Introduces a search term extraction mechanism working for arbitrary search providers.

chrome/browser/search_engines/template_url_unittest.cc

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "base/base_paths.h"
 #include "base/string_util.h"
 #include "base/utf_string_conversions.h"
 #include "chrome/browser/browser_process.h"
 #include "chrome/browser/rlz/rlz.h"
 #include "chrome/browser/search_engines/search_terms_data.h"
@@ skipping 623 matching lines @@
   TemplateURL url(NULL, data);
   TemplateURLRef::Replacements replacements;
   bool valid = false;
   EXPECT_EQ("{",
             url.url_ref().ParseURL("{{searchTerms}", &replacements, &valid));
   ASSERT_EQ(1U, replacements.size());
   EXPECT_EQ(1U, replacements[0].index);
   EXPECT_EQ(TemplateURLRef::SEARCH_TERMS, replacements[0].type);
   EXPECT_TRUE(valid);
 }
+
+TEST_F(TemplateURLTest, DeserializeAndSetAlternateURLs) {
+  TemplateURLData data;
+  EXPECT_TRUE(data.alternate_urls().empty());
+
+  data.DeserializeAndSetAlternateURLs("http://google.com?q={searchTerms},"
+      "{google:baseURL}/#q={searchTerms}");
+  EXPECT_EQ(2U, data.alternate_urls().size());
+  EXPECT_EQ("http://google.com?q={searchTerms}", data.alternate_urls()[0]);
+  EXPECT_EQ("{google:baseURL}/#q={searchTerms}", data.alternate_urls()[1]);
+}
+
+
+TEST_F(TemplateURLTest, DeserializeAndSetEmptyAlternateURLs) {
+  TemplateURLData data;
+  EXPECT_TRUE(data.alternate_urls().empty());
+
+  data.DeserializeAndSetAlternateURLs("");
+  EXPECT_TRUE(data.alternate_urls().empty());
+}
+
+TEST_F(TemplateURLTest, SerializeAlternateURLs) {
+  TemplateURLData data;
+  EXPECT_TRUE(data.alternate_urls().empty());
+  data.alternate_urls_.push_back("{google:baseURL}/#q={searchTerms}");
+  data.alternate_urls_.push_back("http://google.com?q={searchTerms}");
+
+  std::string result = data.SerializeAlternateURLs();
+  EXPECT_EQ("{google:baseURL}/#q={searchTerms},"
+            "http://google.com?q={searchTerms}", result);
+}
+
+TEST_F(TemplateURLTest, SerializeEmptyAlternateURLs) {
+  TemplateURLData data;
+  EXPECT_TRUE(data.alternate_urls().empty());
+
+  std::string result = data.SerializeAlternateURLs();
+  EXPECT_EQ("", result);
+}
+
+TEST_F(TemplateURLTest, GetURL) {
+  TemplateURLData data;
+  data.SetURL("http://google.com/?q={searchTerms}");
+  data.suggestions_url = "http://google.com/suggest?q={searchTerms}";
+  data.instant_url = "http://google.com/instant#q={searchTerms}";
+  data.DeserializeAndSetAlternateURLs("http://google.com/alt?q={searchTerms},"
+      "{google:baseURL}/alt/#q={searchTerms}");
+  TemplateURL url(NULL, data);
+  ASSERT_EQ(4U, url.URLCount());
+  EXPECT_EQ("http://google.com/alt?q={searchTerms}", url.GetURL(0));
+  EXPECT_EQ("{google:baseURL}/alt/#q={searchTerms}", url.GetURL(1));
+  EXPECT_EQ("http://google.com/instant#q={searchTerms}", url.GetURL(2));
+  EXPECT_EQ("http://google.com/?q={searchTerms}", url.GetURL(3));
+}
+
+TEST_F(TemplateURLTest, GetURLNoInstantURL) {
+  TemplateURLData data;
+  data.SetURL("http://google.com/?q={searchTerms}");
+  data.suggestions_url = "http://google.com/suggest?q={searchTerms}";
+  data.DeserializeAndSetAlternateURLs("http://google.com/alt?q={searchTerms},"
+      "{google:baseURL}/alt/#q={searchTerms}");
+  TemplateURL url(NULL, data);
+  ASSERT_EQ(3U, url.URLCount());
+  EXPECT_EQ("http://google.com/alt?q={searchTerms}", url.GetURL(0));
+  EXPECT_EQ("{google:baseURL}/alt/#q={searchTerms}", url.GetURL(1));
+  EXPECT_EQ("http://google.com/?q={searchTerms}", url.GetURL(2));
+}
+
+TEST_F(TemplateURLTest, GetURLNoSuggestionsURL) {
+  TemplateURLData data;
+  data.SetURL("http://google.com/?q={searchTerms}");
+  data.instant_url = "http://google.com/instant#q={searchTerms}";
+  data.DeserializeAndSetAlternateURLs("http://google.com/alt?q={searchTerms},"
+      "{google:baseURL}/alt/#q={searchTerms}");
+  TemplateURL url(NULL, data);
+  ASSERT_EQ(4U, url.URLCount());
+  EXPECT_EQ("http://google.com/alt?q={searchTerms}", url.GetURL(0));
+  EXPECT_EQ("{google:baseURL}/alt/#q={searchTerms}", url.GetURL(1));
+  EXPECT_EQ("http://google.com/instant#q={searchTerms}", url.GetURL(2));
+  EXPECT_EQ("http://google.com/?q={searchTerms}", url.GetURL(3));
+}
+
+TEST_F(TemplateURLTest, GetURLOnlyOneURL) {
+  TemplateURLData data;
+  data.SetURL("http://www.google.co.uk/");
+  TemplateURL url(NULL, data);
+  ASSERT_EQ(1U, url.URLCount());
+  EXPECT_EQ("http://www.google.co.uk/", url.GetURL(0));
+}
+
+TEST_F(TemplateURLTest, ExtractSearchTermsFromURL) {

