ie_importer_win.cc revision 58537e28ecd584eab876aee8be7156509866d23a 
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/file_util.h"
19#include "base/files/file_enumerator.h"
20#include "base/files/file_path.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/common/url_constants.h"
40#include "chrome/utility/importer/favicon_reencode.h"
41#include "components/autofill/core/common/password_form.h"
42#include "grit/generated_resources.h"
43#include "ui/base/l10n/l10n_util.h"
44#include "url/gurl.h"
45
46namespace {
47
48// Registry key paths from which we import IE settings.
49const char16 kSearchScopePath[] =
50  L"Software\\Microsoft\\Internet Explorer\\SearchScopes";
51const char16 kIEVersionKey[] =
52  L"Software\\Microsoft\\Internet Explorer";
53const char16 kIEToolbarKey[] =
54  L"Software\\Microsoft\\Internet Explorer\\Toolbar";
55
56// NTFS stream name of favicon image data.
57const char16 kFaviconStreamName[] = L":favicon:$DATA";
58
59// A struct that hosts the information of AutoComplete data in PStore.
60struct AutoCompleteInfo {
61  string16 key;
62  std::vector<string16> data;
63  bool is_url;
64};
65
66// Gets the creation time of the given file or directory.
67base::Time GetFileCreationTime(const 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 (GetFileTime(file_handle, &creation_filetime, NULL, NULL))
76    creation_time = base::Time::FromFileTime(creation_filetime);
77  return creation_time;
78}
79
80// Safely read an object of type T from a raw sequence of bytes.
81template<typename T>
82bool BinaryRead(T* data, size_t offset, const std::vector<uint8>& blob) {
83  if (offset + sizeof(T) > blob.size())
84    return false;
85  memcpy(data, &blob[offset], sizeof(T));
86  return true;
87}
88
89// Safely read an ITEMIDLIST from a raw sequence of bytes.
90//
91// An ITEMIDLIST is a list of SHITEMIDs, terminated by a SHITEMID with
92// .cb = 0. Here, before simply casting &blob[offset] to LPITEMIDLIST,
93// we verify that the list structure is not overrunning the boundary of
94// the binary blob.
95LPCITEMIDLIST BinaryReadItemIDList(size_t offset, size_t idlist_size,
96                                   const std::vector<uint8>& blob) {
97  size_t head = 0;
98  while (true) {
99    // Use a USHORT instead of SHITEMID to avoid buffer over read.
100    USHORT id_cb;
101    if (head >= idlist_size || !BinaryRead(&id_cb, offset + head, blob))
102      return NULL;
103    if (id_cb == 0)
104      break;
105    head += id_cb;
106  }
107  return reinterpret_cast<LPCITEMIDLIST>(&blob[offset]);
108}
109
110// Compares the two bookmarks in the order of IE's Favorites menu.
111// Returns true if rhs should come later than lhs (lhs < rhs).
112struct IEOrderBookmarkComparator {
113  bool operator()(const ImportedBookmarkEntry& lhs,
114                  const ImportedBookmarkEntry& rhs) const {
115    static const uint32 kNotSorted = 0xfffffffb; // IE uses this magic value.
116    base::FilePath lhs_prefix;
117    base::FilePath rhs_prefix;
118    for (size_t i = 0; i <= lhs.path.size() && i <= rhs.path.size(); ++i) {
119      const base::FilePath::StringType lhs_i =
120        (i < lhs.path.size() ? lhs.path[i] : lhs.title + L".url");
121      const base::FilePath::StringType rhs_i =
122        (i < rhs.path.size() ? rhs.path[i] : rhs.title + L".url");
123      lhs_prefix = lhs_prefix.Append(lhs_i);
124      rhs_prefix = rhs_prefix.Append(rhs_i);
125      if (lhs_i == rhs_i)
126        continue;
127      // The first path element that differs between the two.
128      std::map<base::FilePath, uint32>::const_iterator lhs_iter =
129        sort_index_->find(lhs_prefix);
130      std::map<base::FilePath, uint32>::const_iterator rhs_iter =
131        sort_index_->find(rhs_prefix);
132      uint32 lhs_sort_index = (lhs_iter == sort_index_->end() ? kNotSorted
133        : lhs_iter->second);
134      uint32 rhs_sort_index = (rhs_iter == sort_index_->end() ? kNotSorted
135        : rhs_iter->second);
136      if (lhs_sort_index != rhs_sort_index)
137        return lhs_sort_index < rhs_sort_index;
138      // If they have the same sort order, sort alphabetically.
