MimeData.cpp 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203
  1. /*
  2. * Copyright (c) 2020, Andreas Kling <kling@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <AK/LexicalPath.h>
  7. #include <AK/StringBuilder.h>
  8. #include <LibCore/File.h>
  9. #include <LibCore/MimeData.h>
  10. namespace Core {
  11. Vector<DeprecatedString> MimeData::formats() const
  12. {
  13. Vector<DeprecatedString> mime_types;
  14. mime_types.ensure_capacity(m_data.size());
  15. for (auto& it : m_data)
  16. mime_types.unchecked_append(it.key);
  17. return mime_types;
  18. }
  19. Vector<URL> MimeData::urls() const
  20. {
  21. auto it = m_data.find("text/uri-list");
  22. if (it == m_data.end())
  23. return {};
  24. Vector<URL> urls;
  25. for (auto& line : StringView(it->value).split_view('\n')) {
  26. urls.append(URL(line));
  27. }
  28. return urls;
  29. }
  30. ErrorOr<void> MimeData::set_urls(Vector<URL> const& urls)
  31. {
  32. StringBuilder builder;
  33. for (auto& url : urls) {
  34. TRY(builder.try_append(url.to_deprecated_string()));
  35. TRY(builder.try_append('\n'));
  36. }
  37. set_data("text/uri-list", TRY(builder.to_byte_buffer()));
  38. return {};
  39. }
  40. DeprecatedString MimeData::text() const
  41. {
  42. return DeprecatedString::copy(m_data.get("text/plain").value_or({}));
  43. }
  44. void MimeData::set_text(DeprecatedString const& text)
  45. {
  46. set_data("text/plain", text.to_byte_buffer());
  47. }
  48. StringView guess_mime_type_based_on_filename(StringView path)
  49. {
  50. if (path.ends_with(".pbm"sv, CaseSensitivity::CaseInsensitive))
  51. return "image/x‑portable‑bitmap"sv;
  52. if (path.ends_with(".pgm"sv, CaseSensitivity::CaseInsensitive))
  53. return "image/x‑portable‑graymap"sv;
  54. if (path.ends_with(".png"sv, CaseSensitivity::CaseInsensitive))
  55. return "image/png"sv;
  56. if (path.ends_with(".ppm"sv, CaseSensitivity::CaseInsensitive))
  57. return "image/x‑portable‑pixmap"sv;
  58. if (path.ends_with(".gif"sv, CaseSensitivity::CaseInsensitive))
  59. return "image/gif"sv;
  60. if (path.ends_with(".bmp"sv, CaseSensitivity::CaseInsensitive))
  61. return "image/bmp"sv;
  62. if (path.ends_with(".jpg"sv, CaseSensitivity::CaseInsensitive) || path.ends_with(".jpeg"sv, CaseSensitivity::CaseInsensitive))
  63. return "image/jpeg"sv;
  64. if (path.ends_with(".qoi"sv, CaseSensitivity::CaseInsensitive))
  65. return "image/x-qoi"sv;
  66. if (path.ends_with(".svg"sv, CaseSensitivity::CaseInsensitive))
  67. return "image/svg+xml"sv;
  68. if (path.ends_with(".tga"sv, CaseSensitivity::CaseInsensitive))
  69. return "image/x-targa"sv;
  70. if (path.ends_with(".webp"sv, CaseSensitivity::CaseInsensitive))
  71. return "image/webp"sv;
  72. if (path.ends_with(".md"sv, CaseSensitivity::CaseInsensitive))
  73. return "text/markdown"sv;
  74. if (path.ends_with(".html"sv, CaseSensitivity::CaseInsensitive) || path.ends_with(".htm"sv, CaseSensitivity::CaseInsensitive))
  75. return "text/html"sv;
  76. if (path.ends_with(".css"sv, CaseSensitivity::CaseInsensitive))
  77. return "text/css"sv;
  78. if (path.ends_with(".icc"sv, CaseSensitivity::CaseInsensitive) || path.ends_with(".icm"sv, CaseSensitivity::CaseInsensitive))
  79. return "application/vnd.iccprofile"sv;
  80. if (path.ends_with(".js"sv, CaseSensitivity::CaseInsensitive))
  81. return "application/javascript"sv;
  82. if (path.ends_with(".json"sv, CaseSensitivity::CaseInsensitive))
  83. return "application/json"sv;
  84. if (path.ends_with(".zip"sv, CaseSensitivity::CaseInsensitive))
  85. return "application/zip"sv;
  86. if (path.ends_with(".md"sv, CaseSensitivity::CaseInsensitive))
  87. return "text/markdown"sv;
  88. if (path.ends_with("/"sv, CaseSensitivity::CaseInsensitive))
  89. return "text/html"sv;
  90. if (path.ends_with(".csv"sv, CaseSensitivity::CaseInsensitive))
  91. return "text/csv"sv;
  92. if (path.ends_with(".sheets"sv, CaseSensitivity::CaseInsensitive))
  93. return "application/x-sheets+json"sv;
  94. // FIXME: Share this, TextEditor and HackStudio language detection somehow.
