OLD | NEW |
1 // Copyright (c) 2012, the Dart project authors. Please see the AUTHORS file | 1 // Copyright (c) 2012, the Dart project authors. Please see the AUTHORS file |
2 // for details. All rights reserved. Use of this source code is governed by a | 2 // for details. All rights reserved. Use of this source code is governed by a |
3 // BSD-style license that can be found in the LICENSE file. | 3 // BSD-style license that can be found in the LICENSE file. |
4 | 4 |
5 // Interface for decoders decoding binary data into string data. The | 5 // Interface for decoders decoding binary data into string data. The |
6 // decoder keeps track of line breaks during decoding. | 6 // decoder keeps track of line breaks during decoding. |
7 interface _StringDecoder { | 7 interface _StringDecoder { |
8 // Add more binary data to be decoded. The ownership of the buffer | 8 // Add more binary data to be decoded. The ownership of the buffer |
9 // is transfered to the decoder and the caller most not modify it any more. | 9 // is transfered to the decoder and the caller most not modify it any more. |
10 int write(List<int> buffer); | 10 int write(List<int> buffer); |
(...skipping 10 matching lines...) Expand all Loading... |
21 String get decoded(); | 21 String get decoded(); |
22 | 22 |
23 // Get the string data decoded since the last call to [decode] or | 23 // Get the string data decoded since the last call to [decode] or |
24 // [decodeLine] up to the next line break present. Returns null if | 24 // [decodeLine] up to the next line break present. Returns null if |
25 // no line break is present. The line break character sequence is | 25 // no line break is present. The line break character sequence is |
26 // discarded. | 26 // discarded. |
27 String get decodedLine(); | 27 String get decodedLine(); |
28 } | 28 } |
29 | 29 |
30 | 30 |
| 31 class _StringDecoders { |
| 32 static _StringDecoder decoder(Encoding encoding) { |
| 33 if (encoding == Encoding.UTF_8) { |
| 34 return new _UTF8Decoder(); |
| 35 } else if (encoding == Encoding.ISO_8859_1) { |
| 36 return new _Latin1Decoder(); |
| 37 } else if (encoding == Encoding.ASCII) { |
| 38 return new _AsciiDecoder(); |
| 39 } else { |
| 40 throw new StreamException("Unsupported encoding ${encoding.name}"); |
| 41 } |
| 42 } |
| 43 } |
| 44 |
| 45 |
31 class DecoderException implements Exception { | 46 class DecoderException implements Exception { |
32 const DecoderException([String this.message]); | 47 const DecoderException([String this.message]); |
33 String toString() => "DecoderException: $message"; | 48 String toString() => "DecoderException: $message"; |
34 final String message; | 49 final String message; |
35 } | 50 } |
36 | 51 |
37 | 52 |
38 // Utility class for decoding UTF-8 from data delivered as a stream of | 53 // Utility class for decoding UTF-8 from data delivered as a stream of |
39 // bytes. | 54 // bytes. |
40 class _StringDecoderBase implements _StringDecoder { | 55 class _StringDecoderBase implements _StringDecoder { |
(...skipping 94 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
135 Queue<int> _lineBreakEnds; // Character position of known line breaks. | 150 Queue<int> _lineBreakEnds; // Character position of known line breaks. |
136 int _charOffset = 0; // Character number of the first character in the list. | 151 int _charOffset = 0; // Character number of the first character in the list. |
137 int _charCount = 0; // Total number of characters decoded. | 152 int _charCount = 0; // Total number of characters decoded. |
138 int _lastCharCode = -1; | 153 int _lastCharCode = -1; |
139 | 154 |
140 final int LF = 10; | 155 final int LF = 10; |
141 final int CR = 13; | 156 final int CR = 13; |
142 } | 157 } |
143 | 158 |
144 | 159 |
| 160 // Utility class for decoding UTF-8 from data delivered as a stream of |
| 161 // bytes. |
| 162 class _UTF8Decoder extends _StringDecoderBase { |
| 163 // Process the next UTF-8 encoded character. |