139      return lhs_i < rhs_i;
140    }
141    return lhs.path.size() < rhs.path.size();
142  }
143  const std::map<base::FilePath, uint32>* sort_index_;
144};
145
146// IE stores the order of the Favorites menu in registry under:
147// HKCU\Software\Microsoft\Windows\CurrentVersion\Explorer\MenuOrder\Favorites.
148// The folder hierarchy of Favorites menu is directly mapped to the key
149// hierarchy in the registry.
150//
151// If the order of the items in a folder is customized by user, the order is
152// recorded in the REG_BINARY value named "Order" of the corresponding key.
153// The content of the "Order" value is a raw binary dump of an array of the
154// following data structure
155//   struct {
156//     uint32 size;  // Note that ITEMIDLIST is variably-sized.
157//     uint32 sort_index;  // 0 means this is the first item, 1 the second, ...
158//     ITEMIDLIST item_id;
159//   };
160// where each item_id should correspond to a favorites link file (*.url) in
161// the current folder.
162bool ParseFavoritesOrderBlob(
163    const Importer* importer,
164    const std::vector<uint8>& blob,
165    const base::FilePath& path,
166    std::map<base::FilePath, uint32>* sort_index) WARN_UNUSED_RESULT {
167  static const int kItemCountOffset = 16;
168  static const int kItemListStartOffset = 20;
169
170  // Read the number of items.
171  uint32 item_count = 0;
172  if (!BinaryRead(&item_count, kItemCountOffset, blob))
173    return false;
174
175  // Traverse over the items.
176  size_t base_offset = kItemListStartOffset;
177  for (uint32 i = 0; i < item_count && !importer->cancelled(); ++i) {
178    static const int kSizeOffset = 0;
179    static const int kSortIndexOffset = 4;
180    static const int kItemIDListOffset = 8;
181
182    // Read the size (number of bytes) of the current item.
183    uint32 item_size = 0;
184    if (!BinaryRead(&item_size, base_offset + kSizeOffset, blob) ||
185        base_offset + item_size <= base_offset || // checking overflow
186        base_offset + item_size > blob.size())
187      return false;
188
189    // Read the sort index of the current item.
190    uint32 item_sort_index = 0;
191    if (!BinaryRead(&item_sort_index, base_offset + kSortIndexOffset, blob))
192      return false;
193
194    // Read the file name from the ITEMIDLIST structure.
195    LPCITEMIDLIST idlist = BinaryReadItemIDList(
196      base_offset + kItemIDListOffset, item_size - kItemIDListOffset, blob);
197    TCHAR item_filename[MAX_PATH];
198    if (!idlist || FAILED(SHGetPathFromIDList(idlist, item_filename)))
199      return false;
200    base::FilePath item_relative_path =
201      path.Append(base::FilePath(item_filename).BaseName());
202
203    // Record the retrieved information and go to the next item.
204    sort_index->insert(std::make_pair(item_relative_path, item_sort_index));
205    base_offset += item_size;
206  }
207  return true;
208}
209
210bool ParseFavoritesOrderRegistryTree(
211    const Importer* importer,
212    const base::win::RegKey& key,
213    const base::FilePath& path,
214    std::map<base::FilePath, uint32>* sort_index) WARN_UNUSED_RESULT {
215  // Parse the order information of the current folder.
216  DWORD blob_length = 0;
217  if (key.ReadValue(L"Order", NULL, &blob_length, NULL) == ERROR_SUCCESS) {
218    std::vector<uint8> blob(blob_length);
219    if (blob_length > 0 &&
220        key.ReadValue(L"Order", reinterpret_cast<DWORD*>(&blob[0]),
221                      &blob_length, NULL) == ERROR_SUCCESS) {
222      if (!ParseFavoritesOrderBlob(importer, blob, path, sort_index))
223        return false;
224    }
225  }
226
227  // Recursively parse subfolders.
228  for (base::win::RegistryKeyIterator child(key.Handle(), L"");
229       child.Valid() && !importer->cancelled();
230       ++child) {
231    base::win::RegKey subkey(key.Handle(), child.Name(), KEY_READ);
232    if (subkey.Valid()) {
233      base::FilePath subpath(path.Append(child.Name()));
234      if (!ParseFavoritesOrderRegistryTree(importer, subkey, subpath,
235                                           sort_index)) {
236        return false;
237      }
238    }
239  }
240  return true;
241}
242
243bool ParseFavoritesOrderInfo(
244    const Importer* importer,
245    std::map<base::FilePath, uint32>* sort_index) WARN_UNUSED_RESULT {
246  base::string16 key_path(importer::GetIEFavoritesOrderKey());
247  base::win::RegKey key(HKEY_CURRENT_USER, key_path.c_str(), KEY_READ);
248  if (!key.Valid())
249    return false;
250  return ParseFavoritesOrderRegistryTree(importer, key, base::FilePath(),
251                                         sort_index);
252}
253
254// Reads the sort order from registry. If failed, we don't touch the list
255// and use the default (alphabetical) order.
