io.netty.util.AsciiString Maven / Gradle / Ivy
/*
* Copyright 2014 The Netty Project
*
* The Netty Project licenses this file to you 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:
*
* https://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 io.netty.util;
import io.netty.util.internal.EmptyArrays;
import io.netty.util.internal.InternalThreadLocalMap;
import io.netty.util.internal.ObjectUtil;
import io.netty.util.internal.PlatformDependent;
import java.nio.ByteBuffer;
import java.nio.CharBuffer;
import java.nio.charset.Charset;
import java.nio.charset.CharsetEncoder;
import java.util.Arrays;
import java.util.Collection;
import java.util.List;
import java.util.regex.Pattern;
import java.util.regex.PatternSyntaxException;
import static io.netty.util.internal.MathUtil.isOutOfBounds;
import static io.netty.util.internal.ObjectUtil.checkNotNull;
/**
* A string which has been encoded into a character encoding whose character always takes a single byte, similarly to
* ASCII. It internally keeps its content in a byte array unlike {@link String}, which uses a character array, for
* reduced memory footprint and faster data transfer from/to byte-based data structures such as a byte array and
* {@link ByteBuffer}. It is often used in conjunction with {@code Headers} that require a {@link CharSequence}.
*
* This class was designed to provide an immutable array of bytes, and caches some internal state based upon the value
* of this array. However underlying access to this byte array is provided via not copying the array on construction or
* {@link #array()}. If any changes are made to the underlying byte array it is the user's responsibility to call
* {@link #arrayChanged()} so the state of this class can be reset.
*/
public final class AsciiString implements CharSequence, Comparable {
public static final AsciiString EMPTY_STRING = cached("");
private static final char MAX_CHAR_VALUE = 255;
public static final int INDEX_NOT_FOUND = -1;
/**
* If this value is modified outside the constructor then call {@link #arrayChanged()}.
*/
private final byte[] value;
/**
* Offset into {@link #value} that all operations should use when acting upon {@link #value}.
*/
private final int offset;
/**
* Length in bytes for {@link #value} that we care about. This is independent from {@code value.length}
* because we may be looking at a subsection of the array.
*/
private final int length;
/**
* The hash code is cached after it is first computed. It can be reset with {@link #arrayChanged()}.
*/
private int hash;
/**
* Used to cache the {@link #toString()} value.
*/
private String string;
/**
* Initialize this byte string based upon a byte array. A copy will be made.
*/
public AsciiString(byte[] value) {
this(value, true);
}
/**
* Initialize this byte string based upon a byte array.
* {@code copy} determines if a copy is made or the array is shared.
*/
public AsciiString(byte[] value, boolean copy) {
this(value, 0, value.length, copy);
}
/**
* Construct a new instance from a {@code byte[]} array.
* @param copy {@code true} then a copy of the memory will be made. {@code false} the underlying memory
* will be shared.
*/
public AsciiString(byte[] value, int start, int length, boolean copy) {
if (copy) {
this.value = Arrays.copyOfRange(value, start, start + length);
this.offset = 0;
} else {
if (isOutOfBounds(start, length, value.length)) {
throw new IndexOutOfBoundsException("expected: " + "0 <= start(" + start + ") <= start + length(" +
length + ") <= " + "value.length(" + value.length + ')');
}
this.value = value;
this.offset = start;
}
this.length = length;
}
/**
* Create a copy of the underlying storage from {@code value}.
* The copy will start at {@link ByteBuffer#position()} and copy {@link ByteBuffer#remaining()} bytes.
*/
public AsciiString(ByteBuffer value) {
this(value, true);
}
/**
* Initialize an instance based upon the underlying storage from {@code value}.
* There is a potential to share the underlying array storage if {@link ByteBuffer#hasArray()} is {@code true}.
* if {@code copy} is {@code true} a copy will be made of the memory.
* if {@code copy} is {@code false} the underlying storage will be shared, if possible.
*/
public AsciiString(ByteBuffer value, boolean copy) {
this(value, value.position(), value.remaining(), copy);
}
/**
* Initialize an {@link AsciiString} based upon the underlying storage from {@code value}.
* There is a potential to share the underlying array storage if {@link ByteBuffer#hasArray()} is {@code true}.
