Home | History | Annotate | Download | only in libziparchive
      1 /*
      2  * Copyright (C) 2015 The Android Open Source Project
      3  *
      4  * Licensed under the Apache License, Version 2.0 (the "License");
      5  * you may not use this file except in compliance with the License.
      6  * You may obtain a copy of the License at
      7  *
      8  *      http://www.apache.org/licenses/LICENSE-2.0
      9  *
     10  * Unless required by applicable law or agreed to in writing, software
     11  * distributed under the License is distributed on an "AS IS" BASIS,
     12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
     13  * See the License for the specific language governing permissions and
     14  * limitations under the License.
     15  */
     16 
     17 #include "ziparchive/zip_writer.h"
     18 
     19 #include <cstdio>
     20 #include <sys/param.h>
     21 #include <sys/stat.h>
     22 #include <zlib.h>
     23 #define DEF_MEM_LEVEL 8                // normally in zutil.h?
     24 
     25 #include <memory>
     26 #include <vector>
     27 
     28 #include "android-base/logging.h"
     29 #include "utils/Compat.h"
     30 #include "utils/Log.h"
     31 
     32 #include "entry_name_utils-inl.h"
     33 #include "zip_archive_common.h"
     34 
     35 #if !defined(powerof2)
     36 #define powerof2(x) ((((x)-1)&(x))==0)
     37 #endif
     38 
     39 /* Zip compression methods we support */
     40 enum {
     41   kCompressStored     = 0,        // no compression
     42   kCompressDeflated   = 8,        // standard deflate
     43 };
     44 
     45 // Size of the output buffer used for compression.
     46 static const size_t kBufSize = 32768u;
     47 
     48 // No error, operation completed successfully.
     49 static const int32_t kNoError = 0;
     50 
     51 // The ZipWriter is in a bad state.
     52 static const int32_t kInvalidState = -1;
     53 
     54 // There was an IO error while writing to disk.
     55 static const int32_t kIoError = -2;
     56 
     57 // The zip entry name was invalid.
     58 static const int32_t kInvalidEntryName = -3;
     59 
     60 // An error occurred in zlib.
     61 static const int32_t kZlibError = -4;
     62 
     63 // The start aligned function was called with the aligned flag.
     64 static const int32_t kInvalidAlign32Flag = -5;
     65 
     66 // The alignment parameter is not a power of 2.
     67 static const int32_t kInvalidAlignment = -6;
     68 
     69 static const char* sErrorCodes[] = {
     70     "Invalid state",
     71     "IO error",
     72     "Invalid entry name",
     73     "Zlib error",
     74 };
     75 
     76 const char* ZipWriter::ErrorCodeString(int32_t error_code) {
     77   if (error_code < 0 && (-error_code) < static_cast<int32_t>(arraysize(sErrorCodes))) {
     78     return sErrorCodes[-error_code];
     79   }
     80   return nullptr;
     81 }
     82 
     83 static void DeleteZStream(z_stream* stream) {
     84   deflateEnd(stream);
     85   delete stream;
     86 }
     87 
     88 ZipWriter::ZipWriter(FILE* f) : file_(f), seekable_(false), current_offset_(0),
     89                                 state_(State::kWritingZip), z_stream_(nullptr, DeleteZStream),
     90                                 buffer_(kBufSize) {
     91   // Check if the file is seekable (regular file). If fstat fails, that's fine, subsequent calls
     92   // will fail as well.