256void SortBookmarksInIEOrder(
257    const Importer* importer,
258    std::vector<ImportedBookmarkEntry>* bookmarks) {
259  std::map<base::FilePath, uint32> sort_index;
260  if (!ParseFavoritesOrderInfo(importer, &sort_index))
261    return;
262  IEOrderBookmarkComparator compare = {&sort_index};
263  std::sort(bookmarks->begin(), bookmarks->end(), compare);
264}
265
266// Reads an internet shortcut (*.url) |file| and returns a COM object
267// representing it.
268bool LoadInternetShortcut(
269    const string16& file,
270    base::win::ScopedComPtr<IUniformResourceLocator>* shortcut) {
271  base::win::ScopedComPtr<IUniformResourceLocator> url_locator;
272  if (FAILED(url_locator.CreateInstance(CLSID_InternetShortcut, NULL,
273                                        CLSCTX_INPROC_SERVER)))
274    return false;
275
276  base::win::ScopedComPtr<IPersistFile> persist_file;
277  if (FAILED(persist_file.QueryFrom(url_locator)))
278    return false;
279
280  // Loads the Internet Shortcut from persistent storage.
281  if (FAILED(persist_file->Load(file.c_str(), STGM_READ)))
282    return false;
283
284  std::swap(url_locator, *shortcut);
285  return true;
286}
287
288// Reads the URL stored in the internet shortcut.
289GURL ReadURLFromInternetShortcut(IUniformResourceLocator* url_locator) {
290  base::win::ScopedCoMem<wchar_t> url;
291  // GetURL can return S_FALSE (FAILED(S_FALSE) is false) when url == NULL.
292  return (FAILED(url_locator->GetURL(&url)) || !url) ?
293      GURL() : GURL(WideToUTF16(std::wstring(url)));
294}
295
296// Reads the URL of the favicon of the internet shortcut.
297GURL ReadFaviconURLFromInternetShortcut(IUniformResourceLocator* url_locator) {
298  base::win::ScopedComPtr<IPropertySetStorage> property_set_storage;
299  if (FAILED(property_set_storage.QueryFrom(url_locator)))
300    return GURL();
301
302  base::win::ScopedComPtr<IPropertyStorage> property_storage;
303  if (FAILED(property_set_storage->Open(FMTID_Intshcut, STGM_READ,
304                                        property_storage.Receive()))) {
305    return GURL();
306  }
307
308  PROPSPEC properties[] = {{PRSPEC_PROPID, PID_IS_ICONFILE}};
309  // ReadMultiple takes a non-const array of PROPVARIANTs, but since this code
310  // only needs an array of size 1: a non-const pointer to |output| is
311  // equivalent.
312  base::win::ScopedPropVariant output;
313  // ReadMultiple can return S_FALSE (FAILED(S_FALSE) is false) when the
314  // property is not found, in which case output[0].vt is set to VT_EMPTY.
315  if (FAILED(property_storage->ReadMultiple(1, properties, output.Receive())) ||
316      output.get().vt != VT_LPWSTR)
317    return GURL();
318  return GURL(WideToUTF16(output.get().pwszVal));
319}
320
321// Reads the favicon imaga data in an NTFS alternate data stream. This is where
322// IE7 and above store the data.
323bool ReadFaviconDataFromInternetShortcut(const string16& file,
324                                         std::string* data) {
325  return base::ReadFileToString(
326      base::FilePath(file + kFaviconStreamName), data);
327}
328
329// Reads the favicon imaga data in the Internet cache. IE6 doesn't hold the data
330// explicitly, but it might be found in the cache.
331bool ReadFaviconDataFromCache(const GURL& favicon_url, std::string* data) {
332  std::wstring url_wstring(UTF8ToWide(favicon_url.spec()));
333  DWORD info_size = 0;
334  GetUrlCacheEntryInfoEx(url_wstring.c_str(), NULL, &info_size, NULL, NULL,
335                         NULL, 0);
336  if (GetLastError() != ERROR_INSUFFICIENT_BUFFER)
337    return false;
338
339  std::vector<char> buf(info_size);
340  INTERNET_CACHE_ENTRY_INFO* cache =
341      reinterpret_cast<INTERNET_CACHE_ENTRY_INFO*>(&buf[0]);
342  if (!GetUrlCacheEntryInfoEx(url_wstring.c_str(), cache, &info_size, NULL,
343                              NULL, NULL, 0)) {
344    return false;
345  }
346  return base::ReadFileToString(base::FilePath(cache->lpszLocalFileName), data);
347}
348
349// Reads the binary image data of favicon of an internet shortcut file |file|.
