static boolean check(CharsetDecoder dec, byte[] bytes, boolean direct, int[] flow) { int inPos = flow[0]; int inLen = flow[1]; int outPos = flow[2]; int outLen = flow[3]; int expedInPos = flow[4]; int expedOutPos = flow[5]; CoderResult expedCR = (flow[6] == 0) ? CoderResult.UNDERFLOW : CoderResult.OVERFLOW; ByteBuffer bbf; CharBuffer cbf; if (direct) { bbf = ByteBuffer.allocateDirect(inPos + bytes.length); cbf = ByteBuffer.allocateDirect((outPos + outLen) * 2).asCharBuffer(); } else { bbf = ByteBuffer.allocate(inPos + bytes.length); cbf = CharBuffer.allocate(outPos + outLen); } bbf.position(inPos); bbf.put(bytes).flip().position(inPos).limit(inPos + inLen); cbf.position(outPos); dec.reset(); CoderResult cr = dec.decode(bbf, cbf, false); if (cr != expedCR || bbf.position() != expedInPos || cbf.position() != expedOutPos) { System.out.printf("Expected(direct=%5b): [", direct); for (int i : flow) System.out.print(" " + i); System.out.println( "] CR=" + cr + ", inPos=" + bbf.position() + ", outPos=" + cbf.position()); return false; } return true; }
static char[] decode(byte[] bb, Charset cs, boolean testDirect, Time t) throws Exception { String csn = cs.name(); CharsetDecoder dec = cs.newDecoder(); ByteBuffer bbf; CharBuffer cbf; if (testDirect) { bbf = ByteBuffer.allocateDirect(bb.length); cbf = ByteBuffer.allocateDirect(bb.length * 2).asCharBuffer(); bbf.put(bb); } else { bbf = ByteBuffer.wrap(bb); cbf = CharBuffer.allocate(bb.length); } CoderResult cr = null; long t1 = System.nanoTime() / 1000; for (int i = 0; i < iteration; i++) { bbf.rewind(); cbf.clear(); dec.reset(); cr = dec.decode(bbf, cbf, true); } long t2 = System.nanoTime() / 1000; t.t = (t2 - t1) / iteration; if (cr != CoderResult.UNDERFLOW) { System.out.println("DEC-----------------"); int pos = bbf.position(); System.out.printf( " cr=%s, bbf.pos=%d, bb[pos]=%x,%x,%x,%x%n", cr.toString(), pos, bb[pos++] & 0xff, bb[pos++] & 0xff, bb[pos++] & 0xff, bb[pos++] & 0xff); throw new RuntimeException("Decoding err: " + csn); } char[] cc = new char[cbf.position()]; cbf.flip(); cbf.get(cc); return cc; }
static CoderResult decodeCR(byte[] bb, Charset cs, boolean testDirect) throws Exception { CharsetDecoder dec = cs.newDecoder(); ByteBuffer bbf; CharBuffer cbf; if (testDirect) { bbf = ByteBuffer.allocateDirect(bb.length); cbf = ByteBuffer.allocateDirect(bb.length * 2).asCharBuffer(); bbf.put(bb).flip(); } else { bbf = ByteBuffer.wrap(bb); cbf = CharBuffer.allocate(bb.length); } CoderResult cr = null; for (int i = 0; i < iteration; i++) { bbf.rewind(); cbf.clear(); dec.reset(); cr = dec.decode(bbf, cbf, true); } return cr; }
/** * Decode file charset. * * @param f File to process. * @return File charset. * @throws IOException in case of error. */ public static Charset decode(File f) throws IOException { SortedMap<String, Charset> charsets = Charset.availableCharsets(); String[] firstCharsets = { Charset.defaultCharset().name(), "US-ASCII", "UTF-8", "UTF-16BE", "UTF-16LE" }; Collection<Charset> orderedCharsets = U.newLinkedHashSet(charsets.size()); for (String c : firstCharsets) if (charsets.containsKey(c)) orderedCharsets.add(charsets.get(c)); orderedCharsets.addAll(charsets.values()); try (RandomAccessFile raf = new RandomAccessFile(f, "r")) { FileChannel ch = raf.getChannel(); ByteBuffer buf = ByteBuffer.allocate(4096); ch.read(buf); buf.flip(); for (Charset charset : orderedCharsets) { CharsetDecoder decoder = charset.newDecoder(); decoder.reset(); try { decoder.decode(buf); return charset; } catch (CharacterCodingException ignored) { } } } return Charset.defaultCharset(); }