* if {@code copy} is {@code true} a copy will be made of the memory.
* if {@code copy} is {@code false} the underlying storage will be shared, if possible.
*/
public AsciiString(ByteBuffer value, int start, int length, boolean copy) {
if (isOutOfBounds(start, length, value.capacity())) {
throw new IndexOutOfBoundsException("expected: " + "0 <= start(" + start + ") <= start + length(" + length
+ ") <= " + "value.capacity(" + value.capacity() + ')');
}
if (value.hasArray()) {
if (copy) {
final int bufferOffset = value.arrayOffset() + start;
this.value = Arrays.copyOfRange(value.array(), bufferOffset, bufferOffset + length);
offset = 0;
} else {
this.value = value.array();
this.offset = start;
}
} else {
this.value = PlatformDependent.allocateUninitializedArray(length);
int oldPos = value.position();
value.get(this.value, 0, length);
value.position(oldPos);
this.offset = 0;
}
this.length = length;
}
/**
* Create a copy of {@code value} into this instance assuming ASCII encoding.
*/
public AsciiString(char[] value) {
this(value, 0, value.length);
}
/**
* Create a copy of {@code value} into this instance assuming ASCII encoding.
* The copy will start at index {@code start} and copy {@code length} bytes.
*/
public AsciiString(char[] value, int start, int length) {
if (isOutOfBounds(start, length, value.length)) {
throw new IndexOutOfBoundsException("expected: " + "0 <= start(" + start + ") <= start + length(" + length
+ ") <= " + "value.length(" + value.length + ')');
}
this.value = PlatformDependent.allocateUninitializedArray(length);
for (int i = 0, j = start; i < length; i++, j++) {
this.value[i] = c2b(value[j]);
}
this.offset = 0;
this.length = length;
}
/**
* Create a copy of {@code value} into this instance using the encoding type of {@code charset}.
*/
public AsciiString(char[] value, Charset charset) {
this(value, charset, 0, value.length);
}
/**
* Create a copy of {@code value} into a this instance using the encoding type of {@code charset}.
* The copy will start at index {@code start} and copy {@code length} bytes.
*/
public AsciiString(char[] value, Charset charset, int start, int length) {
CharBuffer cbuf = CharBuffer.wrap(value, start, length);
CharsetEncoder encoder = CharsetUtil.encoder(charset);
ByteBuffer nativeBuffer = ByteBuffer.allocate((int) (encoder.maxBytesPerChar() * length));
encoder.encode(cbuf, nativeBuffer, true);
final int bufferOffset = nativeBuffer.arrayOffset();
this.value = Arrays.copyOfRange(nativeBuffer.array(), bufferOffset, bufferOffset + nativeBuffer.position());
this.offset = 0;
this.length = this.value.length;
}
/**
* Create a copy of {@code value} into this instance assuming ASCII encoding.
*/
public AsciiString(CharSequence value) {
this(value, 0, value.length());
}
/**
* Create a copy of {@code value} into this instance assuming ASCII encoding.
* The copy will start at index {@code start} and copy {@code length} bytes.
*/
public AsciiString(CharSequence value, int start, int length) {
if (isOutOfBounds(start, length, value.length())) {
throw new IndexOutOfBoundsException("expected: " + "0 <= start(" + start + ") <= start + length(" + length
+ ") <= " + "value.length(" + value.length() + ')');
}
this.value = PlatformDependent.allocateUninitializedArray(length);
for (int i = 0, j = start; i < length; i++, j++) {
this.value[i] = c2b(value.charAt(j));
}
this.offset = 0;
this.length = length;
}
/**
* Create a copy of {@code value} into this instance using the encoding type of {@code charset}.
*/
public AsciiString(CharSequence value, Charset charset) {
this(value, charset, 0, value.length());
}
/**
* Create a copy of {@code value} into this instance using the encoding type of {@code charset}.
* The copy will start at index {@code start} and copy {@code length} bytes.
