[Intl] Move parseExtension from JS to C++

Bug: v8:7979
Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng
Change-Id: I367e7a7e93de267cf9b0718f193f5de3abfbf692
Reviewed-on: https://chromium-review.googlesource.com/1152556
Commit-Queue: Frank Tang <ftang@chromium.org>
Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
Cr-Commit-Position: refs/heads/master@{#54992}

src/js/intl.js

@@ -483,61 +483,6 @@ function bestFitMatcher(service, requestedLocales) {
   return lookupMatcher(service, requestedLocales);
 }
 
-
-/**
- * Parses Unicode extension into key - value map.
- * Returns empty object if the extension string is invalid.
- * We are not concerned with the validity of the values at this point.
- * 'attribute' in RFC 6047 is not supported. Keys without explicit
- * values are assigned UNDEFINED.
- * TODO(jshin): Fix the handling of 'attribute' (in RFC 6047, but none
- * has been defined so that it's not used) and boolean keys without
- * an explicit value.
- */
-function parseExtension(extension) {
-  var extensionSplit = %StringSplit(extension, '-', kMaxUint32);
-
-  // Assume ['', 'u', ...] input, but don't throw.
-  if (extensionSplit.length <= 2 ||
-      (extensionSplit[0] !== '' && extensionSplit[1] !== 'u')) {
-    return {__proto__: null};
-  }
-
-  // Key is {2}alphanum, value is {3,8}alphanum.
-  // Some keys may not have explicit values (booleans).
-  var extensionMap = {__proto__: null};
-  var key = UNDEFINED;
-  var value = UNDEFINED;
-  for (var i = 2; i < extensionSplit.length; ++i) {
-    var length = extensionSplit[i].length;
-    var element = extensionSplit[i];
-    if (length === 2) {
-      if (!IS_UNDEFINED(key)) {
-        if (!(key in extensionMap)) {
-          extensionMap[key] = value;
-        }
-        value = UNDEFINED;
-      }
-      key = element;
-    } else if (length >= 3 && length <= 8 && !IS_UNDEFINED(key)) {
-      if (IS_UNDEFINED(value)) {
-        value = element;
-      } else {
-        value = value + "-" + element;
-      }
-    } else {
-      // There is a value that's too long, or that doesn't have a key.
-      return {__proto__: null};
-    }
-  }
-  if (!IS_UNDEFINED(key) && !(key in extensionMap)) {
-    extensionMap[key] = value;
-  }
-
-  return extensionMap;
-}
-
-
 /**
  * Populates internalOptions object with boolean key-value pairs
  * from extensionMap and options.
@@ -928,7 +873,7 @@ function CreateNumberFormat(locales, options) {
 
   // ICU prefers options to be passed using -u- extension key/values for
   // number format, so we need to build that.
-  var extensionMap = parseExtension(locale.extension);
+  var extensionMap = %ParseExtension(locale.extension);
 
   /**
    * Map of Unicode extensions to option properties, and their values and types,
@@ -1277,7 +1222,7 @@ function CreateDateTimeFormat(locales, options) {
   // ICU prefers options to be passed using -u- extension key/values, so
   // we need to build that.
   var internalOptions = {__proto__: null};
-  var extensionMap = parseExtension(locale.extension);
+  var extensionMap = %ParseExtension(locale.extension);
 
   /**
    * Map of Unicode extensions to option properties, and their values and types,

src/objects/intl-objects.cc

@@ -1640,6 +1640,50 @@ MaybeHandle<JSObject> Intl::CreateNumberFormat(Isolate* isolate,
   return local_object;
 }
 
