Zip.h 7.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248
  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. const u8* 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. struct [[gnu::packed]] CentralDirectoryRecord {
  63. static constexpr Array<u8, signature_length> signature = { 0x50, 0x4b, 0x01, 0x02 }; // 'PK\x01\x02'
  64. u16 made_by_version;
  65. u16 minimum_version;
  66. u16 general_purpose_flags;
  67. u16 compression_method;
  68. u16 modification_time;
  69. u16 modification_date;
  70. u32 crc32;
  71. u32 compressed_size;
  72. u32 uncompressed_size;
  73. u16 name_length;
  74. u16 extra_data_length;
  75. u16 comment_length;
  76. u16 start_disk;
  77. u16 internal_attributes;
  78. u32 external_attributes;
  79. u32 local_file_header_offset;
  80. const u8* name;
  81. const u8* extra_data;
  82. const u8* comment;
  83. bool read(ReadonlyBytes buffer)
  84. {
  85. constexpr auto fields_size = sizeof(CentralDirectoryRecord) - (sizeof(u8*) * 3);
  86. if (!read_helper<fields_size>(buffer, this))
  87. return false;
  88. if (buffer.size() < size())
  89. return false;
  90. name = buffer.data() + signature.size() + fields_size;
  91. extra_data = name + name_length;
  92. comment = extra_data + extra_data_length;
  93. return true;
  94. }
  95. void write(OutputStream& stream) const
  96. {
  97. stream.write_or_error(signature);
  98. stream << made_by_version;
  99. stream << minimum_version;
  100. stream << general_purpose_flags;
  101. stream << compression_method;
  102. stream << modification_time;
  103. stream << modification_date;
  104. stream << crc32;
  105. stream << compressed_size;
  106. stream << uncompressed_size;
  107. stream << name_length;
  108. stream << extra_data_length;
  109. stream << comment_length;
  110. stream << start_disk;
  111. stream << internal_attributes;
  112. stream << external_attributes;
  113. stream << local_file_header_offset;
  114. if (name_length > 0)
  115. stream.write_or_error({ name, name_length });
  116. if (extra_data_length > 0)
  117. stream.write_or_error({ extra_data, extra_data_length });
  118. if (comment_length > 0)
  119. stream.write_or_error({ comment, comment_length });
  120. }
  121. [[nodiscard]] size_t size() const
  122. {
  123. return signature.size() + (sizeof(CentralDirectoryRecord) - (sizeof(u8*) * 3)) + name_length + extra_data_length + comment_length;
  124. }
  125. };
  126. static constexpr u32 zip_directory_external_attribute = 1 << 4;
  127. struct [[gnu::packed]] LocalFileHeader {
  128. static constexpr Array<u8, signature_length> signature = { 0x50, 0x4b, 0x03, 0x04 }; // 'PK\x03\x04'
  129. u16 minimum_version;
  130. u16 general_purpose_flags;
  131. u16 compression_method;
  132. u16 modification_time;
  133. u16 modification_date;
  134. u32 crc32;
  135. u32 compressed_size;
  136. u32 uncompressed_size;
  137. u16 name_length;
  138. u16 extra_data_length;
  139. const u8* name;
  140. const u8* extra_data;
  141. const u8* compressed_data;
  142. bool read(ReadonlyBytes buffer)
  143. {
  144. constexpr auto fields_size = sizeof(LocalFileHeader) - (sizeof(u8*) * 3);
  145. if (!read_helper<fields_size>(buffer, this))
  146. return false;
  147. if (buffer.size() < signature.size() + fields_size + name_length + extra_data_length + compressed_size)
  148. return false;
  149. name = buffer.data() + signature.size() + fields_size;
  150. extra_data = name + name_length;
  151. compressed_data = extra_data + extra_data_length;
  152. return true;
  153. }
  154. void write(OutputStream& stream) const
  155. {
  156. stream.write_or_error(signature);
  157. stream << minimum_version;
  158. stream << general_purpose_flags;
  159. stream << compression_method;
  160. stream << modification_time;
  161. stream << modification_date;
  162. stream << crc32;
  163. stream << compressed_size;
  164. stream << uncompressed_size;
  165. stream << name_length;
  166. stream << extra_data_length;
  167. if (name_length > 0)
  168. stream.write_or_error({ name, name_length });
  169. if (extra_data_length > 0)
  170. stream.write_or_error({ extra_data, extra_data_length });
  171. if (compressed_size > 0)
  172. stream.write_or_error({ compressed_data, compressed_size });
  173. }
  174. };
  175. enum ZipCompressionMethod : u16 {
  176. Store = 0,
  177. Shrink = 1,
  178. Reduce1 = 2,
  179. Reduce2 = 3,
  180. Reduce3 = 4,
  181. Reduce4 = 5,
  182. Implode = 6,
  183. Reserved = 7,
  184. Deflate = 8
  185. };
  186. struct ZipMember {
  187. String name;
  188. ReadonlyBytes compressed_data; // TODO: maybe the decompression/compression should be handled by LibArchive instead of the user?
  189. ZipCompressionMethod compression_method;
  190. u32 uncompressed_size;
  191. u32 crc32;
  192. bool is_directory;
  193. };
  194. class Zip {
  195. public:
  196. static Optional<Zip> try_create(ReadonlyBytes buffer);
  197. bool for_each_member(Function<IterationDecision(const ZipMember&)>);
  198. private:
  199. static bool find_end_of_central_directory_offset(ReadonlyBytes, size_t& offset);
  200. Zip(u16 member_count, size_t members_start_offset, ReadonlyBytes input_data)
  201. : m_member_count { member_count }
  202. , m_members_start_offset { members_start_offset }
  203. , m_input_data { input_data }
  204. {
  205. }
  206. u16 m_member_count { 0 };
  207. size_t m_members_start_offset { 0 };
  208. ReadonlyBytes m_input_data;
  209. };
  210. class ZipOutputStream {
  211. public:
  212. ZipOutputStream(OutputStream&);
  213. void add_member(const ZipMember&);
  214. void finish();
  215. private:
  216. OutputStream& m_stream;
  217. Vector<ZipMember> m_members;
  218. bool m_finished { false };
  219. };
  220. }