350// |favicon_url| read by ReadFaviconURLFromInternetShortcut is also needed to
351// examine the IE cache.
352bool ReadReencodedFaviconData(const string16& file,
353                              const GURL& favicon_url,
354                              std::vector<unsigned char>* data) {
355  std::string image_data;
356  if (!ReadFaviconDataFromInternetShortcut(file, &image_data) &&
357      !ReadFaviconDataFromCache(favicon_url, &image_data)) {
358    return false;
359  }
360
361  const unsigned char* ptr =
362      reinterpret_cast<const unsigned char*>(image_data.c_str());
363  return importer::ReencodeFavicon(ptr, image_data.size(), data);
364}
365
366// Loads favicon image data and registers to |favicon_map|.
367void UpdateFaviconMap(
368    const string16& url_file,
369    const GURL& url,
370    IUniformResourceLocator* url_locator,
371    std::map<GURL, ImportedFaviconUsage>* favicon_map) {
372  GURL favicon_url = ReadFaviconURLFromInternetShortcut(url_locator);
373  if (!favicon_url.is_valid())
374    return;
375
376  std::map<GURL, ImportedFaviconUsage>::iterator it =
377    favicon_map->find(favicon_url);
378  if (it != favicon_map->end()) {
379    // Known favicon URL.
380    it->second.urls.insert(url);
381  } else {
382    // New favicon URL. Read the image data and store.
383    ImportedFaviconUsage usage;
384    if (ReadReencodedFaviconData(url_file, favicon_url, &usage.png_data)) {
385      usage.favicon_url = favicon_url;
386      usage.urls.insert(url);
387      favicon_map->insert(std::make_pair(favicon_url, usage));
388    }
389  }
390}
391
392}  // namespace
393
394// static
395// {E161255A-37C3-11D2-BCAA-00C04fD929DB}
396const GUID IEImporter::kPStoreAutocompleteGUID = {
397    0xe161255a, 0x37c3, 0x11d2,
398    { 0xbc, 0xaa, 0x00, 0xc0, 0x4f, 0xd9, 0x29, 0xdb }
399};
400// {A79029D6-753E-4e27-B807-3D46AB1545DF}
401const GUID IEImporter::kUnittestGUID = {
402    0xa79029d6, 0x753e, 0x4e27,
403    { 0xb8, 0x7, 0x3d, 0x46, 0xab, 0x15, 0x45, 0xdf }
404};
405
406IEImporter::IEImporter() {
407}
408
409void IEImporter::StartImport(const importer::SourceProfile& source_profile,
410                             uint16 items,
411                             ImporterBridge* bridge) {
412  bridge_ = bridge;
413  source_path_ = source_profile.source_path;
414
415  bridge_->NotifyStarted();
416
417  if ((items & importer::HOME_PAGE) && !cancelled()) {
418    bridge_->NotifyItemStarted(importer::HOME_PAGE);
419    ImportHomepage();  // Doesn't have a UI item.
420    bridge_->NotifyItemEnded(importer::HOME_PAGE);
421  }
422  // The order here is important!
423  if ((items & importer::HISTORY) && !cancelled()) {
424    bridge_->NotifyItemStarted(importer::HISTORY);
425    ImportHistory();
426    bridge_->NotifyItemEnded(importer::HISTORY);
427  }
428  if ((items & importer::FAVORITES) && !cancelled()) {
429    bridge_->NotifyItemStarted(importer::FAVORITES);
430    ImportFavorites();
431    bridge_->NotifyItemEnded(importer::FAVORITES);
432  }
433  if ((items & importer::SEARCH_ENGINES) && !cancelled()) {
434    bridge_->NotifyItemStarted(importer::SEARCH_ENGINES);
435    ImportSearchEngines();
436    bridge_->NotifyItemEnded(importer::SEARCH_ENGINES);
437  }
438  if ((items & importer::PASSWORDS) && !cancelled()) {
439    bridge_->NotifyItemStarted(importer::PASSWORDS);
440    // Always import IE6 passwords.