*/
public AsciiString(CharSequence value, Charset charset, int start, int length) {
CharBuffer cbuf = CharBuffer.wrap(value, start, start + length);
CharsetEncoder encoder = CharsetUtil.encoder(charset);
ByteBuffer nativeBuffer = ByteBuffer.allocate((int) (encoder.maxBytesPerChar() * length));
encoder.encode(cbuf, nativeBuffer, true);
final int offset = nativeBuffer.arrayOffset();
this.value = Arrays.copyOfRange(nativeBuffer.array(), offset, offset + nativeBuffer.position());
this.offset = 0;
this.length = this.value.length;
}
/**
* Iterates over the readable bytes of this buffer with the specified {@code processor} in ascending order.
*
* @return {@code -1} if the processor iterated to or beyond the end of the readable bytes.
* The last-visited index If the {@link ByteProcessor#process(byte)} returned {@code false}.
*/
public int forEachByte(ByteProcessor visitor) throws Exception {
return forEachByte0(0, length(), visitor);
}
/**
* Iterates over the specified area of this buffer with the specified {@code processor} in ascending order.
* (i.e. {@code index}, {@code (index + 1)}, .. {@code (index + length - 1)}).
*
* @return {@code -1} if the processor iterated to or beyond the end of the specified area.
* The last-visited index If the {@link ByteProcessor#process(byte)} returned {@code false}.
*/
public int forEachByte(int index, int length, ByteProcessor visitor) throws Exception {
if (isOutOfBounds(index, length, length())) {
throw new IndexOutOfBoundsException("expected: " + "0 <= index(" + index + ") <= start + length(" + length
+ ") <= " + "length(" + length() + ')');
}
return forEachByte0(index, length, visitor);
}
private int forEachByte0(int index, int length, ByteProcessor visitor) throws Exception {
final int len = offset + index + length;
for (int i = offset + index; i < len; ++i) {
if (!visitor.process(value[i])) {
return i - offset;
}
}
return -1;
}
/**
* Iterates over the readable bytes of this buffer with the specified {@code processor} in descending order.
*
* @return {@code -1} if the processor iterated to or beyond the beginning of the readable bytes.
* The last-visited index If the {@link ByteProcessor#process(byte)} returned {@code false}.
*/
public int forEachByteDesc(ByteProcessor visitor) throws Exception {
return forEachByteDesc0(0, length(), visitor);
}
/**
* Iterates over the specified area of this buffer with the specified {@code processor} in descending order.
* (i.e. {@code (index + length - 1)}, {@code (index + length - 2)}, ... {@code index}).
*
* @return {@code -1} if the processor iterated to or beyond the beginning of the specified area.
* The last-visited index If the {@link ByteProcessor#process(byte)} returned {@code false}.
*/
public int forEachByteDesc(int index, int length, ByteProcessor visitor) throws Exception {
if (isOutOfBounds(index, length, length())) {
throw new IndexOutOfBoundsException("expected: " + "0 <= index(" + index + ") <= start + length(" + length
+ ") <= " + "length(" + length() + ')');
}
return forEachByteDesc0(index, length, visitor);
}
private int forEachByteDesc0(int index, int length, ByteProcessor visitor) throws Exception {
final int end = offset + index;
for (int i = offset + index + length - 1; i >= end; --i) {
if (!visitor.process(value[i])) {
return i - offset;
}
}
return -1;
}
public byte byteAt(int index) {
// We must do a range check here to enforce the access does not go outside our sub region of the array.
// We rely on the array access itself to pick up the array out of bounds conditions
if (index < 0 || index >= length) {
throw new IndexOutOfBoundsException("index: " + index + " must be in the range [0," + length + ")");
}
// Try to use unsafe to avoid double checking the index bounds
if (PlatformDependent.hasUnsafe()) {
return PlatformDependent.getByte(value, index + offset);
}
return value[index + offset];
}
/**
* Determine if this instance has 0 length.
*/
public boolean isEmpty() {
return length == 0;
}
/**
* The length in bytes of this instance.
*/
@Override
public int length() {
return length;
}
/**
* During normal use cases the {@link AsciiString} should be immutable, but if the underlying array is shared,
* and changes then this needs to be called.
*/
public void arrayChanged() {
string = null;
hash = 0;
}
/**
* This gives direct access to the underlying storage array.
* The {@link #toByteArray()} should be preferred over this method.
* If the return value is changed then {@link #arrayChanged()} must be called.
