com.sun.grizzly.util.buf.CharChunk Maven / Gradle / Ivy
/*
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS HEADER.
*
* Copyright (c) 2007-2010 Oracle and/or its affiliates. All rights reserved.
*
* The contents of this file are subject to the terms of either the GNU
* General Public License Version 2 only ("GPL") or the Common Development
* and Distribution License("CDDL") (collectively, the "License"). You
* may not use this file except in compliance with the License. You can
* obtain a copy of the License at
* https://glassfish.dev.java.net/public/CDDL+GPL_1_1.html
* or packager/legal/LICENSE.txt. See the License for the specific
* language governing permissions and limitations under the License.
*
* When distributing the software, include this License Header Notice in each
* file and include the License file at packager/legal/LICENSE.txt.
*
* GPL Classpath Exception:
* Oracle designates this particular file as subject to the "Classpath"
* exception as provided by Oracle in the GPL Version 2 section of the License
* file that accompanied this code.
*
* Modifications:
* If applicable, add the following below the License Header, with the fields
* enclosed by brackets [] replaced by your own identifying information:
* "Portions Copyright [year] [name of copyright owner]"
*
* Contributor(s):
* If you wish your version of this file to be governed by only the CDDL or
* only the GPL Version 2, indicate your decision by adding "[Contributor]
* elects to include this software in this distribution under the [CDDL or GPL
* Version 2] license." If you don't indicate a single choice of license, a
* recipient has the option to distribute your version of this file under
* either the CDDL, the GPL Version 2 or to extend the choice of license to
* its licensees as provided above. However, if you add GPL Version 2 code
* and therefore, elected the GPL Version 2 license, then the option applies
* only if the new code is made subject to such option by the copyright
* holder.
*
*
* This file incorporates work covered by the following copyright and
* permission notice:
*
* Copyright 2004 The Apache Software Foundation
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package com.sun.grizzly.util.buf;
import java.io.IOException;
import java.io.Serializable;
/**
* Utilities to manipluate char chunks. While String is
* the easiest way to manipulate chars ( search, substrings, etc),
* it is known to not be the most efficient solution - Strings are
* designed as imutable and secure objects.
*
* @author [email protected]
* @author James Todd [[email protected]]
* @author Costin Manolache
* @author Remy Maucherat
*/
public final class CharChunk implements Cloneable, Serializable {
// Input interface, used when the buffer is emptied.
public static interface CharInputChannel {
/**
* Read new bytes ( usually the internal conversion buffer ).
* The implementation is allowed to ignore the parameters,
* and mutate the chunk if it wishes to implement its own buffering.
*/
public int realReadChars(char cbuf[], int off, int len)
throws IOException;
}
/**
* When we need more space we'll either
* grow the buffer ( up to the limit ) or send it to a channel.
*/
public static interface CharOutputChannel {
/** Send the bytes ( usually the internal conversion buffer ).
* Expect 8k output if the buffer is full.
*/
public void realWriteChars(char cbuf[], int off, int len)
throws IOException;
}
// --------------------
// char[]
private char buff[];
private int start;
private int end;
private boolean isSet=false; // XXX
// -1: grow undefinitely
// maximum amount to be cached
private int limit=-1;
private transient CharInputChannel in = null;
private transient CharOutputChannel out = null;
private boolean optimizedWrite=true;
/**
* Creates a new, uninitialized CharChunk object.
*/
public CharChunk() {
}
public CharChunk(int size) {
allocate( size, -1 );
}
// --------------------
public CharChunk getClone() {
try {
return (CharChunk)this.clone();
} catch( Exception ex) {
return null;
}
}
public boolean isNull() {
if( end > 0 ) return false;
return !isSet; //XXX
}
/**
* Resets the message bytes to an uninitialized state.
