Fri, 23 Aug 2013 09:57:21 +0100
8022885: Update JAX-WS RI integration to 2.2.9-b14140
8013016: Rebase 8009009 against the latest jdk8/jaxws
Reviewed-by: alanb, chegar
ohair@286 | 1 | /* |
mkos@397 | 2 | * Copyright (c) 2009, Oracle and/or its affiliates. All rights reserved. |
ohair@286 | 3 | * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. |
ohair@286 | 4 | * |
ohair@286 | 5 | * This code is free software; you can redistribute it and/or modify it |
ohair@286 | 6 | * under the terms of the GNU General Public License version 2 only, as |
ohair@286 | 7 | * published by the Free Software Foundation. Oracle designates this |
ohair@286 | 8 | * particular file as subject to the "Classpath" exception as provided |
ohair@286 | 9 | * by Oracle in the LICENSE file that accompanied this code. |
ohair@286 | 10 | * |
ohair@286 | 11 | * This code is distributed in the hope that it will be useful, but WITHOUT |
ohair@286 | 12 | * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or |
ohair@286 | 13 | * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License |
ohair@286 | 14 | * version 2 for more details (a copy is included in the LICENSE file that |
ohair@286 | 15 | * accompanied this code). |
ohair@286 | 16 | * |
ohair@286 | 17 | * You should have received a copy of the GNU General Public License version |
ohair@286 | 18 | * 2 along with this work; if not, write to the Free Software Foundation, |
ohair@286 | 19 | * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. |
ohair@286 | 20 | * |
ohair@286 | 21 | * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA |
ohair@286 | 22 | * or visit www.oracle.com if you need additional information or have any |
ohair@286 | 23 | * questions. |
ohair@286 | 24 | */ |
ohair@286 | 25 | |
ohair@286 | 26 | package com.sun.xml.internal.dtdparser; |
ohair@286 | 27 | |
ohair@286 | 28 | import java.io.ByteArrayInputStream; |
ohair@286 | 29 | import java.io.CharConversionException; |
ohair@286 | 30 | import java.io.IOException; |
ohair@286 | 31 | import java.io.InputStream; |
ohair@286 | 32 | import java.io.InputStreamReader; |
ohair@286 | 33 | import java.io.PushbackInputStream; |
ohair@286 | 34 | import java.io.Reader; |
ohair@286 | 35 | import java.util.Hashtable; |
ohair@286 | 36 | |
ohair@286 | 37 | |
ohair@286 | 38 | // NOTE: Add I18N support to this class when JDK gets the ability to |
ohair@286 | 39 | // defer selection of locale for exception messages ... use the same |
ohair@286 | 40 | // technique for both. |
ohair@286 | 41 | |
ohair@286 | 42 | |
ohair@286 | 43 | /** |
ohair@286 | 44 | * This handles several XML-related tasks that normal java.io Readers |
ohair@286 | 45 | * don't support, inluding use of IETF standard encoding names and |
ohair@286 | 46 | * automatic detection of most XML encodings. The former is needed |
ohair@286 | 47 | * for interoperability; the latter is needed to conform with the XML |
ohair@286 | 48 | * spec. This class also optimizes reading some common encodings by |
ohair@286 | 49 | * providing low-overhead unsynchronized Reader support. |
ohair@286 | 50 | * <p/> |
ohair@286 | 51 | * <P> Note that the autodetection facility should be used only on |
ohair@286 | 52 | * data streams which have an unknown character encoding. For example, |
ohair@286 | 53 | * it should never be used on MIME text/xml entities. |
ohair@286 | 54 | * <p/> |
ohair@286 | 55 | * <P> Note that XML processors are only required to support UTF-8 and |
ohair@286 | 56 | * UTF-16 character encodings. Autodetection permits the underlying Java |
ohair@286 | 57 | * implementation to provide support for many other encodings, such as |
ohair@286 | 58 | * US-ASCII, ISO-8859-5, Shift_JIS, EUC-JP, and ISO-2022-JP. |
ohair@286 | 59 | * |
ohair@286 | 60 | * @author David Brownell |
ohair@286 | 61 | * @author Janet Koenig |
ohair@286 | 62 | * @version 1.3 00/02/24 |
ohair@286 | 63 | */ |
ohair@286 | 64 | // package private |
ohair@286 | 65 | final class XmlReader extends Reader { |
ohair@286 | 66 | private static final int MAXPUSHBACK = 512; |
ohair@286 | 67 | |
ohair@286 | 68 | private Reader in; |
ohair@286 | 69 | private String assignedEncoding; |
ohair@286 | 70 | private boolean closed; |
ohair@286 | 71 | |
ohair@286 | 72 | // |
ohair@286 | 73 | // This class always delegates I/O to a reader, which gets |