* @see #arrayOffset()
* @see #isEntireArrayUsed()
*/
public byte[] array() {
return value;
}
/**
* The offset into {@link #array()} for which data for this ByteString begins.
* @see #array()
* @see #isEntireArrayUsed()
*/
public int arrayOffset() {
return offset;
}
/**
* Determine if the storage represented by {@link #array()} is entirely used.
* @see #array()
*/
public boolean isEntireArrayUsed() {
return offset == 0 && length == value.length;
}
/**
* Converts this string to a byte array.
*/
public byte[] toByteArray() {
return toByteArray(0, length());
}
/**
* Converts a subset of this string to a byte array.
* The subset is defined by the range [{@code start}, {@code end}).
*/
public byte[] toByteArray(int start, int end) {
return Arrays.copyOfRange(value, start + offset, end + offset);
}
/**
* Copies the content of this string to a byte array.
*
* @param srcIdx the starting offset of characters to copy.
* @param dst the destination byte array.
* @param dstIdx the starting offset in the destination byte array.
* @param length the number of characters to copy.
*/
public void copy(int srcIdx, byte[] dst, int dstIdx, int length) {
if (isOutOfBounds(srcIdx, length, length())) {
throw new IndexOutOfBoundsException("expected: " + "0 <= srcIdx(" + srcIdx + ") <= srcIdx + length("
+ length + ") <= srcLen(" + length() + ')');
}
System.arraycopy(value, srcIdx + offset, checkNotNull(dst, "dst"), dstIdx, length);
}
@Override
public char charAt(int index) {
return b2c(byteAt(index));
}
/**
* Determines if this {@code String} contains the sequence of characters in the {@code CharSequence} passed.
*
* @param cs the character sequence to search for.
* @return {@code true} if the sequence of characters are contained in this string, otherwise {@code false}.
*/
public boolean contains(CharSequence cs) {
return indexOf(cs) >= 0;
}
/**
* Compares the specified string to this string using the ASCII values of the characters. Returns 0 if the strings
* contain the same characters in the same order. Returns a negative integer if the first non-equal character in
* this string has an ASCII value which is less than the ASCII value of the character at the same position in the
* specified string, or if this string is a prefix of the specified string. Returns a positive integer if the first
* non-equal character in this string has a ASCII value which is greater than the ASCII value of the character at
* the same position in the specified string, or if the specified string is a prefix of this string.
*
* @param string the string to compare.
* @return 0 if the strings are equal, a negative integer if this string is before the specified string, or a
* positive integer if this string is after the specified string.
* @throws NullPointerException if {@code string} is {@code null}.
*/
@Override
public int compareTo(CharSequence string) {
if (this == string) {
return 0;
}
int result;
int length1 = length();
int length2 = string.length();
int minLength = Math.min(length1, length2);
for (int i = 0, j = arrayOffset(); i < minLength; i++, j++) {
result = b2c(value[j]) - string.charAt(i);
if (result != 0) {
return result;
}
}
return length1 - length2;
}
/**
* Concatenates this string and the specified string.
*
* @param string the string to concatenate
* @return a new string which is the concatenation of this string and the specified string.
*/
public AsciiString concat(CharSequence string) {
int thisLen = length();
int thatLen = string.length();
if (thatLen == 0) {
return this;
}
if (string instanceof AsciiString) {
AsciiString that = (AsciiString) string;
if (isEmpty()) {
return that;
}
byte[] newValue = PlatformDependent.allocateUninitializedArray(thisLen + thatLen);
System.arraycopy(value, arrayOffset(), newValue, 0, thisLen);
System.arraycopy(that.value, that.arrayOffset(), newValue, thisLen, thatLen);
return new AsciiString(newValue, false);
}
if (isEmpty()) {
return new AsciiString(string);
}
byte[] newValue = PlatformDependent.allocateUninitializedArray(thisLen + thatLen);
System.arraycopy(value, arrayOffset(), newValue, 0, thisLen);
for (int i = thisLen, j = 0; i < newValue.length; i++, j++) {
newValue[i] = c2b(string.charAt(j));
}
return new AsciiString(newValue, false);
}
/**
* Compares the specified string to this string to determine if the specified string is a suffix.
*
* @param suffix the suffix to look for.