*/
public void recycle() {
// buff=null;
isSet=false; // XXX
start=0;
end=0;
}
public void reset() {
buff=null;
}
// -------------------- Setup --------------------
public void allocate( int initial, int limit ) {
if( buff==null || buff.length < initial ) {
buff=new char[initial];
}
this.limit=limit;
start=0;
end=0;
isSet=true;
}
public void setOptimizedWrite(boolean optimizedWrite) {
this.optimizedWrite = optimizedWrite;
}
public void setChars( char[] c, int off, int len ) {
buff=c;
start=off;
end=start + len;
isSet=true;
}
/** Maximum amount of data in this buffer.
*
* If -1 or not set, the buffer will grow undefinitely.
* Can be smaller than the current buffer size ( which will not shrink ).
* When the limit is reached, the buffer will be flushed ( if out is set )
* or throw exception.
*/
public void setLimit(int limit) {
this.limit=limit;
}
public int getLimit() {
return limit;
}
/**
* When the buffer is empty, read the data from the input channel.
*/
public void setCharInputChannel(CharInputChannel in) {
this.in = in;
}
/** When the buffer is full, write the data to the output channel.
* Also used when large amount of data is appended.
*
* If not set, the buffer will grow to the limit.
*/
public void setCharOutputChannel(CharOutputChannel out) {
this.out=out;
}
// compat
public char[] getChars() {
return getBuffer();
}
public char[] getBuffer() {
return buff;
}
/**
* Returns the start offset of the bytes.
* For output this is the end of the buffer.
*/
public int getStart() {
return start;
}
public int getOffset() {
return start;
}
/**
* Returns the start offset of the bytes.
*/
public void setOffset(int off) {
start=off;
}
/**
* Returns the length of the bytes.
*/
public int getLength() {
return end-start;
}
public int getEnd() {
return end;
}
public void setEnd( int i ) {
end=i;
}
// -------------------- Adding data --------------------
public void append( char b ) throws IOException {
makeSpace( 1 );
// couldn't make space
if( limit >0 && end >= limit ) {
flushBuffer();
}
buff[end++]=b;
}
public void append( CharChunk src ) throws IOException {
append( src.getBuffer(), src.getOffset(), src.getLength());
}
/** Add data to the buffer
*/
public void append( char src[], int off, int len ) throws IOException {
// will grow, up to limit
makeSpace( len );
// if we don't have limit: makeSpace can grow as it wants
if( limit < 0 ) {
// assert: makeSpace made enough space
System.arraycopy( src, off, buff, end, len );
end+=len;
return;
}
// Optimize on a common case.
// If the source is going to fill up all the space in buffer, may
// as well write it directly to the output, and avoid an extra copy
if ( optimizedWrite && len == limit && end == start) {
out.realWriteChars( src, off, len );
return;
}
// if we have limit and we're below
if( len <= limit - end ) {
// makeSpace will grow the buffer to the limit,
// so we have space
System.arraycopy( src, off, buff, end, len );
end+=len;
return;
}
// need more space than we can afford, need to flush
// buffer
// the buffer is already at ( or bigger than ) limit
// Optimization:
// If len-avail < length ( i.e. after we fill the buffer with
// what we can, the remaining will fit in the buffer ) we'll just
// copy the first part, flush, then copy the second part - 1 write
// and still have some space for more. We'll still have 2 writes, but
// we write more on the first.
if( len + end < 2 * limit ) {
/* If the request length exceeds the size of the output buffer,
flush the output buffer and then write the data directly.
