Zip.h 7.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250
  1. /*
  2. * Copyright (c) 2021, Idan Horowitz <idan.horowitz@serenityos.org>
  3. * Copyright (c) 2022, the SerenityOS developers.
  4. *
  5. * SPDX-License-Identifier: BSD-2-Clause
  6. */
  7. #pragma once
  8. #include <AK/Array.h>
  9. #include <AK/Function.h>
  10. #include <AK/IterationDecision.h>
  11. #include <AK/Stream.h>
  12. #include <AK/String.h>
  13. #include <AK/Vector.h>
  14. #include <string.h>
  15. namespace Archive {
  16. template<size_t fields_size, class T>
  17. static bool read_helper(ReadonlyBytes buffer, T* self)
  18. {
  19. if (buffer.size() < T::signature.size() + fields_size)
  20. return false;
  21. if (buffer.slice(0, T::signature.size()) != T::signature)
  22. return false;
  23. memcpy(self, buffer.data() + T::signature.size(), fields_size);
  24. return true;
  25. }
  26. // NOTE: Due to the format of zip files compression is streamed and decompression is random access.
  27. static constexpr auto signature_length = 4;
  28. struct [[gnu::packed]] EndOfCentralDirectory {
  29. static constexpr Array<u8, signature_length> signature = { 0x50, 0x4b, 0x05, 0x06 }; // 'PK\x05\x06'
  30. u16 disk_number;
  31. u16 central_directory_start_disk;
  32. u16 disk_records_count;
  33. u16 total_records_count;
  34. u32 central_directory_size;
  35. u32 central_directory_offset;
  36. u16 comment_length;
  37. u8 const* comment;
  38. bool read(ReadonlyBytes buffer)
  39. {
  40. constexpr auto fields_size = sizeof(EndOfCentralDirectory) - (sizeof(u8*) * 1);
  41. if (!read_helper<fields_size>(buffer, this))
  42. return false;
  43. if (buffer.size() < signature.size() + fields_size + comment_length)
  44. return false;
  45. comment = buffer.data() + signature.size() + fields_size;
  46. return true;
  47. }
  48. void write(OutputStream& stream) const
  49. {
  50. stream.write_or_error(signature);
  51. stream << disk_number;
  52. stream << central_directory_start_disk;
  53. stream << disk_records_count;
  54. stream << total_records_count;
  55. stream << central_directory_size;
  56. stream << central_directory_offset;
  57. stream << comment_length;
  58. if (comment_length > 0)
  59. stream.write_or_error({ comment, comment_length });
  60. }
  61. };
  62. enum class ZipCompressionMethod : u16 {
  63. Store = 0,
  64. Shrink = 1,
  65. Reduce1 = 2,
  66. Reduce2 = 3,
  67. Reduce3 = 4,
  68. Reduce4 = 5,
  69. Implode = 6,
  70. Reserved = 7,
  71. Deflate = 8
  72. };
  73. OutputStream& operator<<(OutputStream& stream, ZipCompressionMethod method);
  74. struct [[gnu::packed]] CentralDirectoryRecord {
  75. static constexpr Array<u8, signature_length> signature = { 0x50, 0x4b, 0x01, 0x02 }; // 'PK\x01\x02'
  76. u16 made_by_version;
  77. u16 minimum_version;
  78. u16 general_purpose_flags;
  79. ZipCompressionMethod compression_method;
  80. u16 modification_time;
  81. u16 modification_date;
  82. u32 crc32;
  83. u32 compressed_size;
  84. u32 uncompressed_size;
  85. u16 name_length;
  86. u16 extra_data_length;
  87. u16 comment_length;
  88. u16 start_disk;
  89. u16 internal_attributes;
  90. u32 external_attributes;
  91. u32 local_file_header_offset;
  92. u8 const* name;
  93. u8 const* extra_data;
  94. u8 const* comment;
  95. bool read(ReadonlyBytes buffer)
  96. {
  97. constexpr auto fields_size = sizeof(CentralDirectoryRecord) - (sizeof(u8*) * 3);
  98. if (!read_helper<fields_size>(buffer, this))
  99. return false;
  100. if (buffer.size() < size())
  101. return false;
  102. name = buffer.data() + signature.size() + fields_size;
  103. extra_data = name + name_length;
  104. comment = extra_data + extra_data_length;
  105. return true;
  106. }
  107. void write(OutputStream& stream) const
  108. {
  109. stream.write_or_error(signature);
  110. stream << made_by_version;
  111. stream << minimum_version;
  112. stream << general_purpose_flags;
  113. stream << compression_method;