ohair@286 | 74 | // its data from the very beginning of the XML text. It needs |
ohair@286 | 75 | // to use a pushback stream since (a) autodetection can read |
ohair@286 | 76 | // partial UTF-8 characters which need to be fully processed, |
ohair@286 | 77 | // (b) the "Unicode" readers swallow characters that they think |
ohair@286 | 78 | // are byte order marks, so tests fail if they don't see the |
ohair@286 | 79 | // real byte order mark. |
ohair@286 | 80 | // |
ohair@286 | 81 | // It's got do this efficiently: character I/O is solidly on the |
ohair@286 | 82 | // critical path. (So keep buffer length over 2 Kbytes to avoid |
ohair@286 | 83 | // excess buffering. Many URL handlers stuff a BufferedInputStream |
ohair@286 | 84 | // between here and the real data source, and larger buffers keep |
ohair@286 | 85 | // that from slowing you down.) |
ohair@286 | 86 | // |
ohair@286 | 87 | |
ohair@286 | 88 | /** |
mkos@397 | 89 | * Constructs the reader from an input stream, autodetecting |
ohair@286 | 90 | * the encoding to use according to the heuristic specified |
ohair@286 | 91 | * in the XML 1.0 recommendation. |
ohair@286 | 92 | * |
ohair@286 | 93 | * @param in the input stream from which the reader is constructed |
ohair@286 | 94 | * @throws IOException on error, such as unrecognized encoding |
ohair@286 | 95 | */ |
ohair@286 | 96 | public static Reader createReader(InputStream in) throws IOException { |
ohair@286 | 97 | return new XmlReader(in); |
ohair@286 | 98 | } |
ohair@286 | 99 | |
ohair@286 | 100 | /** |
ohair@286 | 101 | * Creates a reader supporting the given encoding, mapping |
ohair@286 | 102 | * from standard encoding names to ones that understood by |
ohair@286 | 103 | * Java where necessary. |
ohair@286 | 104 | * |
ohair@286 | 105 | * @param in the input stream from which the reader is constructed |
ohair@286 | 106 | * @param encoding the IETF standard name of the encoding to use; |
mkos@397 | 107 | * if null, autodetection is used. |
ohair@286 | 108 | * @throws IOException on error, including unrecognized encoding |
ohair@286 | 109 | */ |
ohair@286 | 110 | public static Reader createReader(InputStream in, String encoding) |
ohair@286 | 111 | throws IOException { |
ohair@286 | 112 | if (encoding == null) |
ohair@286 | 113 | return new XmlReader(in); |
ohair@286 | 114 | if ("UTF-8".equalsIgnoreCase(encoding) |
ohair@286 | 115 | || "UTF8".equalsIgnoreCase(encoding)) |
ohair@286 | 116 | return new Utf8Reader(in); |
ohair@286 | 117 | if ("US-ASCII".equalsIgnoreCase(encoding) |
ohair@286 | 118 | || "ASCII".equalsIgnoreCase(encoding)) |
ohair@286 | 119 | return new AsciiReader(in); |
ohair@286 | 120 | if ("ISO-8859-1".equalsIgnoreCase(encoding) |
ohair@286 | 121 | // plus numerous aliases ... |
ohair@286 | 122 | ) |
ohair@286 | 123 | return new Iso8859_1Reader(in); |
ohair@286 | 124 | |
ohair@286 | 125 | // |
ohair@286 | 126 | // What we really want is an administerable resource mapping |
ohair@286 | 127 | // encoding names/aliases to classnames. For example a property |
ohair@286 | 128 | // file resource, "readers/mapping.props", holding and a set |
ohair@286 | 129 | // of readers in that (sub)package... defaulting to this call |
ohair@286 | 130 | // only if no better choice is available. |
ohair@286 | 131 | // |
ohair@286 | 132 | return new InputStreamReader(in, std2java(encoding)); |
ohair@286 | 133 | } |
ohair@286 | 134 | |
ohair@286 | 135 | // |
ohair@286 | 136 | // JDK doesn't know all of the standard encoding names, and |
ohair@286 | 137 | // in particular none of the EBCDIC ones IANA defines (and |
ohair@286 | 138 | // which IBM encourages). |
ohair@286 | 139 | // |
ohair@286 | 140 | static private final Hashtable charsets = new Hashtable(31); |
ohair@286 | 141 | |
ohair@286 | 142 | static { |
ohair@286 | 143 | charsets.put("UTF-16", "Unicode"); |
ohair@286 | 144 | charsets.put("ISO-10646-UCS-2", "Unicode"); |
ohair@286 | 145 | |
ohair@286 | 146 | // NOTE: no support for ISO-10646-UCS-4 yet. |
ohair@286 | 147 | |
ohair@286 | 148 | charsets.put("EBCDIC-CP-US", "cp037"); |
ohair@286 | 149 | charsets.put("EBCDIC-CP-CA", "cp037"); |
ohair@286 | 150 | charsets.put("EBCDIC-CP-NL", "cp037"); |
ohair@286 | 151 | charsets.put("EBCDIC-CP-WT", "cp037"); |
ohair@286 | 152 | |
ohair@286 | 153 | charsets.put("EBCDIC-CP-DK", "cp277"); |
ohair@286 | 154 | charsets.put("EBCDIC-CP-NO", "cp277"); |