We can't avoid 2 writes, but we can write more on the second
*/
int avail=limit-end;
System.arraycopy(src, off, buff, end, avail);
end += avail;
flushBuffer();
System.arraycopy(src, off+avail, buff, end, len - avail);
end+= len - avail;
} else { // len > buf.length + avail
// long write - flush the buffer and write the rest
// directly from source
flushBuffer();
out.realWriteChars( src, off, len );
}
}
/** Add data to the buffer
*/
public void append( StringBuffer sb ) throws IOException {
int len=sb.length();
// will grow, up to limit
makeSpace( len );
// if we don't have limit: makeSpace can grow as it wants
if( limit < 0 ) {
// assert: makeSpace made enough space
sb.getChars(0, len, buff, end );
end+=len;
return;
}
int off=0;
int sbOff = off;
int sbEnd = off + len;
while (sbOff < sbEnd) {
int d = min(limit - end, sbEnd - sbOff);
sb.getChars( sbOff, sbOff+d, buff, end);
sbOff += d;
end += d;
if (end >= limit)
flushBuffer();
}
}
/** Append a string to the buffer
*/
public void append(String s) throws IOException {
append(s, 0, s.length());
}
/** Append a string to the buffer
*/
public void append(String s, int off, int len) throws IOException {
if (s==null) return;
// will grow, up to limit
makeSpace( len );
// if we don't have limit: makeSpace can grow as it wants
if( limit < 0 ) {
// assert: makeSpace made enough space
s.getChars(off, off+len, buff, end );
end+=len;
return;
}
int sOff = off;
int sEnd = off + len;
while (sOff < sEnd) {
int d = min(limit - end, sEnd - sOff);
s.getChars( sOff, sOff+d, buff, end);
sOff += d;
end += d;
if (end >= limit)
flushBuffer();
}
}
// -------------------- Removing data from the buffer --------------------
public int substract()
throws IOException {
if ((end - start) == 0) {
if (in == null)
return -1;
int n = in.realReadChars(buff, end, buff.length - end);
if (n < 0)
return -1;
}
return (buff[start++]);
}
public int substract(CharChunk src)
throws IOException {
if ((end - start) == 0) {
if (in == null)
return -1;
int n = in.realReadChars( buff, end, buff.length - end);
if (n < 0)
return -1;
}
int len = getLength();
src.append(buff, start, len);
start = end;
return len;
}
public int substract( char src[], int off, int len )
throws IOException {
if ((end - start) == 0) {
if (in == null)
return -1;
int n = in.realReadChars( buff, end, buff.length - end);
if (n < 0)
return -1;
}
int n = len;
if (len > getLength()) {
n = getLength();
}
System.arraycopy(buff, start, src, off, n);
start += n;
return n;
}
public void flushBuffer() throws IOException {
//assert out!=null
if( out==null ) {
throw new IOException( "Buffer overflow, no sink " + limit + " " +
buff.length );
}
out.realWriteChars( buff, start, end - start );
end=start;
}
/** Make space for len chars. If len is small, allocate
* a reserve space too. Never grow bigger than limit.
*/
void makeSpace(int count) {
char[] tmp = null;
int newSize;
int desiredSize=end + count;
// Can't grow above the limit
if( limit > 0 &&
desiredSize > limit) {
desiredSize=limit;
}
if( buff==null ) {
if( desiredSize < 256 ) desiredSize=256; // take a minimum
buff=new char[desiredSize];
}
// limit < buf.length ( the buffer is already big )
// or we already have space XXX
if( desiredSize <= buff.length) {
return;
}
// grow in larger chunks
if( desiredSize < 2 * buff.length ) {
newSize= buff.length * 2;
if( limit >0 &&
newSize > limit ) newSize=limit;
tmp=new char[newSize];
} else {
newSize= buff.length * 2 + count ;
if( limit > 0 &&
newSize > limit ) newSize=limit;
tmp=new char[newSize];
}
System.arraycopy(buff, start, tmp, start, end-start);
buff = tmp;
tmp = null;
}
// -------------------- Conversion and getters --------------------
public String toString() {
if (null == buff) {
return null;
} else if (end-start == 0) {
return "";
}
return StringCache.toString(this);
}
public String toStringInternal() {
return new String(buff, start, end-start);
}
public int getInt() {
return Ascii.parseInt(buff, start, end-start);
}
// -------------------- equals --------------------
/**
* Compares the message bytes to the specified String object.