     93   struct stat file_stats;
     94   if (fstat(fileno(f), &file_stats) == 0) {
     95     seekable_ = S_ISREG(file_stats.st_mode);
     96   }
     97 }
     98 
     99 ZipWriter::ZipWriter(ZipWriter&& writer) : file_(writer.file_),
    100                                            seekable_(writer.seekable_),
    101                                            current_offset_(writer.current_offset_),
    102                                            state_(writer.state_),
    103                                            files_(std::move(writer.files_)),
    104                                            z_stream_(std::move(writer.z_stream_)),
    105                                            buffer_(std::move(writer.buffer_)){
    106   writer.file_ = nullptr;
    107   writer.state_ = State::kError;
    108 }
    109 
    110 ZipWriter& ZipWriter::operator=(ZipWriter&& writer) {
    111   file_ = writer.file_;
    112   seekable_ = writer.seekable_;
    113   current_offset_ = writer.current_offset_;
    114   state_ = writer.state_;
    115   files_ = std::move(writer.files_);
    116   z_stream_ = std::move(writer.z_stream_);
    117   buffer_ = std::move(writer.buffer_);
    118   writer.file_ = nullptr;
    119   writer.state_ = State::kError;
    120   return *this;
    121 }
    122 
    123 int32_t ZipWriter::HandleError(int32_t error_code) {
    124   state_ = State::kError;
    125   z_stream_.reset();
    126   return error_code;
    127 }
    128 
    129 int32_t ZipWriter::StartEntry(const char* path, size_t flags) {
    130   uint32_t alignment = 0;
    131   if (flags & kAlign32) {
    132     flags &= ~kAlign32;
    133     alignment = 4;
    134   }
    135   return StartAlignedEntryWithTime(path, flags, time_t(), alignment);
    136 }
    137 
    138 int32_t ZipWriter::StartAlignedEntry(const char* path, size_t flags, uint32_t alignment) {
    139   return StartAlignedEntryWithTime(path, flags, time_t(), alignment);
    140 }
    141 
    142 int32_t ZipWriter::StartEntryWithTime(const char* path, size_t flags, time_t time) {
    143   uint32_t alignment = 0;
    144   if (flags & kAlign32) {
    145     flags &= ~kAlign32;
    146     alignment = 4;
    147   }
    148   return StartAlignedEntryWithTime(path, flags, time, alignment);
    149 }
    150 
    151 static void ExtractTimeAndDate(time_t when, uint16_t* out_time, uint16_t* out_date) {
    152   /* round up to an even number of seconds */
    153   when = static_cast<time_t>((static_cast<unsigned long>(when) + 1) & (~1));
    154 
    155   struct tm* ptm;
    156 #if !defined(_WIN32)
    157     struct tm tm_result;
    158     ptm = localtime_r(&when, &tm_result);
    159 #else
    160     ptm = localtime(&when);
    161 #endif
    162 
    163   int year = ptm->tm_year;
    164   if (year < 80) {
    165     year = 80;
    166   }
    167 
    168   *out_date = (year - 80) << 9 | (ptm->tm_mon + 1) << 5 | ptm->tm_mday;
    169   *out_time = ptm->tm_hour << 11 | ptm->tm_min << 5 | ptm->tm_sec >> 1;
    170 }
    171 
    172 static void CopyFromFileEntry(const ZipWriter::FileEntry& src, bool use_data_descriptor,
    173                               LocalFileHeader* dst) {
    174   dst->lfh_signature = LocalFileHeader::kSignature;
    175   if (use_data_descriptor) {
    176     // Set this flag to denote that a DataDescriptor struct will appear after the data,
    177     // containing the crc and size fields.
    178     dst->gpb_flags |= kGPBDDFlagMask;
    179 
    180     // The size and crc fields must be 0.
    181     dst->compressed_size = 0u;
    182     dst->uncompressed_size = 0u;
    183     dst->crc32 = 0u;
    184   } else {
    185     dst->compressed_size = src.compressed_size;
    186     dst->uncompressed_size = src.uncompressed_size;
    187     dst->crc32 = src.crc32;
    188   }
    189   dst->compression_method = src.compression_method;
    190   dst->last_mod_time = src.last_mod_time;
    191   dst->last_mod_date = src.last_mod_date;
    192   dst->file_name_length = src.path.size();
    193   dst->extra_field_length = src.padding_length;
    194 }
    195 
    196 int32_t ZipWriter::StartAlignedEntryWithTime(const char* path, size_t flags,
    197                                              time_t time, uint32_t alignment) {
    198   if (state_ != State::kWritingZip) {
    199     return kInvalidState;
    200   }
    201 
    202   if (flags & kAlign32) {
    203     return kInvalidAlign32Flag;
    204   }
    205 
    206   if (powerof2(alignment) == 0) {
    207     return kInvalidAlignment;
    208   }
    209 
    210   FileEntry file_entry = {};
    211   file_entry.local_file_header_offset = current_offset_;
    212   file_entry.path = path;
    213 
    214   if (!IsValidEntryName(reinterpret_cast<const uint8_t*>(file_entry.path.data()),
    215                         file_entry.path.size())) {
    216     return kInvalidEntryName;
    217   }
    218 
    219   if (flags & ZipWriter::kCompress) {
    220     file_entry.compression_method = kCompressDeflated;
    221 
    222     int32_t result = PrepareDeflate();
    223     if (result != kNoError) {
    224       return result;
    225     }
    226   } else {
    227     file_entry.compression_method = kCompressStored;
    228   }
    229 
    230   ExtractTimeAndDate(time, &file_entry.last_mod_time, &file_entry.last_mod_date);
    231 
    232   off_t offset = current_offset_ + sizeof(LocalFileHeader) + file_entry.path.size();
    233   std::vector<char> zero_padding;
    234   if (alignment != 0 && (offset & (alignment - 1))) {
    235     // Pad the extra field so the data will be aligned.
