1 // Copyright 2013 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/bookmark_html_reader.h" 6 7 #include "base/callback.h" 8 #include "base/file_util.h" 9 #include "base/i18n/icu_string_conversions.h" 10 #include "base/strings/string_number_conversions.h" 11 #include "base/strings/string_split.h" 12 #include "base/strings/string_util.h" 13 #include "base/time/time.h" 14 #include "chrome/common/importer/imported_bookmark_entry.h" 15 #include "chrome/common/importer/imported_favicon_usage.h" 16 #include "chrome/utility/importer/favicon_reencode.h" 17 #include "content/public/common/url_constants.h" 18 #include "net/base/data_url.h" 19 #include "net/base/escape.h" 20 #include "url/gurl.h" 21 22 namespace { 23 24 // Fetches the given |attribute| value from the |attribute_list|. Returns true 25 // if successful, and |value| will contain the value. 26 bool GetAttribute(const std::string& attribute_list, 27 const std::string& attribute, 28 std::string* value) { 29 const char kQuote[] = "\""; 30 31 size_t begin = attribute_list.find(attribute + "=" + kQuote); 32 if (begin == std::string::npos) 33 return false; // Can't find the attribute. 34 35 begin += attribute.size() + 2; 36 size_t end = begin + 1; 37 38 while (end < attribute_list.size()) { 39 if (attribute_list[end] == '"' && 40 attribute_list[end - 1] != '\\') { 41 break; 42 } 43 end++; 44 } 45 46 if (end == attribute_list.size()) 47 return false; // The value is not quoted. 48 49 *value = attribute_list.substr(begin, end - begin); 50 return true; 51 } 52 53 // Given the URL of a page and a favicon data URL, adds an appropriate record 54 // to the given favicon usage vector. 55 void DataURLToFaviconUsage( 56 const GURL& link_url, 57 const GURL& favicon_data, 58 std::vector<ImportedFaviconUsage>* favicons) { 59 if (!link_url.is_valid() || !favicon_data.is_valid() || 60 !favicon_data.SchemeIs(chrome::kDataScheme)) 61 return; 62 63 // Parse the data URL. 64 std::string mime_type, char_set, data; 65 if (!net::DataURL::Parse(favicon_data, &mime_type, &char_set, &data) || 66 data.empty()) 67 return; 68 69 ImportedFaviconUsage usage; 70 if (!importer::ReencodeFavicon( 71 reinterpret_cast<const unsigned char*>(&data[0]), 72 data.size(), &usage.png_data)) 73 return; // Unable to decode. 74 75 // We need to make up a URL for the favicon. We use a version of the page's 76 // URL so that we can be sure it will not collide. 77 usage.favicon_url = GURL(std::string("made-up-favicon:") + link_url.spec()); 78 79 // We only have one URL per favicon for Firefox 2 bookmarks. 80 usage.urls.insert(link_url); 81 82 favicons->push_back(usage); 83 } 84 85 } // namespace 86 87 namespace bookmark_html_reader { 88 89 void ImportBookmarksFile( 90 const base::Callback<bool(void)>& cancellation_callback, 91 const base::Callback<bool(const GURL&)>& valid_url_callback, 92 const base::FilePath& file_path, 93 std::vector<ImportedBookmarkEntry>* bookmarks, 94 std::vector<ImportedFaviconUsage>* favicons) { 95 std::string content; 96 file_util::ReadFileToString(file_path, &content); 97 std::vector<std::string> lines; 98 base::SplitString(content, '\n', &lines); 99 100 base::string16 last_folder; 101 bool last_folder_on_toolbar = false; 102 bool last_folder_is_empty = true; 103 bool has_subfolder = false; 104 base::Time last_folder_add_date; 105 std::vector<base::string16> path; 106 size_t toolbar_folder_index = 0; 107 std::string charset; 108 for (size_t i = 0; 109 i < lines.size() && 110 (cancellation_callback.is_null() || !cancellation_callback.Run()); 111 ++i) { 112 std::string line; 113 TrimString(lines[i], " ", &line); 114 115 // Get the encoding of the bookmark file. 116 if (internal::ParseCharsetFromLine(line, &charset)) 117 continue; 118 119 // Get the folder name. 120 if (internal::ParseFolderNameFromLine(line, 121 charset, 122 &last_folder, 123 &last_folder_on_toolbar, 124 &last_folder_add_date)) { 125 continue; 126 } 127 128 // Get the bookmark entry. 129 base::string16 title; 130 base::string16 shortcut; 131 GURL url, favicon; 132 base::Time add_date; 133 base::string16 post_data; 134 bool is_bookmark; 135 // TODO(jcampan): http://b/issue?id=1196285 we do not support POST based 136 // keywords yet. 137 is_bookmark = 138 internal::ParseBookmarkFromLine(line, charset, &title, 139 &url, &favicon, &shortcut, 140 &add_date, &post_data) || 141 internal::ParseMinimumBookmarkFromLine(line, charset, &title, &url); 142 143 if (is_bookmark) 144 last_folder_is_empty = false; 145 146 if (is_bookmark && 147 post_data.empty() && 148 (valid_url_callback.is_null() || valid_url_callback.Run(url))) { 149 if (toolbar_folder_index > path.size() && !path.empty()) { 150 NOTREACHED(); // error in parsing. 151 break; 152 } 153 154 ImportedBookmarkEntry entry; 155 entry.creation_time = add_date; 156 entry.url = url; 157 entry.title = title; 158 159 if (toolbar_folder_index) { 160 // The toolbar folder should be at the top level. 161 entry.in_toolbar = true; 162 entry.path.assign(path.begin() + toolbar_folder_index - 1, path.end()); 163 } else { 164 // Add this bookmark to the list of |bookmarks|. 165 if (!has_subfolder && !last_folder.empty()) { 166 path.push_back(last_folder); 167 last_folder.clear(); 168 } 169 entry.path.assign(path.begin(), path.end()); 170 } 171 bookmarks->push_back(entry); 172 173 // Save the favicon. DataURLToFaviconUsage will handle the case where 174 // there is no favicon. 175 if (favicons) 176 DataURLToFaviconUsage(url, favicon, favicons); 177 178 continue; 179 } 180 181 // Bookmarks in sub-folder are encapsulated with <DL> tag. 182 if (StartsWithASCII(line, "<DL>", false)) { 183 has_subfolder = true; 184 if (!last_folder.empty()) { 185 path.push_back(last_folder); 186 last_folder.clear(); 187 } 188 if (last_folder_on_toolbar && !toolbar_folder_index) 189 toolbar_folder_index = path.size(); 190 191 // Mark next folder empty as initial state. 192 last_folder_is_empty = true; 193 } else if (StartsWithASCII(line, "</DL>", false)) { 194 if (path.empty()) 195 break; // Mismatch <DL>. 196 197 base::string16 folder_title = path.back(); 198 path.pop_back(); 199 200 if (last_folder_is_empty) { 201 // Empty folder should be added explicitly. 202 ImportedBookmarkEntry entry; 203 entry.is_folder = true; 204 entry.creation_time = last_folder_add_date; 205 entry.title = folder_title; 206 if (toolbar_folder_index) { 207 // The toolbar folder should be at the top level. 208 // Make sure we don't add the toolbar folder itself if it is empty. 209 if (toolbar_folder_index <= path.size()) { 210 entry.in_toolbar = true; 211 entry.path.assign(path.begin() + toolbar_folder_index - 1, 212 path.end()); 213 bookmarks->push_back(entry); 214 } 215 } else { 216 // Add this folder to the list of |bookmarks|. 