ohair@286 | 155 | charsets.put("EBCDIC-CP-FI", "cp278"); |
ohair@286 | 156 | charsets.put("EBCDIC-CP-SE", "cp278"); |
ohair@286 | 157 | |
ohair@286 | 158 | charsets.put("EBCDIC-CP-IT", "cp280"); |
ohair@286 | 159 | charsets.put("EBCDIC-CP-ES", "cp284"); |
ohair@286 | 160 | charsets.put("EBCDIC-CP-GB", "cp285"); |
ohair@286 | 161 | charsets.put("EBCDIC-CP-FR", "cp297"); |
ohair@286 | 162 | |
ohair@286 | 163 | charsets.put("EBCDIC-CP-AR1", "cp420"); |
ohair@286 | 164 | charsets.put("EBCDIC-CP-HE", "cp424"); |
ohair@286 | 165 | charsets.put("EBCDIC-CP-BE", "cp500"); |
ohair@286 | 166 | charsets.put("EBCDIC-CP-CH", "cp500"); |
ohair@286 | 167 | |
ohair@286 | 168 | charsets.put("EBCDIC-CP-ROECE", "cp870"); |
ohair@286 | 169 | charsets.put("EBCDIC-CP-YU", "cp870"); |
ohair@286 | 170 | charsets.put("EBCDIC-CP-IS", "cp871"); |
ohair@286 | 171 | charsets.put("EBCDIC-CP-AR2", "cp918"); |
ohair@286 | 172 | |
ohair@286 | 173 | // IANA also defines two that JDK 1.2 doesn't handle: |
ohair@286 | 174 | // EBCDIC-CP-GR --> CP423 |
ohair@286 | 175 | // EBCDIC-CP-TR --> CP905 |
ohair@286 | 176 | } |
ohair@286 | 177 | |
ohair@286 | 178 | // returns an encoding name supported by JDK >= 1.1.6 |
ohair@286 | 179 | // for some cases required by the XML spec |
ohair@286 | 180 | private static String std2java(String encoding) { |
ohair@286 | 181 | String temp = encoding.toUpperCase(); |
ohair@286 | 182 | temp = (String) charsets.get(temp); |
ohair@286 | 183 | return temp != null ? temp : encoding; |
ohair@286 | 184 | } |
ohair@286 | 185 | |
ohair@286 | 186 | /** |
ohair@286 | 187 | * Returns the standard name of the encoding in use |
ohair@286 | 188 | */ |
ohair@286 | 189 | public String getEncoding() { |
ohair@286 | 190 | return assignedEncoding; |
ohair@286 | 191 | } |
ohair@286 | 192 | |
ohair@286 | 193 | private XmlReader(InputStream stream) throws IOException { |
ohair@286 | 194 | super(stream); |
ohair@286 | 195 | |
ohair@286 | 196 | PushbackInputStream pb; |
ohair@286 | 197 | byte buf []; |
ohair@286 | 198 | int len; |
ohair@286 | 199 | |
ohair@286 | 200 | if (stream instanceof PushbackInputStream) |
ohair@286 | 201 | pb = (PushbackInputStream) stream; |
ohair@286 | 202 | else |
ohair@286 | 203 | pb = new PushbackInputStream(stream, MAXPUSHBACK); |
ohair@286 | 204 | |
ohair@286 | 205 | // |
ohair@286 | 206 | // See if we can figure out the character encoding used |
ohair@286 | 207 | // in this file by peeking at the first few bytes. |
ohair@286 | 208 | // |
ohair@286 | 209 | buf = new byte[4]; |
ohair@286 | 210 | len = pb.read(buf); |
ohair@286 | 211 | if (len > 0) |
ohair@286 | 212 | pb.unread(buf, 0, len); |
ohair@286 | 213 | |
ohair@286 | 214 | if (len == 4) |
ohair@286 | 215 | switch (buf[0] & 0x0ff) { |
ohair@286 | 216 | case 0: |
ohair@286 | 217 | // 00 3c 00 3f == illegal UTF-16 big-endian |
ohair@286 | 218 | if (buf[1] == 0x3c && buf[2] == 0x00 && buf[3] == 0x3f) { |
ohair@286 | 219 | setEncoding(pb, "UnicodeBig"); |
ohair@286 | 220 | return; |
ohair@286 | 221 | } |
ohair@286 | 222 | // else it's probably UCS-4 |
ohair@286 | 223 | break; |
ohair@286 | 224 | |
ohair@286 | 225 | case '<': // 0x3c: the most common cases! |
ohair@286 | 226 | switch (buf[1] & 0x0ff) { |
ohair@286 | 227 | // First character is '<'; could be XML without |
ohair@286 | 228 | // an XML directive such as "<hello>", "<!-- ...", |
ohair@286 | 229 | // and so on. |
ohair@286 | 230 | default: |
ohair@286 | 231 | break; |
ohair@286 | 232 | |
ohair@286 | 233 | // 3c 00 3f 00 == illegal UTF-16 little endian |
ohair@286 | 234 | case 0x00: |
ohair@286 | 235 | if (buf[2] == 0x3f && buf[3] == 0x00) { |
ohair@286 | 236 | setEncoding(pb, "UnicodeLittle"); |
ohair@286 | 237 | return; |
ohair@286 | 238 | } |
ohair@286 | 239 | // else probably UCS-4 |
ohair@286 | 240 | break; |
ohair@286 | 241 | |
ohair@286 | 242 | // 3c 3f 78 6d == ASCII and supersets '<?xm' |
ohair@286 | 243 | case '?': |
ohair@286 | 244 | if (buf[2] != 'x' || buf[3] != 'm') |
ohair@286 | 245 | break; |
ohair@286 | 246 | // |
ohair@286 | 247 | // One of several encodings could be used: |
ohair@286 | 248 | // Shift-JIS, ASCII, UTF-8, ISO-8859-*, etc |
ohair@286 | 249 | // |
ohair@286 | 250 | useEncodingDecl(pb, "UTF8"); |
ohair@286 | 251 | return; |
ohair@286 | 252 | } |
ohair@286 | 253 | break; |
ohair@286 | 254 | |