441    ImportPasswordsIE6();
442
443    if (CurrentIEVersion() >= 7)
444      ImportPasswordsIE7();
445    bridge_->NotifyItemEnded(importer::PASSWORDS);
446  }
447  bridge_->NotifyEnded();
448}
449
450IEImporter::~IEImporter() {
451}
452
453void IEImporter::ImportFavorites() {
454  FavoritesInfo info;
455  if (!GetFavoritesInfo(&info))
456    return;
457
458  BookmarkVector bookmarks;
459  std::vector<ImportedFaviconUsage> favicons;
460  ParseFavoritesFolder(info, &bookmarks, &favicons);
461
462  if (!bookmarks.empty() && !cancelled()) {
463    const string16& first_folder_name =
464        l10n_util::GetStringUTF16(IDS_BOOKMARK_GROUP_FROM_IE);
465    bridge_->AddBookmarks(bookmarks, first_folder_name);
466  }
467  if (!favicons.empty() && !cancelled())
468    bridge_->SetFavicons(favicons);
469}
470
471void IEImporter::ImportHistory() {
472  const std::string kSchemes[] = {chrome::kHttpScheme,
473                                  content::kHttpsScheme,
474                                  chrome::kFtpScheme,
475                                  chrome::kFileScheme};
476  int total_schemes = arraysize(kSchemes);
477
478  base::win::ScopedComPtr<IUrlHistoryStg2> url_history_stg2;
479  HRESULT result;
480  result = url_history_stg2.CreateInstance(CLSID_CUrlHistory, NULL,
481                                           CLSCTX_INPROC_SERVER);
482  if (FAILED(result))
483    return;
484  base::win::ScopedComPtr<IEnumSTATURL> enum_url;
485  if (SUCCEEDED(result = url_history_stg2->EnumUrls(enum_url.Receive()))) {
486    std::vector<ImporterURLRow> rows;
487    STATURL stat_url;
488    ULONG fetched;
489    while (!cancelled() &&
490           (result = enum_url->Next(1, &stat_url, &fetched)) == S_OK) {
491      string16 url_string;
492      if (stat_url.pwcsUrl) {
493        url_string = stat_url.pwcsUrl;
494        CoTaskMemFree(stat_url.pwcsUrl);
495      }
496      string16 title_string;
497      if (stat_url.pwcsTitle) {
498        title_string = stat_url.pwcsTitle;
499        CoTaskMemFree(stat_url.pwcsTitle);
500      }
501
502      GURL url(url_string);
503      // Skips the URLs that are invalid or have other schemes.
504      if (!url.is_valid() ||
505          (std::find(kSchemes, kSchemes + total_schemes, url.scheme()) ==
506           kSchemes + total_schemes))
507        continue;
508
509      ImporterURLRow row(url);
510      row.title = title_string;
511      row.last_visit = base::Time::FromFileTime(stat_url.ftLastVisited);
512      if (stat_url.dwFlags == STATURL_QUERYFLAG_TOPLEVEL) {
513        row.visit_count = 1;
514        row.hidden = false;
515      } else {
516        row.hidden = true;
517      }
518
519      rows.push_back(row);
520    }
521
522    if (!rows.empty() && !cancelled()) {
523      bridge_->SetHistoryItems(rows, importer::VISIT_SOURCE_IE_IMPORTED);
524    }
525  }
526}
527
528void IEImporter::ImportPasswordsIE6() {
529  GUID AutocompleteGUID = kPStoreAutocompleteGUID;
530  if (!source_path_.empty()) {
531    // We supply a fake GUID for testting.
532    AutocompleteGUID = kUnittestGUID;
533  }
534
535  // The PStoreCreateInstance function retrieves an interface pointer
536  // to a storage provider. But this function has no associated import
537  // library or header file, we must call it using the LoadLibrary()
538  // and GetProcAddress() functions.
539  typedef HRESULT (WINAPI *PStoreCreateFunc)(IPStore**, DWORD, DWORD, DWORD);
540  HMODULE pstorec_dll = LoadLibrary(L"pstorec.dll");
541  if (!pstorec_dll)
542    return;
543  PStoreCreateFunc PStoreCreateInstance =
544      (PStoreCreateFunc)GetProcAddress(pstorec_dll, "PStoreCreateInstance");
545  if (!PStoreCreateInstance) {
546    FreeLibrary(pstorec_dll);
547    return;
548  }
549
550  base::win::ScopedComPtr<IPStore, &IID_IPStore> pstore;
551  HRESULT result = PStoreCreateInstance(pstore.Receive(), 0, 0, 0);
552  if (result != S_OK) {
553    FreeLibrary(pstorec_dll);
554    return;
555  }
556
557  std::vector<AutoCompleteInfo> ac_list;
558
559  // Enumerates AutoComplete items in the protected database.