  95. auto basename = LexicalPath::basename(path);
  96. if (path.ends_with(".cpp"sv, CaseSensitivity::CaseInsensitive)
  97. || path.ends_with(".c"sv, CaseSensitivity::CaseInsensitive)
  98. || path.ends_with(".hpp"sv, CaseSensitivity::CaseInsensitive)
  99. || path.ends_with(".h"sv, CaseSensitivity::CaseInsensitive)
  100. || path.ends_with(".gml"sv, CaseSensitivity::CaseInsensitive)
  101. || path.ends_with(".ini"sv, CaseSensitivity::CaseInsensitive)
  102. || path.ends_with(".ipc"sv, CaseSensitivity::CaseInsensitive)
  103. || path.ends_with(".txt"sv, CaseSensitivity::CaseInsensitive)
  104. || basename == "CMakeLists.txt"
  105. || basename == ".history"
  106. || basename == ".shellrc")
  107. return "text/plain"sv;
  108. return "application/octet-stream"sv;
  109. }
  110. #define ENUMERATE_HEADER_CONTENTS \
  111. __ENUMERATE_MIME_TYPE_HEADER(blend, "extra/blender", 0, 7, 'B', 'L', 'E', 'N', 'D', 'E', 'R') \
  112. __ENUMERATE_MIME_TYPE_HEADER(bmp, "image/bmp", 0, 2, 'B', 'M') \
  113. __ENUMERATE_MIME_TYPE_HEADER(bzip2, "application/x-bzip2", 0, 3, 'B', 'Z', 'h') \
  114. __ENUMERATE_MIME_TYPE_HEADER(compressed_iso, "extra/isz", 0, 4, 'I', 's', 'Z', '!') \
  115. __ENUMERATE_MIME_TYPE_HEADER(elf, "extra/elf", 0, 4, 0x7F, 'E', 'L', 'F') \
  116. __ENUMERATE_MIME_TYPE_HEADER(ext, "extra/ext", 0x438, 2, 0x53, 0xEF) \
  117. __ENUMERATE_MIME_TYPE_HEADER(flac, "audio/flac", 0, 4, 'f', 'L', 'a', 'C') \
  118. __ENUMERATE_MIME_TYPE_HEADER(gif_87, "image/gif", 0, 6, 'G', 'I', 'F', '8', '7', 'a') \
  119. __ENUMERATE_MIME_TYPE_HEADER(gif_89, "image/gif", 0, 6, 'G', 'I', 'F', '8', '9', 'a') \
  120. __ENUMERATE_MIME_TYPE_HEADER(gzip, "application/gzip", 0, 2, 0x1F, 0x8B) \
  121. __ENUMERATE_MIME_TYPE_HEADER(icc, "application/vnd.iccprofile", 36, 4, 'a', 'c', 's', 'p') \
  122. __ENUMERATE_MIME_TYPE_HEADER(iso9660_0, "extra/iso-9660", 0x8001, 5, 0x43, 0x44, 0x30, 0x30, 0x31) \
  123. __ENUMERATE_MIME_TYPE_HEADER(iso9660_1, "extra/iso-9660", 0x8801, 5, 0x43, 0x44, 0x30, 0x30, 0x31) \
  124. __ENUMERATE_MIME_TYPE_HEADER(iso9660_2, "extra/iso-9660", 0x9001, 5, 0x43, 0x44, 0x30, 0x30, 0x31) \
  125. __ENUMERATE_MIME_TYPE_HEADER(jpeg, "image/jpeg", 0, 4, 0xFF, 0xD8, 0xFF, 0xDB) \
  126. __ENUMERATE_MIME_TYPE_HEADER(jpeg_huh, "image/jpeg", 0, 4, 0xFF, 0xD8, 0xFF, 0xEE) \
