Convert page contents grabbing from wide to UTF16. The current code is a bit...

chrome/browser/history/text_database_manager.cc

 // Copyright (c) 2010 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 "chrome/browser/history/text_database_manager.h"
 
 #include <string>
 
 #include "base/compiler_specific.h"
 #include "base/file_util.h"
@@ skipping 10 matching lines @@
 using base::TimeDelta;
 using base::TimeTicks;
 
 namespace history {
 
 namespace {
 
 // The number of database files we will be attached to at once.
 const int kCacheDBSize = 5;
 
-std::string ConvertStringForIndexer(
-    const std::wstring& input) {
+std::string ConvertStringForIndexer(const std::wstring& input) {
   // TODO(evanm): other transformations here?

[Nico, 2010/06/10 21:03:35]

Should this call the string16 version?

[brettw, 2010/06/11 21:01:04]

This is currently only used for titles. I'm OK having this very small amount of
duplication for now until we also fix titles.

   return WideToUTF8(CollapseWhitespace(input, false));
 }
 
+std::string ConvertStringForIndexer(const string16& input) {
+  // TODO(evanm): other transformations here?
+  return UTF16ToUTF8(CollapseWhitespace(input, false));
+}
+
 // Data older than this will be committed to the full text index even if we
 // haven't gotten a title and/or body.
 const int kExpirationSec = 20;
 
 }  // namespace
 
 // TextDatabaseManager::PageInfo -----------------------------------------------
 
 TextDatabaseManager::PageInfo::PageInfo(URLID url_id,
                                         VisitID visit_id,
                                         Time visit_time)
     : url_id_(url_id),
       visit_id_(visit_id),
       visit_time_(visit_time) {
   added_time_ = TimeTicks::Now();
 }
 
 void TextDatabaseManager::PageInfo::set_title(const std::wstring& ttl) {
   if (ttl.empty())  // Make the title nonempty when we set it for EverybodySet.
     title_ = L" ";
   else
     title_ = ttl;
 }
 
-void TextDatabaseManager::PageInfo::set_body(const std::wstring& bdy) {
+void TextDatabaseManager::PageInfo::set_body(const string16& bdy) {
   if (bdy.empty())  // Make the body nonempty when we set it for EverybodySet.
-    body_ = L" ";
+    body_ = ASCIIToUTF16(" ");
   else
     body_ = bdy;
 }
 
 bool TextDatabaseManager::PageInfo::Expired(TimeTicks now) const {
   return now - added_time_ > TimeDelta::FromSeconds(kExpirationSec);
 }
 
 // TextDatabaseManager ---------------------------------------------------------
 
@@ skipping 128 matching lines @@
       // current body, or have a special setter for only the title, but this is
       // not worth it for this edge case.
       //
       // It will be almost impossible for the title to take longer than
       // kExpirationSec yet we got a body in less than that time, since the
       // title should always come in first.
       return;
     }
 
     AddPageData(url, url_row.id(), visit.visit_id, visit.visit_time,
-                title, std::wstring());
+                title, string16());
     return;  // We don't know about this page, give up.
   }
 
   PageInfo& info = found->second;
   if (info.has_body()) {
     // This info is complete, write to the database.
     AddPageData(url, info.url_id(), info.visit_id(), info.visit_time(),
                 title, info.body());
     recent_changes_.Erase(found);
     return;
   }
 
   info.set_title(title);
 }
 
 void TextDatabaseManager::AddPageContents(const GURL& url,
-                                          const std::wstring& body) {
+                                          const string16& body) {
   RecentChangeList::iterator found = recent_changes_.Peek(url);
   if (found == recent_changes_.end()) {
     // This page is not in our cache of recent pages. This means that the page
     // took more than kExpirationSec to load. Often, this will be the result of
     // a very slow iframe or other resource on the page that makes us think its
     // still loading.
     //
     // As a fallback, set the most recent visit's contents using the input, and
     // use the last set title in the URL table as the title to index.
     URLRow url_row;
@@ skipping 19 matching lines @@
   }
 
   info.set_body(body);
 }
 
 bool TextDatabaseManager::AddPageData(const GURL& url,
                                       URLID url_id,
                                       VisitID visit_id,
                                       Time visit_time,
                                       const std::wstring& title,
-                                      const std::wstring& body) {
+                                      const string16& body) {
   TextDatabase* db = GetDBForTime(visit_time, true);
   if (!db)
     return false;
 
   TimeTicks beginning_time = TimeTicks::Now();
 
   // First delete any recently-indexed data for this page. This will delete
   // anything in the main database, but we don't bother looking through the
   // archived database.
   VisitVector visits;
@@ skipping 264 matching lines @@
   while (i != recent_changes_.rend() && i->second.Expired(now)) {
     AddPageData(i->first, i->second.url_id(), i->second.visit_id(),
                 i->second.visit_time(), i->second.title(), i->second.body());
     i = recent_changes_.Erase(i);
   }
 
   ScheduleFlushOldChanges();
 }
 
 }  // namespace history