* @return {@code true} if the specified string is a suffix of this string, {@code false} otherwise.
* @throws NullPointerException if {@code suffix} is {@code null}.
*/
public boolean endsWith(CharSequence suffix) {
int suffixLen = suffix.length();
return regionMatches(length() - suffixLen, suffix, 0, suffixLen);
}
/**
* Compares the specified string to this string ignoring the case of the characters and returns true if they are
* equal.
*
* @param string the string to compare.
* @return {@code true} if the specified string is equal to this string, {@code false} otherwise.
*/
public boolean contentEqualsIgnoreCase(CharSequence string) {
if (this == string) {
return true;
}
if (string == null || string.length() != length()) {
return false;
}
if (string instanceof AsciiString) {
AsciiString rhs = (AsciiString) string;
for (int i = arrayOffset(), j = rhs.arrayOffset(), end = i + length(); i < end; ++i, ++j) {
if (!equalsIgnoreCase(value[i], rhs.value[j])) {
return false;
}
}
return true;
}
for (int i = arrayOffset(), j = 0, end = length(); j < end; ++i, ++j) {
if (!equalsIgnoreCase(b2c(value[i]), string.charAt(j))) {
return false;
}
}
return true;
}
/**
* Copies the characters in this string to a character array.
*
* @return a character array containing the characters of this string.
*/
public char[] toCharArray() {
return toCharArray(0, length());
}
/**
* Copies the characters in this string to a character array.
*
* @return a character array containing the characters of this string.
*/
public char[] toCharArray(int start, int end) {
int length = end - start;
if (length == 0) {
return EmptyArrays.EMPTY_CHARS;
}
if (isOutOfBounds(start, length, length())) {
throw new IndexOutOfBoundsException("expected: " + "0 <= start(" + start + ") <= srcIdx + length("
+ length + ") <= srcLen(" + length() + ')');
}
final char[] buffer = new char[length];
for (int i = 0, j = start + arrayOffset(); i < length; i++, j++) {
buffer[i] = b2c(value[j]);
}
return buffer;
}
/**
* Copied the content of this string to a character array.
*
* @param srcIdx the starting offset of characters to copy.
* @param dst the destination character array.
* @param dstIdx the starting offset in the destination byte array.
* @param length the number of characters to copy.
*/
public void copy(int srcIdx, char[] dst, int dstIdx, int length) {
ObjectUtil.checkNotNull(dst, "dst");
if (isOutOfBounds(srcIdx, length, length())) {
throw new IndexOutOfBoundsException("expected: " + "0 <= srcIdx(" + srcIdx + ") <= srcIdx + length("
+ length + ") <= srcLen(" + length() + ')');
}
final int dstEnd = dstIdx + length;
for (int i = dstIdx, j = srcIdx + arrayOffset(); i < dstEnd; i++, j++) {
dst[i] = b2c(value[j]);
}
}
/**
* Copies a range of characters into a new string.
* @param start the offset of the first character (inclusive).
* @return a new string containing the characters from start to the end of the string.
* @throws IndexOutOfBoundsException if {@code start < 0} or {@code start > length()}.
*/
public AsciiString subSequence(int start) {
return subSequence(start, length());
}
/**
* Copies a range of characters into a new string.
* @param start the offset of the first character (inclusive).
* @param end The index to stop at (exclusive).
* @return a new string containing the characters from start to the end of the string.
* @throws IndexOutOfBoundsException if {@code start < 0} or {@code start > length()}.
*/
@Override
public AsciiString subSequence(int start, int end) {
return subSequence(start, end, true);
}
/**
* Either copy or share a subset of underlying sub-sequence of bytes.
* @param start the offset of the first character (inclusive).
* @param end The index to stop at (exclusive).
* @param copy If {@code true} then a copy of the underlying storage will be made.
* If {@code false} then the underlying storage will be shared.
* @return a new string containing the characters from start to the end of the string.
* @throws IndexOutOfBoundsException if {@code start < 0} or {@code start > length()}.