217 entry.path.assign(path.begin(), path.end()); 218 bookmarks->push_back(entry); 219 } 220 221 // Parent folder include current one, so it's not empty. 222 last_folder_is_empty = false; 223 } 224 225 if (toolbar_folder_index > path.size()) 226 toolbar_folder_index = 0; 227 } 228 } 229 } 230 231 namespace internal { 232 233 bool ParseCharsetFromLine(const std::string& line, std::string* charset) { 234 const char kCharset[] = "charset="; 235 if (StartsWithASCII(line, "<META", false) && 236 (line.find("CONTENT=\"") != std::string::npos || 237 line.find("content=\"") != std::string::npos)) { 238 size_t begin = line.find(kCharset); 239 if (begin == std::string::npos) 240 return false; 241 begin += std::string(kCharset).size(); 242 size_t end = line.find_first_of('\"', begin); 243 *charset = line.substr(begin, end - begin); 244 return true; 245 } 246 return false; 247 } 248 249 bool ParseFolderNameFromLine(const std::string& line, 250 const std::string& charset, 251 base::string16* folder_name, 252 bool* is_toolbar_folder, 253 base::Time* add_date) { 254 const char kFolderOpen[] = "<DT><H3"; 255 const char kFolderClose[] = "</H3>"; 256 const char kToolbarFolderAttribute[] = "PERSONAL_TOOLBAR_FOLDER"; 257 const char kAddDateAttribute[] = "ADD_DATE"; 258 259 if (!StartsWithASCII(line, kFolderOpen, true)) 260 return false; 261 262 size_t end = line.find(kFolderClose); 263 size_t tag_end = line.rfind('>', end) + 1; 264 // If no end tag or start tag is broken, we skip to find the folder name. 265 if (end == std::string::npos || tag_end < arraysize(kFolderOpen)) 266 return false; 267 268 base::CodepageToUTF16(line.substr(tag_end, end - tag_end), charset.c_str(), 269 base::OnStringConversionError::SKIP, folder_name); 270 *folder_name = net::UnescapeForHTML(*folder_name); 271 272 std::string attribute_list = line.substr(arraysize(kFolderOpen), 273 tag_end - arraysize(kFolderOpen) - 1); 274 std::string value; 275 276 // Add date 277 if (GetAttribute(attribute_list, kAddDateAttribute, &value)) { 278 int64 time; 279 base::StringToInt64(value, &time); 280 // Upper bound it at 32 bits. 281 if (0 < time && time < (1LL << 32)) 282 *add_date = base::Time::FromTimeT(time); 283 } 284 285 if (GetAttribute(attribute_list, kToolbarFolderAttribute, &value) && 286 LowerCaseEqualsASCII(value, "true")) 287 *is_toolbar_folder = true; 288 else 289 *is_toolbar_folder = false; 290 291 return true; 292 } 293 294 bool ParseBookmarkFromLine(const std::string& line, 295 const std::string& charset, 296 base::string16* title, 297 GURL* url, 298 GURL* favicon, 299 base::string16* shortcut, 300 base::Time* add_date, 301 base::string16* post_data) { 302 const char kItemOpen[] = "<DT><A"; 303 const char kItemClose[] = "</A>"; 304 const char kFeedURLAttribute[] = "FEEDURL"; 305 const char kHrefAttribute[] = "HREF"; 306 const char kIconAttribute[] = "ICON"; 307 const char kShortcutURLAttribute[] = "SHORTCUTURL"; 308 const char kAddDateAttribute[] = "ADD_DATE"; 309 const char kPostDataAttribute[] = "POST_DATA"; 310 311 title->clear(); 312 *url = GURL(); 313 *favicon = GURL(); 314 shortcut->clear(); 315 post_data->clear(); 316 *add_date = base::Time(); 317 318 if (!StartsWithASCII(line, kItemOpen, true)) 319 return false; 320 321 size_t end = line.find(kItemClose); 322 size_t tag_end = line.rfind('>', end) + 1; 323 if (end == std::string::npos || tag_end < arraysize(kItemOpen)) 324 return false; // No end tag or start tag is broken. 