560  base::win::ScopedComPtr<IEnumPStoreItems, &IID_IEnumPStoreItems> item;
561  result = pstore->EnumItems(0, &AutocompleteGUID,
562                             &AutocompleteGUID, 0, item.Receive());
563  if (result != PST_E_OK) {
564    pstore.Release();
565    FreeLibrary(pstorec_dll);
566    return;
567  }
568
569  wchar_t* item_name;
570  while (!cancelled() && SUCCEEDED(item->Next(1, &item_name, 0))) {
571    DWORD length = 0;
572    unsigned char* buffer = NULL;
573    result = pstore->ReadItem(0, &AutocompleteGUID, &AutocompleteGUID,
574                              item_name, &length, &buffer, NULL, 0);
575    if (SUCCEEDED(result)) {
576      AutoCompleteInfo ac;
577      ac.key = item_name;
578      string16 data;
579      data.insert(0, reinterpret_cast<wchar_t*>(buffer),
580                  length / sizeof(wchar_t));
581
582      // The key name is always ended with ":StringData".
583      const wchar_t kDataSuffix[] = L":StringData";
584      size_t i = ac.key.rfind(kDataSuffix);
585      if (i != string16::npos && ac.key.substr(i) == kDataSuffix) {
586        ac.key.erase(i);
587        ac.is_url = (ac.key.find(L"://") != string16::npos);
588        ac_list.push_back(ac);
589        base::SplitString(data, L'\0', &ac_list[ac_list.size() - 1].data);
590      }
591      CoTaskMemFree(buffer);
592    }
593    CoTaskMemFree(item_name);
594  }
595  // Releases them before unload the dll.
596  item.Release();
597  pstore.Release();
598  FreeLibrary(pstorec_dll);
599
600  size_t i;
601  for (i = 0; i < ac_list.size(); i++) {
602    if (!ac_list[i].is_url || ac_list[i].data.size() < 2)
603      continue;
604
605    GURL url(ac_list[i].key.c_str());
606    if (!(LowerCaseEqualsASCII(url.scheme(), chrome::kHttpScheme) ||
607        LowerCaseEqualsASCII(url.scheme(), content::kHttpsScheme))) {
608      continue;
609    }
610
611    autofill::PasswordForm form;
612    GURL::Replacements rp;
613    rp.ClearUsername();
614    rp.ClearPassword();
615    rp.ClearQuery();
616    rp.ClearRef();
617    form.origin = url.ReplaceComponents(rp);
618    form.username_value = ac_list[i].data[0];
619    form.password_value = ac_list[i].data[1];
620    form.signon_realm = url.GetOrigin().spec();
621
622    // This is not precise, because a scheme of https does not imply a valid
623    // certificate was presented; however we assign it this way so that if we
624    // import a password from IE whose scheme is https, we give it the benefit
625    // of the doubt and DONT auto-fill it unless the form appears under
626    // valid SSL conditions.
627    form.ssl_valid = url.SchemeIsSecure();
628
629    // Goes through the list to find out the username field
630    // of the web page.
631    size_t list_it, item_it;
632    for (list_it = 0; list_it < ac_list.size(); ++list_it) {
633      if (ac_list[list_it].is_url)
634        continue;
635
636      for (item_it = 0; item_it < ac_list[list_it].data.size(); ++item_it)
637        if (ac_list[list_it].data[item_it] == form.username_value) {
638          form.username_element = ac_list[list_it].key;
639          break;
640        }
641    }
642
643    bridge_->SetPasswordForm(form);
644  }
645}
646
647void IEImporter::ImportPasswordsIE7() {
648  base::string16 key_path(importer::GetIE7PasswordsKey());
649  base::win::RegKey key(HKEY_CURRENT_USER, key_path.c_str(), KEY_READ);
650  base::win::RegistryValueIterator reg_iterator(HKEY_CURRENT_USER,
651                                                key_path.c_str());
652  importer::ImporterIE7PasswordInfo password_info;
653  while (reg_iterator.Valid() && !cancelled()) {
654    // Get the size of the encrypted data.
655    DWORD value_len = 0;
656    key.ReadValue(reg_iterator.Name(), NULL, &value_len, NULL);
657    if (value_len) {
658      // Query the encrypted data.
659      password_info.encrypted_data.resize(value_len);
660      if (key.ReadValue(reg_iterator.Name(),
661                        &password_info.encrypted_data.front(),
662                        &value_len, NULL) == ERROR_SUCCESS) {
663        password_info.url_hash = reg_iterator.Name();
664        password_info.date_created = base::Time::Now();
665
666        bridge_->AddIE7PasswordInfo(password_info);
667      }
668    }
669
670    ++reg_iterator;
671  }
672}
673
674void IEImporter::ImportSearchEngines() {
675  // On IE, search engines are stored in the registry, under:
676  // Software\Microsoft\Internet Explorer\SearchScopes
677  // Each key represents a search engine. The URL value contains the URL and
678  // the DisplayName the name.