ohair@286 | 255 | // 4c 6f a7 94 ... some EBCDIC code page |
ohair@286 | 256 | case 0x4c: |
ohair@286 | 257 | if (buf[1] == 0x6f |
ohair@286 | 258 | && (0x0ff & buf[2]) == 0x0a7 |
ohair@286 | 259 | && (0x0ff & buf[3]) == 0x094) { |
ohair@286 | 260 | useEncodingDecl(pb, "CP037"); |
ohair@286 | 261 | return; |
ohair@286 | 262 | } |
ohair@286 | 263 | // whoops, treat as UTF-8 |
ohair@286 | 264 | break; |
ohair@286 | 265 | |
ohair@286 | 266 | // UTF-16 big-endian |
ohair@286 | 267 | case 0xfe: |
ohair@286 | 268 | if ((buf[1] & 0x0ff) != 0xff) |
ohair@286 | 269 | break; |
ohair@286 | 270 | setEncoding(pb, "UTF-16"); |
ohair@286 | 271 | return; |
ohair@286 | 272 | |
ohair@286 | 273 | // UTF-16 little-endian |
ohair@286 | 274 | case 0xff: |
ohair@286 | 275 | if ((buf[1] & 0x0ff) != 0xfe) |
ohair@286 | 276 | break; |
ohair@286 | 277 | setEncoding(pb, "UTF-16"); |
ohair@286 | 278 | return; |
ohair@286 | 279 | |
ohair@286 | 280 | // default ... no XML declaration |
ohair@286 | 281 | default: |
ohair@286 | 282 | break; |
ohair@286 | 283 | } |
ohair@286 | 284 | |
ohair@286 | 285 | // |
ohair@286 | 286 | // If all else fails, assume XML without a declaration, and |
ohair@286 | 287 | // using UTF-8 encoding. |
ohair@286 | 288 | // |
ohair@286 | 289 | setEncoding(pb, "UTF-8"); |
ohair@286 | 290 | } |
ohair@286 | 291 | |
ohair@286 | 292 | /* |
ohair@286 | 293 | * Read the encoding decl on the stream, knowing that it should |
ohair@286 | 294 | * be readable using the specified encoding (basically, ASCII or |
ohair@286 | 295 | * EBCDIC). The body of the document may use a wider range of |
ohair@286 | 296 | * characters than the XML/Text decl itself, so we switch to use |
ohair@286 | 297 | * the specified encoding as soon as we can. (ASCII is a subset |
ohair@286 | 298 | * of UTF-8, ISO-8859-*, ISO-2022-JP, EUC-JP, and more; EBCDIC |
ohair@286 | 299 | * has a variety of "code pages" that have these characters as |
ohair@286 | 300 | * a common subset.) |
ohair@286 | 301 | */ |
ohair@286 | 302 | private void useEncodingDecl(PushbackInputStream pb, String encoding) |
ohair@286 | 303 | throws IOException { |
ohair@286 | 304 | byte buffer [] = new byte[MAXPUSHBACK]; |
ohair@286 | 305 | int len; |
ohair@286 | 306 | Reader r; |
ohair@286 | 307 | int c; |
ohair@286 | 308 | |
ohair@286 | 309 | // |
ohair@286 | 310 | // Buffer up a bunch of input, and set up to read it in |
ohair@286 | 311 | // the specified encoding ... we can skip the first four |
ohair@286 | 312 | // bytes since we know that "<?xm" was read to determine |
ohair@286 | 313 | // what encoding to use! |
ohair@286 | 314 | // |
ohair@286 | 315 | len = pb.read(buffer, 0, buffer.length); |
ohair@286 | 316 | pb.unread(buffer, 0, len); |
ohair@286 | 317 | r = new InputStreamReader(new ByteArrayInputStream(buffer, 4, len), |
ohair@286 | 318 | encoding); |
ohair@286 | 319 | |
ohair@286 | 320 | // |
ohair@286 | 321 | // Next must be "l" (and whitespace) else we conclude |
ohair@286 | 322 | // error and choose UTF-8. |
ohair@286 | 323 | // |
ohair@286 | 324 | if ((c = r.read()) != 'l') { |
ohair@286 | 325 | setEncoding(pb, "UTF-8"); |
ohair@286 | 326 | return; |
ohair@286 | 327 | } |
ohair@286 | 328 | |
ohair@286 | 329 | // |
ohair@286 | 330 | // Then, we'll skip any |
ohair@286 | 331 | // S version="..." [or single quotes] |
ohair@286 | 332 | // bit and get any subsequent |
ohair@286 | 333 | // S encoding="..." [or single quotes] |
ohair@286 | 334 | // |
ohair@286 | 335 | // We put an arbitrary size limit on how far we read; lots |
ohair@286 | 336 | // of space will break this algorithm. |
ohair@286 | 337 | // |
ohair@286 | 338 | StringBuffer buf = new StringBuffer(); |
ohair@286 | 339 | StringBuffer keyBuf = null; |
ohair@286 | 340 | String key = null; |
ohair@286 | 341 | boolean sawEq = false; |
ohair@286 | 342 | char quoteChar = 0; |
ohair@286 | 343 | boolean sawQuestion = false; |
ohair@286 | 344 | |
ohair@286 | 345 | XmlDecl: |
ohair@286 | 346 | for (int i = 0; i < MAXPUSHBACK - 5; ++i) { |
ohair@286 | 347 | if ((c = r.read()) == -1) |
ohair@286 | 348 | break; |
ohair@286 | 349 | |
ohair@286 | 350 | // ignore whitespace before/between "key = 'value'" |
ohair@286 | 351 | if (c == ' ' || c == '\t' || c == '\n' || c == '\r') |
ohair@286 | 352 | continue; |
ohair@286 | 353 | |
ohair@286 | 354 | // ... but require at least a little! |
ohair@286 | 355 | if (i == 0) |