    236     uint16_t padding = alignment - (offset % alignment);
    237     file_entry.padding_length = padding;
    238     offset += padding;
    239     zero_padding.resize(padding, 0);
    240   }
    241 
    242   LocalFileHeader header = {};
    243   // Always start expecting a data descriptor. When the data has finished being written,
    244   // if it is possible to seek back, the GPB flag will reset and the sizes written.
    245   CopyFromFileEntry(file_entry, true /*use_data_descriptor*/, &header);
    246 
    247   if (fwrite(&header, sizeof(header), 1, file_) != 1) {
    248     return HandleError(kIoError);
    249   }
    250 
    251   if (fwrite(path, sizeof(*path), file_entry.path.size(), file_) != file_entry.path.size()) {
    252     return HandleError(kIoError);
    253   }
    254 
    255   if (file_entry.padding_length != 0 &&
    256       fwrite(zero_padding.data(), 1, file_entry.padding_length, file_)
    257       != file_entry.padding_length) {
    258     return HandleError(kIoError);
    259   }
    260 
    261   current_file_entry_ = std::move(file_entry);
    262   current_offset_ = offset;
    263   state_ = State::kWritingEntry;
    264   return kNoError;
    265 }
    266 
    267 int32_t ZipWriter::DiscardLastEntry() {
    268   if (state_ != State::kWritingZip || files_.empty()) {
    269     return kInvalidState;
    270   }
    271 
    272   FileEntry& last_entry = files_.back();
    273   current_offset_ = last_entry.local_file_header_offset;
    274   if (fseeko(file_, current_offset_, SEEK_SET) != 0) {
    275     return HandleError(kIoError);
    276   }
    277   files_.pop_back();
    278   return kNoError;
    279 }
    280 
    281 int32_t ZipWriter::GetLastEntry(FileEntry* out_entry) {
    282   CHECK(out_entry != nullptr);
    283 
    284   if (files_.empty()) {
    285     return kInvalidState;
    286   }
    287   *out_entry = files_.back();
    288   return kNoError;
    289 }
    290 
    291 int32_t ZipWriter::PrepareDeflate() {
    292   CHECK(state_ == State::kWritingZip);
    293 
    294   // Initialize the z_stream for compression.
    295   z_stream_ = std::unique_ptr<z_stream, void(*)(z_stream*)>(new z_stream(), DeleteZStream);
    296 
    297 #pragma GCC diagnostic push
    298 #pragma GCC diagnostic ignored "-Wold-style-cast"
    299   int zerr = deflateInit2(z_stream_.get(), Z_BEST_COMPRESSION, Z_DEFLATED, -MAX_WBITS,
    300                           DEF_MEM_LEVEL, Z_DEFAULT_STRATEGY);
    301 #pragma GCC diagnostic pop
    302 
    303   if (zerr != Z_OK) {
    304     if (zerr == Z_VERSION_ERROR) {
    305       ALOGE("Installed zlib is not compatible with linked version (%s)", ZLIB_VERSION);
    306       return HandleError(kZlibError);
    307     } else {
    308       ALOGE("deflateInit2 failed (zerr=%d)", zerr);
    309       return HandleError(kZlibError);
    310     }
    311   }
    312 
    313   z_stream_->next_out = buffer_.data();
    314   z_stream_->avail_out = buffer_.size();
    315   return kNoError;
    316 }
    317 
    318 int32_t ZipWriter::WriteBytes(const void* data, size_t len) {
    319   if (state_ != State::kWritingEntry) {
    320     return HandleError(kInvalidState);
    321   }
    322 
    323   int32_t result = kNoError;
    324   if (current_file_entry_.compression_method & kCompressDeflated) {
    325     result = CompressBytes(&current_file_entry_, data, len);
    326   } else {
    327     result = StoreBytes(&current_file_entry_, data, len);
    328   }
    329 
    330   if (result != kNoError) {
    331     return result;
    332   }
    333 
    334   current_file_entry_.crc32 = crc32(current_file_entry_.crc32,
    335                                     reinterpret_cast<const Bytef*>(data), len);
    336   current_file_entry_.uncompressed_size += len;
