Zip.cpp 8.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179
  1. /*
  2. * Copyright (c) 2021, Idan Horowitz <idan.horowitz@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <LibArchive/Zip.h>
  7. namespace Archive {
  8. bool Zip::find_end_of_central_directory_offset(const ReadonlyBytes& buffer, size_t& offset)
  9. {
  10. for (size_t backwards_offset = 0; backwards_offset <= UINT16_MAX; backwards_offset++) // the file may have a trailing comment of an arbitrary 16 bit length
  11. {
  12. if (buffer.size() < (sizeof(EndOfCentralDirectory) - sizeof(u8*)) + backwards_offset)
  13. return false;
  14. auto signature_offset = (buffer.size() - (sizeof(EndOfCentralDirectory) - sizeof(u8*)) - backwards_offset);
  15. if (memcmp(buffer.data() + signature_offset, end_of_central_directory_signature, sizeof(end_of_central_directory_signature)) == 0) {
  16. offset = signature_offset;
  17. return true;
  18. }
  19. }
  20. return false;
  21. }
  22. Optional<Zip> Zip::try_create(const ReadonlyBytes& buffer)
  23. {
  24. size_t end_of_central_directory_offset;
  25. if (!find_end_of_central_directory_offset(buffer, end_of_central_directory_offset))
  26. return {};
  27. EndOfCentralDirectory end_of_central_directory {};
  28. if (!end_of_central_directory.read(buffer.slice(end_of_central_directory_offset)))
  29. return {};
  30. if (end_of_central_directory.disk_number != 0 || end_of_central_directory.central_directory_start_disk != 0 || end_of_central_directory.disk_records_count != end_of_central_directory.total_records_count)
  31. return {}; // TODO: support multi-volume zip archives
  32. size_t member_offset = end_of_central_directory.central_directory_offset;
  33. for (size_t i = 0; i < end_of_central_directory.total_records_count; i++) {
  34. CentralDirectoryRecord central_directory_record {};
  35. if (member_offset > buffer.size())
  36. return {};
  37. if (!central_directory_record.read(buffer.slice(member_offset)))
  38. return {};
  39. if (central_directory_record.general_purpose_flags & 1)
  40. return {}; // TODO: support encrypted zip members
  41. if (central_directory_record.general_purpose_flags & 3)
  42. return {}; // TODO: support zip data descriptors
  43. if (central_directory_record.compression_method != ZipCompressionMethod::Store && central_directory_record.compression_method != ZipCompressionMethod::Deflate)
  44. return {}; // TODO: support obsolete zip compression methods
  45. if (central_directory_record.compression_method == ZipCompressionMethod::Store && central_directory_record.uncompressed_size != central_directory_record.compressed_size)
  46. return {};
  47. if (central_directory_record.start_disk != 0)
  48. return {}; // TODO: support multi-volume zip archives
  49. if (memchr(central_directory_record.name, 0, central_directory_record.name_length) != nullptr)
  50. return {};
  51. LocalFileHeader local_file_header {};
  52. if (central_directory_record.local_file_header_offset > buffer.size())
  53. return {};
  54. if (!local_file_header.read(buffer.slice(central_directory_record.local_file_header_offset)))
  55. return {};
  56. if (buffer.size() - (local_file_header.compressed_data - buffer.data()) < central_directory_record.compressed_size)
  57. return {};
  58. member_offset += central_directory_record.size();
  59. }
  60. Zip zip;
  61. zip.m_input_data = buffer;
  62. zip.member_count = end_of_central_directory.total_records_count;
  63. zip.members_start_offset = end_of_central_directory.central_directory_offset;
  64. return zip;
  65. }
  66. bool Zip::for_each_member(Function<IterationDecision(const ZipMember&)> callback)
  67. {
  68. size_t member_offset = members_start_offset;
  69. for (size_t i = 0; i < member_count; i++) {
  70. CentralDirectoryRecord central_directory_record {};
  71. VERIFY(central_directory_record.read(m_input_data.slice(member_offset)));
  72. LocalFileHeader local_file_header {};
  73. VERIFY(local_file_header.read(m_input_data.slice(central_directory_record.local_file_header_offset)));
  74. ZipMember member;
  75. char null_terminated_name[central_directory_record.name_length + 1];
  76. memcpy(null_terminated_name, central_directory_record.name, central_directory_record.name_length);
  77. null_terminated_name[central_directory_record.name_length] = 0;
  78. member.name = String { null_terminated_name };
  79. member.compressed_data = { local_file_header.compressed_data, central_directory_record.compressed_size };
  80. member.compression_method = static_cast<ZipCompressionMethod>(central_directory_record.compression_method);
