1 // Copyright (c) 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/test/chromedriver/chrome/zip_reader.h" 6 7 #include "base/file_util.h" 8 #include "base/logging.h" 9 #include "base/strings/string_util.h" 10 #include "base/strings/utf_string_conversions.h" 11 #include "chrome/test/chromedriver/chrome/zip_internal.h" 12 #include "net/base/file_stream.h" 13 14 #if defined(USE_SYSTEM_MINIZIP) 15 #include <minizip/unzip.h> 16 #else 17 #include "third_party/zlib/contrib/minizip/unzip.h" 18 #if defined(OS_WIN) 19 #include "third_party/zlib/contrib/minizip/iowin32.h" 20 #endif // defined(OS_WIN) 21 #endif // defined(USE_SYSTEM_MINIZIP) 22 23 namespace zip { 24 25 // TODO(satorux): The implementation assumes that file names in zip files 26 // are encoded in UTF-8. This is true for zip files created by Zip() 27 // function in zip.h, but not true for user-supplied random zip files. 28 ZipReader::EntryInfo::EntryInfo(const std::string& file_name_in_zip, 29 const unz_file_info& raw_file_info) 30 : file_path_(base::FilePath::FromUTF8Unsafe(file_name_in_zip)), 31 is_directory_(false) { 32 original_size_ = raw_file_info.uncompressed_size; 33 34 // Directory entries in zip files end with "/". 35 is_directory_ = EndsWith(file_name_in_zip, "/", false); 36 37 // Check the file name here for directory traversal issues. In the name of 38 // simplicity and security, we might reject a valid file name such as "a..b". 39 is_unsafe_ = file_name_in_zip.find("..") != std::string::npos; 40 41 // We also consider that the file name is unsafe, if it's invalid UTF-8. 42 string16 file_name_utf16; 43 if (!UTF8ToUTF16(file_name_in_zip.data(), file_name_in_zip.size(), 44 &file_name_utf16)) { 45 is_unsafe_ = true; 46 } 47 48 // We also consider that the file name is unsafe, if it's absolute. 49 // On Windows, IsAbsolute() returns false for paths starting with "/". 50 if (file_path_.IsAbsolute() || StartsWithASCII(file_name_in_zip, "/", false)) 51 is_unsafe_ = true; 52 53 // Construct the last modified time. The timezone info is not present in 54 // zip files, so we construct the time as local time. 55 base::Time::Exploded exploded_time = {}; // Zero-clear. 56 exploded_time.year = raw_file_info.tmu_date.tm_year; 57 // The month in zip file is 0-based, whereas ours is 1-based. 58 exploded_time.month = raw_file_info.tmu_date.tm_mon + 1; 59 exploded_time.day_of_month = raw_file_info.tmu_date.tm_mday; 60 exploded_time.hour = raw_file_info.tmu_date.tm_hour; 61 exploded_time.minute = raw_file_info.tmu_date.tm_min; 62 exploded_time.second = raw_file_info.tmu_date.tm_sec; 63 exploded_time.millisecond = 0; 64 if (exploded_time.HasValidValues()) { 65 last_modified_ = base::Time::FromLocalExploded(exploded_time); 66 } else { 67 // Use Unix time epoch if the time stamp data is invalid. 68 last_modified_ = base::Time::UnixEpoch(); 69 } 70 } 71 72 ZipReader::ZipReader() { 73 Reset(); 74 } 75 76 ZipReader::~ZipReader() { 77 Close(); 78 } 79 80 bool ZipReader::Open(const base::FilePath& zip_file_path) { 81 DCHECK(!zip_file_); 82 83 // Use of "Unsafe" function does not look good, but there is no way to do 84 // this safely on Linux. See file_util.h for details. 85 zip_file_ = internal::OpenForUnzipping(zip_file_path.AsUTF8Unsafe()); 86 if (!zip_file_) { 87 return false; 88 } 89 90 return OpenInternal(); 91 } 92 93 bool ZipReader::OpenFromPlatformFile(base::PlatformFile zip_fd) { 94 DCHECK(!zip_file_); 95 96 #if defined(OS_POSIX) 97 zip_file_ = internal::OpenFdForUnzipping(zip_fd); 98 #elif defined(OS_WIN) 99 zip_file_ = internal::OpenHandleForUnzipping(zip_fd); 100 #endif 101 if (!zip_file_) { 102 return false; 103 } 104 105 return OpenInternal(); 106 } 107 108 bool ZipReader::OpenFromString(const std::string& data) { 109 zip_file_ = internal::PreprareMemoryForUnzipping(data); 110 if (!zip_file_) 111 return false; 112 return OpenInternal(); 113 } 114 115 void ZipReader::Close() { 116 if (zip_file_) { 117 unzClose(zip_file_); 118 } 119 Reset(); 120 } 121 122 bool ZipReader::HasMore() { 123 return !reached_end_; 124 } 125 126 bool ZipReader::AdvanceToNextEntry() { 127 DCHECK(zip_file_); 128 129 // Should not go further if we already reached the end. 130 if (reached_end_) 131 return false; 132 133 unz_file_pos position = {}; 134 if (unzGetFilePos(zip_file_, &position) != UNZ_OK) 135 return false; 136 const int current_entry_index = position.num_of_file; 137 // If we are currently at the last entry, then the next position is the 138 // end of the zip file, so mark that we reached the end. 139 if (current_entry_index + 1 == num_entries_) { 140 reached_end_ = true; 141 } else { 142 DCHECK_LT(current_entry_index + 1, num_entries_); 143 if (unzGoToNextFile(zip_file_) != UNZ_OK) { 144 return false; 145 } 146 } 147 current_entry_info_.reset(); 148 return true; 149 } 150 151 bool ZipReader::OpenCurrentEntryInZip() { 152 DCHECK(zip_file_); 153 154 unz_file_info raw_file_info = {}; 155 char raw_file_name_in_zip[internal::kZipMaxPath] = {}; 156 const int result = unzGetCurrentFileInfo(zip_file_, 157 &raw_file_info, 158 raw_file_name_in_zip, 159 sizeof(raw_file_name_in_zip) - 1, 160 NULL, // extraField. 