Importing certain bookmarks from firefox and HTML file as search engines.

chrome/utility/importer/bookmark_html_reader.cc

Index: chrome/utility/importer/bookmark_html_reader.cc
diff --git a/chrome/utility/importer/bookmark_html_reader.cc b/chrome/utility/importer/bookmark_html_reader.cc
index e11dc1d7761f68631aede24d93a78fe37839f162..c04bac9be1e2b11dca3f6606402d07269d7c8939 100644
--- a/chrome/utility/importer/bookmark_html_reader.cc
+++ b/chrome/utility/importer/bookmark_html_reader.cc
@@ -10,10 +10,13 @@
 #include "base/strings/string_number_conversions.h"
 #include "base/strings/string_split.h"
 #include "base/strings/string_util.h"
+#include "base/strings/utf_string_conversions.h"
 #include "base/time/time.h"
 #include "chrome/common/importer/imported_bookmark_entry.h"
 #include "chrome/common/importer/imported_favicon_usage.h"
 #include "chrome/utility/importer/favicon_reencode.h"
+#include "components/search_engines/search_terms_data.h"
+#include "components/search_engines/template_url.h"
 #include "net/base/data_url.h"
 #include "net/base/escape.h"
 #include "url/gurl.h"
@@ -91,6 +94,7 @@ void ImportBookmarksFile(
       const base::Callback<bool(const GURL&)>& valid_url_callback,
       const base::FilePath& file_path,
       std::vector<ImportedBookmarkEntry>* bookmarks,
+      std::vector<importer::SearchEngineInfo>* search_engines,
       std::vector<ImportedFaviconUsage>* favicons) {
   std::string content;
   base::ReadFileToString(file_path, &content);
@@ -150,6 +154,21 @@ void ImportBookmarksFile(
                                         &add_date, &post_data) ||
         internal::ParseMinimumBookmarkFromLine(line, charset, &title, &url);
 
+    std::string search_engine_url;

[Peter Kasting, 2014/11/17 21:30:36]

Nit: Declare this below the comment instead of above it

[Tapu Ghose, 2014/12/07 02:49:54]

Done.

+
+    // If bookmark contains a valid replaceable url and a keyword then import
+    // it as search engine.
+    if (is_bookmark && post_data.empty() &&
+        CanImportURLAsSearchEngine(url, &search_engine_url) &&
+            !shortcut.empty()) {
+      importer::SearchEngineInfo search_engine_info;
+      search_engine_info.url.assign(base::UTF8ToUTF16(search_engine_url));
+      search_engine_info.keyword = shortcut;
+      search_engine_info.display_name = title;
+      search_engines->push_back(search_engine_info);
+      continue;
+    }
+
     if (is_bookmark)
       last_folder_is_empty = false;
 
@@ -238,6 +257,26 @@ void ImportBookmarksFile(
   }
 }
 
+bool CanImportURLAsSearchEngine(const GURL& url,
+                                std::string* search_engine_url) {
+  std::string url_spec = url.possibly_invalid_spec();
+
+  if (url_spec.empty())
+    return false;
+
+  url_spec = net::UnescapeURLComponent(url_spec,
+                                       net::UnescapeRule::URL_SPECIAL_CHARS);

[Peter Kasting, 2014/11/17 21:30:36]

Do you have a testcase that will fail without this unescaping?  If not, please
add one.

[Tapu Ghose, 2014/12/07 02:49:54]

As I mentioned in Patch Set 4, unescaping the url is necessary. For instance,
url "http://example.%s.com" without unescape rule appears as
"http://example.%25s.com". If I skip unescaping the url then
|SupportsReplacement| at line 277 always returns false, because it looks for
"%s" (not %25s). In addition, unescaping is needed to show proper url in the UI
("Manager Search Engines" from chrome://settings).

Hence, all test cases that are supposed to pass would fail, if unescaping the
url is omitted.

+
+  // Replace replacement terms ("%s") in |url_spec| with {searchTerms}.
+  url_spec =
+      TemplateURLRef::DisplayURLToURLRef(base::UTF8ToUTF16(url_spec));
+
+  TemplateURLData data;
+  data.SetURL(url_spec);
+  *search_engine_url = url_spec;
+  return TemplateURL(data).SupportsReplacement(SearchTermsData());
+}
+
 namespace internal {
 
 bool ParseCharsetFromLine(const std::string& line, std::string* charset) {