::Lock& js) {
+  KJ_IF_MAYBE(cf, value) {
+    KJ_SWITCH_ONEOF(*cf) {
+      KJ_CASE_ONEOF(parsed, jsg::V8Ref<v8::Object>) {
+        auto ref = parsed.deepClone(js);
+        return CfProperty(kj::mv(ref));
+      }
+      KJ_CASE_ONEOF(unparsed, kj::String) {
+        return CfProperty(unparsed.asPtr());
+      }
+    }
+  }
+
+  return nullptr;
+}
+
+void CfProperty::visitForGc(jsg::GcVisitor& visitor) {
+  KJ_IF_MAYBE(cf, value) {
+    KJ_SWITCH_ONEOF(*cf) {
+      KJ_CASE_ONEOF(parsed, jsg::V8Ref<v8::Object>) {
+        visitor.visit(parsed);
+      }
+      KJ_CASE_ONEOF_DEFAULT {}
+    }
+  }
+}
+
+} // namespace workerd::api

src/workerd/api/cf-property.h

@@ -0,0 +1,56 @@
+// Copyright (c) 2017-2022 Cloudflare, Inc.
+// Licensed under the Apache 2.0 license found in the LICENSE file or at:
+//     https://opensource.org/licenses/Apache-2.0
+
+#pragma once
+
+// Common functionality to manage cf headers and properties.
+
+#include <workerd/jsg/jsg.h>
+
+namespace workerd::api {
+
+class CfProperty {
+  // A holder for Cf header property value.
+  // The string header is parsed on demand and the parsed value cached.
+
+public:
+  KJ_DISALLOW_COPY(CfProperty);
+
+  explicit CfProperty() {}
+  CfProperty(decltype(nullptr)) {}
+  CfProperty(CfProperty&&) = default;
+  CfProperty& operator=(CfProperty&&) = default;
+
+  explicit CfProperty(kj::Maybe<kj::StringPtr> unparsed) {
+    KJ_IF_MAYBE(str, unparsed) {
+      value = kj::str(*str);
+    }
+  }
+
+  explicit CfProperty(kj::Maybe<jsg::V8Ref<v8::Object>>&& parsed) {
+    KJ_IF_MAYBE(v, parsed) {
+      value = kj::mv(*v);
+    }
+  }
+
+  jsg::Optional<v8::Local<v8::Object>> get(jsg::Lock& js);
+  // Get parsed value
+
+  jsg::Optional<jsg::V8Ref<v8::Object>> getRef(jsg::Lock& js);
+  // Get parsed value as a global ref
+
+  kj::Maybe<kj::String> serialize(jsg::Lock& js);
+  // Serialize to string
+
+  CfProperty deepClone(jsg::Lock& js);
+  // Clone by deep cloning parsed v8 object (if any).
+
+  void visitForGc(jsg::GcVisitor& visitor);
+
+private:
+  kj::Maybe<kj::OneOf<kj::String, jsg::V8Ref<v8::Object>>> value;
+};
+
+
+} // namespace workerd::api

src/workerd/api/global-scope.c++

@@ -92,42 +92,6 @@ private:
   }
 };
 
-static constexpr auto kDefaultBotManagementValue = R"DATA({
-  "corporateProxy": false,
-  "verifiedBot": false,
-  "jsDetection": { "passed": false },
-  "staticResource": false,
-  "detectionIds": {},
-  "score": 99
-})DATA";
-
-void handleDefaultBotManagement(jsg::Lock& js, jsg::Value& cf) {
-  // When the cfBotManagementNoOp compatibility flag is set, we'll check the
-  // request cf blob to see if it contains a botManagement field. If it does
-  // *not* we will add it using the following default fields.
-  // Note that if the botManagement team changes any of the fields they provide,
-  // this default value may need to be changed also.
-  auto context = js.v8Context();
-  auto handle = cf.getHandle(js).As<v8::Object>();
-  if (!js.v8Has(handle, "botManagement"_kj)) {
-    auto sym = v8::Private::ForApi(js.v8Isolate,
-        jsg::v8StrIntern(js.v8Isolate, "botManagement"_kj));
-    // For performance reasons, we only want to construct the default values
-    // once per isolate so we cache the constructed value using an internal
-    // private field on the global scope. Whenever we need to use it again we
-    // pull the exact same value.
-    auto defaultBm = jsg::check(context->Global()->GetPrivate(context, sym));
-    if (defaultBm->IsUndefined()) {
-      auto bm = js.parseJson(kj::StringPtr(kDefaultBotManagementValue));
-      KJ_DASSERT(bm.getHandle(js)->IsObject());
-      js.recursivelyFreeze(bm);
-      defaultBm = bm.getHandle(js);
-      jsg::check(context->Global()->SetPrivate(context, sym, defaultBm));
-    }
-    js.v8Set(handle, "botManagement"_kj, defaultBm);
-  }
-}
-
 kj::String getEventName(v8::PromiseRejectEvent type) {
   switch (type) {
     case v8::PromiseRejectEvent::kPromiseRejectWithNoHandler:
@@ -200,21 +164,7 @@ kj::Promise<DeferredProxy<void>> ServiceWorkerGlobalScope::request(
   auto& ioContext = IoContext::current();
   jsg::Lock& js = lock;
 
-  kj::Maybe<jsg::V8Ref<v8::Object>> cf;
-
-  KJ_IF_MAYBE(c, cfBlobJson) {
-    auto handle = js.parseJson(*c);
-    KJ_ASSERT(handle.getHandle(js)->IsObject());
-
-    if (!FeatureFlags::get(js).getNoCfBotManagementDefault()) {
-      handleDefaultBotManagement(js, handle);
-    }
-
-    // For the inbound request, we make the `cf` blob immutable.
-    js.recursivelyFreeze(handle);
-
-    cf = handle.cast<v8::Object>(js);
-  }
+  CfProperty cf(cfBlobJson);
 
   auto jsHeaders = jsg::alloc<Headers>(headers, Headers::Guard::REQUEST);
   // We do not automatically decode gzipped request bodies because the fetch() standard doesn't