Omnibox: Improve HQP Scoring for Terms that Start with Punctuation

components/omnibox/browser/scored_history_match_unittest.cc

Index: components/omnibox/browser/scored_history_match_unittest.cc
diff --git a/components/omnibox/browser/scored_history_match_unittest.cc b/components/omnibox/browser/scored_history_match_unittest.cc
index a4d776115fcc7dfe11775c195f60a585c35a309c..5e7a22f2887879c335559052bb52a710417ad5ff 100644
--- a/components/omnibox/browser/scored_history_match_unittest.cc
+++ b/components/omnibox/browser/scored_history_match_unittest.cc
@@ -9,6 +9,7 @@
 
 #include "base/auto_reset.h"
 #include "base/bind.h"
+#include "base/i18n/break_iterator.h"
 #include "base/memory/ptr_util.h"
 #include "base/strings/string16.h"
 #include "base/strings/utf_string_conversions.h"
@@ -119,8 +120,14 @@ float ScoredHistoryMatchTest::GetTopicalityScoreOfTermAgainstURLAndTitle(
   RowWordStarts word_starts;
   String16SetFromString16(url, &word_starts.url_word_starts_);
   String16SetFromString16(title, &word_starts.title_word_starts_);
-  WordStarts one_word_no_offset(1, 0u);
-  return scored_match.GetTopicalityScore(1, url, one_word_no_offset,
+  WordStarts term_word_starts(1, 0u);
+  base::i18n::BreakIterator iter(term, base::i18n::BreakIterator::BREAK_WORD);
+  if (iter.Init()) {
+    // Find the first word start.
+    while (iter.Advance() && !iter.IsWord()) {}
+    term_word_starts[0] = iter.prev();
+  }
+  return scored_match.GetTopicalityScore(1, url, term_word_starts,
                                          word_starts);
 }
 
@@ -554,23 +561,38 @@ TEST_F(ScoredHistoryMatchTest, GetTopicalityScore) {
   const float hostname_mid_word_score =
       GetTopicalityScoreOfTermAgainstURLAndTitle(ASCIIToUTF16("bc"), url,
                                                  title);
+  const float hostname_score_preceeding_punctuation =
+      GetTopicalityScoreOfTermAgainstURLAndTitle(ASCIIToUTF16("://abc"), url,
+                                                 title);
   const float domain_name_score = GetTopicalityScoreOfTermAgainstURLAndTitle(
       ASCIIToUTF16("def"), url, title);
   const float domain_name_mid_word_score =
       GetTopicalityScoreOfTermAgainstURLAndTitle(ASCIIToUTF16("ef"), url,
                                                  title);
+  const float domain_name_score_preceeding_dot =
+      GetTopicalityScoreOfTermAgainstURLAndTitle(ASCIIToUTF16(".def"), url,
+                                                 title);
   const float tld_score = GetTopicalityScoreOfTermAgainstURLAndTitle(
       ASCIIToUTF16("com"), url, title);
   const float tld_mid_word_score = GetTopicalityScoreOfTermAgainstURLAndTitle(
       ASCIIToUTF16("om"), url, title);
+  const float tld_score_preceeding_dot =
+      GetTopicalityScoreOfTermAgainstURLAndTitle(ASCIIToUTF16(".com"), url,
+                                                 title);
   const float path_score = GetTopicalityScoreOfTermAgainstURLAndTitle(
       ASCIIToUTF16("path1"), url, title);
   const float path_mid_word_score = GetTopicalityScoreOfTermAgainstURLAndTitle(
       ASCIIToUTF16("ath1"), url, title);
+  const float path_score_preceeding_slash =
+      GetTopicalityScoreOfTermAgainstURLAndTitle(ASCIIToUTF16("/path1"), url,
+                                                 title);
   const float arg_score = GetTopicalityScoreOfTermAgainstURLAndTitle(
-      ASCIIToUTF16("arg2"), url, title);
+      ASCIIToUTF16("arg1"), url, title);
   const float arg_mid_word_score = GetTopicalityScoreOfTermAgainstURLAndTitle(
-      ASCIIToUTF16("rg2"), url, title);
+      ASCIIToUTF16("rg1"), url, title);
+  const float arg_score_preceeding_question_mark =
+      GetTopicalityScoreOfTermAgainstURLAndTitle(ASCIIToUTF16("?arg1"), url,
+                                                 title);
   const float protocol_score = GetTopicalityScoreOfTermAgainstURLAndTitle(
       ASCIIToUTF16("htt"), url, title);
   const float protocol_mid_word_score =
@@ -584,6 +606,12 @@ TEST_F(ScoredHistoryMatchTest, GetTopicalityScore) {
   EXPECT_GT(hostname_score, path_score);
   EXPECT_GT(domain_name_score, path_score);
   EXPECT_GT(path_score, arg_score);
+  // Verify leading punctuation doesn't confuse scoring.
+  EXPECT_EQ(hostname_score, hostname_score_preceeding_punctuation);
+  EXPECT_EQ(domain_name_score, domain_name_score_preceeding_dot);
+  EXPECT_EQ(tld_score, tld_score_preceeding_dot);
+  EXPECT_EQ(path_score, path_score_preceeding_slash);
+  EXPECT_EQ(arg_score, arg_score_preceeding_question_mark);
   // Verify that domain name > path and domain name > arg for non-word
   // boundaries.
   EXPECT_GT(hostname_mid_word_score, path_mid_word_score);