*/
public AsciiString subSequence(int start, int end, boolean copy) {
if (isOutOfBounds(start, end - start, length())) {
throw new IndexOutOfBoundsException("expected: 0 <= start(" + start + ") <= end (" + end + ") <= length("
+ length() + ')');
}
if (start == 0 && end == length()) {
return this;
}
if (end == start) {
return EMPTY_STRING;
}
return new AsciiString(value, start + offset, end - start, copy);
}
/**
* Searches in this string for the first index of the specified string. The search for the string starts at the
* beginning and moves towards the end of this string.
*
* @param string the string to find.
* @return the index of the first character of the specified string in this string, -1 if the specified string is
* not a substring.
* @throws NullPointerException if {@code string} is {@code null}.
*/
public int indexOf(CharSequence string) {
return indexOf(string, 0);
}
/**
* Searches in this string for the index of the specified string. The search for the string starts at the specified
* offset and moves towards the end of this string.
*
* @param subString the string to find.
* @param start the starting offset.
* @return the index of the first character of the specified string in this string, -1 if the specified string is
* not a substring.
* @throws NullPointerException if {@code subString} is {@code null}.
*/
public int indexOf(CharSequence subString, int start) {
final int subCount = subString.length();
if (start < 0) {
start = 0;
}
if (subCount <= 0) {
return start < length ? start : length;
}
if (subCount > length - start) {
return INDEX_NOT_FOUND;
}
final char firstChar = subString.charAt(0);
if (firstChar > MAX_CHAR_VALUE) {
return INDEX_NOT_FOUND;
}
final byte firstCharAsByte = c2b0(firstChar);
final int len = offset + length - subCount;
for (int i = start + offset; i <= len; ++i) {
if (value[i] == firstCharAsByte) {
int o1 = i, o2 = 0;
while (++o2 < subCount && b2c(value[++o1]) == subString.charAt(o2)) {
// Intentionally empty
}
if (o2 == subCount) {
return i - offset;
}
}
}
return INDEX_NOT_FOUND;
}
/**
* Searches in this string for the index of the specified char {@code ch}.
* The search for the char starts at the specified offset {@code start} and moves towards the end of this string.
*
* @param ch the char to find.
* @param start the starting offset.
* @return the index of the first occurrence of the specified char {@code ch} in this string,
* -1 if found no occurrence.
*/
public int indexOf(char ch, int start) {
if (ch > MAX_CHAR_VALUE) {
return INDEX_NOT_FOUND;
}
if (start < 0) {
start = 0;
}
final byte chAsByte = c2b0(ch);
final int len = offset + length;
for (int i = start + offset; i < len; ++i) {
if (value[i] == chAsByte) {
return i - offset;
}
}
return INDEX_NOT_FOUND;
}
/**
* Searches in this string for the last index of the specified string. The search for the string starts at the end
* and moves towards the beginning of this string.
*
* @param string the string to find.
* @return the index of the first character of the specified string in this string, -1 if the specified string is
* not a substring.
* @throws NullPointerException if {@code string} is {@code null}.
*/
public int lastIndexOf(CharSequence string) {
// Use count instead of count - 1 so lastIndexOf("") answers count
return lastIndexOf(string, length);
}
/**
* Searches in this string for the index of the specified string. The search for the string starts at the specified
* offset and moves towards the beginning of this string.
*
* @param subString the string to find.
* @param start the starting offset.
* @return the index of the first character of the specified string in this string , -1 if the specified string is
* not a substring.
* @throws NullPointerException if {@code subString} is {@code null}.
*/
public int lastIndexOf(CharSequence subString, int start) {
final int subCount = subString.length();
start = Math.min(start, length - subCount);
if (start < 0) {
return INDEX_NOT_FOUND;
}
if (subCount == 0) {
return start;
}
final char firstChar = subString.charAt(0);
if (firstChar > MAX_CHAR_VALUE) {
return INDEX_NOT_FOUND;
}
final byte firstCharAsByte = c2b0(firstChar);
for (int i = offset + start; i >= 0; --i) {
if (value[i] == firstCharAsByte) {
int o1 = i, o2 = 0;
while (++o2 < subCount && b2c(value[++o1]) == subString.charAt(o2)) {
// Intentionally empty
}
if (o2 == subCount) {
return i - offset;
}
}
}
return INDEX_NOT_FOUND;
}
/**
* Compares the specified string to this string and compares the specified range of characters to determine if they
* are the same.
*
* @param thisStart the starting offset in this string.
