chrome/utility/importer/ie_importer_win.cc

   1 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "chrome/utility/importer/ie_importer_win.h"
   6
   7 #include <ole2.h>
   8 #include <intshcut.h>
   9 #include <shlobj.h>
  10 #include <urlhist.h>
  11 #include <wininet.h>
  12
  13 #include <algorithm>
  14 #include <map>
  15 #include <string>
  16 #include <vector>
  17
  18 #include "base/files/file_enumerator.h"
  19 #include "base/files/file_path.h"
  20 #include "base/files/file_util.h"
  21 #include "base/strings/string16.h"
  22 #include "base/strings/string_split.h"
  23 #include "base/strings/string_util.h"
  24 #include "base/strings/utf_string_conversions.h"
  25 #include "base/time/time.h"
  26 #include "base/win/registry.h"
  27 #include "base/win/scoped_co_mem.h"
  28 #include "base/win/scoped_comptr.h"
  29 #include "base/win/scoped_handle.h"
  30 #include "base/win/scoped_propvariant.h"
  31 #include "base/win/windows_version.h"
  32 #include "chrome/common/importer/ie_importer_utils_win.h"
  33 #include "chrome/common/importer/imported_bookmark_entry.h"
  34 #include "chrome/common/importer/imported_favicon_usage.h"
  35 #include "chrome/common/importer/importer_bridge.h"
  36 #include "chrome/common/importer/importer_data_types.h"
  37 #include "chrome/common/importer/importer_url_row.h"
  38 #include "chrome/common/importer/pstore_declarations.h"
  39 #include "chrome/grit/generated_resources.h"
  40 #include "chrome/utility/importer/favicon_reencode.h"
  41 #include "components/autofill/core/common/password_form.h"
  42 #include "ui/base/l10n/l10n_util.h"
  43 #include "url/gurl.h"
  44 #include "url/url_constants.h"
  45
  46 namespace {
  47
  48 // Registry key paths from which we import IE settings.
  49 const base::char16 kSearchScopePath[] =
  50   L"Software\\Microsoft\\Internet Explorer\\SearchScopes";
  51 const base::char16 kIEVersionKey[] =
  52   L"Software\\Microsoft\\Internet Explorer";
  53 const base::char16 kIEToolbarKey[] =
  54   L"Software\\Microsoft\\Internet Explorer\\Toolbar";
  55
  56 // NTFS stream name of favicon image data.
  57 const base::char16 kFaviconStreamName[] = L":favicon:$DATA";
  58
  59 // A struct that hosts the information of AutoComplete data in PStore.
  60 struct AutoCompleteInfo {
  61   base::string16 key;
  62   std::vector<base::string16> data;
  63   bool is_url;
  64 };
  65
  66 // Gets the creation time of the given file or directory.
  67 base::Time GetFileCreationTime(const base::string16& file) {
  68   base::Time creation_time;
  69   base::win::ScopedHandle file_handle(
  70       CreateFile(file.c_str(), GENERIC_READ,
  71                  FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE,
  72                  NULL, OPEN_EXISTING,
  73                  FILE_ATTRIBUTE_NORMAL | FILE_FLAG_BACKUP_SEMANTICS, NULL));
  74   FILETIME creation_filetime;
  75   if (!file_handle.IsValid())
  76     return creation_time;
  77   if (GetFileTime(file_handle.Get(), &creation_filetime, NULL, NULL))
  78     creation_time = base::Time::FromFileTime(creation_filetime);
  79   return creation_time;
  80 }
  81
  82 // Safely read an object of type T from a raw sequence of bytes.
  83 template<typename T>
  84 bool BinaryRead(T* data, size_t offset, const std::vector<uint8>& blob) {
  85   if (offset + sizeof(T) > blob.size())
  86     return false;
  87   memcpy(data, &blob[offset], sizeof(T));
  88   return true;
  89 }
  90
  91 // Safely read an ITEMIDLIST from a raw sequence of bytes.
  92 //
  93 // An ITEMIDLIST is a list of SHITEMIDs, terminated by a SHITEMID with
  94 // .cb = 0. Here, before simply casting &blob[offset] to LPITEMIDLIST,
  95 // we verify that the list structure is not overrunning the boundary of
  96 // the binary blob.
  97 LPCITEMIDLIST BinaryReadItemIDList(size_t offset, size_t idlist_size,
  98                                    const std::vector<uint8>& blob) {
  99   size_t head = 0;
 100   while (true) {
 101     // Use a USHORT instead of SHITEMID to avoid buffer over read.
 102     USHORT id_cb;
 103     if (head >= idlist_size || !BinaryRead(&id_cb, offset + head, blob))
 104       return NULL;
 105     if (id_cb == 0)
 106       break;
 107     head += id_cb;
 108   }
 109   return reinterpret_cast<LPCITEMIDLIST>(&blob[offset]);
 110 }
 111
 112 // Compares the two bookmarks in the order of IE's Favorites menu.
 113 // Returns true if rhs should come later than lhs (lhs < rhs).
 114 struct IEOrderBookmarkComparator {
 115   bool operator()(const ImportedBookmarkEntry& lhs,
 116                   const ImportedBookmarkEntry& rhs) const {
 117     static const uint32 kNotSorted = 0xfffffffb; // IE uses this magic value.
