Move the email regex from HTMLInputElement to DocumentData

This reduces the overall memory. And hopefully it could fix the
regression introduced by
https://chromium-review.googlesource.com/c/chromium/src/+/2324204,
which could allocate a regex for every input element.

Bug: 1116302
Change-Id: I00f1fa7ad07664d5ab799e760483034c1a1594fe
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2357464
Commit-Queue: Yao Xiao <yaoxia@chromium.org>
Reviewed-by: Mason Freed <masonfreed@chromium.org>
Cr-Commit-Position: refs/heads/master@{#802754}

third_party/blink/renderer/core/dom/document.cc

@@ -198,6 +198,7 @@
 #include "third_party/blink/renderer/core/html/custom/v0_custom_element_registration_context.h"
 #include "third_party/blink/renderer/core/html/document_all_name_collection.h"
 #include "third_party/blink/renderer/core/html/document_name_collection.h"
+#include "third_party/blink/renderer/core/html/forms/email_input_type.h"
 #include "third_party/blink/renderer/core/html/forms/form_controller.h"
 #include "third_party/blink/renderer/core/html/forms/html_form_element.h"
 #include "third_party/blink/renderer/core/html/forms/html_input_element.h"
@@ -7301,6 +7302,12 @@ Document::EnsureDocumentExplicitRootIntersectionObserverData() {
   return *document_explicit_root_intersection_observer_data_;
 }
 
+const ScriptRegexp& Document::EnsureEmailRegexp() const {
+  if (!data_->email_regexp_)
+    data_->email_regexp_ = EmailInputType::CreateEmailRegexp();
+  return *data_->email_regexp_;
+}
+
 void Document::AddConsoleMessage(ConsoleMessage* message,
                                  bool discard_duplicates) const {
   // Don't let non-attached Documents spam the console.

third_party/blink/renderer/core/dom/document.h

@@ -188,6 +188,7 @@ class Text;
 class TrustedHTML;
 class ScriptElementBase;
 class ScriptPromise;
+class ScriptRegexp;
 class ScriptRunner;
 class ScriptableDocumentParser;
 class ScriptedAnimationController;
@@ -986,6 +987,8 @@ class CORE_EXPORT Document : public ContainerNode,
   ElementIntersectionObserverData&
   EnsureDocumentExplicitRootIntersectionObserverData();
 
+  const ScriptRegexp& EnsureEmailRegexp() const;
+
   // Returns the owning element in the parent document. Returns nullptr if
   // this is the top level document or the owner is remote.
   HTMLFrameOwnerElement* LocalOwner() const;

third_party/blink/renderer/core/dom/document_data.h

@@ -7,6 +7,7 @@
 
 #include "services/network/public/mojom/trust_tokens.mojom-blink.h"
 #include "third_party/blink/public/mojom/permissions/permission.mojom-blink.h"
+#include "third_party/blink/renderer/bindings/core/v8/script_regexp.h"
 #include "third_party/blink/renderer/platform/mojo/heap_mojo_remote.h"
 
 namespace blink {
@@ -19,7 +20,7 @@ namespace blink {
 //   Other instances should not have strong references to the DocumentData.
 // Lifetime: A DocumentData instance is created on a Document creation, and
 //   is never destructed before the Document.
-class DocumentData : public GarbageCollected<DocumentData> {
+class DocumentData final : public GarbageCollected<DocumentData> {
  public:
   explicit DocumentData(ExecutionContext* context)
       : permission_service_(context), has_trust_tokens_answerer_(context) {}
@@ -51,6 +52,9 @@ class DocumentData : public GarbageCollected<DocumentData> {
   HeapHashSet<Member<ScriptPromiseResolver>>
       pending_has_trust_tokens_resolvers_;
 
+  // To do email regex checks.
+  std::unique_ptr<ScriptRegexp> email_regexp_;
+
   friend class Document;
 };
 

third_party/blink/renderer/core/html/forms/email_input_type.cc

@@ -175,24 +175,25 @@ const AtomicString& EmailInputType::FormControlType() const {
   return input_type_names::kEmail;
 }
 
