Fix crash in IdnSpoofChecker with certain characters

IdnSpoofChecker incorrectly assumed that a skeleton string shouldn't contain any dots. This isn't true, skeletons of some characters such as Extended Arabic-Indic Digit Zero (۰) is indeed a dot.

Remove this assumption and add a test case.

Bug: 877045
Change-Id: I6e73c4d2b850f614b8a685f5068d9cd1028f1b24
Reviewed-on: https://chromium-review.googlesource.com/c/1284058Reviewed-by: Tommy Li <tommycli@chromium.org>
Commit-Queue: Mustafa Emre Acer <meacer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#600528}

components/url_formatter/idn_spoof_checker.cc

@@ -88,7 +88,8 @@ IDNSpoofChecker::HuffmanTrieParams g_trie_params{
 
 std::string LookupMatchInTopDomains(const std::string& skeleton) {
   DCHECK(!skeleton.empty());
-  DCHECK_NE(skeleton.back(), '.');
+  // There are no other guarantees about a skeleton string such as not including
+  // a dot. Skeleton of certain characters are dots (e.g. "۰" (U+06F0)).
   TopDomainPreloadDecoder preload_decoder(
       g_trie_params.huffman_tree, g_trie_params.huffman_tree_size,
       g_trie_params.trie, g_trie_params.trie_bits,
@@ -366,6 +367,7 @@ bool IDNSpoofChecker::SafeToDisplayAsUnicode(base::StringPiece16 label,
 }
 
 std::string IDNSpoofChecker::GetSimilarTopDomain(base::StringPiece16 hostname) {
+  DCHECK(!hostname.empty());
   for (const std::string& skeleton : GetSkeletons(hostname)) {
     DCHECK(!skeleton.empty());
     std::string matching_top_domain = LookupMatchInTopDomains(skeleton);

components/url_formatter/url_formatter_unittest.cc

@@ -956,6 +956,10 @@ const IDNTestCase idn_cases[] = {
     {"xn--kx8a.com", L"\xa661.com", false},
     // Cyrillic Ext C: ᲂ.com (Narrow o)
     {"xn--43f.com", L"\x1c82.com", false},
+
+    // The skeleton of Extended Arabic-Indic Digit Zero (۰) is a dot. Check that
+    // this is handled correctly (crbug/877045).
+    {"xn--dmb", L"\x06f0", true},
 };
 
 struct AdjustOffsetCase {