    337   return kNoError;
    338 }
    339 
    340 int32_t ZipWriter::StoreBytes(FileEntry* file, const void* data, size_t len) {
    341   CHECK(state_ == State::kWritingEntry);
    342 
    343   if (fwrite(data, 1, len, file_) != len) {
    344     return HandleError(kIoError);
    345   }
    346   file->compressed_size += len;
    347   current_offset_ += len;
    348   return kNoError;
    349 }
    350 
    351 int32_t ZipWriter::CompressBytes(FileEntry* file, const void* data, size_t len) {
    352   CHECK(state_ == State::kWritingEntry);
    353   CHECK(z_stream_);
    354   CHECK(z_stream_->next_out != nullptr);
    355   CHECK(z_stream_->avail_out != 0);
    356 
    357   // Prepare the input.
    358   z_stream_->next_in = reinterpret_cast<const uint8_t*>(data);
    359   z_stream_->avail_in = len;
    360 
    361   while (z_stream_->avail_in > 0) {
    362     // We have more data to compress.
    363     int zerr = deflate(z_stream_.get(), Z_NO_FLUSH);
    364     if (zerr != Z_OK) {
    365       return HandleError(kZlibError);
    366     }
    367 
    368     if (z_stream_->avail_out == 0) {
    369       // The output is full, let's write it to disk.
    370       size_t write_bytes = z_stream_->next_out - buffer_.data();
    371       if (fwrite(buffer_.data(), 1, write_bytes, file_) != write_bytes) {
    372         return HandleError(kIoError);
    373       }
    374       file->compressed_size += write_bytes;
    375       current_offset_ += write_bytes;
    376 
    377       // Reset the output buffer for the next input.
    378       z_stream_->next_out = buffer_.data();
    379       z_stream_->avail_out = buffer_.size();
    380     }
    381   }
    382   return kNoError;
    383 }
    384 
    385 int32_t ZipWriter::FlushCompressedBytes(FileEntry* file) {
    386   CHECK(state_ == State::kWritingEntry);
    387   CHECK(z_stream_);
    388   CHECK(z_stream_->next_out != nullptr);
    389   CHECK(z_stream_->avail_out != 0);
    390 
    391   // Keep deflating while there isn't enough space in the buffer to
    392   // to complete the compress.
    393   int zerr;
    394   while ((zerr = deflate(z_stream_.get(), Z_FINISH)) == Z_OK) {
    395     CHECK(z_stream_->avail_out == 0);
    396     size_t write_bytes = z_stream_->next_out - buffer_.data();
    397     if (fwrite(buffer_.data(), 1, write_bytes, file_) != write_bytes) {
    398       return HandleError(kIoError);
    399     }
    400     file->compressed_size += write_bytes;
    401     current_offset_ += write_bytes;
    402 
    403     z_stream_->next_out = buffer_.data();
    404     z_stream_->avail_out = buffer_.size();
    405   }
    406   if (zerr != Z_STREAM_END) {
    407     return HandleError(kZlibError);
    408   }
    409 
    410   size_t write_bytes = z_stream_->next_out - buffer_.data();
    411   if (write_bytes != 0) {
    412     if (fwrite(buffer_.data(), 1, write_bytes, file_) != write_bytes) {
    413       return HandleError(kIoError);
    414     }
    415     file->compressed_size += write_bytes;
    416     current_offset_ += write_bytes;
    417   }
    418   z_stream_.reset();
    419   return kNoError;
    420 }
    421 
    422 int32_t ZipWriter::FinishEntry() {
    423   if (state_ != State::kWritingEntry) {
    424     return kInvalidState;
    425   }
    426 
    427   if (current_file_entry_.compression_method & kCompressDeflated) {
    428     int32_t result = FlushCompressedBytes(&current_file_entry_);
    429     if (result != kNoError) {
    430       return result;
    431     }
    432   }
    433 
    434   if ((current_file_entry_.compression_method & kCompressDeflated) || !seekable_) {
    435     // Some versions of ZIP don't allow STORED data to have a trailing DataDescriptor.
