Tweaking WebPageSerializerImpl to emit a BOM for UTF16/32.

chrome/browser/download/save_page_browsertest.cc

Index: chrome/browser/download/save_page_browsertest.cc
diff --git a/chrome/browser/download/save_page_browsertest.cc b/chrome/browser/download/save_page_browsertest.cc
index 6bd94ca14d58794d9d23d67e680f4f2d682f1565..23b3dd5f2cce4bbf7457d99f5065c274d29ba695 100644
--- a/chrome/browser/download/save_page_browsertest.cc
+++ b/chrome/browser/download/save_page_browsertest.cc
@@ -973,16 +973,27 @@ class SavePageMultiFrameBrowserTest
           nullptr, nullptr);
 
       EXPECT_EQ(1, actual_number_of_matches)
-          << "Verifying if \"" << expected_substring << "\" appears "
-          << "exactly once in the web-contents text";
+          << "Verifying that \"" << expected_substring << "\" appears "
+          << "exactly once in the text of web contents";
     }
 
-    int actual_number_of_errors = ui_test_utils::FindInPage(
-        GetCurrentTab(browser()), base::UTF8ToUTF16("err"),
-        true,   // |forward|
-        false,  // |case_sensitive|
-        nullptr, nullptr);
-    EXPECT_EQ(0, actual_number_of_errors);
+    std::vector<std::string> forbidden_substrings{
+        "head"  // Html markup should not be visible.
+        "err",  // "err" is a prefix of error messages + is included as text
+                // content of <object> elements in some test files (text content
+                // would be rendered in case the object itself doesn't work).
+    };
+    for (const auto& forbidden_substring : forbidden_substrings) {
+      int actual_number_of_matches = ui_test_utils::FindInPage(
+          GetCurrentTab(browser()), base::UTF8ToUTF16(forbidden_substring),
+          true,  // |forward|
+          true,  // |case_sensitive|
+          nullptr, nullptr);
+
+      EXPECT_EQ(0, actual_number_of_matches)
+          << "Verifying that \"" << forbidden_substring << "\" doesn't "
+          << "appear in the text of web contents";
+    }
   }
 
   static void IncrementInteger(int* i, content::RenderFrameHost* /* unused */) {
@@ -1076,6 +1087,40 @@ IN_PROC_BROWSER_TEST_P(SavePageMultiFrameBrowserTest, NestedFrames) {
   TestMultiFramePage(save_page_type, url, 3, expected_substrings);
 }
 
+// Test for saving frames with various encodings (crbug.com/541699).
+// - iso-8859-2: encoding declared via <meta> element
+// - utf16-le-bom.htm, utf16-be-bom.htm: encoding detected via BOM
+// - utf16-le-nobom.htm, utf16-le-nobom.htm, utf32.htm - encoding declared via
+//                                                       mocked http headers
+IN_PROC_BROWSER_TEST_P(SavePageMultiFrameBrowserTest, Encoding) {
+  content::SavePageType save_page_type = GetParam();
+
+  std::vector<std::string> expected_substrings{
+      "frames-encodings.htm: f53295dd-a95b-4b32-85f5-b6e15377fb20",
+      "iso-8859-2.htm: Zażółć gęślą jaźń",
+      "utf16-le-nobom.htm: Zażółć gęślą jaźń",
+      "utf16-le-bom.htm: Zażółć gęślą jaźń",
+      "utf16-be-nobom.htm: Zażółć gęślą jaźń",
+      "utf16-be-bom.htm: Zażółć gęślą jaźń",
+      // TODO(lukasza): Fix UTF32 (otoh it is low-priority compared with UTF16).
+      // "utf32.htm: Zażółć gęślą jaźń",
+  };
+
+  GURL url(embedded_test_server()->GetURL("a.com",
+                                          "/save_page/frames-encodings.htm"));
+
+  // TODO(lukasza): crbug.com/541699: MHTML needs to handle multi-byte encodings
+  // by either:
+  // 1. Continuing to preserve the original encoding, but starting to round-trip
+  //    the encoding declaration (in Content-Type MIME/MHTML header?)
+  // 2. Saving html docs in UTF8.
+  // 3. Saving the BOM (this won't help for all cases though).
+  if (save_page_type == content::SAVE_PAGE_TYPE_AS_MHTML)
+    return;
+
+  TestMultiFramePage(save_page_type, url, 7, expected_substrings);
+}
+
 INSTANTIATE_TEST_CASE_P(
     ,
     SavePageMultiFrameBrowserTest,