* @param string the string to compare.
* @param start the starting offset in the specified string.
* @param length the number of characters to compare.
* @return {@code true} if the ranges of characters are equal, {@code false} otherwise
* @throws NullPointerException if {@code string} is {@code null}.
*/
public boolean regionMatches(int thisStart, CharSequence string, int start, int length) {
ObjectUtil.checkNotNull(string, "string");
if (start < 0 || string.length() - start < length) {
return false;
}
final int thisLen = length();
if (thisStart < 0 || thisLen - thisStart < length) {
return false;
}
if (length <= 0) {
return true;
}
final int thatEnd = start + length;
for (int i = start, j = thisStart + arrayOffset(); i < thatEnd; i++, j++) {
if (b2c(value[j]) != string.charAt(i)) {
return false;
}
}
return true;
}
/**
* Compares the specified string to this string and compares the specified range of characters to determine if they
* are the same. When ignoreCase is true, the case of the characters is ignored during the comparison.
*
* @param ignoreCase specifies if case should be ignored.
* @param thisStart the starting offset in this string.
* @param string the string to compare.
* @param start the starting offset in the specified string.
* @param length the number of characters to compare.
* @return {@code true} if the ranges of characters are equal, {@code false} otherwise.
* @throws NullPointerException if {@code string} is {@code null}.
*/
public boolean regionMatches(boolean ignoreCase, int thisStart, CharSequence string, int start, int length) {
if (!ignoreCase) {
return regionMatches(thisStart, string, start, length);
}
ObjectUtil.checkNotNull(string, "string");
final int thisLen = length();
if (thisStart < 0 || length > thisLen - thisStart) {
return false;
}
if (start < 0 || length > string.length() - start) {
return false;
}
thisStart += arrayOffset();
final int thisEnd = thisStart + length;
while (thisStart < thisEnd) {
if (!equalsIgnoreCase(b2c(value[thisStart++]), string.charAt(start++))) {
return false;
}
}
return true;
}
/**
* Copies this string replacing occurrences of the specified character with another character.
*
* @param oldChar the character to replace.
* @param newChar the replacement character.
* @return a new string with occurrences of oldChar replaced by newChar.
*/
public AsciiString replace(char oldChar, char newChar) {
if (oldChar > MAX_CHAR_VALUE) {
return this;
}
final byte oldCharAsByte = c2b0(oldChar);
final byte newCharAsByte = c2b(newChar);
final int len = offset + length;
for (int i = offset; i < len; ++i) {
if (value[i] == oldCharAsByte) {
byte[] buffer = PlatformDependent.allocateUninitializedArray(length());
System.arraycopy(value, offset, buffer, 0, i - offset);
buffer[i - offset] = newCharAsByte;
++i;
for (; i < len; ++i) {
byte oldValue = value[i];
buffer[i - offset] = oldValue != oldCharAsByte ? oldValue : newCharAsByte;
}
return new AsciiString(buffer, false);
}
}
return this;
}
/**
* Compares the specified string to this string to determine if the specified string is a prefix.
*
* @param prefix the string to look for.
* @return {@code true} if the specified string is a prefix of this string, {@code false} otherwise
* @throws NullPointerException if {@code prefix} is {@code null}.
*/
public boolean startsWith(CharSequence prefix) {
return startsWith(prefix, 0);
}
/**
* Compares the specified string to this string, starting at the specified offset, to determine if the specified
* string is a prefix.
*
* @param prefix the string to look for.
* @param start the starting offset.
* @return {@code true} if the specified string occurs in this string at the specified offset, {@code false}
* otherwise.
* @throws NullPointerException if {@code prefix} is {@code null}.
*/
public boolean startsWith(CharSequence prefix, int start) {
return regionMatches(start, prefix, 0, prefix.length());
}
/**
* Converts the characters in this string to lowercase, using the default Locale.
*
* @return a new string containing the lowercase characters equivalent to the characters in this string.