ohair@286 | 356 | break; |
ohair@286 | 357 | |
ohair@286 | 358 | // terminate the loop ASAP |
ohair@286 | 359 | if (c == '?') |
ohair@286 | 360 | sawQuestion = true; |
ohair@286 | 361 | else if (sawQuestion) { |
ohair@286 | 362 | if (c == '>') |
ohair@286 | 363 | break; |
ohair@286 | 364 | sawQuestion = false; |
ohair@286 | 365 | } |
ohair@286 | 366 | |
ohair@286 | 367 | // did we get the "key =" bit yet? |
ohair@286 | 368 | if (key == null || !sawEq) { |
ohair@286 | 369 | if (keyBuf == null) { |
ohair@286 | 370 | if (Character.isWhitespace((char) c)) |
ohair@286 | 371 | continue; |
ohair@286 | 372 | keyBuf = buf; |
ohair@286 | 373 | buf.setLength(0); |
ohair@286 | 374 | buf.append((char) c); |
ohair@286 | 375 | sawEq = false; |
ohair@286 | 376 | } else if (Character.isWhitespace((char) c)) { |
ohair@286 | 377 | key = keyBuf.toString(); |
ohair@286 | 378 | } else if (c == '=') { |
ohair@286 | 379 | if (key == null) |
ohair@286 | 380 | key = keyBuf.toString(); |
ohair@286 | 381 | sawEq = true; |
ohair@286 | 382 | keyBuf = null; |
ohair@286 | 383 | quoteChar = 0; |
ohair@286 | 384 | } else |
ohair@286 | 385 | keyBuf.append((char) c); |
ohair@286 | 386 | continue; |
ohair@286 | 387 | } |
ohair@286 | 388 | |
ohair@286 | 389 | // space before quoted value |
ohair@286 | 390 | if (Character.isWhitespace((char) c)) |
ohair@286 | 391 | continue; |
ohair@286 | 392 | if (c == '"' || c == '\'') { |
ohair@286 | 393 | if (quoteChar == 0) { |
ohair@286 | 394 | quoteChar = (char) c; |
ohair@286 | 395 | buf.setLength(0); |
ohair@286 | 396 | continue; |
ohair@286 | 397 | } else if (c == quoteChar) { |
ohair@286 | 398 | if ("encoding".equals(key)) { |
ohair@286 | 399 | assignedEncoding = buf.toString(); |
ohair@286 | 400 | |
ohair@286 | 401 | // [81] Encname ::= [A-Za-z] ([A-Za-z0-9._]|'-')* |
ohair@286 | 402 | for (i = 0; i < assignedEncoding.length(); i++) { |
ohair@286 | 403 | c = assignedEncoding.charAt(i); |
ohair@286 | 404 | if ((c >= 'A' && c <= 'Z') |
ohair@286 | 405 | || (c >= 'a' && c <= 'z')) |
ohair@286 | 406 | continue; |
ohair@286 | 407 | if (i == 0) |
ohair@286 | 408 | break XmlDecl; |
ohair@286 | 409 | if (i > 0 && (c == '-' |
ohair@286 | 410 | || (c >= '0' && c <= '9') |
ohair@286 | 411 | || c == '.' || c == '_')) |
ohair@286 | 412 | continue; |
ohair@286 | 413 | // map illegal names to UTF-8 default |
ohair@286 | 414 | break XmlDecl; |
ohair@286 | 415 | } |
ohair@286 | 416 | |
ohair@286 | 417 | setEncoding(pb, assignedEncoding); |
ohair@286 | 418 | return; |
ohair@286 | 419 | |
ohair@286 | 420 | } else { |
ohair@286 | 421 | key = null; |
ohair@286 | 422 | continue; |
ohair@286 | 423 | } |
ohair@286 | 424 | } |
ohair@286 | 425 | } |
ohair@286 | 426 | buf.append((char) c); |
ohair@286 | 427 | } |
ohair@286 | 428 | |
ohair@286 | 429 | setEncoding(pb, "UTF-8"); |
ohair@286 | 430 | } |
ohair@286 | 431 | |
ohair@286 | 432 | private void setEncoding(InputStream stream, String encoding) |
ohair@286 | 433 | throws IOException { |
ohair@286 | 434 | assignedEncoding = encoding; |
ohair@286 | 435 | in = createReader(stream, encoding); |
ohair@286 | 436 | } |
ohair@286 | 437 | |
ohair@286 | 438 | /** |
ohair@286 | 439 | * Reads the number of characters read into the buffer, or -1 on EOF. |
ohair@286 | 440 | */ |
ohair@286 | 441 | public int read(char buf [], int off, int len) throws IOException { |
ohair@286 | 442 | int val; |
ohair@286 | 443 | |
ohair@286 | 444 | if (closed) |
ohair@286 | 445 | return -1; // throw new IOException ("closed"); |
ohair@286 | 446 | val = in.read(buf, off, len); |
ohair@286 | 447 | if (val == -1) |
ohair@286 | 448 | close(); |
ohair@286 | 449 | return val; |
ohair@286 | 450 | } |
ohair@286 | 451 | |
ohair@286 | 452 | /** |
ohair@286 | 453 | * Reads a single character. |
ohair@286 | 454 | */ |
ohair@286 | 455 | public int read() throws IOException { |
ohair@286 | 456 | int val; |
ohair@286 | 457 | |
ohair@286 | 458 | if (closed) |
ohair@286 | 459 | throw new IOException("closed"); |
ohair@286 | 460 | val = in.read(); |
ohair@286 | 461 | if (val == -1) |
ohair@286 | 462 | close(); |
ohair@286 | 463 | return val; |
ohair@286 | 464 | } |
ohair@286 | 465 | |
ohair@286 | 466 | /** |
ohair@286 | 467 | * Returns true iff the reader supports mark/reset. |
ohair@286 | 468 | */ |
ohair@286 | 469 | public boolean markSupported() { |