+/**
+ * Parses Unicode extension into key - value map.
+ * Returns empty object if the extension string is invalid.
+ * We are not concerned with the validity of the values at this point.
+ * 'attribute' in RFC 6047 is not supported. Keys without explicit
+ * values are assigned UNDEFINED.
+ * TODO(jshin): Fix the handling of 'attribute' (in RFC 6047, but none
+ * has been defined so that it's not used) and boolean keys without
+ * an explicit value.
+ */
+void Intl::ParseExtension(Isolate* isolate, const std::string& extension,
+                          std::map<std::string, std::string>& out) {
+  if (extension.compare(0, 3, "-u-") != 0) return;
+
+  // Key is {2}alphanum, value is {3,8}alphanum.
+  // Some keys may not have explicit values (booleans).
+  std::string key;
+  std::string value;
+  // Skip the "-u-".
+  size_t start = 3;
+  size_t end;
+  do {
+    end = extension.find("-", start);
+    size_t length =
+        (end == std::string::npos) ? extension.length() - start : end - start;
+    std::string element = extension.substr(start, length);
+    // Key is {2}alphanum
+    if (length == 2) {
+      if (!key.empty()) {
+        out.insert(std::pair<std::string, std::string>(key, value));
+        value.clear();
+      }
+      key = element;
+      // value is {3,8}alphanum.
+    } else if (length >= 3 && length <= 8 && !key.empty()) {
+      value = value.empty() ? element : (value + "-" + element);
+    } else {
+      return;
+    }
+    start = end + 1;
+  } while (end != std::string::npos);
+  if (!key.empty()) out.insert(std::pair<std::string, std::string>(key, value));
+}
+
 namespace {
 
 bool IsAToZ(char ch) {

src/objects/intl-objects.h

@@ -9,6 +9,7 @@
 #ifndef V8_OBJECTS_INTL_OBJECTS_H_
 #define V8_OBJECTS_INTL_OBJECTS_H_
 
+#include <map>
 #include <set>
 #include <string>
 
@@ -276,6 +277,10 @@ class Intl {
   // The currency is expected to an all upper case string value.
   static Handle<Smi> CurrencyDigits(Isolate* isolate, Handle<String> currency);
 
+  // TODO(ftang): Remove this and use ICU to the conversion in the future
+  static void ParseExtension(Isolate* isolate, const std::string& extension,
+                             std::map<std::string, std::string>& out);
+
   V8_WARN_UNUSED_RESULT static MaybeHandle<JSObject> CreateNumberFormat(
       Isolate* isolate, Handle<String> locale, Handle<JSObject> options,
       Handle<JSObject> resolved);

src/runtime/runtime-intl.cc

@@ -329,6 +329,25 @@ RUNTIME_FUNCTION(Runtime_PluralRulesResolvedOptions) {
   return *JSPluralRules::ResolvedOptions(isolate, plural_rules);
 }
 
+RUNTIME_FUNCTION(Runtime_ParseExtension) {
+  Factory* factory = isolate->factory();
+  HandleScope scope(isolate);
+  DCHECK_EQ(1, args.length());
+  CONVERT_ARG_HANDLE_CHECKED(String, extension, 0);
+  std::map<std::string, std::string> map;
+  Intl::ParseExtension(isolate, std::string(extension->ToCString().get()), map);
+  Handle<JSObject> extension_map =
+      isolate->factory()->NewJSObjectWithNullProto();
+  for (std::map<std::string, std::string>::iterator it = map.begin();
+       it != map.end(); it++) {
+    JSObject::AddProperty(
+        isolate, extension_map,
+        factory->NewStringFromAsciiChecked(it->first.c_str()),
+        factory->NewStringFromAsciiChecked(it->second.c_str()), NONE);
+  }
+  return *extension_map;
+}
+
 RUNTIME_FUNCTION(Runtime_PluralRulesSelect) {
   HandleScope scope(isolate);
 

src/runtime/runtime.h

@@ -228,6 +228,7 @@ namespace internal {
   F(IsInitializedIntlObjectOfType, 2, 1)     \
   F(IsWellFormedCurrencyCode, 1, 1)          \
   F(MarkAsInitializedIntlObjectOfType, 2, 1) \
+  F(ParseExtension, 1, 1)                    \
   F(PluralRulesResolvedOptions, 1, 1)        \
   F(PluralRulesSelect, 2, 1)                 \
   F(StringToLowerCaseIntl, 1, 1)             \