161 0, // extraFieldBufferSize. 162 NULL, // szComment. 163 0); // commentBufferSize. 164 if (result != UNZ_OK) 165 return false; 166 if (raw_file_name_in_zip[0] == '\0') 167 return false; 168 current_entry_info_.reset( 169 new EntryInfo(raw_file_name_in_zip, raw_file_info)); 170 return true; 171 } 172 173 bool ZipReader::LocateAndOpenEntry(const base::FilePath& path_in_zip) { 174 DCHECK(zip_file_); 175 176 current_entry_info_.reset(); 177 reached_end_ = false; 178 const int kDefaultCaseSensivityOfOS = 0; 179 const int result = unzLocateFile(zip_file_, 180 path_in_zip.AsUTF8Unsafe().c_str(), 181 kDefaultCaseSensivityOfOS); 182 if (result != UNZ_OK) 183 return false; 184 185 // Then Open the entry. 186 return OpenCurrentEntryInZip(); 187 } 188 189 bool ZipReader::ExtractCurrentEntryToFilePath( 190 const base::FilePath& output_file_path) { 191 DCHECK(zip_file_); 192 193 // If this is a directory, just create it and return. 194 if (current_entry_info()->is_directory()) 195 return file_util::CreateDirectory(output_file_path); 196 197 const int open_result = unzOpenCurrentFile(zip_file_); 198 if (open_result != UNZ_OK) 199 return false; 200 201 // We can't rely on parent directory entries being specified in the 202 // zip, so we make sure they are created. 203 base::FilePath output_dir_path = output_file_path.DirName(); 204 if (!file_util::CreateDirectory(output_dir_path)) 205 return false; 206 207 net::FileStream stream(NULL); 208 const int flags = (base::PLATFORM_FILE_CREATE_ALWAYS | 209 base::PLATFORM_FILE_WRITE); 210 if (stream.OpenSync(output_file_path, flags) != 0) 211 return false; 212 213 bool success = true; // This becomes false when something bad happens. 214 while (true) { 215 char buf[internal::kZipBufSize]; 216 const int num_bytes_read = unzReadCurrentFile(zip_file_, buf, 217 internal::kZipBufSize); 218 if (num_bytes_read == 0) { 219 // Reached the end of the file. 220 break; 221 } else if (num_bytes_read < 0) { 222 // If num_bytes_read < 0, then it's a specific UNZ_* error code. 223 success = false; 224 break; 225 } else if (num_bytes_read > 0) { 226 // Some data is read. Write it to the output file. 227 if (num_bytes_read != stream.WriteSync(buf, num_bytes_read)) { 228 success = false; 229 break; 230 } 231 } 232 } 233 234 unzCloseCurrentFile(zip_file_); 235 return success; 236 } 237 238 bool ZipReader::ExtractCurrentEntryIntoDirectory( 239 const base::FilePath& output_directory_path) { 240 DCHECK(current_entry_info_.get()); 241 242 base::FilePath output_file_path = output_directory_path.Append( 243 current_entry_info()->file_path()); 244 return ExtractCurrentEntryToFilePath(output_file_path); 245 } 246 247 #if defined(OS_POSIX) 248 bool ZipReader::ExtractCurrentEntryToFd(const int fd) { 249 DCHECK(zip_file_); 250 251 // If this is a directory, there's nothing to extract to the file descriptor, 252 // so return false. 253 if (current_entry_info()->is_directory()) 254 return false; 255 256 const int open_result = unzOpenCurrentFile(zip_file_); 257 if (open_result != UNZ_OK) 258 return false; 259 260 bool success = true; // This becomes false when something bad happens. 261 while (true) { 262 char buf[internal::kZipBufSize]; 263 const int num_bytes_read = unzReadCurrentFile(zip_file_, buf, 264 internal::kZipBufSize); 265 if (num_bytes_read == 0) { 266 // Reached the end of the file. 267 break; 268 } else if (num_bytes_read < 0) { 269 // If num_bytes_read < 0, then it's a specific UNZ_* error code. 270 success = false; 271 break; 272 } else if (num_bytes_read > 0) { 273 // Some data is read. Write it to the output file descriptor. 274 if (num_bytes_read != 275 file_util::WriteFileDescriptor(fd, buf, num_bytes_read)) { 276 success = false; 277 break; 278 } 279 } 280 } 281 282 unzCloseCurrentFile(zip_file_); 283 return success; 284 } 285 #endif // defined(OS_POSIX) 286 287 bool ZipReader::OpenInternal() { 288 DCHECK(zip_file_); 289 290 unz_global_info zip_info = {}; // Zero-clear. 291 if (unzGetGlobalInfo(zip_file_, &zip_info) != UNZ_OK) { 292 return false; 293 } 294 num_entries_ = zip_info.number_entry; 295 if (num_entries_ < 0) 296 return false; 297 298 // We are already at the end if the zip file is empty. 299 reached_end_ = (num_entries_ == 0); 300 return true; 301 } 302 303 void ZipReader::Reset() { 304 zip_file_ = NULL; 305 num_entries_ = 0; 306 reached_end_ = false; 307 current_entry_info_.reset(); 308 } 309 310 } // namespace zip 311