ohair@286 | 470 | return in == null ? false : in.markSupported(); |
ohair@286 | 471 | } |
ohair@286 | 472 | |
ohair@286 | 473 | /** |
ohair@286 | 474 | * Sets a mark allowing a limited number of characters to |
ohair@286 | 475 | * be "peeked", by reading and then resetting. |
ohair@286 | 476 | * |
ohair@286 | 477 | * @param value how many characters may be "peeked". |
ohair@286 | 478 | */ |
ohair@286 | 479 | public void mark(int value) throws IOException { |
ohair@286 | 480 | if (in != null) in.mark(value); |
ohair@286 | 481 | } |
ohair@286 | 482 | |
ohair@286 | 483 | /** |
ohair@286 | 484 | * Resets the current position to the last marked position. |
ohair@286 | 485 | */ |
ohair@286 | 486 | public void reset() throws IOException { |
ohair@286 | 487 | if (in != null) in.reset(); |
ohair@286 | 488 | } |
ohair@286 | 489 | |
ohair@286 | 490 | /** |
ohair@286 | 491 | * Skips a specified number of characters. |
ohair@286 | 492 | */ |
ohair@286 | 493 | public long skip(long value) throws IOException { |
ohair@286 | 494 | return in == null ? 0 : in.skip(value); |
ohair@286 | 495 | } |
ohair@286 | 496 | |
ohair@286 | 497 | /** |
ohair@286 | 498 | * Returns true iff input characters are known to be ready. |
ohair@286 | 499 | */ |
ohair@286 | 500 | public boolean ready() throws IOException { |
ohair@286 | 501 | return in == null ? false : in.ready(); |
ohair@286 | 502 | } |
ohair@286 | 503 | |
ohair@286 | 504 | /** |
ohair@286 | 505 | * Closes the reader. |
ohair@286 | 506 | */ |
ohair@286 | 507 | public void close() throws IOException { |
ohair@286 | 508 | if (closed) |
ohair@286 | 509 | return; |
ohair@286 | 510 | in.close(); |
ohair@286 | 511 | in = null; |
ohair@286 | 512 | closed = true; |
ohair@286 | 513 | } |
ohair@286 | 514 | |
ohair@286 | 515 | // |
ohair@286 | 516 | // Delegating to a converter module will always be slower than |
ohair@286 | 517 | // direct conversion. Use a similar approach for any other |
ohair@286 | 518 | // readers that need to be particularly fast; only block I/O |
ohair@286 | 519 | // speed matters to this package. For UTF-16, separate readers |
ohair@286 | 520 | // for big and little endian streams make a difference, too; |
ohair@286 | 521 | // fewer conditionals in the critical path! |
ohair@286 | 522 | // |
ohair@286 | 523 | static abstract class BaseReader extends Reader { |
ohair@286 | 524 | protected InputStream instream; |
ohair@286 | 525 | protected byte buffer []; |
ohair@286 | 526 | protected int start, finish; |
ohair@286 | 527 | |
ohair@286 | 528 | BaseReader(InputStream stream) { |
ohair@286 | 529 | super(stream); |
ohair@286 | 530 | |
ohair@286 | 531 | instream = stream; |
ohair@286 | 532 | buffer = new byte[8192]; |
ohair@286 | 533 | } |
ohair@286 | 534 | |
ohair@286 | 535 | public boolean ready() throws IOException { |
ohair@286 | 536 | return instream == null |
ohair@286 | 537 | || (finish - start) > 0 |
ohair@286 | 538 | || instream.available() != 0; |
ohair@286 | 539 | } |
ohair@286 | 540 | |
ohair@286 | 541 | // caller shouldn't read again |
ohair@286 | 542 | public void close() throws IOException { |
ohair@286 | 543 | if (instream != null) { |
ohair@286 | 544 | instream.close(); |
ohair@286 | 545 | start = finish = 0; |
ohair@286 | 546 | buffer = null; |
ohair@286 | 547 | instream = null; |
ohair@286 | 548 | } |
ohair@286 | 549 | } |
ohair@286 | 550 | } |
ohair@286 | 551 | |
ohair@286 | 552 | // |
ohair@286 | 553 | // We want this reader, to make the default encoding be as fast |
ohair@286 | 554 | // as we can make it. JDK's "UTF8" (not "UTF-8" till JDK 1.2) |
ohair@286 | 555 | // InputStreamReader works, but 20+% slower speed isn't OK for |
ohair@286 | 556 | // the default/primary encoding. |
ohair@286 | 557 | // |
ohair@286 | 558 | static final class Utf8Reader extends BaseReader { |
ohair@286 | 559 | // 2nd half of UTF-8 surrogate pair |
ohair@286 | 560 | private char nextChar; |
ohair@286 | 561 | |
ohair@286 | 562 | Utf8Reader(InputStream stream) { |
ohair@286 | 563 | super(stream); |
ohair@286 | 564 | } |
ohair@286 | 565 | |
ohair@286 | 566 | public int read(char buf [], int offset, int len) throws IOException { |
ohair@286 | 567 | int i = 0, c = 0; |
ohair@286 | 568 | |
ohair@286 | 569 | if (len <= 0) |
ohair@286 | 570 | return 0; |
ohair@286 | 571 | |
ohair@286 | 572 | // Consume remaining half of any surrogate pair immediately |
