@NotNull
private static Pair<CharSequence, String> convertLineSeparators(@NotNull CharBuffer buffer) {
int dst = 0;
char prev = ' ';
int crCount = 0;
int lfCount = 0;
int crlfCount = 0;
final int length = buffer.length();
final char[] bufferArray = CharArrayUtil.fromSequenceWithoutCopying(buffer);
for (int src = 0; src < length; src++) {
char c = bufferArray != null ? bufferArray[src] : buffer.charAt(src);
switch (c) {
case '\r':
if (bufferArray != null) bufferArray[dst++] = '\n';
else buffer.put(dst++, '\n');
crCount++;
break;
case '\n':
if (prev == '\r') {
crCount--;
crlfCount++;
} else {
if (bufferArray != null) bufferArray[dst++] = '\n';
else buffer.put(dst++, '\n');
lfCount++;
}
break;
default:
if (bufferArray != null) bufferArray[dst++] = c;
else buffer.put(dst++, c);
break;
}
prev = c;
}
String detectedLineSeparator = null;
if (crlfCount > crCount && crlfCount > lfCount) {
detectedLineSeparator = "\r\n";
} else if (crCount > lfCount) {
detectedLineSeparator = "\r";
} else if (lfCount > 0) {
detectedLineSeparator = "\n";
}
CharSequence result;
if (buffer.length() == dst) {
result = buffer;
} else {
// in Mac JDK CharBuffer.subSequence() signature differs from Oracle's
// more than that, the signature has changed between jd6 and jdk7,
// so use more generic CharSequence.subSequence() just in case
@SuppressWarnings("UnnecessaryLocalVariable")
CharSequence seq = buffer;
result = seq.subSequence(0, dst);
}
return Pair.create(result, detectedLineSeparator);
}
private void checkSplitSequence() {
int len = buffer.position();
buffer.position(0);
// Search for the first & or \ in the last 10 (or less) characters
prevBuf = null;
int j = 0;
for (int i = len - 1; ((i >= 0) && (j < 10)); i--) {
if ((buffer.charAt(i) == '&') || (buffer.charAt(i) == '\\')) {
prevBuf = buffer.subSequence(i, len).toString();
len = i;
break;
}
j++;
}
buffer.position(len);
buffer.limit(len);
}
/**
* Check if the first X characters of a byte stream match a String.
*
* @param data The byte array to process
* @param pattern The String to match
* @return True if the pattern was found, false otherwise
*/
private static boolean bytesEqualsString(byte[] data, String pattern) {
byte[] bytes = new byte[pattern.length()];
Charset csets = Charset.forName("US-ASCII");
boolean fin = false;
int currChar = 0;
// remove any CR and/or LF characters at the beginning of the article
// data
while (!fin) {
if (currChar >= data.length) break;
byte in = data[currChar];
ByteBuffer bb = ByteBuffer.wrap(new byte[] {(byte) in});
CharBuffer cb = csets.decode(bb);
char c = cb.charAt(0);
if (data.length > 0 && (c == '\n' || c == '\r')) currChar++;
else fin = true;
if (data.length == 0) fin = true;
}
// extract bytes (chars) to check from article data
for (int i = 0; i < bytes.length && i < data.length; i++, currChar++) {
byte in = data[currChar];
bytes[i] = (byte) in;
}
// decode byte data to characters
ByteBuffer bb = ByteBuffer.wrap(bytes);
CharBuffer cb = csets.decode(bb);
// compare these characters to the pattern String
for (int i = 0; i < pattern.length(); i++) if (cb.charAt(i) != pattern.charAt(i)) return false;
return true;
}
public static void main(String[] args) throws Exception {
System.out.println(">>> StringCharBufferSliceTest-main: testing the slice method...");
final String in = "for testing";
System.out.println(">>> StringCharBufferSliceTest-main: testing with the position 0.");
CharBuffer buff = CharBuffer.wrap(in);
test(buff, buff.slice());
System.out.println(">>> StringCharBufferSliceTest-main: testing with new position.");
buff.position(2);
test(buff, buff.slice());
System.out.println(
">>> StringCharBufferSliceTest-main: testing with non zero initial position.");
buff = CharBuffer.wrap(in, 3, in.length());
test(buff, buff.slice());
System.out.println(">>> StringCharBufferSliceTest-main: testing slice result with get()");
buff.position(4);
buff.limit(7);
CharBuffer slice = buff.slice();
for (int i = 0; i < 3; i++) {
if (slice.get() != buff.get()) {
throw new RuntimeException("Wrong characters in slice result.");
}
}
System.out.println(">>> StringCharBufferSliceTest-main: testing slice result with get(int)");
buff.position(4);
buff.limit(7);
slice = buff.slice();
for (int i = 0; i < 3; i++) {
if (slice.get(i) != buff.get(4 + i)) {
throw new RuntimeException("Wrong characters in slice result.");
}
}
System.out.println(">>> StringCharBufferSliceTest-main: testing slice with result of slice");
buff.position(0);
buff.limit(buff.capacity());
slice = buff.slice();
for (int i = 0; i < 4; i++) {
slice.position(i);
CharBuffer nextSlice = slice.slice();
if (nextSlice.position() != 0)
throw new RuntimeException("New buffer's position should be zero");
if (!nextSlice.equals(slice)) throw new RuntimeException("New buffer should be equal");
slice = nextSlice;
}
System.out.println(">>> StringCharBufferSliceTest-main: testing toString.");
buff.position(4);
buff.limit(7);
slice = buff.slice();
if (!slice.toString().equals("tes")) {
throw new RuntimeException("bad toString() after slice(): " + slice.toString());
}
System.out.println(">>> StringCharBufferSliceTest-main: testing subSequence.");
buff.position(4);
buff.limit(8);
slice = buff.slice();
CharSequence subSeq = slice.subSequence(1, 3);
if (subSeq.charAt(0) != 'e' || subSeq.charAt(1) != 's') {
throw new RuntimeException("bad subSequence() after slice(): '" + subSeq + "'");
}
System.out.println(">>> StringCharBufferSliceTest-main: testing duplicate.");
buff.position(4);
buff.limit(8);
slice = buff.slice();
CharBuffer dupe = slice.duplicate();
if (dupe.charAt(0) != 't'
|| dupe.charAt(1) != 'e'
|| dupe.charAt(2) != 's'
|| dupe.charAt(3) != 't') {
throw new RuntimeException("bad duplicate() after slice(): '" + dupe + "'");
}
System.out.println(">>> StringCharBufferSliceTest-main: done!");
}