    436     // If this file is not seekable, or if the data is compressed, write a DataDescriptor.
    437     const uint32_t sig = DataDescriptor::kOptSignature;
    438     if (fwrite(&sig, sizeof(sig), 1, file_) != 1) {
    439       return HandleError(kIoError);
    440     }
    441 
    442     DataDescriptor dd = {};
    443     dd.crc32 = current_file_entry_.crc32;
    444     dd.compressed_size = current_file_entry_.compressed_size;
    445     dd.uncompressed_size = current_file_entry_.uncompressed_size;
    446     if (fwrite(&dd, sizeof(dd), 1, file_) != 1) {
    447       return HandleError(kIoError);
    448     }
    449     current_offset_ += sizeof(DataDescriptor::kOptSignature) + sizeof(dd);
    450   } else {
    451     // Seek back to the header and rewrite to include the size.
    452     if (fseeko(file_, current_file_entry_.local_file_header_offset, SEEK_SET) != 0) {
    453       return HandleError(kIoError);
    454     }
    455 
    456     LocalFileHeader header = {};
    457     CopyFromFileEntry(current_file_entry_, false /*use_data_descriptor*/, &header);
    458 
    459     if (fwrite(&header, sizeof(header), 1, file_) != 1) {
    460       return HandleError(kIoError);
    461     }
    462 
    463     if (fseeko(file_, current_offset_, SEEK_SET) != 0) {
    464       return HandleError(kIoError);
    465     }
    466   }
    467 
    468   files_.emplace_back(std::move(current_file_entry_));
    469   state_ = State::kWritingZip;
    470   return kNoError;
    471 }
    472 
    473 int32_t ZipWriter::Finish() {
    474   if (state_ != State::kWritingZip) {
    475     return kInvalidState;
    476   }
    477 
    478   off_t startOfCdr = current_offset_;
    479   for (FileEntry& file : files_) {
    480     CentralDirectoryRecord cdr = {};
    481     cdr.record_signature = CentralDirectoryRecord::kSignature;
    482     if ((file.compression_method & kCompressDeflated) || !seekable_) {
    483       cdr.gpb_flags |= kGPBDDFlagMask;
    484     }
    485     cdr.compression_method = file.compression_method;
    486     cdr.last_mod_time = file.last_mod_time;
    487     cdr.last_mod_date = file.last_mod_date;
    488     cdr.crc32 = file.crc32;
    489     cdr.compressed_size = file.compressed_size;
    490     cdr.uncompressed_size = file.uncompressed_size;
    491     cdr.file_name_length = file.path.size();
    492     cdr.local_file_header_offset = static_cast<uint32_t>(file.local_file_header_offset);
    493     if (fwrite(&cdr, sizeof(cdr), 1, file_) != 1) {
    494       return HandleError(kIoError);
    495     }
    496 
    497     if (fwrite(file.path.data(), 1, file.path.size(), file_) != file.path.size()) {
    498       return HandleError(kIoError);
    499     }
    500 
    501     current_offset_ += sizeof(cdr) + file.path.size();
    502   }
    503 
    504   EocdRecord er = {};
    505   er.eocd_signature = EocdRecord::kSignature;
    506   er.disk_num = 0;
    507   er.cd_start_disk = 0;
    508   er.num_records_on_disk = files_.size();
    509   er.num_records = files_.size();
    510   er.cd_size = current_offset_ - startOfCdr;
    511   er.cd_start_offset = startOfCdr;
    512 
    513   if (fwrite(&er, sizeof(er), 1, file_) != 1) {
    514     return HandleError(kIoError);
    515   }
    516 
    517   current_offset_ += sizeof(er);
    518 
    519   // Since we can BackUp() and potentially finish writing at an offset less than one we had
    520   // already written at, we must truncate the file.
    521 
    522   if (ftruncate(fileno(file_), current_offset_) != 0) {
    523     return HandleError(kIoError);
    524   }
    525 
    526   if (fflush(file_) != 0) {
    527     return HandleError(kIoError);
    528   }
    529 
    530   state_ = State::kDone;
    531   return kNoError;
    532 }
    533