RELNOTES[INC]: goog.format.EmailAddress.parse now ignores unicode bid…

…i markers PiperOrigin-RevId: 331777392
google · Sep 17, 2020 · ac1c88d · ac1c88d
1 parent 4c3621c
commit ac1c88d
Show file tree

Hide file tree

Showing 3 changed files with 46 additions and 15 deletions.
diff --git a/closure/goog/deps.js b/closure/goog/deps.js
diff --git a/closure/goog/format/emailaddress.js b/closure/goog/format/emailaddress.js
@@ -10,10 +10,10 @@
 
 goog.provide('goog.format.EmailAddress');
 
+goog.require('goog.i18n.bidi');
+goog.require('goog.object');
 goog.require('goog.string');
 
-
-
 /**
  * Formats an email address string for display, and allows for extraction of
  * the individual components of the address.
@@ -133,7 +133,6 @@ goog.format.EmailAddress.LOCAL_PART_REGEXP_STR_ =
 goog.format.EmailAddress.DOMAIN_PART_REGEXP_STR_ =
     '([a-zA-Z0-9-]+\\.)+[a-zA-Z0-9]{2,63}';
 
-
 /**
  * A RegExp to match the local part of an email address.
  * @private {!RegExp}
@@ -158,6 +157,14 @@ goog.format.EmailAddress.EMAIL_ADDRESS_ = new RegExp(
     '^' + goog.format.EmailAddress.LOCAL_PART_REGEXP_STR_ + '@' +
     goog.format.EmailAddress.DOMAIN_PART_REGEXP_STR_ + '$');
 
+/**
+ * Regular expression for bidi format character replacement in text.
+ * @type {!RegExp}
+ * @private
+ */
+goog.format.EmailAddress.ALL_BIDI_FORMAT_CHARS_ = new RegExp(
+    '[' + goog.object.getValues(goog.i18n.bidi.Format).join('') + ']', 'g');
+
 
 /**
  * Get the name associated with the email address.
@@ -341,7 +348,7 @@ goog.format.EmailAddress.isValidDomainPartSpec = function(str) {
  */
 goog.format.EmailAddress.parseInternal = function(addr, ctor) {
   'use strict';
-  // TODO(ecattell): Strip bidi markers.
+  addr = goog.format.EmailAddress.stripBidiChars_(addr);
   var name = '';
   var address = '';
   for (var i = 0; i < addr.length;) {
@@ -504,3 +511,15 @@ goog.format.EmailAddress.isAddressSeparator = function(ch) {
   'use strict';
   return goog.string.contains(goog.format.EmailAddress.ADDRESS_SEPARATORS_, ch);
 };
+
+/**
+ * Returns the input text without Unicode formatting characters
+ * and directionality string constants as defined in {@link
+ * goog.i18n.bidi.Format}.
+ * @param {string} str The given string.
+ * @return {string} The given string cleaned of formatting characters.
+ * @private
+ */
+goog.format.EmailAddress.stripBidiChars_ = function(str) {
+  return str.replace(goog.format.EmailAddress.ALL_BIDI_FORMAT_CHARS_, '');
+};
diff --git a/closure/goog/format/emailaddress_test.js b/closure/goog/format/emailaddress_test.js
@@ -8,8 +8,14 @@ goog.module('goog.format.EmailAddressTest');
 goog.setTestOnly();
 
 const EmailAddress = goog.require('goog.format.EmailAddress');
+const Format = goog.require('goog.i18n.bidi.Format');
 const googArray = goog.require('goog.array');
 const testSuite = goog.require('goog.testing.testSuite');
+const LRM = Format.LRM;
+const RLM = Format.RLM;
+const LRE = Format.LRE;
+const RLE = Format.RLE;
+const PDF = Format.PDF;
 
 function doIsValidTest(testFunc, valid, invalid) {
   googArray.forEach(valid, (str) => {
@@ -140,6 +146,18 @@ testSuite({
         'Failed to parse pre-OS X Mac newlines');
   },
 
+  testparseListBidiMarks() {
+    // These bidi marks can be copy pasted from an RTL formatted email
+    assertParsedList(
+        `ab ${LRE}${PDF}${RLM}<${LRE}a@b.com${PDF}${RLM}>` +
+            `${PDF}${RLM},c@d.com${PDF}${RLM} `,
+        ['a@b.com', 'c@d.com']);
+    assertParsedList(
+        `${PDF}ab ${RLE}${PDF}${LRM}<${RLE}a@b.com${PDF}${LRM}>` +
+            `${PDF}${LRM},c@d.com${PDF}${LRM} `,
+        ['a@b.com', 'c@d.com']);
+  },
+
   testToString() {
     const f = (str) => EmailAddress.parse(str).toString();
 
@@ -197,17 +215,11 @@ testSuite({
 
   testIsValid() {
     const valid = [
-      'e@b.eu',
-      '<a.b+foo@c.com>',
-      'eric <e@b.com>',
-      '"e" <e@b.com>',
-      'a@FOO.MUSEUM',
-      'bla@b.co.ac.uk',
-      'bla@a.b.com',
-      'o\'hara@gm.com',
-      'plus+is+allowed@gmail.com',
-      '!/#$%&\'*+-=~|`{}?^_@expample.com',
+      'e@b.eu', '<a.b+foo@c.com>', 'eric <e@b.com>', '"e" <e@b.com>',
+      'a@FOO.MUSEUM', 'bla@b.co.ac.uk', 'bla@a.b.com', 'o\'hara@gm.com',
+      'plus+is+allowed@gmail.com', '!/#$%&\'*+-=~|`{}?^_@expample.com',
       'confirm-bhk=modulo.org@yahoogroups.com',
+      `blah blahson ${LRE}${PDF}${RLM}<${LRE}blah@blah.com${PDF}${RLM}>`
     ];
     const invalid = [
       'e',