[atoms] Fix getText atom for unicode charater middle of word (#8736)

* [atoms] Fix getText atom for unicode charater middle of word The change to getText atom by commit c065dda does not handle case when unicode character is in the middle of a word, and unicode character will be incorrectly capitalized (see https://bugs.chromium.org/p/chromedriver/issues/detail?id=3611). The problem is \b mark the boundary between word character and unicode character as a boundary. This is fixed by explicitly using unicode flag and specifying unicode character and unicode symbol. * [atoms] Add meta tag for charset in test html Add charset for utf-8 in text_test.html
SeleniumHQ · Oct 20, 2020 · cf49ba2 · cf49ba2
1 parent 004be30
commit cf49ba2
Show file tree

Hide file tree

Showing 2 changed files with 30 additions and 1 deletion.
diff --git a/javascript/atoms/dom.js b/javascript/atoms/dom.js
@@ -1175,7 +1175,7 @@ bot.dom.appendVisibleTextLinesFromTextNode_ = function(textNode, lines,
   }
 
   if (textTransform == 'capitalize') {
-    text = text.replace(/(^|\s|\b)(\S)/g, function() {
+    text = text.replace(/(^|[^\d\p{L}\p{S}])([\p{Ll}|\p{S}])/gu, function() {
       return arguments[1] + arguments[2].toUpperCase();
     });
   } else if (textTransform == 'uppercase') {

diff --git a/javascript/atoms/test/text_test.html b/javascript/atoms/test/text_test.html
@@ -18,6 +18,7 @@
 <html>
 <head>
   <title>text_test.html</title>
+  <meta charset="UTF-8">
   <script src="test_bootstrap.js"></script>
   <script type="text/javascript">
     goog.require('bot.dom');
@@ -262,6 +263,23 @@
       assertEquals("hello, world! bla-bla-bla", text);
       text = getVisibleTextByElementId("uppercased");
       assertEquals("HELLO, WORLD! BLA-BLA-BLA", text);
+
+      text = getVisibleTextByElementId("capitalized-1");
+      assertEquals("Äåìî", text);
+      text = getVisibleTextByElementId("capitalized-2");
+      assertEquals("Manipulowanie Przepływem", text);
+      text = getVisibleTextByElementId("capitalized-3");
+      assertEquals("Manipulowanie Przepływem", text);
+      text = getVisibleTextByElementId("capitalized-4");
+      assertEquals("Manipulowanie Pr0123z4epływem", text);
+      text = getVisibleTextByElementId("capitalized-5");
+      assertEquals("Lorem Ipsum Dolor Sit Amet, Consectetur Adipisicing Elit,", text);
+      text = getVisibleTextByElementId("capitalized-6");
+      assertEquals("Ⓐⓑⓒ (Ⓓⓔⓕ) —Ⓖⓗⓘ— Ⓙkl", text);
+      text = getVisibleTextByElementId("capitalized-7");
+      assertEquals('(This) “Is” [A] –Short– -Test- «For» *The* _Css_ ¿Capitalize? ?¡Transfor', text);
+      text = getVisibleTextByElementId("capitalized-8");
+      assertEquals('The Dutch Word: "Ijsland" Starts With A Digraph', text);
     }
 
     function getVisibleTextByElementId(id) {
@@ -403,5 +421,16 @@
     <a id="uppercased" style="text-transform: uppercase">hello, world! bla-bla-BLA</a><br/>
   </div>
 
+  <div>
+    <a lang="ru" id="capitalized-1" style="text-transform: capitalize">äåìî</a><br/>
+    <a id="capitalized-2" style="text-transform: capitalize">Manipulowanie przepływem</a><br/>
+    <a id="capitalized-3" style="text-transform: capitalize">manipulowanie przep&#x142ywem</a><br/>
+    <a id="capitalized-4" style="text-transform: capitalize">Manipulowanie pr0123z4epływem</a><br/>
+    <a id="capitalized-5" style="text-transform: capitalize">Lorem ipsum dolor sit amet, consectetur adipisicing elit,</a><br/>
+    <a id="capitalized-6" style="text-transform: capitalize">ⓐⓑⓒ (ⓓⓔⓕ) —ⓖⓗⓘ— ⓙkl</a><br/>
+    <a id="capitalized-7" style="text-transform: capitalize">(this) “is” [a] –short– -test- «for» *the* _css_ ¿capitalize? ?¡transfor</a><br/>
+    <a id="capitalized-8" style="text-transform: capitalize">The Dutch word: "ijsland" starts with a digraph</a><br/>
+  </div>
+
 </body>
 </html>