325 326 std::string attribute_list = line.substr(arraysize(kItemOpen), 327 tag_end - arraysize(kItemOpen) - 1); 328 329 // We don't import Live Bookmark folders, which is Firefox's RSS reading 330 // feature, since the user never necessarily bookmarked them and we don't 331 // have this feature to update their contents. 332 std::string value; 333 if (GetAttribute(attribute_list, kFeedURLAttribute, &value)) 334 return false; 335 336 // Title 337 base::CodepageToUTF16(line.substr(tag_end, end - tag_end), charset.c_str(), 338 base::OnStringConversionError::SKIP, title); 339 *title = net::UnescapeForHTML(*title); 340 341 // URL 342 if (GetAttribute(attribute_list, kHrefAttribute, &value)) { 343 base::string16 url16; 344 base::CodepageToUTF16(value, charset.c_str(), 345 base::OnStringConversionError::SKIP, &url16); 346 url16 = net::UnescapeForHTML(url16); 347 348 *url = GURL(url16); 349 } 350 351 // Favicon 352 if (GetAttribute(attribute_list, kIconAttribute, &value)) 353 *favicon = GURL(value); 354 355 // Keyword 356 if (GetAttribute(attribute_list, kShortcutURLAttribute, &value)) { 357 base::CodepageToUTF16(value, charset.c_str(), 358 base::OnStringConversionError::SKIP, shortcut); 359 *shortcut = net::UnescapeForHTML(*shortcut); 360 } 361 362 // Add date 363 if (GetAttribute(attribute_list, kAddDateAttribute, &value)) { 364 int64 time; 365 base::StringToInt64(value, &time); 366 // Upper bound it at 32 bits. 367 if (0 < time && time < (1LL << 32)) 368 *add_date = base::Time::FromTimeT(time); 369 } 370 371 // Post data. 372 if (GetAttribute(attribute_list, kPostDataAttribute, &value)) { 373 base::CodepageToUTF16(value, charset.c_str(), 374 base::OnStringConversionError::SKIP, post_data); 375 *post_data = net::UnescapeForHTML(*post_data); 376 } 377 378 return true; 379 } 380 381 bool ParseMinimumBookmarkFromLine(const std::string& line, 382 const std::string& charset, 383 base::string16* title, 384 GURL* url) { 385 const char kItemOpen[] = "<DT><A"; 386 const char kItemClose[] = "</"; 387 const char kHrefAttributeUpper[] = "HREF"; 388 const char kHrefAttributeLower[] = "href"; 389 390 title->clear(); 391 *url = GURL(); 392 393 // Case-insensitive check of open tag. 394 if (!StartsWithASCII(line, kItemOpen, false)) 395 return false; 396 397 // Find any close tag. 398 size_t end = line.find(kItemClose); 399 size_t tag_end = line.rfind('>', end) + 1; 400 if (end == std::string::npos || tag_end < arraysize(kItemOpen)) 401 return false; // No end tag or start tag is broken. 402 403 std::string attribute_list = line.substr(arraysize(kItemOpen), 404 tag_end - arraysize(kItemOpen) - 1); 405 406 // Title 407 base::CodepageToUTF16(line.substr(tag_end, end - tag_end), charset.c_str(), 408 base::OnStringConversionError::SKIP, title); 409 *title = net::UnescapeForHTML(*title); 410 411 // URL 412 std::string value; 413 if (GetAttribute(attribute_list, kHrefAttributeUpper, &value) || 414 GetAttribute(attribute_list, kHrefAttributeLower, &value)) { 415 if (charset.length() != 0) { 416 base::string16 url16; 417 base::CodepageToUTF16(value, charset.c_str(), 418 base::OnStringConversionError::SKIP, &url16); 419 url16 = net::UnescapeForHTML(url16); 420 421 *url = GURL(url16); 422 } else { 423 *url = GURL(value); 424 } 425 } 426 427 return true; 428 } 429 430 } // namespace internal 431 432 } // namespace bookmark_html_reader 433