 118     base::FilePath lhs_prefix;
 119     base::FilePath rhs_prefix;
 120     for (size_t i = 0; i <= lhs.path.size() && i <= rhs.path.size(); ++i) {
 121       const base::FilePath::StringType lhs_i =
 122         (i < lhs.path.size() ? lhs.path[i] : lhs.title + L".url");
 123       const base::FilePath::StringType rhs_i =
 124         (i < rhs.path.size() ? rhs.path[i] : rhs.title + L".url");
 125       lhs_prefix = lhs_prefix.Append(lhs_i);
 126       rhs_prefix = rhs_prefix.Append(rhs_i);
 127       if (lhs_i == rhs_i)
 128         continue;
 129       // The first path element that differs between the two.
 130       std::map<base::FilePath, uint32>::const_iterator lhs_iter =
 131         sort_index_->find(lhs_prefix);
 132       std::map<base::FilePath, uint32>::const_iterator rhs_iter =
 133         sort_index_->find(rhs_prefix);
 134       uint32 lhs_sort_index = (lhs_iter == sort_index_->end() ? kNotSorted
 135         : lhs_iter->second);
 136       uint32 rhs_sort_index = (rhs_iter == sort_index_->end() ? kNotSorted
 137         : rhs_iter->second);
 138       if (lhs_sort_index != rhs_sort_index)
 139         return lhs_sort_index < rhs_sort_index;
 140       // If they have the same sort order, sort alphabetically.
 141       return lhs_i < rhs_i;
 142     }
 143     return lhs.path.size() < rhs.path.size();
 144   }
 145   const std::map<base::FilePath, uint32>* sort_index_;
 146 };
 147
 148 // IE stores the order of the Favorites menu in registry under:
 149 // HKCU\Software\Microsoft\Windows\CurrentVersion\Explorer\MenuOrder\Favorites.
 150 // The folder hierarchy of Favorites menu is directly mapped to the key
 151 // hierarchy in the registry.
 152 //
 153 // If the order of the items in a folder is customized by user, the order is
 154 // recorded in the REG_BINARY value named "Order" of the corresponding key.
 155 // The content of the "Order" value is a raw binary dump of an array of the
 156 // following data structure
 157 //   struct {
 158 //     uint32 size;  // Note that ITEMIDLIST is variably-sized.
 159 //     uint32 sort_index;  // 0 means this is the first item, 1 the second, ...
 160 //     ITEMIDLIST item_id;
 161 //   };
 162 // where each item_id should correspond to a favorites link file (*.url) in
 163 // the current folder.
 164 bool ParseFavoritesOrderBlob(
 165     const Importer* importer,
 166     const std::vector<uint8>& blob,
 167     const base::FilePath& path,
 168     std::map<base::FilePath, uint32>* sort_index) WARN_UNUSED_RESULT {
 169   static const int kItemCountOffset = 16;
 170   static const int kItemListStartOffset = 20;
 171
 172   // Read the number of items.
 173   uint32 item_count = 0;
 174   if (!BinaryRead(&item_count, kItemCountOffset, blob))
 175     return false;
 176
 177   // Traverse over the items.
 178   size_t base_offset = kItemListStartOffset;
 179   for (uint32 i = 0; i < item_count && !importer->cancelled(); ++i) {
 180     static const int kSizeOffset = 0;
 181     static const int kSortIndexOffset = 4;
 182     static const int kItemIDListOffset = 8;
 183
 184     // Read the size (number of bytes) of the current item.
 185     uint32 item_size = 0;
 186     if (!BinaryRead(&item_size, base_offset + kSizeOffset, blob) ||
 187         base_offset + item_size <= base_offset || // checking overflow
 188         base_offset + item_size > blob.size())
 189       return false;
 190
 191     // Read the sort index of the current item.
 192     uint32 item_sort_index = 0;
 193     if (!BinaryRead(&item_sort_index, base_offset + kSortIndexOffset, blob))
 194       return false;
 195
 196     // Read the file name from the ITEMIDLIST structure.
 197     LPCITEMIDLIST idlist = BinaryReadItemIDList(
 198       base_offset + kItemIDListOffset, item_size - kItemIDListOffset, blob);
 199     TCHAR item_filename[MAX_PATH];
 200     if (!idlist || !SHGetPathFromIDList(idlist, item_filename))
 201       return false;
 202     base::FilePath item_relative_path =
 203       path.Append(base::FilePath(item_filename).BaseName());
 204
 205     // Record the retrieved information and go to the next item.
 206     sort_index->insert(std::make_pair(item_relative_path, item_sort_index));
 207     base_offset += item_size;
 208   }
 209   return true;
 210 }
 211
 212 bool ParseFavoritesOrderRegistryTree(
 213     const Importer* importer,
 214     const base::win::RegKey& key,
 215     const base::FilePath& path,
 216     std::map<base::FilePath, uint32>* sort_index) WARN_UNUSED_RESULT {
 217   // Parse the order information of the current folder.