[dominich, 2012/09/12 22:23:12]

you should also test for blank search strings in both the query and ref
parameters.

[beaudoin, 2012/09/13 18:16:45]

Added them.

However, more importantly: what is the expected behavior in this situation?
Currently, the full URL will show since the empty string is used to represent
"no match".

[Peter Kasting, 2012/09/13 18:22:08]

I don't have the full context here.  Is the question what should happen with a
URL like "google.com/search?q="?  If so I would say we should fail to extract
search terms from that URL, as opposed to successfully extracting the empty
string.

[dominich, 2012/09/13 18:30:30]

Agreed. We don't want to have an empty string in the Omnibox.

[beaudoin, 2012/09/13 18:31:49]

Ok, right now it fails to match which means it returns "". The slight difference
between that and a successful match is that:

google.com/search?q=123#q=

will match "123".

+  TemplateURLData data;
+  data.SetURL("http://google.com/?q={searchTerms}");
+  data.instant_url = "http://google.com/instant#q={searchTerms}";
+  data.DeserializeAndSetAlternateURLs(
+      "http://google.com/alt/#q={searchTerms},"
+      "http://google.com/alt/?ext=foo&q={searchTerms}#ref=bar");
+  TemplateURL url(NULL, data);
+  EXPECT_EQ(ASCIIToUTF16("something"),
+      url.ExtractSearchTermsFromURL(GURL("http://google.com/?q=something")));
+  EXPECT_EQ(string16(),
+      url.ExtractSearchTermsFromURL(GURL("http://google.ca/?q=something")));
+  EXPECT_EQ(string16(),
+      url.ExtractSearchTermsFromURL(GURL("http://google.com/foo/?q=foo")));
+  EXPECT_EQ(string16(),
+      url.ExtractSearchTermsFromURL(GURL("https://google.com/?q=foo")));
+  EXPECT_EQ(string16(),
+      url.ExtractSearchTermsFromURL(GURL("http://google.com:8080/?q=foo")));
+  EXPECT_EQ(ASCIIToUTF16("1 2 3"),
+      url.ExtractSearchTermsFromURL(GURL("http://google.com/?q=1+2+3&b=456")));
+  EXPECT_EQ(ASCIIToUTF16("456"), url.ExtractSearchTermsFromURL(
+      GURL("http://google.com/alt/?q=123#q=456")));
+  EXPECT_EQ(ASCIIToUTF16("123"), url.ExtractSearchTermsFromURL(
+      GURL("http://google.com/alt/?a=012&q=123&b=456#f=789")));
+  EXPECT_EQ(ASCIIToUTF16("789"), url.ExtractSearchTermsFromURL(
+      GURL("http://google.com/alt/?a=012&q=123&b=456#j=abc&q=789&h=def")));
+
+}