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