 218   DWORD blob_length = 0;
 219   if (key.ReadValue(L"Order", NULL, &blob_length, NULL) == ERROR_SUCCESS) {
 220     std::vector<uint8> blob(blob_length);
 221     if (blob_length > 0 &&
 222         key.ReadValue(L"Order", reinterpret_cast<DWORD*>(&blob[0]),
 223                       &blob_length, NULL) == ERROR_SUCCESS) {
 224       if (!ParseFavoritesOrderBlob(importer, blob, path, sort_index))
 225         return false;
 226     }
 227   }
 228
 229   // Recursively parse subfolders.
 230   for (base::win::RegistryKeyIterator child(key.Handle(), L"");
 231        child.Valid() && !importer->cancelled();
 232        ++child) {
 233     base::win::RegKey subkey(key.Handle(), child.Name(), KEY_READ);
 234     if (subkey.Valid()) {
 235       base::FilePath subpath(path.Append(child.Name()));
 236       if (!ParseFavoritesOrderRegistryTree(importer, subkey, subpath,
 237                                            sort_index)) {
 238         return false;
 239       }
 240     }
 241   }
 242   return true;
 243 }
 244
 245 bool ParseFavoritesOrderInfo(
 246     const Importer* importer,
 247     std::map<base::FilePath, uint32>* sort_index) WARN_UNUSED_RESULT {
 248   base::string16 key_path(importer::GetIEFavoritesOrderKey());
 249   base::win::RegKey key(HKEY_CURRENT_USER, key_path.c_str(), KEY_READ);
 250   if (!key.Valid())
 251     return false;
 252   return ParseFavoritesOrderRegistryTree(importer, key, base::FilePath(),
 253                                          sort_index);
 254 }
 255
 256 // Reads the sort order from registry. If failed, we don't touch the list
 257 // and use the default (alphabetical) order.
 258 void SortBookmarksInIEOrder(
 259     const Importer* importer,
 260     std::vector<ImportedBookmarkEntry>* bookmarks) {
 261   std::map<base::FilePath, uint32> sort_index;
 262   if (!ParseFavoritesOrderInfo(importer, &sort_index))
 263     return;
 264   IEOrderBookmarkComparator compare = {&sort_index};
 265   std::sort(bookmarks->begin(), bookmarks->end(), compare);
 266 }
 267
 268 // Reads an internet shortcut (*.url) |file| and returns a COM object
 269 // representing it.
 270 bool LoadInternetShortcut(
 271     const base::string16& file,
 272     base::win::ScopedComPtr<IUniformResourceLocator>* shortcut) {
 273   base::win::ScopedComPtr<IUniformResourceLocator> url_locator;
 274   if (FAILED(url_locator.CreateInstance(CLSID_InternetShortcut, NULL,
 275                                         CLSCTX_INPROC_SERVER)))
 276     return false;
 277
 278   base::win::ScopedComPtr<IPersistFile> persist_file;
 279   if (FAILED(persist_file.QueryFrom(url_locator)))
 280     return false;
 281
 282   // Loads the Internet Shortcut from persistent storage.
 283   if (FAILED(persist_file->Load(file.c_str(), STGM_READ)))
 284     return false;
 285
 286   std::swap(url_locator, *shortcut);
 287   return true;
 288 }
 289
 290 // Reads the URL stored in the internet shortcut.
 291 GURL ReadURLFromInternetShortcut(IUniformResourceLocator* url_locator) {
 292   base::win::ScopedCoMem<wchar_t> url;
 293   // GetURL can return S_FALSE (FAILED(S_FALSE) is false) when url == NULL.
 294   return (FAILED(url_locator->GetURL(&url)) || !url) ?
 295       GURL() : GURL(base::WideToUTF16(std::wstring(url)));
 296 }
 297
 298 // Reads the URL of the favicon of the internet shortcut.
 299 GURL ReadFaviconURLFromInternetShortcut(IUniformResourceLocator* url_locator) {
 300   base::win::ScopedComPtr<IPropertySetStorage> property_set_storage;
 301   if (FAILED(property_set_storage.QueryFrom(url_locator)))
 302     return GURL();
 303
 304   base::win::ScopedComPtr<IPropertyStorage> property_storage;
 305   if (FAILED(property_set_storage->Open(FMTID_Intshcut, STGM_READ,
 306                                         property_storage.Receive()))) {
 307     return GURL();
 308   }
 309
 310   PROPSPEC properties[] = {{PRSPEC_PROPID, PID_IS_ICONFILE}};
 311   // ReadMultiple takes a non-const array of PROPVARIANTs, but since this code
 312   // only needs an array of size 1: a non-const pointer to |output| is
 313   // equivalent.
 314   base::win::ScopedPropVariant output;
 315   // ReadMultiple can return S_FALSE (FAILED(S_FALSE) is false) when the
 316   // property is not found, in which case output[0].vt is set to VT_EMPTY.
 317   if (FAILED(property_storage->ReadMultiple(1, properties, output.Receive())) ||
 318       output.get().vt != VT_LPWSTR)
 319     return GURL();
 320   return GURL(base::WideToUTF16(output.get().pwszVal));
 321 }
 322
 323 // Reads the favicon imaga data in an NTFS alternate data stream. This is where
 324 // IE7 and above store the data.