  127. __ENUMERATE_MIME_TYPE_HEADER(jpeg_jfif, "image/jpeg", 0, 12, 0xFF, 0xD8, 0xFF, 0xE0, 0x00, 0x10, 'J', 'F', 'I', 'F', 0x00, 0x01) \
  128. __ENUMERATE_MIME_TYPE_HEADER(lua_bytecode, "extra/lua-bytecode", 0, 4, 0x1B, 'L', 'u', 'a') \
  129. __ENUMERATE_MIME_TYPE_HEADER(midi, "audio/midi", 0, 4, 0x4D, 0x54, 0x68, 0x64) \
  130. __ENUMERATE_MIME_TYPE_HEADER(mkv, "extra/matroska", 0, 4, 0x1A, 0x45, 0xDF, 0xA3) \
  131. __ENUMERATE_MIME_TYPE_HEADER(mp3, "audio/mpeg", 0, 2, 0xFF, 0xFB) \
  132. __ENUMERATE_MIME_TYPE_HEADER(nesrom, "extra/nes-rom", 0, 4, 'N', 'E', 'S', 0x1A) \
  133. __ENUMERATE_MIME_TYPE_HEADER(pbm, "image/x-portable-bitmap", 0, 3, 0x50, 0x31, 0x0A) \
  134. __ENUMERATE_MIME_TYPE_HEADER(pdf, "application/pdf", 0, 5, 0x25, 'P', 'D', 'F', 0x2D) \
  135. __ENUMERATE_MIME_TYPE_HEADER(pgm, "image/x-portable-graymap", 0, 3, 0x50, 0x32, 0x0A) \
  136. __ENUMERATE_MIME_TYPE_HEADER(png, "image/png", 0, 8, 0x89, 'P', 'N', 'G', 0x0D, 0x0A, 0x1A, 0x0A) \
  137. __ENUMERATE_MIME_TYPE_HEADER(ppm, "image/x-portable-pixmap", 0, 3, 0x50, 0x33, 0x0A) \
  138. __ENUMERATE_MIME_TYPE_HEADER(qcow, "extra/qcow", 0, 3, 'Q', 'F', 'I') \
  139. __ENUMERATE_MIME_TYPE_HEADER(qoa, "audio/qoa", 0, 4, 'q', 'o', 'a', 'f') \
  140. __ENUMERATE_MIME_TYPE_HEADER(qoi, "image/x-qoi", 0, 4, 'q', 'o', 'i', 'f') \
  141. __ENUMERATE_MIME_TYPE_HEADER(rtf, "application/rtf", 0, 6, 0x7B, 0x5C, 0x72, 0x74, 0x66, 0x31) \
  142. __ENUMERATE_MIME_TYPE_HEADER(sevenzip, "application/x-7z-compressed", 0, 6, 0x37, 0x7A, 0xBC, 0xAF, 0x27, 0x1C) \
  143. __ENUMERATE_MIME_TYPE_HEADER(shell, "text/x-shellscript", 0, 10, '#', '!', '/', 'b', 'i', 'n', '/', 's', 'h', '\n') \
  144. __ENUMERATE_MIME_TYPE_HEADER(sqlite, "extra/sqlite", 0, 16, 'S', 'Q', 'L', 'i', 't', 'e', ' ', 'f', 'o', 'r', 'm', 'a', 't', ' ', '3', 0x00) \
  145. __ENUMERATE_MIME_TYPE_HEADER(tar, "application/tar", 0x101, 5, 0x75, 0x73, 0x74, 0x61, 0x72) \
  146. __ENUMERATE_MIME_TYPE_HEADER(zip, "application/zip", 0, 2, 0x50, 0x4B) \
  147. __ENUMERATE_MIME_TYPE_HEADER(tiff, "image/tiff", 0, 4, 'I', 'I', '*', 0x00) \
  148. __ENUMERATE_MIME_TYPE_HEADER(tiff_bigendian, "image/tiff", 0, 4, 'M', 'M', 0x00, '*') \
  149. __ENUMERATE_MIME_TYPE_HEADER(wasm, "application/wasm", 0, 4, 0x00, 'a', 's', 'm') \
  150. __ENUMERATE_MIME_TYPE_HEADER(wav, "audio/wav", 8, 4, 'W', 'A', 'V', 'E') \