ohair@286 | 573 | if (nextChar != 0) { |
ohair@286 | 574 | buf[offset + i++] = nextChar; |
ohair@286 | 575 | nextChar = 0; |
ohair@286 | 576 | } |
ohair@286 | 577 | |
ohair@286 | 578 | while (i < len) { |
ohair@286 | 579 | // stop or read data if needed |
ohair@286 | 580 | if (finish <= start) { |
ohair@286 | 581 | if (instream == null) { |
ohair@286 | 582 | c = -1; |
ohair@286 | 583 | break; |
ohair@286 | 584 | } |
ohair@286 | 585 | start = 0; |
ohair@286 | 586 | finish = instream.read(buffer, 0, buffer.length); |
ohair@286 | 587 | if (finish <= 0) { |
ohair@286 | 588 | this.close(); |
ohair@286 | 589 | c = -1; |
ohair@286 | 590 | break; |
ohair@286 | 591 | } |
ohair@286 | 592 | } |
ohair@286 | 593 | |
ohair@286 | 594 | // |
ohair@286 | 595 | // RFC 2279 describes UTF-8; there are six encodings. |
ohair@286 | 596 | // Each encoding takes a fixed number of characters |
ohair@286 | 597 | // (1-6 bytes) and is flagged by a bit pattern in the |
ohair@286 | 598 | // first byte. The five and six byte-per-character |
ohair@286 | 599 | // encodings address characters which are disallowed |
ohair@286 | 600 | // in XML documents, as do some four byte ones. |
ohair@286 | 601 | // |
ohair@286 | 602 | |
ohair@286 | 603 | // |
ohair@286 | 604 | // Single byte == ASCII. Common; optimize. |
ohair@286 | 605 | // |
ohair@286 | 606 | c = buffer[start] & 0x0ff; |
ohair@286 | 607 | if ((c & 0x80) == 0x00) { |
ohair@286 | 608 | // 0x0000 <= c <= 0x007f |
ohair@286 | 609 | start++; |
ohair@286 | 610 | buf[offset + i++] = (char) c; |
ohair@286 | 611 | continue; |
ohair@286 | 612 | } |
ohair@286 | 613 | |
ohair@286 | 614 | // |
ohair@286 | 615 | // Multibyte chars -- check offsets optimistically, |
ohair@286 | 616 | // ditto the "10xx xxxx" format for subsequent bytes |
ohair@286 | 617 | // |
ohair@286 | 618 | int off = start; |
ohair@286 | 619 | |
ohair@286 | 620 | try { |
ohair@286 | 621 | // 2 bytes |
ohair@286 | 622 | if ((buffer[off] & 0x0E0) == 0x0C0) { |
ohair@286 | 623 | c = (buffer[off++] & 0x1f) << 6; |
ohair@286 | 624 | c += buffer[off++] & 0x3f; |
ohair@286 | 625 | |
ohair@286 | 626 | // 0x0080 <= c <= 0x07ff |
ohair@286 | 627 | |
ohair@286 | 628 | // 3 bytes |
ohair@286 | 629 | } else if ((buffer[off] & 0x0F0) == 0x0E0) { |
ohair@286 | 630 | c = (buffer[off++] & 0x0f) << 12; |
ohair@286 | 631 | c += (buffer[off++] & 0x3f) << 6; |
ohair@286 | 632 | c += buffer[off++] & 0x3f; |
ohair@286 | 633 | |
ohair@286 | 634 | // 0x0800 <= c <= 0xffff |
ohair@286 | 635 | |
ohair@286 | 636 | // 4 bytes |
ohair@286 | 637 | } else if ((buffer[off] & 0x0f8) == 0x0F0) { |
ohair@286 | 638 | c = (buffer[off++] & 0x07) << 18; |
ohair@286 | 639 | c += (buffer[off++] & 0x3f) << 12; |
ohair@286 | 640 | c += (buffer[off++] & 0x3f) << 6; |
ohair@286 | 641 | c += buffer[off++] & 0x3f; |
ohair@286 | 642 | |
ohair@286 | 643 | // 0x0001 0000 <= c <= 0x001f ffff |
ohair@286 | 644 | |
ohair@286 | 645 | // Unicode supports c <= 0x0010 ffff ... |
ohair@286 | 646 | if (c > 0x0010ffff) |
ohair@286 | 647 | throw new CharConversionException("UTF-8 encoding of character 0x00" |
ohair@286 | 648 | + Integer.toHexString(c) |
ohair@286 | 649 | + " can't be converted to Unicode."); |
ohair@286 | 650 | |
ohair@286 | 651 | // Convert UCS-4 char to surrogate pair (UTF-16) |
ohair@286 | 652 | c -= 0x10000; |
ohair@286 | 653 | nextChar = (char) (0xDC00 + (c & 0x03ff)); |
ohair@286 | 654 | c = 0xD800 + (c >> 10); |
ohair@286 | 655 | |
ohair@286 | 656 | // 5 and 6 byte versions are XML WF errors, but |
ohair@286 | 657 | // typically come from mislabeled encodings |
ohair@286 | 658 | } else |
ohair@286 | 659 | throw new CharConversionException("Unconvertible UTF-8 character" |
ohair@286 | 660 | + " beginning with 0x" |
ohair@286 | 661 | + Integer.toHexString(buffer[start] & 0xff)); |
ohair@286 | 662 | |
ohair@286 | 663 | } catch (ArrayIndexOutOfBoundsException e) { |
ohair@286 | 664 | // off > length && length >= buffer.length |
ohair@286 | 665 | c = 0; |
ohair@286 | 666 | } |
ohair@286 | 667 | |
ohair@286 | 668 | // |
ohair@286 | 669 | // if the buffer held only a partial character, |
ohair@286 | 670 | // compact it and try to read the rest of the |
ohair@286 | 671 | // character. worst case involves three |
ohair@286 | 672 | // single-byte reads -- quite rare. |
ohair@286 | 673 | // |
ohair@286 | 674 | if (off > finish) { |