 325 bool ReadFaviconDataFromInternetShortcut(const base::string16& file,
 326                                          std::string* data) {
 327   return base::ReadFileToString(
 328       base::FilePath(file + kFaviconStreamName), data);
 329 }
 330
 331 // Reads the favicon imaga data in the Internet cache. IE6 doesn't hold the data
 332 // explicitly, but it might be found in the cache.
 333 bool ReadFaviconDataFromCache(const GURL& favicon_url, std::string* data) {
 334   std::wstring url_wstring(base::UTF8ToWide(favicon_url.spec()));
 335   DWORD info_size = 0;
 336   GetUrlCacheEntryInfoEx(url_wstring.c_str(), NULL, &info_size, NULL, NULL,
 337                          NULL, 0);
 338   if (GetLastError() != ERROR_INSUFFICIENT_BUFFER)
 339     return false;
 340
 341   std::vector<char> buf(info_size);
 342   INTERNET_CACHE_ENTRY_INFO* cache =
 343       reinterpret_cast<INTERNET_CACHE_ENTRY_INFO*>(&buf[0]);
 344   if (!GetUrlCacheEntryInfoEx(url_wstring.c_str(), cache, &info_size, NULL,
 345                               NULL, NULL, 0)) {
 346     return false;
 347   }
 348   return base::ReadFileToString(base::FilePath(cache->lpszLocalFileName), data);
 349 }
 350
 351 // Reads the binary image data of favicon of an internet shortcut file |file|.
 352 // |favicon_url| read by ReadFaviconURLFromInternetShortcut is also needed to
 353 // examine the IE cache.
 354 bool ReadReencodedFaviconData(const base::string16& file,
 355                               const GURL& favicon_url,
 356                               std::vector<unsigned char>* data) {
 357   std::string image_data;
 358   if (!ReadFaviconDataFromInternetShortcut(file, &image_data) &&
 359       !ReadFaviconDataFromCache(favicon_url, &image_data)) {
 360     return false;
 361   }
 362
 363   const unsigned char* ptr =
 364       reinterpret_cast<const unsigned char*>(image_data.c_str());
 365   return importer::ReencodeFavicon(ptr, image_data.size(), data);
 366 }
 367
 368 // Loads favicon image data and registers to |favicon_map|.
 369 void UpdateFaviconMap(const base::string16& url_file,
 370                       const GURL& url,
 371                       IUniformResourceLocator* url_locator,
 372                       std::map<GURL, ImportedFaviconUsage>* favicon_map) {
 373   GURL favicon_url = ReadFaviconURLFromInternetShortcut(url_locator);
 374   if (!favicon_url.is_valid())
 375     return;
 376
 377   std::map<GURL, ImportedFaviconUsage>::iterator it =
 378     favicon_map->find(favicon_url);
 379   if (it != favicon_map->end()) {
 380     // Known favicon URL.
 381     it->second.urls.insert(url);
 382   } else {
 383     // New favicon URL. Read the image data and store.
 384     ImportedFaviconUsage usage;
 385     if (ReadReencodedFaviconData(url_file, favicon_url, &usage.png_data)) {
 386       usage.favicon_url = favicon_url;
 387       usage.urls.insert(url);
 388       favicon_map->insert(std::make_pair(favicon_url, usage));
 389     }
 390   }
 391 }
 392
 393 }  // namespace
 394
 395 // static
 396 // {E161255A-37C3-11D2-BCAA-00C04fD929DB}
 397 const GUID IEImporter::kPStoreAutocompleteGUID = {
 398     0xe161255a, 0x37c3, 0x11d2,
 399     { 0xbc, 0xaa, 0x00, 0xc0, 0x4f, 0xd9, 0x29, 0xdb }
 400 };
 401 // {A79029D6-753E-4e27-B807-3D46AB1545DF}
 402 const GUID IEImporter::kUnittestGUID = {
 403     0xa79029d6, 0x753e, 0x4e27,
 404     { 0xb8, 0x7, 0x3d, 0x46, 0xab, 0x15, 0x45, 0xdf }
 405 };
 406
 407 IEImporter::IEImporter() {
 408 }
 409
 410 void IEImporter::StartImport(const importer::SourceProfile& source_profile,
 411                              uint16 items,
 412                              ImporterBridge* bridge) {
 413   bridge_ = bridge;
 414   source_path_ = source_profile.source_path;
 415
 416   bridge_->NotifyStarted();
 417
 418   if ((items & importer::HOME_PAGE) && !cancelled()) {
 419     bridge_->NotifyItemStarted(importer::HOME_PAGE);
 420     ImportHomepage();  // Doesn't have a UI item.
 421     bridge_->NotifyItemEnded(importer::HOME_PAGE);
 422   }
 423   // The order here is important!