  151. __ENUMERATE_MIME_TYPE_HEADER(webp, "image/webp", 8, 4, 'W', 'E', 'B', 'P') \
  152. __ENUMERATE_MIME_TYPE_HEADER(win_31x_archive, "extra/win-31x-compressed", 0, 4, 'K', 'W', 'A', 'J') \
  153. __ENUMERATE_MIME_TYPE_HEADER(win_95_archive, "extra/win-95-compressed", 0, 4, 'S', 'Z', 'D', 'D') \
  154. __ENUMERATE_MIME_TYPE_HEADER(zlib_0, "extra/raw-zlib", 0, 2, 0x78, 0x01) \
  155. __ENUMERATE_MIME_TYPE_HEADER(zlib_1, "extra/raw-zlib", 0, 2, 0x78, 0x5E) \
  156. __ENUMERATE_MIME_TYPE_HEADER(zlib_2, "extra/raw-zlib", 0, 2, 0x78, 0x9C) \
  157. __ENUMERATE_MIME_TYPE_HEADER(zlib_3, "extra/raw-zlib", 0, 2, 0x78, 0xDA) \
  158. __ENUMERATE_MIME_TYPE_HEADER(zlib_4, "extra/raw-zlib", 0, 2, 0x78, 0x20) \
  159. __ENUMERATE_MIME_TYPE_HEADER(zlib_5, "extra/raw-zlib", 0, 2, 0x78, 0x7D) \
  160. __ENUMERATE_MIME_TYPE_HEADER(zlib_6, "extra/raw-zlib", 0, 2, 0x78, 0xBB) \
  161. __ENUMERATE_MIME_TYPE_HEADER(zlib_7, "extra/raw-zlib", 0, 2, 0x78, 0xF9)
  162. #define __ENUMERATE_MIME_TYPE_HEADER(var_name, mime_type, pattern_offset, pattern_size, ...) \
  163. static const u8 var_name##_arr[pattern_size] = { __VA_ARGS__ }; \
  164. static constexpr ReadonlyBytes var_name = ReadonlyBytes { var_name##_arr, pattern_size };
  165. ENUMERATE_HEADER_CONTENTS
  166. #undef __ENUMERATE_MIME_TYPE_HEADER
  167. Optional<DeprecatedString> guess_mime_type_based_on_sniffed_bytes(ReadonlyBytes bytes)
  168. {
  169. #define __ENUMERATE_MIME_TYPE_HEADER(var_name, mime_type, pattern_offset, pattern_size, ...) \
  170. if (static_cast<ssize_t>(bytes.size()) >= pattern_offset && bytes.slice(pattern_offset).starts_with(var_name)) \
  171. return mime_type;
  172. ENUMERATE_HEADER_CONTENTS;
  173. #undef __ENUMERATE_MIME_TYPE_HEADER
  174. return {};
  175. }
  176. Optional<DeprecatedString> guess_mime_type_based_on_sniffed_bytes(Core::File& file)
  177. {
  178. // Read accounts for longest possible offset + signature we currently match against (extra/iso-9660)
  179. auto maybe_buffer = ByteBuffer::create_uninitialized(0x9006);
  180. if (maybe_buffer.is_error())
  181. return {};
  182. auto maybe_bytes = file.read_some(maybe_buffer.value());
  183. if (maybe_bytes.is_error())
  184. return {};
  185. return Core::guess_mime_type_based_on_sniffed_bytes(maybe_bytes.value());
  186. }
  187. }