| 164 bool _processNext() { |
| 165 // Peek the next byte to calculate the number of bytes required for |
| 166 // the next character. |
| 167 int value = _bufferList.peek() & 0xFF; |
| 168 if ((value & 0x80) == 0x80) { |
| 169 int additionalBytes; |
| 170 if ((value & 0xe0) == 0xc0) { // 110xxxxx |
| 171 value = value & 0x1F; |
| 172 additionalBytes = 1; |
| 173 } else if ((value & 0xf0) == 0xe0) { // 1110xxxx |
| 174 value = value & 0x0F; |
| 175 additionalBytes = 2; |
| 176 } else { // 11110xxx |
| 177 value = value & 0x07; |
| 178 additionalBytes = 3; |
| 179 } |
| 180 // Check if there are enough bytes to decode the character. Otherwise |
| 181 // return false. |
| 182 if (_bufferList.length < additionalBytes + 1) { |
| 183 return false; |
| 184 } |
| 185 // Remove the value peeked from the buffer list. |
| 186 _bufferList.next(); |
| 187 for (int i = 0; i < additionalBytes; i++) { |
| 188 int byte = _bufferList.next(); |
| 189 value = value << 6 | (byte & 0x3F); |
| 190 } |
| 191 } else { |
| 192 // Remove the value peeked from the buffer list. |
| 193 _bufferList.next(); |
| 194 } |
| 195 addChar(value); |
| 196 return true; |
| 197 } |
| 198 } |
| 199 |
| 200 |
145 // Utility class for decoding ascii data delivered as a stream of | 201 // Utility class for decoding ascii data delivered as a stream of |
146 // bytes. | 202 // bytes. |
147 class _AsciiDecoder extends _StringDecoderBase { | 203 class _AsciiDecoder extends _StringDecoderBase { |
148 // Process the next ascii encoded character. | 204 // Process the next ascii encoded character. |
149 bool _processNext() { | 205 bool _processNext() { |
150 while (_bufferList.length > 0) { | 206 while (_bufferList.length > 0) { |
151 int byte = _bufferList.next(); | 207 int byte = _bufferList.next(); |
152 if (byte > 127) { | 208 if (byte > 127) { |
153 throw new DecoderException("Illegal ASCII character $byte"); | 209 throw new DecoderException("Illegal ASCII character $byte"); |
154 } | 210 } |
(...skipping 11 matching lines...) Expand all Loading... |
166 bool _processNext() { | 222 bool _processNext() { |
167 while (_bufferList.length > 0) { | 223 while (_bufferList.length > 0) { |
168 int byte = _bufferList.next(); | 224 int byte = _bufferList.next(); |
169 addChar(byte); | 225 addChar(byte); |
170 } | 226 } |
171 return true; | 227 return true; |
172 } | 228 } |
173 } | 229 } |
174 | 230 |
175 | 231 |
176 // Utility class for decoding UTF-8 from data delivered as a stream of | 232 // Interface for encoders encoding string data into binary data. |
177 // bytes. | 233 interface _StringEncoder { |
178 class _UTF8Decoder extends _StringDecoderBase { | 234 List<int> encodeString(String string); |
179 // Process the next UTF-8 encoded character. | 235 } |
180 bool _processNext() { | 236 |
181 // Peek the next byte to calculate the number of bytes required for | 237 |
182 // the next character. | 238 // Utility class for encoding a string into UTF-8 byte stream. |
183 int value = _bufferList.peek() & 0xFF; | 239 class _UTF8Encoder implements _StringEncoder { |
184 if ((value & 0x80) == 0x80) { | 240 List<int> encodeString(String string) { |
| 241 int size = _encodingSize(string); |
| 242 ByteArray result = new ByteArray(size); |
| 243 _encodeString(string, result); |
| 244 return result; |
| 245 } |
| 246 |
| 247 static int _encodingSize(String string) => _encodeString(string, null); |
| 248 |
| 249 static int _encodeString(String string, List<int> buffer) { |
| 250 int pos = 0; |
| 251 int length = string.length; |
| 252 for (int i = 0; i < length; i++) { |
185 int additionalBytes; | 253 int additionalBytes; |
186 if ((value & 0xe0) == 0xc0) { // 110xxxxx | 254 int charCode = string.charCodeAt(i); |
187 value = value & 0x1F; | 255 if (charCode <= 0x007F) { |
| 256 additionalBytes = 0; |
| 257 if (buffer != null) buffer[pos] = charCode; |
| 258 } else if (charCode <= 0x07FF) { |
| 259 // 110xxxxx (xxxxx is top 5 bits). |