 424   if ((items & importer::HISTORY) && !cancelled()) {
 425     bridge_->NotifyItemStarted(importer::HISTORY);
 426     ImportHistory();
 427     bridge_->NotifyItemEnded(importer::HISTORY);
 428   }
 429   if ((items & importer::FAVORITES) && !cancelled()) {
 430     bridge_->NotifyItemStarted(importer::FAVORITES);
 431     ImportFavorites();
 432     bridge_->NotifyItemEnded(importer::FAVORITES);
 433   }
 434   if ((items & importer::SEARCH_ENGINES) && !cancelled()) {
 435     bridge_->NotifyItemStarted(importer::SEARCH_ENGINES);
 436     ImportSearchEngines();
 437     bridge_->NotifyItemEnded(importer::SEARCH_ENGINES);
 438   }
 439   if ((items & importer::PASSWORDS) && !cancelled()) {
 440     bridge_->NotifyItemStarted(importer::PASSWORDS);
 441     // Always import IE6 passwords.
 442     ImportPasswordsIE6();
 443
 444     if (CurrentIEVersion() >= 7)
 445       ImportPasswordsIE7();
 446     bridge_->NotifyItemEnded(importer::PASSWORDS);
 447   }
 448   bridge_->NotifyEnded();
 449 }
 450
 451 IEImporter::~IEImporter() {
 452 }
 453
 454 void IEImporter::ImportFavorites() {
 455   FavoritesInfo info;
 456   if (!GetFavoritesInfo(&info))
 457     return;
 458
 459   BookmarkVector bookmarks;
 460   std::vector<ImportedFaviconUsage> favicons;
 461   ParseFavoritesFolder(info, &bookmarks, &favicons);
 462
 463   if (!bookmarks.empty() && !cancelled()) {
 464     const base::string16& first_folder_name =
 465         l10n_util::GetStringUTF16(IDS_BOOKMARK_GROUP_FROM_IE);
 466     bridge_->AddBookmarks(bookmarks, first_folder_name);
 467   }
 468   if (!favicons.empty() && !cancelled())
 469     bridge_->SetFavicons(favicons);
 470 }
 471
 472 void IEImporter::ImportHistory() {
 473   const std::string kSchemes[] = {url::kHttpScheme,
 474                                   url::kHttpsScheme,
 475                                   url::kFtpScheme,
 476                                   url::kFileScheme};
 477   int total_schemes = arraysize(kSchemes);
 478
 479   base::win::ScopedComPtr<IUrlHistoryStg2> url_history_stg2;
 480   HRESULT result;
 481   result = url_history_stg2.CreateInstance(CLSID_CUrlHistory, NULL,
 482                                            CLSCTX_INPROC_SERVER);
 483   if (FAILED(result))
 484     return;
 485   base::win::ScopedComPtr<IEnumSTATURL> enum_url;
 486   if (SUCCEEDED(result = url_history_stg2->EnumUrls(enum_url.Receive()))) {
 487     std::vector<ImporterURLRow> rows;
 488     STATURL stat_url;
 489     ULONG fetched;
 490     while (!cancelled() &&
 491            (result = enum_url->Next(1, &stat_url, &fetched)) == S_OK) {
 492       base::string16 url_string;
 493       if (stat_url.pwcsUrl) {
 494         url_string = stat_url.pwcsUrl;
 495         CoTaskMemFree(stat_url.pwcsUrl);
 496       }
 497       base::string16 title_string;
 498       if (stat_url.pwcsTitle) {
 499         title_string = stat_url.pwcsTitle;
 500         CoTaskMemFree(stat_url.pwcsTitle);
 501       }
 502
 503       GURL url(url_string);
 504       // Skips the URLs that are invalid or have other schemes.
 505       if (!url.is_valid() ||
 506           (std::find(kSchemes, kSchemes + total_schemes, url.scheme()) ==
 507            kSchemes + total_schemes))
 508         continue;
 509
 510       ImporterURLRow row(url);
 511       row.title = title_string;
 512       row.last_visit = base::Time::FromFileTime(stat_url.ftLastVisited);
 513       if (stat_url.dwFlags == STATURL_QUERYFLAG_TOPLEVEL) {
 514         row.visit_count = 1;
 515         row.hidden = false;
 516       } else {
 517         row.hidden = true;
 518       }
 519
 520       rows.push_back(row);
 521     }
 522
 523     if (!rows.empty() && !cancelled()) {
 524       bridge_->SetHistoryItems(rows, importer::VISIT_SOURCE_IE_IMPORTED);
 525     }
 526   }
 527 }
 528
 529 void IEImporter::ImportPasswordsIE6() {
 530   GUID AutocompleteGUID = kPStoreAutocompleteGUID;
 531   if (!source_path_.empty()) {
 532     // We supply a fake GUID for testting.
 533     AutocompleteGUID = kUnittestGUID;
 534   }
 535
 536   // The PStoreCreateInstance function retrieves an interface pointer
 537   // to a storage provider. But this function has no associated import
 538   // library or header file, we must call it using the LoadLibrary()
 539   // and GetProcAddress() functions.
 540   typedef HRESULT (WINAPI *PStoreCreateFunc)(IPStore**, DWORD, DWORD, DWORD);
 541   HMODULE pstorec_dll = LoadLibrary(L"pstorec.dll");
 542   if (!pstorec_dll)
 543     return;
 544   PStoreCreateFunc PStoreCreateInstance =
 545       (PStoreCreateFunc)GetProcAddress(pstorec_dll, "PStoreCreateInstance");
 546   if (!PStoreCreateInstance) {
 547     FreeLibrary(pstorec_dll);
 548     return;
 549   }
 550
 551   base::win::ScopedComPtr<IPStore, &IID_IPStore> pstore;
 552   HRESULT result = PStoreCreateInstance(pstore.Receive(), 0, 0, 0);
 553   if (result != S_OK) {
 554     FreeLibrary(pstorec_dll);
 555     return;
 556   }
 557
 558   std::vector<AutoCompleteInfo> ac_list;
 559
 560   // Enumerates AutoComplete items in the protected database.