  81. member.uncompressed_size = central_directory_record.uncompressed_size;
  82. member.crc32 = central_directory_record.crc32;
  83. member.is_directory = central_directory_record.external_attributes & zip_directory_external_attribute || member.name.ends_with('/'); // FIXME: better directory detection
  84. if (callback(member) == IterationDecision::Break)
  85. return false;
  86. member_offset += central_directory_record.size();
  87. }
  88. return true;
  89. }
  90. ZipOutputStream::ZipOutputStream(OutputStream& stream)
  91. : m_stream(stream)
  92. {
  93. }
  94. void ZipOutputStream::add_member(const ZipMember& member)
  95. {
  96. VERIFY(!m_finished);
  97. VERIFY(member.name.length() <= UINT16_MAX);
  98. VERIFY(member.compressed_data.size() <= UINT32_MAX);
  99. m_members.append(member);
  100. LocalFileHeader local_file_header {};
  101. local_file_header.minimum_version = member.compression_method == ZipCompressionMethod::Deflate ? 20 : 10; // Deflate was added in PKZip 2.0
  102. local_file_header.general_purpose_flags = 0;
  103. local_file_header.compression_method = static_cast<u16>(member.compression_method);
  104. local_file_header.modification_time = 0; // TODO: support modification time
  105. local_file_header.modification_date = 0;
  106. local_file_header.crc32 = member.crc32;
  107. local_file_header.compressed_size = member.compressed_data.size();
  108. local_file_header.uncompressed_size = member.uncompressed_size;
  109. local_file_header.name_length = member.name.length();
  110. local_file_header.extra_data_length = 0;
  111. local_file_header.name = (const u8*)(member.name.characters());
  112. local_file_header.extra_data = nullptr;
  113. local_file_header.compressed_data = member.compressed_data.data();
  114. local_file_header.write(m_stream);
  115. }
  116. void ZipOutputStream::finish()
  117. {
  118. VERIFY(!m_finished);
  119. m_finished = true;
  120. auto file_header_offset = 0;
  121. auto central_directory_size = 0;
  122. for (const ZipMember& member : m_members) {
  123. CentralDirectoryRecord central_directory_record {};
  124. auto zip_version = member.compression_method == ZipCompressionMethod::Deflate ? 20 : 10; // Deflate was added in PKZip 2.0
  125. central_directory_record.made_by_version = zip_version;
  126. central_directory_record.minimum_version = zip_version;
  127. central_directory_record.general_purpose_flags = 0;
  128. central_directory_record.compression_method = static_cast<u16>(member.compression_method);
  129. central_directory_record.modification_time = 0; // TODO: support modification time
  130. central_directory_record.modification_date = 0;
  131. central_directory_record.crc32 = member.crc32;
  132. central_directory_record.compressed_size = member.compressed_data.size();
  133. central_directory_record.uncompressed_size = member.uncompressed_size;
  134. central_directory_record.name_length = member.name.length();
  135. central_directory_record.extra_data_length = 0;
  136. central_directory_record.comment_length = 0;
  137. central_directory_record.start_disk = 0;
  138. central_directory_record.internal_attributes = 0;
  139. central_directory_record.external_attributes = member.is_directory ? zip_directory_external_attribute : 0;
  140. central_directory_record.local_file_header_offset = file_header_offset; // FIXME: we assume the wrapped output stream was never written to before us
  141. file_header_offset += sizeof(local_file_header_signature) + (sizeof(LocalFileHeader) - (sizeof(u8*) * 3)) + member.name.length() + member.compressed_data.size();
  142. central_directory_record.name = (const u8*)(member.name.characters());
  143. central_directory_record.extra_data = nullptr;
  144. central_directory_record.comment = nullptr;
  145. central_directory_record.write(m_stream);
  146. central_directory_size += central_directory_record.size();
  147. }
  148. EndOfCentralDirectory end_of_central_directory {};
  149. end_of_central_directory.disk_number = 0;
  150. end_of_central_directory.central_directory_start_disk = 0;
  151. end_of_central_directory.disk_records_count = m_members.size();
  152. end_of_central_directory.total_records_count = m_members.size();
  153. end_of_central_directory.central_directory_size = central_directory_size;
  154. end_of_central_directory.central_directory_offset = file_header_offset;
  155. end_of_central_directory.comment_length = 0;
  156. end_of_central_directory.comment = nullptr;
  157. end_of_central_directory.write(m_stream);
  158. }
  159. }