| 260 if (buffer != null) buffer[pos] = ((charCode >> 6) & 0x1F) | 0xC0; |
188 additionalBytes = 1; | 261 additionalBytes = 1; |
189 } else if ((value & 0xf0) == 0xe0) { // 1110xxxx | 262 } else if (charCode <= 0xFFFF) { |
190 value = value & 0x0F; | 263 // 1110xxxx (xxxx is top 4 bits) |
| 264 if (buffer != null) buffer[pos] = ((charCode >> 12) & 0x0F)| 0xE0; |
191 additionalBytes = 2; | 265 additionalBytes = 2; |
192 } else { // 11110xxx | 266 } else { |
193 value = value & 0x07; | 267 // 11110xxx (xxx is top 3 bits) |
| 268 if (buffer != null) buffer[pos] = ((charCode >> 18) & 0x07) | 0xF0; |
194 additionalBytes = 3; | 269 additionalBytes = 3; |
195 } | 270 } |
196 // Check if there are enough bytes to decode the character. Otherwise | 271 pos++; |
197 // return false. | 272 if (buffer != null) { |
198 if (_bufferList.length < additionalBytes + 1) { | 273 for (int i = additionalBytes; i > 0; i--) { |
199 return false; | 274 // 10xxxxxx (xxxxxx is next 6 bits from the top). |
| 275 buffer[pos++] = ((charCode >> (6 * (i - 1))) & 0x3F) | 0x80; |
| 276 } |
| 277 } else { |
| 278 pos += additionalBytes; |
200 } | 279 } |
201 // Remove the value peeked from the buffer list. | |
202 _bufferList.next(); | |
203 for (int i = 0; i < additionalBytes; i++) { | |
204 int byte = _bufferList.next(); | |
205 value = value << 6 | (byte & 0x3F); | |
206 } | |
207 } else { | |
208 // Remove the value peeked from the buffer list. | |
209 _bufferList.next(); | |
210 } | 280 } |
211 addChar(value); | 281 return pos; |
212 return true; | |
213 } | 282 } |
214 } | 283 } |
215 | 284 |
216 | 285 |
| 286 // Utility class for encoding a string into a Latin1 byte stream. |
| 287 class _Latin1Encoder implements _StringEncoder { |
| 288 List<int> encodeString(String string) { |
| 289 ByteArray result = new ByteArray(string.length); |
| 290 for (int i = 0; i < string.length; i++) { |
| 291 int charCode = string.charCodeAt(i); |
| 292 if (charCode > 255) { |
| 293 throw new EncoderException( |
| 294 "No ISO_8859_1 encoding for code point $charCode"); |
| 295 } |
| 296 result[i] = charCode; |
| 297 } |
| 298 return result; |
| 299 } |
| 300 } |
| 301 |
| 302 |
| 303 // Utility class for encoding a string into an ASCII byte stream. |
| 304 class _AsciiEncoder implements _StringEncoder { |
| 305 List<int> encodeString(String string) { |
| 306 ByteArray result = new ByteArray(string.length); |
| 307 for (int i = 0; i < string.length; i++) { |
| 308 int charCode = string.charCodeAt(i); |
| 309 if (charCode > 127) { |
| 310 throw new EncoderException( |
| 311 "No ASCII encoding for code point $charCode"); |
| 312 } |
| 313 result[i] = charCode; |
| 314 } |
| 315 return result; |
| 316 } |
| 317 } |
| 318 |
| 319 |
| 320 class _StringEncoders { |
| 321 static _StringEncoder encoder(Encoding encoding) { |
| 322 if (encoding == Encoding.UTF_8) { |
| 323 return new _UTF8Encoder(); |
| 324 } else if (encoding == Encoding.ISO_8859_1) { |
| 325 return new _Latin1Encoder(); |
| 326 } else if (encoding == Encoding.ASCII) { |
| 327 return new _AsciiEncoder(); |
| 328 } else { |
| 329 throw new StreamException("Unsupported encoding ${encoding.name}"); |
| 330 } |
| 331 } |
| 332 } |
| 333 |
| 334 |
| 335 class EncoderException implements Exception { |
| 336 const EncoderException([String this.message]); |
| 337 String toString() => "EncoderException: $message"; |
| 338 final String message; |
| 339 } |
| 340 |
| 341 |
217 class _StringInputStream implements StringInputStream { | 342 class _StringInputStream implements StringInputStream { |
218 _StringInputStream(InputStream this._input, [String encoding]) | 343 _StringInputStream(InputStream this._input, |
| 344 [Encoding encoding = Encoding.UTF_8]) |
219 : _encoding = encoding { | 345 : _encoding = encoding { |
220 if (_encoding === null) { | 346 _decoder = _StringDecoders.decoder(encoding); |
221 _encoding = "UTF-8"; | |
222 } | |
223 if (_encoding == "UTF-8") { | |
224 _decoder = new _UTF8Decoder(); | |