* @param s the String to compare
* @return true if the comparison succeeded, false otherwise
*/
public boolean equals(String s) {
char[] c = buff;
int len = end-start;
if (c == null || len != s.length()) {
return false;
}
int off = start;
for (int i = 0; i < len; i++) {
if (c[off++] != s.charAt(i)) {
return false;
}
}
return true;
}
/**
* Compares the message bytes to the specified String object.
* @param s the String to compare
* @return true if the comparison succeeded, false otherwise
*/
public boolean equalsIgnoreCase(String s) {
char[] c = buff;
int len = end-start;
if (c == null || len != s.length()) {
return false;
}
int off = start;
for (int i = 0; i < len; i++) {
if (Ascii.toLower( c[off++] ) != Ascii.toLower( s.charAt(i))) {
return false;
}
}
return true;
}
public boolean equals(CharChunk cc) {
return equals( cc.getChars(), cc.getOffset(), cc.getLength());
}
public boolean equals(char b2[], int off2, int len2) {
char b1[]=buff;
if( b1==null && b2==null ) return true;
if (b1== null || b2==null || end-start != len2) {
return false;
}
int off1 = start;
int len=end-start;
while ( len-- > 0) {
if (b1[off1++] != b2[off2++]) {
return false;
}
}
return true;
}
public boolean equals(byte b2[], int off2, int len2) {
char b1[]=buff;
if( b2==null && b1==null ) return true;
if (b1== null || b2==null || end-start != len2) {
return false;
}
int off1 = start;
int len=end-start;
while ( len-- > 0) {
if ( b1[off1++] != (char)b2[off2++]) {
return false;
}
}
return true;
}
/**
* Returns true if the message bytes starts with the specified string.
* @param s the string
*/
public boolean startsWith(String s) {
char[] c = buff;
int len = s.length();
if (c == null || len > end-start) {
return false;
}
int off = start;
for (int i = 0; i < len; i++) {
if (c[off++] != s.charAt(i)) {
return false;
}
}
return true;
}
/**
* Returns true if the message bytes starts with the specified string.
* @param s the string
*/
public boolean startsWithIgnoreCase(String s, int pos) {
char[] c = buff;
int len = s.length();
if (c == null || len+pos > end-start) {
return false;
}
int off = start+pos;
for (int i = 0; i < len; i++) {
if (Ascii.toLower( c[off++] ) != Ascii.toLower( s.charAt(i))) {
return false;
}
}
return true;
}
// -------------------- Hash code --------------------
// normal hash.
public int hash() {
int code=0;
for (int i = start; i < start + end-start; i++) {
code = code * 37 + buff[i];
}
return code;
}
// hash ignoring case
public int hashIgnoreCase() {
int code=0;
for (int i = start; i < end; i++) {
code = code * 37 + Ascii.toLower(buff[i]);
}
return code;
}
public int indexOf(char c) {
return indexOf( c, start);
}
/**
* Returns true if the message bytes starts with the specified string.
* @param c the character
*/
public int indexOf(char c, int starting) {
int ret = indexOf( buff, start+starting, end, c );
return (ret >= start) ? ret - start : -1;
}
public static int indexOf( char chars[], int off, int cend, char qq ) {
while( off < cend ) {
char b=chars[off];
if( b==qq )
return off;
off++;
}
return -1;
}
public int indexOf( String src, int srcOff, int srcLen, int myOff ) {
char first=src.charAt( srcOff );
// Look for first char
int srcEnd = srcOff + srcLen;
for( int i=myOff+start; i <= (end - srcLen); i++ ) {
if( buff[i] != first ) continue;
// found first char, now look for a match
int myPos=i+1;
for( int srcPos=srcOff + 1; srcPos< srcEnd; ) {
if( buff[myPos++] != src.charAt( srcPos++ ))
break;
if( srcPos==srcEnd ) return i-start; // found it
}
}
return -1;
}
// -------------------- utils
private int min(int a, int b) {
if (a < b) return a;
return b;
}
}
© 2015 - 2024 Weber Informatics LLC | Privacy Policy