 561   base::win::ScopedComPtr<IEnumPStoreItems, &IID_IEnumPStoreItems> item;
 562   result = pstore->EnumItems(0, &AutocompleteGUID,
 563                              &AutocompleteGUID, 0, item.Receive());
 564   if (result != PST_E_OK) {
 565     pstore.Release();
 566     FreeLibrary(pstorec_dll);
 567     return;
 568   }
 569
 570   wchar_t* item_name;
 571   while (!cancelled() && SUCCEEDED(item->Next(1, &item_name, 0))) {
 572     DWORD length = 0;
 573     unsigned char* buffer = NULL;
 574     result = pstore->ReadItem(0, &AutocompleteGUID, &AutocompleteGUID,
 575                               item_name, &length, &buffer, NULL, 0);
 576     if (SUCCEEDED(result)) {
 577       AutoCompleteInfo ac;
 578       ac.key = item_name;
 579       base::string16 data;
 580       data.insert(0, reinterpret_cast<wchar_t*>(buffer),
 581                   length / sizeof(wchar_t));
 582
 583       // The key name is always ended with ":StringData".
 584       const wchar_t kDataSuffix[] = L":StringData";
 585       size_t i = ac.key.rfind(kDataSuffix);
 586       if (i != base::string16::npos && ac.key.substr(i) == kDataSuffix) {
 587         ac.key.erase(i);
 588         ac.is_url = (ac.key.find(L"://") != base::string16::npos);
 589         ac_list.push_back(ac);
 590         base::SplitString(data, L'\0', &ac_list[ac_list.size() - 1].data);
 591       }
 592       CoTaskMemFree(buffer);
 593     }
 594     CoTaskMemFree(item_name);
 595   }
 596   // Releases them before unload the dll.
 597   item.Release();
 598   pstore.Release();
 599   FreeLibrary(pstorec_dll);
 600
 601   size_t i;
 602   for (i = 0; i < ac_list.size(); i++) {
 603     if (!ac_list[i].is_url || ac_list[i].data.size() < 2)
 604       continue;
 605
 606     GURL url(ac_list[i].key.c_str());
 607     if (!(LowerCaseEqualsASCII(url.scheme(), url::kHttpScheme) ||
 608           LowerCaseEqualsASCII(url.scheme(), url::kHttpsScheme))) {
 609       continue;
 610     }
 611
 612     autofill::PasswordForm form;
 613     GURL::Replacements rp;
 614     rp.ClearUsername();
 615     rp.ClearPassword();
 616     rp.ClearQuery();
 617     rp.ClearRef();
 618     form.origin = url.ReplaceComponents(rp);
 619     form.username_value = ac_list[i].data[0];
 620     form.password_value = ac_list[i].data[1];
 621     form.signon_realm = url.GetOrigin().spec();
 622
 623     // This is not precise, because a scheme of https does not imply a valid
 624     // certificate was presented; however we assign it this way so that if we
 625     // import a password from IE whose scheme is https, we give it the benefit
 626     // of the doubt and DONT auto-fill it unless the form appears under
 627     // valid SSL conditions.
 628     form.ssl_valid = url.SchemeIsSecure();
 629
 630     // Goes through the list to find out the username field
 631     // of the web page.
 632     size_t list_it, item_it;
 633     for (list_it = 0; list_it < ac_list.size(); ++list_it) {
 634       if (ac_list[list_it].is_url)
 635         continue;
 636
 637       for (item_it = 0; item_it < ac_list[list_it].data.size(); ++item_it)
 638         if (ac_list[list_it].data[item_it] == form.username_value) {
 639           form.username_element = ac_list[list_it].key;
 640           break;
 641         }
 642     }
 643
 644     bridge_->SetPasswordForm(form);
 645   }
 646 }
 647
 648 void IEImporter::ImportPasswordsIE7() {
 649   base::string16 key_path(importer::GetIE7PasswordsKey());
 650   base::win::RegKey key(HKEY_CURRENT_USER, key_path.c_str(), KEY_READ);
 651   base::win::RegistryValueIterator reg_iterator(HKEY_CURRENT_USER,
 652                                                 key_path.c_str());
 653   importer::ImporterIE7PasswordInfo password_info;
 654   while (reg_iterator.Valid() && !cancelled()) {
 655     // Get the size of the encrypted data.
 656     DWORD value_len = 0;
 657     key.ReadValue(reg_iterator.Name(), NULL, &value_len, NULL);
 658     if (value_len) {
 659       // Query the encrypted data.
