|
| 1 | +// Copyright (c) .NET Foundation. All rights reserved. |
| 2 | +// Licensed under the Apache License, Version 2.0. See License.txt in the project root for license information. |
| 3 | + |
| 4 | +using Microsoft.AspNet.Server.Kestrel.Infrastructure; |
| 5 | + |
| 6 | +namespace Microsoft.AspNet.Server.Kestrel.Http |
| 7 | +{ |
| 8 | + public class UrlPathDecoder |
| 9 | + { |
| 10 | + /// <summary> |
| 11 | + /// Unescapes the string between given memory iterators in place. |
| 12 | + /// </summary> |
| 13 | + /// <param name="start">The iterator points to the beginning of the sequence.</param> |
| 14 | + /// <param name="end">The iterator points to the byte behind the end of the sequence.</param> |
| 15 | + /// <returns>The iterator points to the byte behind the end of the processed sequence.</returns> |
| 16 | + public static MemoryPoolIterator2 Unescape(MemoryPoolIterator2 start, MemoryPoolIterator2 end) |
| 17 | + { |
| 18 | + // the slot to read the input |
| 19 | + var reader = start; |
| 20 | + |
| 21 | + // the slot to write the unescaped byte |
| 22 | + var writer = reader; |
| 23 | + |
| 24 | + while (true) |
| 25 | + { |
| 26 | + if (CompareIterators(ref reader, ref end)) |
| 27 | + { |
| 28 | + return writer; |
| 29 | + } |
| 30 | + |
| 31 | + if (reader.Peek() == '%') |
| 32 | + { |
| 33 | + var decodeReader = reader; |
| 34 | + |
| 35 | + // If decoding process successes, the writer iterator will be moved |
| 36 | + // to the next write-ready location. On the other hand if the scanned |
| 37 | + // percent-encodings cannot be interpreted as sequence of UTF-8 octets, |
| 38 | + // these bytes should be copied to output as is. |
| 39 | + // The decodeReader iterator is always moved to the first byte not yet |
| 40 | + // be scanned after the process. A failed decoding means the chars |
| 41 | + // between the reader and decodeReader can be copied to output untouched. |
| 42 | + if (!DecodeCore(ref decodeReader, ref writer, end)) |
| 43 | + { |
| 44 | + Copy(reader, decodeReader, ref writer); |
| 45 | + } |
| 46 | + |
| 47 | + reader = decodeReader; |
| 48 | + } |
| 49 | + else |
| 50 | + { |
| 51 | + writer.Put((byte)reader.Take()); |
| 52 | + } |
| 53 | + } |
| 54 | + } |
| 55 | + |
| 56 | + /// <summary> |
| 57 | + /// Unescape the percent-encodings |
| 58 | + /// </summary> |
| 59 | + /// <param name="reader">The iterator point to the first % char</param> |
| 60 | + /// <param name="writer">The place to write to</param> |
| 61 | + /// <param name="end">The end of the sequence</param> |
| 62 | + private static bool DecodeCore(ref MemoryPoolIterator2 reader, ref MemoryPoolIterator2 writer, MemoryPoolIterator2 end) |
| 63 | + { |
| 64 | + // preserves the original head. if the percent-encodings cannot be interpreted as sequence of UTF-8 octets, |
| 65 | + // bytes from this till the last scanned one will be copied to the memory pointed by writer. |
| 66 | + var byte1 = UnescapePercentEncoding(ref reader, end); |
| 67 | + if (byte1 == -1) |
| 68 | + { |
| 69 | + return false; |
| 70 | + } |
| 71 | + |
| 72 | + if (byte1 <= 0x7F) |
| 73 | + { |
| 74 | + // first byte < U+007f, it is a single byte ASCII |
| 75 | + writer.Put((byte)byte1); |
| 76 | + return true; |
| 77 | + } |
| 78 | + |
| 79 | + int byte2 = 0, byte3 = 0, byte4 = 0; |
| 80 | + |
| 81 | + // anticipate more bytes |
| 82 | + var currentDecodeBits = 0; |
| 83 | + var byteCount = 1; |
| 84 | + var expectValueMin = 0; |
| 85 | + if ((byte1 & 0xE0) == 0xC0) |
| 86 | + { |
| 87 | + // 110x xxxx, expect one more byte |