225 } else if (_encoding == "ISO-8859-1") { | |
226 _decoder = new _Latin1Decoder(); | |
227 } else if (_encoding == "ASCII") { | |
228 _decoder = new _AsciiDecoder(); | |
229 } else { | |
230 throw new StreamException("Unsupported encoding $_encoding"); | |
231 } | |
232 _input.onData = _onData; | 347 _input.onData = _onData; |
233 _input.onClosed = _onClosed; | 348 _input.onClosed = _onClosed; |
234 } | 349 } |
235 | 350 |
236 String read() { | 351 String read() { |
237 String result = _decoder.decoded; | 352 String result = _decoder.decoded; |
238 _checkInstallDataHandler(); | 353 _checkInstallDataHandler(); |
239 return result; | 354 return result; |
240 } | 355 } |
241 | 356 |
242 String readLine() { | 357 String readLine() { |
243 String decodedLine = _decoder.decodedLine; | 358 String decodedLine = _decoder.decodedLine; |
244 if (decodedLine == null) { | 359 if (decodedLine == null) { |
245 if (_inputClosed) { | 360 if (_inputClosed) { |
246 // Last line might not have a line separator. | 361 // Last line might not have a line separator. |
247 decodedLine = _decoder.decoded; | 362 decodedLine = _decoder.decoded; |
248 if (decodedLine != null && | 363 if (decodedLine != null && |
249 decodedLine[decodedLine.length - 1] == '\r') { | 364 decodedLine[decodedLine.length - 1] == '\r') { |
250 decodedLine = decodedLine.substring(0, decodedLine.length - 1); | 365 decodedLine = decodedLine.substring(0, decodedLine.length - 1); |
251 } | 366 } |
252 } | 367 } |
253 } | 368 } |
254 _checkInstallDataHandler(); | 369 _checkInstallDataHandler(); |
255 return decodedLine; | 370 return decodedLine; |
256 } | 371 } |
257 | 372 |
258 int available() => _decoder.available(); | 373 int available() => _decoder.available(); |
259 | 374 |
260 String get encoding() => _encoding; | 375 Encoding get encoding() => _encoding; |
261 | 376 |
262 bool get closed() => _inputClosed && _decoder.isEmpty(); | 377 bool get closed() => _inputClosed && _decoder.isEmpty(); |
263 | 378 |
264 void set onData(void callback()) { | 379 void set onData(void callback()) { |
265 _clientDataHandler = callback; | 380 _clientDataHandler = callback; |
266 _clientLineHandler = null; | 381 _clientLineHandler = null; |
267 _checkInstallDataHandler(); | 382 _checkInstallDataHandler(); |
268 _checkScheduleCallback(); | 383 _checkScheduleCallback(); |
269 } | 384 } |
270 | 385 |
(...skipping 112 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
383 // Schedule close callback if no more data and input is closed. | 498 // Schedule close callback if no more data and input is closed. |
384 if (_decoder.isEmpty() && | 499 if (_decoder.isEmpty() && |
385 _inputClosed && | 500 _inputClosed && |
386 _scheduledCloseCallback == null) { | 501 _scheduledCloseCallback == null) { |
387 _scheduledCloseCallback = new Timer(0, issueCloseCallback); | 502 _scheduledCloseCallback = new Timer(0, issueCloseCallback); |
388 } | 503 } |
389 } | 504 } |
390 } | 505 } |
391 | 506 |
392 InputStream _input; | 507 InputStream _input; |
393 String _encoding; | 508 Encoding _encoding; |
394 _StringDecoder _decoder; | 509 _StringDecoder _decoder; |
395 bool _inputClosed = false; // Is the underlying input stream closed? | 510 bool _inputClosed = false; // Is the underlying input stream closed? |
396 bool _closed = false; // Is this stream closed. | 511 bool _closed = false; // Is this stream closed. |
397 bool _eof = false; // Has all data been read from the decoder? | 512 bool _eof = false; // Has all data been read from the decoder? |
398 Timer _scheduledDataCallback; | 513 Timer _scheduledDataCallback; |
399 Timer _scheduledLineCallback; | 514 Timer _scheduledLineCallback; |
400 Timer _scheduledCloseCallback; | 515 Timer _scheduledCloseCallback; |
401 Function _clientDataHandler; | 516 Function _clientDataHandler; |
402 Function _clientLineHandler; | 517 Function _clientLineHandler; |
403 Function _clientCloseHandler; | 518 Function _clientCloseHandler; |
404 } | 519 } |
OLD | NEW |