 660       password_info.encrypted_data.resize(value_len);
 661       if (key.ReadValue(reg_iterator.Name(),
 662                         &password_info.encrypted_data.front(),
 663                         &value_len, NULL) == ERROR_SUCCESS) {
 664         password_info.url_hash = reg_iterator.Name();
 665         password_info.date_created = base::Time::Now();
 666
 667         bridge_->AddIE7PasswordInfo(password_info);
 668       }
 669     }
 670
 671     ++reg_iterator;
 672   }
 673 }
 674
 675 void IEImporter::ImportSearchEngines() {
 676   // On IE, search engines are stored in the registry, under:
 677   // Software\Microsoft\Internet Explorer\SearchScopes
 678   // Each key represents a search engine. The URL value contains the URL and
 679   // the DisplayName the name.
 680   typedef std::map<std::string, base::string16> SearchEnginesMap;
 681   SearchEnginesMap search_engines_map;
 682   for (base::win::RegistryKeyIterator key_iter(HKEY_CURRENT_USER,
 683        kSearchScopePath); key_iter.Valid(); ++key_iter) {
 684     base::string16 sub_key_name = kSearchScopePath;
 685     sub_key_name.append(L"\\").append(key_iter.Name());
 686     base::win::RegKey sub_key(HKEY_CURRENT_USER, sub_key_name.c_str(),
 687                               KEY_READ);
 688     base::string16 wide_url;
 689     if ((sub_key.ReadValue(L"URL", &wide_url) != ERROR_SUCCESS) ||
 690         wide_url.empty()) {
 691       VLOG(1) << "No URL for IE search engine at " << key_iter.Name();
 692       continue;
 693     }
 694     // For the name, we try the default value first (as Live Search uses a
 695     // non displayable name in DisplayName, and the readable name under the
 696     // default value).
 697     base::string16 name;
 698     if ((sub_key.ReadValue(NULL, &name) != ERROR_SUCCESS) || name.empty()) {
 699       // Try the displayable name.
 700       if ((sub_key.ReadValue(L"DisplayName", &name) != ERROR_SUCCESS) ||
 701           name.empty()) {
 702         VLOG(1) << "No name for IE search engine at " << key_iter.Name();
 703         continue;
 704       }
 705     }
 706
 707     std::string url(base::WideToUTF8(wide_url));
 708     SearchEnginesMap::iterator t_iter = search_engines_map.find(url);
 709     if (t_iter == search_engines_map.end()) {
 710       // First time we see that URL.
 711       GURL gurl(url);
 712       if (gurl.is_valid()) {
 713         t_iter = search_engines_map.insert(std::make_pair(url, name)).first;
 714       }
 715     }
 716   }
 717   // ProfileWriter::AddKeywords() requires a vector and we have a map.
 718   std::vector<importer::URLKeywordInfo> url_keywords;
 719   for (SearchEnginesMap::iterator i = search_engines_map.begin();
 720        i != search_engines_map.end(); ++i) {
 721     importer::URLKeywordInfo url_keyword_info;
 722     url_keyword_info.url = GURL(i->first);
 723     url_keyword_info.display_name = i->second;
 724     url_keywords.push_back(url_keyword_info);
 725   }
 726   bridge_->SetKeywords(url_keywords, true);
 727 }
 728
 729 void IEImporter::ImportHomepage() {
 730   const wchar_t* kIEHomepage = L"Start Page";
 731   const wchar_t* kIEDefaultHomepage = L"Default_Page_URL";
 732
 733   base::string16 key_path(importer::GetIESettingsKey());
 734
 735   base::win::RegKey key(HKEY_CURRENT_USER, key_path.c_str(), KEY_READ);
 736   base::string16 homepage_url;
 737   if (key.ReadValue(kIEHomepage, &homepage_url) != ERROR_SUCCESS ||
 738       homepage_url.empty())
 739     return;
 740
 741   GURL homepage = GURL(homepage_url);
 742   if (!homepage.is_valid())
 743     return;
 744
 745   // Check to see if this is the default website and skip import.
 746   base::win::RegKey keyDefault(HKEY_LOCAL_MACHINE, key_path.c_str(), KEY_READ);
 747   base::string16 default_homepage_url;
 748   LONG result = keyDefault.ReadValue(kIEDefaultHomepage, &default_homepage_url);
 749   if (result == ERROR_SUCCESS && !default_homepage_url.empty()) {
 750     if (homepage.spec() == GURL(default_homepage_url).spec())
 751       return;
 752   }
 753   bridge_->AddHomePage(homepage);
 754 }
 755
 756 bool IEImporter::GetFavoritesInfo(IEImporter::FavoritesInfo* info) {
 757   if (!source_path_.empty()) {
 758     // Source path exists during testing.
 759     info->path = source_path_;
 760     info->path = info->path.AppendASCII("Favorites");
 761     info->links_folder = L"Links";
 762     return true;
 763   }
 764
 765   // IE stores the favorites in the Favorites under user profile's folder.
 766   wchar_t buffer[MAX_PATH];
 767   if (FAILED(SHGetFolderPath(NULL, CSIDL_FAVORITES, NULL,
 768                              SHGFP_TYPE_CURRENT, buffer)))
 769     return false;
 770   info->path = base::FilePath(buffer);
 771
 772   // There is a Links folder under Favorites folder in Windows Vista, but it
 773   // is not recording in Vista's registry. So in Vista, we assume the Links
 774   // folder is under Favorites folder since it looks like there is not name
 775   // different in every language version of Windows Vista.