| 88 | + currentDecodeBits = byte1 & 0x1F; |
| 89 | + byteCount = 2; |
| 90 | + expectValueMin = 0x80; |
| 91 | + } |
| 92 | + else if ((byte1 & 0xF0) == 0xE0) |
| 93 | + { |
| 94 | + // 1110 xxxx, expect two more bytes |
| 95 | + currentDecodeBits = byte1 & 0x0F; |
| 96 | + byteCount = 3; |
| 97 | + expectValueMin = 0x800; |
| 98 | + } |
| 99 | + else if ((byte1 & 0xF8) == 0xF0) |
| 100 | + { |
| 101 | + // 1111 0xxx, expect three more bytes |
| 102 | + currentDecodeBits = byte1 & 0x07; |
| 103 | + byteCount = 4; |
| 104 | + expectValueMin = 0x10000; |
| 105 | + } |
| 106 | + else |
| 107 | + { |
| 108 | + // invalid first byte |
| 109 | + return false; |
| 110 | + } |
| 111 | + |
| 112 | + var remainingBytes = byteCount - 1; |
| 113 | + while (remainingBytes > 0) |
| 114 | + { |
| 115 | + // read following three chars |
| 116 | + if (CompareIterators(ref reader, ref end)) |
| 117 | + { |
| 118 | + return false; |
| 119 | + } |
| 120 | + |
| 121 | + var nextByte = UnescapePercentEncoding(ref reader, end); |
| 122 | + if (nextByte == -1) |
| 123 | + { |
| 124 | + return false; |
| 125 | + } |
| 126 | + |
| 127 | + if ((nextByte & 0xC0) != 0x80) |
| 128 | + { |
| 129 | + // the follow up byte is not in form of 10xx xxxx |
| 130 | + return false; |
| 131 | + } |
| 132 | + |
| 133 | + currentDecodeBits = (currentDecodeBits << 6) | (nextByte & 0x3F); |
| 134 | + remainingBytes--; |
| 135 | + |
| 136 | + if (remainingBytes == 1 && currentDecodeBits >= 0x360 && currentDecodeBits <= 0x37F) |
| 137 | + { |
| 138 | + // this is going to end up in the range of 0xD800-0xDFFF UTF-16 surrogates that |
| 139 | + // are not allowed in UTF-8; |
| 140 | + return false; |
| 141 | + } |
| 142 | + |
| 143 | + if (remainingBytes == 2 && currentDecodeBits >= 0x110) |
| 144 | + { |
| 145 | + // this is going to be out of the upper Unicode bound 0x10FFFF. |
| 146 | + return false; |
| 147 | + } |
| 148 | + |
| 149 | + if (byteCount - remainingBytes == 2) |
| 150 | + { |
| 151 | + byte2 = nextByte; |
| 152 | + } |
| 153 | + else if (byteCount - remainingBytes == 3) |
| 154 | + { |
| 155 | + byte3 = nextByte; |
| 156 | + } |
| 157 | + else if (byteCount - remainingBytes == 4) |
| 158 | + { |
| 159 | + byte4 = nextByte; |
| 160 | + } |
| 161 | + } |
| 162 | + |
| 163 | + if (currentDecodeBits < expectValueMin) |
| 164 | + { |
| 165 | + // overlong encoding (e.g. using 2 bytes to encode something that only needed 1). |
| 166 | + return false; |
| 167 | + } |
| 168 | + |
| 169 | + // all bytes are verified, write to the output |
| 170 | + if (byteCount > 0) |
| 171 | + { |
| 172 | + writer.Put((byte)byte1); |
| 173 | + } |
| 174 | + if (byteCount > 1) |
| 175 | + { |
| 176 | + writer.Put((byte)byte2); |
| 177 | + } |
| 178 | + if (byteCount > 2) |
| 179 | + { |
| 180 | + writer.Put((byte)byte3); |
| 181 | + } |
| 182 | + if (byteCount > 3) |
| 183 | + { |
| 184 | + writer.Put((byte)byte4); |
| 185 | + } |
| 186 | + |
| 187 | + return true; |
| 188 | + } |
| 189 | + |
| 190 | + private static void Copy(MemoryPoolIterator2 head, MemoryPoolIterator2 tail, ref MemoryPoolIterator2 writer) |
| 191 | + { |
| 192 | + while (!CompareIterators(ref head, ref tail)) |
| 193 | + { |
| 194 | + writer.Put((byte)head.Take()); |
| 195 | + } |
| 196 | + } |
| 197 | + |
| 198 | + /// <summary> |
| 199 | + /// Read the percent-encoding and try unescape it. |
| 200 | + /// |
| 201 | + /// The operation first peek at the character the <paramref name="scan"/> |
| 202 | + /// iterator points at. If it is % the <paramref name="scan"/> is then |