-ScriptRegexp& EmailInputType::EnsureEmailRegexp() const {
-  return GetElement().EnsureEmailRegexp();
-}
-
 // The return value is an invalid email address string if the specified string
-// contains an invalid email address. Otherwise, null string is returned.
+// contains an invalid email address. Otherwise, an empty string is returned.
 // If an empty string is returned, it means empty address is specified.
 // e.g. "foo@example.com,,bar@example.com" for multiple case.
 String EmailInputType::FindInvalidAddress(const String& value) const {
   if (value.IsEmpty())
     return String();
-  if (!GetElement().Multiple())
-    return IsValidEmailAddress(EnsureEmailRegexp(), value) ? String() : value;
+  if (!GetElement().Multiple()) {
+    return IsValidEmailAddress(GetElement().GetDocument().EnsureEmailRegexp(),
+                               value)
+               ? String()
+               : value;
+  }
   Vector<String> addresses;
   value.Split(',', true, addresses);
   for (const auto& address : addresses) {
     String stripped = StripLeadingAndTrailingHTMLSpaces(address);
-    if (!IsValidEmailAddress(EnsureEmailRegexp(), stripped))
+    if (!IsValidEmailAddress(GetElement().GetDocument().EnsureEmailRegexp(),
+                             stripped))
       return stripped;
   }
   return String();
@@ -282,8 +283,10 @@ String EmailInputType::SanitizeValue(const String& proposed_value) const {
 String EmailInputType::ConvertFromVisibleValue(
     const String& visible_value) const {
   String sanitized_value = SanitizeValue(visible_value);
-  if (!GetElement().Multiple())
-    return ConvertEmailAddressToASCII(EnsureEmailRegexp(), sanitized_value);
+  if (!GetElement().Multiple()) {
+    return ConvertEmailAddressToASCII(
+        GetElement().GetDocument().EnsureEmailRegexp(), sanitized_value);
+  }
   Vector<String> addresses;
   sanitized_value.Split(',', true, addresses);
   StringBuilder builder;
@@ -291,8 +294,8 @@ String EmailInputType::ConvertFromVisibleValue(
   for (wtf_size_t i = 0; i < addresses.size(); ++i) {
     if (i > 0)
       builder.Append(',');
-    builder.Append(
-        ConvertEmailAddressToASCII(EnsureEmailRegexp(), addresses[i]));
+    builder.Append(ConvertEmailAddressToASCII(
+        GetElement().GetDocument().EnsureEmailRegexp(), addresses[i]));
   }
   return builder.ToString();
 }

third_party/blink/renderer/core/html/forms/email_input_type.h

@@ -57,7 +57,6 @@ class EmailInputType final : public BaseTextInputType {
   String ConvertFromVisibleValue(const String&) const override;
   String VisibleValue() const override;
 
-  ScriptRegexp& EnsureEmailRegexp() const;
   String ConvertEmailAddressToUnicode(const String&) const;
   String FindInvalidAddress(const String&) const;
 };

third_party/blink/renderer/core/html/forms/html_input_element.cc

@@ -2030,12 +2030,6 @@ bool HTMLInputElement::IsDraggedSlider() const {
   return input_type_view_->IsDraggedSlider();
 }
 
-ScriptRegexp& HTMLInputElement::EnsureEmailRegexp() const {
-  if (!email_regexp_)
-    email_regexp_ = EmailInputType::CreateEmailRegexp();
-  return *email_regexp_;
-}
-
 void HTMLInputElement::MaybeReportPiiMetrics() {
   // Don't report metrics if the field is empty.
   if (value().IsEmpty())
@@ -2061,7 +2055,8 @@ void HTMLInputElement::MaybeReportPiiMetrics() {
   // https://www.rfc-editor.org/errata_search.php?rfc=3696) in addition to
   // matching with the pattern given by the HTML standard.
   if (value().length() <= kMaxEmailFieldLength &&
-      EmailInputType::IsValidEmailAddress(EnsureEmailRegexp(), value())) {
+      EmailInputType::IsValidEmailAddress(GetDocument().EnsureEmailRegexp(),
+                                          value())) {
     UseCounter::Count(GetDocument(),
                       WebFeature::kEmailFieldDetected_PatternMatch);
   }

third_party/blink/renderer/core/html/forms/html_input_element.h

@@ -347,8 +347,6 @@ class CORE_EXPORT HTMLInputElement
     form_element_pii_type_ = form_element_pii_type;
   }
 
-  ScriptRegexp& EnsureEmailRegexp() const;
-
  protected:
   void DefaultEventHandler(Event&) override;
   void CreateShadowSubtree();
@@ -478,7 +476,6 @@ class CORE_EXPORT HTMLInputElement
   Member<ListAttributeTargetObserver> list_attribute_target_observer_;
 
   FormElementPiiType form_element_pii_type_ = FormElementPiiType::kUnknown;
-  mutable std::unique_ptr<ScriptRegexp> email_regexp_;
 
   FRIEND_TEST_ALL_PREFIXES(HTMLInputElementTest, RadioKeyDownDCHECKFailure);
 };