679  typedef std::map<std::string, string16> SearchEnginesMap;
680  SearchEnginesMap search_engines_map;
681  for (base::win::RegistryKeyIterator key_iter(HKEY_CURRENT_USER,
682       kSearchScopePath); key_iter.Valid(); ++key_iter) {
683    string16 sub_key_name = kSearchScopePath;
684    sub_key_name.append(L"\\").append(key_iter.Name());
685    base::win::RegKey sub_key(HKEY_CURRENT_USER, sub_key_name.c_str(),
686                              KEY_READ);
687    string16 wide_url;
688    if ((sub_key.ReadValue(L"URL", &wide_url) != ERROR_SUCCESS) ||
689        wide_url.empty()) {
690      VLOG(1) << "No URL for IE search engine at " << key_iter.Name();
691      continue;
692    }
693    // For the name, we try the default value first (as Live Search uses a
694    // non displayable name in DisplayName, and the readable name under the
695    // default value).
696    string16 name;
697    if ((sub_key.ReadValue(NULL, &name) != ERROR_SUCCESS) || name.empty()) {
698      // Try the displayable name.
699      if ((sub_key.ReadValue(L"DisplayName", &name) != ERROR_SUCCESS) ||
700          name.empty()) {
701        VLOG(1) << "No name for IE search engine at " << key_iter.Name();
702        continue;
703      }
704    }
705
706    std::string url(WideToUTF8(wide_url));
707    SearchEnginesMap::iterator t_iter = search_engines_map.find(url);
708    if (t_iter == search_engines_map.end()) {
709      // First time we see that URL.
710      GURL gurl(url);
711      if (gurl.is_valid()) {
712        t_iter = search_engines_map.insert(std::make_pair(url, name)).first;
713      }
714    }
715  }
716  // ProfileWriter::AddKeywords() requires a vector and we have a map.
717  std::vector<importer::URLKeywordInfo> url_keywords;
718  for (SearchEnginesMap::iterator i = search_engines_map.begin();
719       i != search_engines_map.end(); ++i) {
720    importer::URLKeywordInfo url_keyword_info;
721    url_keyword_info.url = GURL(i->first);
722    url_keyword_info.display_name = i->second;
723    url_keywords.push_back(url_keyword_info);
724  }
725  bridge_->SetKeywords(url_keywords, true);
726}
727
728void IEImporter::ImportHomepage() {
729  const wchar_t* kIEHomepage = L"Start Page";
730  const wchar_t* kIEDefaultHomepage = L"Default_Page_URL";
731
732  base::string16 key_path(importer::GetIESettingsKey());
733
734  base::win::RegKey key(HKEY_CURRENT_USER, key_path.c_str(), KEY_READ);
735  string16 homepage_url;
736  if (key.ReadValue(kIEHomepage, &homepage_url) != ERROR_SUCCESS ||
737      homepage_url.empty())
738    return;
739
740  GURL homepage = GURL(homepage_url);
741  if (!homepage.is_valid())
742    return;
743
744  // Check to see if this is the default website and skip import.
745  base::win::RegKey keyDefault(HKEY_LOCAL_MACHINE, key_path.c_str(), KEY_READ);
746  string16 default_homepage_url;
747  LONG result = keyDefault.ReadValue(kIEDefaultHomepage, &default_homepage_url);
748  if (result == ERROR_SUCCESS && !default_homepage_url.empty()) {
749    if (homepage.spec() == GURL(default_homepage_url).spec())
750      return;
751  }
752  bridge_->AddHomePage(homepage);
753}
754
755bool IEImporter::GetFavoritesInfo(IEImporter::FavoritesInfo* info) {
756  if (!source_path_.empty()) {
757    // Source path exists during testing.
758    info->path = source_path_;
759    info->path = info->path.AppendASCII("Favorites");
760    info->links_folder = L"Links";
761    return true;
762  }
763
764  // IE stores the favorites in the Favorites under user profile's folder.
765  wchar_t buffer[MAX_PATH];
766  if (FAILED(SHGetFolderPath(NULL, CSIDL_FAVORITES, NULL,
767                             SHGFP_TYPE_CURRENT, buffer)))
768    return false;
769  info->path = base::FilePath(buffer);
770
771  // There is a Links folder under Favorites folder in Windows Vista, but it
772  // is not recording in Vista's registry. So in Vista, we assume the Links
773  // folder is under Favorites folder since it looks like there is not name
774  // different in every language version of Windows Vista.
775  if (base::win::GetVersion() < base::win::VERSION_VISTA) {
776    // The Link folder name is stored in the registry.