| 203 | + /// moved on to scan the following to characters. If the two following |
| 204 | + /// characters are hexadecimal literals they will be unescaped and the |
| 205 | + /// value will be returned. |
| 206 | + /// |
| 207 | + /// If the first character is not % the <paramref name="scan"/> iterator |
| 208 | + /// will be removed beyond the location of % and -1 will be returned. |
| 209 | + /// |
| 210 | + /// If the following two characters can't be successfully unescaped the |
| 211 | + /// <paramref name="scan"/> iterator will be move behind the % and -1 |
| 212 | + /// will be returned. |
| 213 | + /// </summary> |
| 214 | + /// <param name="scan">The value to read</param> |
| 215 | + /// <param name="end">The end of the sequence</param> |
| 216 | + /// <returns>The unescaped byte if success. Otherwise return -1.</returns> |
| 217 | + private static int UnescapePercentEncoding(ref MemoryPoolIterator2 scan, MemoryPoolIterator2 end) |
| 218 | + { |
| 219 | + if (scan.Take() != '%') |
| 220 | + { |
| 221 | + return -1; |
| 222 | + } |
| 223 | + |
| 224 | + var probe = scan; |
| 225 | + |
| 226 | + int value1 = ReadHex(ref probe, end); |
| 227 | + if (value1 == -1) |
| 228 | + { |
| 229 | + return -1; |
| 230 | + } |
| 231 | + |
| 232 | + int value2 = ReadHex(ref probe, end); |
| 233 | + if (value2 == -1) |
| 234 | + { |
| 235 | + return -1; |
| 236 | + } |
| 237 | + |
| 238 | + if (SkipUnescape(value1, value2)) |
| 239 | + { |
| 240 | + return -1; |
| 241 | + } |
| 242 | + |
| 243 | + scan = probe; |
| 244 | + return (value1 << 4) + value2; |
| 245 | + } |
| 246 | + |
| 247 | + /// <summary> |
| 248 | + /// Read the next char and convert it into hexadecimal value. |
| 249 | + /// |
| 250 | + /// The <paramref name="scan"/> iterator will be moved to the next |
| 251 | + /// byte no matter no matter whether the operation successes. |
| 252 | + /// </summary> |
| 253 | + /// <param name="scan">The value to read</param> |
| 254 | + /// <param name="end">The end of the sequence</param> |
| 255 | + /// <returns>The hexadecimal value if successes, otherwise -1.</returns> |
| 256 | + private static int ReadHex(ref MemoryPoolIterator2 scan, MemoryPoolIterator2 end) |
| 257 | + { |
| 258 | + if (CompareIterators(ref scan, ref end)) |
| 259 | + { |
| 260 | + return -1; |
| 261 | + } |
| 262 | + |
| 263 | + var value = scan.Take(); |
| 264 | + var isHead = (((value >= '0') && (value <= '9')) || |
| 265 | + ((value >= 'A') && (value <= 'F')) || |
| 266 | + ((value >= 'a') && (value <= 'f'))); |
| 267 | + |
| 268 | + if (!isHead) |
| 269 | + { |
| 270 | + return -1; |
| 271 | + } |
| 272 | + |
| 273 | + if (value <= '9') |
| 274 | + { |
| 275 | + return value - '0'; |
| 276 | + } |
| 277 | + else if (value <= 'F') |
| 278 | + { |
| 279 | + return (value - 'A') + 10; |
| 280 | + } |
| 281 | + else // a - f |
| 282 | + { |
| 283 | + return (value - 'a') + 10; |
| 284 | + } |
| 285 | + } |
| 286 | + |
| 287 | + private static bool SkipUnescape(int value1, int value2) |
| 288 | + { |
| 289 | + // skip %2F |
| 290 | + if (value1 == 2 && value2 == 15) |
| 291 | + { |
| 292 | + return true; |
| 293 | + } |
| 294 | + |
| 295 | + return false; |
| 296 | + } |
| 297 | + |
| 298 | + private static bool CompareIterators(ref MemoryPoolIterator2 lhs, ref MemoryPoolIterator2 rhs) |
| 299 | + { |
| 300 | + // uses ref parameter to save cost of copying |
| 301 | + return (lhs.Block == rhs.Block) && (lhs.Index == rhs.Index); |
| 302 | + } |
| 303 | + } |
| 304 | +} |
0 commit comments