ohair@286 | 675 | System.arraycopy(buffer, start, |
ohair@286 | 676 | buffer, 0, finish - start); |
ohair@286 | 677 | finish -= start; |
ohair@286 | 678 | start = 0; |
ohair@286 | 679 | off = instream.read(buffer, finish, |
ohair@286 | 680 | buffer.length - finish); |
ohair@286 | 681 | if (off < 0) { |
ohair@286 | 682 | this.close(); |
ohair@286 | 683 | throw new CharConversionException("Partial UTF-8 char"); |
ohair@286 | 684 | } |
ohair@286 | 685 | finish += off; |
ohair@286 | 686 | continue; |
ohair@286 | 687 | } |
ohair@286 | 688 | |
ohair@286 | 689 | // |
ohair@286 | 690 | // check the format of the non-initial bytes |
ohair@286 | 691 | // |
ohair@286 | 692 | for (start++; start < off; start++) { |
ohair@286 | 693 | if ((buffer[start] & 0xC0) != 0x80) { |
ohair@286 | 694 | this.close(); |
ohair@286 | 695 | throw new CharConversionException("Malformed UTF-8 char -- " |
ohair@286 | 696 | + "is an XML encoding declaration missing?"); |
ohair@286 | 697 | } |
ohair@286 | 698 | } |
ohair@286 | 699 | |
ohair@286 | 700 | // |
ohair@286 | 701 | // If this needed a surrogate pair, consume ASAP |
ohair@286 | 702 | // |
ohair@286 | 703 | buf[offset + i++] = (char) c; |
ohair@286 | 704 | if (nextChar != 0 && i < len) { |
ohair@286 | 705 | buf[offset + i++] = nextChar; |
ohair@286 | 706 | nextChar = 0; |
ohair@286 | 707 | } |
ohair@286 | 708 | } |
ohair@286 | 709 | if (i > 0) |
ohair@286 | 710 | return i; |
ohair@286 | 711 | return (c == -1) ? -1 : 0; |
ohair@286 | 712 | } |
ohair@286 | 713 | } |
ohair@286 | 714 | |
ohair@286 | 715 | // |
ohair@286 | 716 | // We want ASCII and ISO-8859 Readers since they're the most common |
ohair@286 | 717 | // encodings in the US and Europe, and we don't want performance |
ohair@286 | 718 | // regressions for them. They're also easy to implement efficiently, |
ohair@286 | 719 | // since they're bitmask subsets of UNICODE. |
ohair@286 | 720 | // |
ohair@286 | 721 | // XXX haven't benchmarked these readers vs what we get out of JDK. |
ohair@286 | 722 | // |
ohair@286 | 723 | static final class AsciiReader extends BaseReader { |
ohair@286 | 724 | AsciiReader(InputStream in) { |
ohair@286 | 725 | super(in); |
ohair@286 | 726 | } |
ohair@286 | 727 | |
ohair@286 | 728 | public int read(char buf [], int offset, int len) throws IOException { |
ohair@286 | 729 | int i, c; |
ohair@286 | 730 | |
ohair@286 | 731 | if (instream == null) |
ohair@286 | 732 | return -1; |
ohair@286 | 733 | |
ohair@286 | 734 | for (i = 0; i < len; i++) { |
ohair@286 | 735 | if (start >= finish) { |
ohair@286 | 736 | start = 0; |
ohair@286 | 737 | finish = instream.read(buffer, 0, buffer.length); |
ohair@286 | 738 | if (finish <= 0) { |
ohair@286 | 739 | if (finish <= 0) |
ohair@286 | 740 | this.close(); |
ohair@286 | 741 | break; |
ohair@286 | 742 | } |
ohair@286 | 743 | } |
ohair@286 | 744 | c = buffer[start++]; |
ohair@286 | 745 | if ((c & 0x80) != 0) |
ohair@286 | 746 | throw new CharConversionException("Illegal ASCII character, 0x" |
ohair@286 | 747 | + Integer.toHexString(c & 0xff)); |
ohair@286 | 748 | buf[offset + i] = (char) c; |
ohair@286 | 749 | } |
ohair@286 | 750 | if (i == 0 && finish <= 0) |
ohair@286 | 751 | return -1; |
ohair@286 | 752 | return i; |
ohair@286 | 753 | } |
ohair@286 | 754 | } |
ohair@286 | 755 | |
ohair@286 | 756 | static final class Iso8859_1Reader extends BaseReader { |
ohair@286 | 757 | Iso8859_1Reader(InputStream in) { |
ohair@286 | 758 | super(in); |
ohair@286 | 759 | } |
ohair@286 | 760 | |
ohair@286 | 761 | public int read(char buf [], int offset, int len) throws IOException { |
ohair@286 | 762 | int i; |
ohair@286 | 763 | |
ohair@286 | 764 | if (instream == null) |
ohair@286 | 765 | return -1; |
ohair@286 | 766 | |
ohair@286 | 767 | for (i = 0; i < len; i++) { |
ohair@286 | 768 | if (start >= finish) { |
ohair@286 | 769 | start = 0; |
ohair@286 | 770 | finish = instream.read(buffer, 0, buffer.length); |
ohair@286 | 771 | if (finish <= 0) { |
ohair@286 | 772 | if (finish <= 0) |
ohair@286 | 773 | this.close(); |
ohair@286 | 774 | break; |
ohair@286 | 775 | } |
ohair@286 | 776 | } |
ohair@286 | 777 | buf[offset + i] = (char) (0x0ff & buffer[start++]); |
ohair@286 | 778 | } |
ohair@286 | 779 | if (i == 0 && finish <= 0) |
ohair@286 | 780 | return -1; |
ohair@286 | 781 | return i; |
ohair@286 | 782 | } |
ohair@286 | 783 | } |
ohair@286 | 784 | } |