777    DWORD buffer_length = sizeof(buffer);
778    base::win::RegKey reg_key(HKEY_CURRENT_USER, kIEToolbarKey, KEY_READ);
779    if (reg_key.ReadValue(L"LinksFolderName", buffer,
780                          &buffer_length, NULL) != ERROR_SUCCESS)
781      return false;
782    info->links_folder = buffer;
783  } else {
784    info->links_folder = L"Links";
785  }
786
787  return true;
788}
789
790void IEImporter::ParseFavoritesFolder(
791    const FavoritesInfo& info,
792    BookmarkVector* bookmarks,
793    std::vector<ImportedFaviconUsage>* favicons) {
794  base::FilePath file;
795  std::vector<base::FilePath::StringType> file_list;
796  base::FilePath favorites_path(info.path);
797  // Favorites path length.  Make sure it doesn't include the trailing \.
798  size_t favorites_path_len =
799      favorites_path.StripTrailingSeparators().value().size();
800  base::FileEnumerator file_enumerator(
801      favorites_path, true, base::FileEnumerator::FILES);
802  while (!(file = file_enumerator.Next()).value().empty() && !cancelled())
803    file_list.push_back(file.value());
804
805  // Keep the bookmarks in alphabetical order.
806  std::sort(file_list.begin(), file_list.end());
807
808  // Map from favicon URLs to the favicon data (the binary image data and the
809  // set of bookmark URLs referring to the favicon).
810  typedef std::map<GURL, ImportedFaviconUsage> FaviconMap;
811  FaviconMap favicon_map;
812
813  for (std::vector<base::FilePath::StringType>::iterator it = file_list.begin();
814       it != file_list.end(); ++it) {
815    base::FilePath shortcut(*it);
816    if (!LowerCaseEqualsASCII(shortcut.Extension(), ".url"))
817      continue;
818
819    // Skip the bookmark with invalid URL.
820    base::win::ScopedComPtr<IUniformResourceLocator> url_locator;
821    if (!LoadInternetShortcut(*it, &url_locator))
822      continue;
823    GURL url = ReadURLFromInternetShortcut(url_locator);
824    if (!url.is_valid())
825      continue;
826    // Skip default bookmarks. go.microsoft.com redirects to
827    // search.microsoft.com, and http://go.microsoft.com/fwlink/?LinkId=XXX,
828    // which URLs IE has as default, to some another sites.
829    // We expect that users will never themselves create bookmarks having this
830    // hostname.
831    if (url.host() == "go.microsoft.com")
832      continue;
833    // Read favicon.
834    UpdateFaviconMap(*it, url, url_locator, &favicon_map);
835
836    // Make the relative path from the Favorites folder, without the basename.
837    // ex. Suppose that the Favorites folder is C:\Users\Foo\Favorites.
838    //   C:\Users\Foo\Favorites\Foo.url -> ""
839    //   C:\Users\Foo\Favorites\Links\Bar\Baz.url -> "Links\Bar"
840    base::FilePath::StringType relative_string =
841        shortcut.DirName().value().substr(favorites_path_len);
842    if (!relative_string.empty() &&
843        base::FilePath::IsSeparator(relative_string[0]))
844      relative_string = relative_string.substr(1);
845    base::FilePath relative_path(relative_string);
846
847    ImportedBookmarkEntry entry;
848    // Remove the dot, the file extension, and the directory path.
849    entry.title = shortcut.RemoveExtension().BaseName().value();
850    entry.url = url;
851    entry.creation_time = GetFileCreationTime(*it);
852    if (!relative_path.empty())
853      relative_path.GetComponents(&entry.path);
854
855    // Add the bookmark.
856    if (!entry.path.empty() && entry.path[0] == info.links_folder) {
857      // Bookmarks in the Link folder should be imported to the toolbar.
858      entry.in_toolbar = true;
859    }
860    bookmarks->push_back(entry);
861  }
862
863  // Reflect the menu order in IE.
864  SortBookmarksInIEOrder(this, bookmarks);
865
866  // Record favicon data.
867  for (FaviconMap::iterator iter = favicon_map.begin();
868       iter != favicon_map.end(); ++iter)
869    favicons->push_back(iter->second);
870}
871
872int IEImporter::CurrentIEVersion() const {
873  static int version = -1;
874  if (version < 0) {
875    wchar_t buffer[128];
876    DWORD buffer_length = sizeof(buffer);
877    base::win::RegKey reg_key(HKEY_LOCAL_MACHINE, kIEVersionKey, KEY_READ);
878    LONG result = reg_key.ReadValue(L"Version", buffer, &buffer_length, NULL);
879    version = ((result == ERROR_SUCCESS)? _wtoi(buffer) : 0);
880  }
881  return version;
882}
883