  114. stream << modification_time;
  115. stream << modification_date;
  116. stream << crc32;
  117. stream << compressed_size;
  118. stream << uncompressed_size;
  119. stream << name_length;
  120. stream << extra_data_length;
  121. stream << comment_length;
  122. stream << start_disk;
  123. stream << internal_attributes;
  124. stream << external_attributes;
  125. stream << local_file_header_offset;
  126. if (name_length > 0)
  127. stream.write_or_error({ name, name_length });
  128. if (extra_data_length > 0)
  129. stream.write_or_error({ extra_data, extra_data_length });
  130. if (comment_length > 0)
  131. stream.write_or_error({ comment, comment_length });
  132. }
  133. [[nodiscard]] size_t size() const
  134. {
  135. return signature.size() + (sizeof(CentralDirectoryRecord) - (sizeof(u8*) * 3)) + name_length + extra_data_length + comment_length;
  136. }
  137. };
  138. static constexpr u32 zip_directory_external_attribute = 1 << 4;
  139. struct [[gnu::packed]] LocalFileHeader {
  140. static constexpr Array<u8, signature_length> signature = { 0x50, 0x4b, 0x03, 0x04 }; // 'PK\x03\x04'
  141. u16 minimum_version;
  142. u16 general_purpose_flags;
  143. u16 compression_method;
  144. u16 modification_time;
  145. u16 modification_date;
  146. u32 crc32;
  147. u32 compressed_size;
  148. u32 uncompressed_size;
  149. u16 name_length;
  150. u16 extra_data_length;
  151. u8 const* name;
  152. u8 const* extra_data;
  153. u8 const* compressed_data;
  154. bool read(ReadonlyBytes buffer)
  155. {
  156. constexpr auto fields_size = sizeof(LocalFileHeader) - (sizeof(u8*) * 3);
  157. if (!read_helper<fields_size>(buffer, this))
  158. return false;
  159. if (buffer.size() < signature.size() + fields_size + name_length + extra_data_length + compressed_size)
  160. return false;
  161. name = buffer.data() + signature.size() + fields_size;
  162. extra_data = name + name_length;
  163. compressed_data = extra_data + extra_data_length;
  164. return true;
  165. }
  166. void write(OutputStream& stream) const
  167. {
  168. stream.write_or_error(signature);
  169. stream << minimum_version;
  170. stream << general_purpose_flags;
  171. stream << compression_method;
  172. stream << modification_time;
  173. stream << modification_date;
  174. stream << crc32;
  175. stream << compressed_size;
  176. stream << uncompressed_size;
  177. stream << name_length;
  178. stream << extra_data_length;
  179. if (name_length > 0)
  180. stream.write_or_error({ name, name_length });
  181. if (extra_data_length > 0)
  182. stream.write_or_error({ extra_data, extra_data_length });
  183. if (compressed_size > 0)
  184. stream.write_or_error({ compressed_data, compressed_size });
  185. }
  186. };
  187. struct ZipMember {
  188. String name;
  189. ReadonlyBytes compressed_data; // TODO: maybe the decompression/compression should be handled by LibArchive instead of the user?
  190. ZipCompressionMethod compression_method;
  191. u32 uncompressed_size;
  192. u32 crc32;
  193. bool is_directory;
  194. };
  195. class Zip {
  196. public:
  197. static Optional<Zip> try_create(ReadonlyBytes buffer);
  198. bool for_each_member(Function<IterationDecision(ZipMember const&)>);
  199. private:
  200. static bool find_end_of_central_directory_offset(ReadonlyBytes, size_t& offset);
  201. Zip(u16 member_count, size_t members_start_offset, ReadonlyBytes input_data)
  202. : m_member_count { member_count }
  203. , m_members_start_offset { members_start_offset }
  204. , m_input_data { input_data }
  205. {
  206. }
  207. u16 m_member_count { 0 };
  208. size_t m_members_start_offset { 0 };
  209. ReadonlyBytes m_input_data;
  210. };
  211. class ZipOutputStream {
  212. public:
  213. ZipOutputStream(OutputStream&);
  214. void add_member(ZipMember const&);
  215. void finish();
  216. private:
  217. OutputStream& m_stream;
  218. Vector<ZipMember> m_members;
  219. bool m_finished { false };
  220. };
  221. }