 776   if (base::win::GetVersion() < base::win::VERSION_VISTA) {
 777     // The Link folder name is stored in the registry.
 778     DWORD buffer_length = sizeof(buffer);
 779     base::win::RegKey reg_key(HKEY_CURRENT_USER, kIEToolbarKey, KEY_READ);
 780     if (reg_key.ReadValue(L"LinksFolderName", buffer,
 781                           &buffer_length, NULL) != ERROR_SUCCESS)
 782       return false;
 783     info->links_folder = buffer;
 784   } else {
 785     info->links_folder = L"Links";
 786   }
 787
 788   return true;
 789 }
 790
 791 void IEImporter::ParseFavoritesFolder(
 792     const FavoritesInfo& info,
 793     BookmarkVector* bookmarks,
 794     std::vector<ImportedFaviconUsage>* favicons) {
 795   base::FilePath file;
 796   std::vector<base::FilePath::StringType> file_list;
 797   base::FilePath favorites_path(info.path);
 798   // Favorites path length.  Make sure it doesn't include the trailing \.
 799   size_t favorites_path_len =
 800       favorites_path.StripTrailingSeparators().value().size();
 801   base::FileEnumerator file_enumerator(
 802       favorites_path, true, base::FileEnumerator::FILES);
 803   while (!(file = file_enumerator.Next()).value().empty() && !cancelled())
 804     file_list.push_back(file.value());
 805
 806   // Keep the bookmarks in alphabetical order.
 807   std::sort(file_list.begin(), file_list.end());
 808
 809   // Map from favicon URLs to the favicon data (the binary image data and the
 810   // set of bookmark URLs referring to the favicon).
 811   typedef std::map<GURL, ImportedFaviconUsage> FaviconMap;
 812   FaviconMap favicon_map;
 813
 814   for (std::vector<base::FilePath::StringType>::iterator it = file_list.begin();
 815        it != file_list.end(); ++it) {
 816     base::FilePath shortcut(*it);
 817     if (!LowerCaseEqualsASCII(shortcut.Extension(), ".url"))
 818       continue;
 819
 820     // Skip the bookmark with invalid URL.
 821     base::win::ScopedComPtr<IUniformResourceLocator> url_locator;
 822     if (!LoadInternetShortcut(*it, &url_locator))
 823       continue;
 824     GURL url = ReadURLFromInternetShortcut(url_locator);
 825     if (!url.is_valid())
 826       continue;
 827     // Skip default bookmarks. go.microsoft.com redirects to
 828     // search.microsoft.com, and http://go.microsoft.com/fwlink/?LinkId=XXX,
 829     // which URLs IE has as default, to some another sites.
 830     // We expect that users will never themselves create bookmarks having this
 831     // hostname.
 832     if (url.host() == "go.microsoft.com")
 833       continue;
 834     // Read favicon.
 835     UpdateFaviconMap(*it, url, url_locator, &favicon_map);
 836
 837     // Make the relative path from the Favorites folder, without the basename.
 838     // ex. Suppose that the Favorites folder is C:\Users\Foo\Favorites.
 839     //   C:\Users\Foo\Favorites\Foo.url -> ""
 840     //   C:\Users\Foo\Favorites\Links\Bar\Baz.url -> "Links\Bar"
 841     base::FilePath::StringType relative_string =
 842         shortcut.DirName().value().substr(favorites_path_len);
 843     if (!relative_string.empty() &&
 844         base::FilePath::IsSeparator(relative_string[0]))
 845       relative_string = relative_string.substr(1);
 846     base::FilePath relative_path(relative_string);
 847
 848     ImportedBookmarkEntry entry;
 849     // Remove the dot, the file extension, and the directory path.
 850     entry.title = shortcut.RemoveExtension().BaseName().value();
 851     entry.url = url;
 852     entry.creation_time = GetFileCreationTime(*it);
 853     if (!relative_path.empty())
 854       relative_path.GetComponents(&entry.path);
 855
 856     // Add the bookmark.
 857     if (!entry.path.empty() && entry.path[0] == info.links_folder) {
 858       // Bookmarks in the Link folder should be imported to the toolbar.
 859       entry.in_toolbar = true;
 860     }
 861     bookmarks->push_back(entry);
 862   }
 863
 864   // Reflect the menu order in IE.
 865   SortBookmarksInIEOrder(this, bookmarks);
 866
 867   // Record favicon data.
 868   for (FaviconMap::iterator iter = favicon_map.begin();
 869        iter != favicon_map.end(); ++iter)
 870     favicons->push_back(iter->second);
 871 }
 872
 873 int IEImporter::CurrentIEVersion() const {
 874   static int version = -1;
 875   if (version < 0) {
 876     wchar_t buffer[128];
 877     DWORD buffer_length = sizeof(buffer);
 878     base::win::RegKey reg_key(HKEY_LOCAL_MACHINE, kIEVersionKey, KEY_READ);
 879     LONG result = reg_key.ReadValue(L"Version", buffer, &buffer_length, NULL);
 880     version = ((result == ERROR_SUCCESS)? _wtoi(buffer) : 0);
 881   }
 882   return version;
 883 }