*/
public AsciiString toLowerCase() {
boolean lowercased = true;
int i, j;
final int len = length() + arrayOffset();
for (i = arrayOffset(); i < len; ++i) {
byte b = value[i];
if (b >= 'A' && b <= 'Z') {
lowercased = false;
break;
}
}
// Check if this string does not contain any uppercase characters.
if (lowercased) {
return this;
}
final byte[] newValue = PlatformDependent.allocateUninitializedArray(length());
for (i = 0, j = arrayOffset(); i < newValue.length; ++i, ++j) {
newValue[i] = toLowerCase(value[j]);
}
return new AsciiString(newValue, false);
}
/**
* Converts the characters in this string to uppercase, using the default Locale.
*
* @return a new string containing the uppercase characters equivalent to the characters in this string.
*/
public AsciiString toUpperCase() {
boolean uppercased = true;
int i, j;
final int len = length() + arrayOffset();
for (i = arrayOffset(); i < len; ++i) {
byte b = value[i];
if (b >= 'a' && b <= 'z') {
uppercased = false;
break;
}
}
// Check if this string does not contain any lowercase characters.
if (uppercased) {
return this;
}
final byte[] newValue = PlatformDependent.allocateUninitializedArray(length());
for (i = 0, j = arrayOffset(); i < newValue.length; ++i, ++j) {
newValue[i] = toUpperCase(value[j]);
}
return new AsciiString(newValue, false);
}
/**
* Copies this string removing white space characters from the beginning and end of the string, and tries not to
* copy if possible.
*
* @param c The {@link CharSequence} to trim.
* @return a new string with characters {@code <= \\u0020} removed from the beginning and the end.
*/
public static CharSequence trim(CharSequence c) {
if (c instanceof AsciiString) {
return ((AsciiString) c).trim();
}
if (c instanceof String) {
return ((String) c).trim();
}
int start = 0, last = c.length() - 1;
int end = last;
while (start <= end && c.charAt(start) <= ' ') {
start++;
}
while (end >= start && c.charAt(end) <= ' ') {
end--;
}
if (start == 0 && end == last) {
return c;
}
return c.subSequence(start, end);
}
/**
* Duplicates this string removing white space characters from the beginning and end of the
* string, without copying.
*
* @return a new string with characters {@code <= \\u0020} removed from the beginning and the end.
*/
public AsciiString trim() {
int start = arrayOffset(), last = arrayOffset() + length() - 1;
int end = last;
while (start <= end && value[start] <= ' ') {
start++;
}
while (end >= start && value[end] <= ' ') {
end--;
}
if (start == 0 && end == last) {
return this;
}
return new AsciiString(value, start, end - start + 1, false);
}
/**
* Compares a {@code CharSequence} to this {@code String} to determine if their contents are equal.
*
* @param a the character sequence to compare to.
* @return {@code true} if equal, otherwise {@code false}
*/
public boolean contentEquals(CharSequence a) {
if (this == a) {
return true;
}
if (a == null || a.length() != length()) {
return false;
}
if (a instanceof AsciiString) {
return equals(a);
}
for (int i = arrayOffset(), j = 0; j < a.length(); ++i, ++j) {
if (b2c(value[i]) != a.charAt(j)) {
return false;
}
}
return true;
}
/**
* Determines whether this string matches a given regular expression.
*
* @param expr the regular expression to be matched.
* @return {@code true} if the expression matches, otherwise {@code false}.
* @throws PatternSyntaxException if the syntax of the supplied regular expression is not valid.
* @throws NullPointerException if {@code expr} is {@code null}.
*/
public boolean matches(String expr) {
return Pattern.matches(expr, this);
}
/**
* Splits this string using the supplied regular expression {@code expr}. The parameter {@code max} controls the
* behavior how many times the pattern is applied to the string.
*
* @param expr the regular expression used to divide the string.
* @param max the number of entries in the resulting array.
* @return an array of Strings created by separating the string along matches of the regular expression.
* @throws NullPointerException if {@code expr} is {@code null}.
* @throws PatternSyntaxException if the syntax of the supplied regular expression is not valid.
* @see Pattern#split(CharSequence, int)
*/
public AsciiString[] split(String expr, int max) {
return toAsciiStringArray(Pattern.compile(expr).split(this, max));
}
/**
* Splits the specified {@link String} with the specified delimiter..
*/
public AsciiString[] split(char delim) {
final List res = InternalThreadLocalMap.get().arrayList();
int start = 0;
final int length = length();
for (int i = start; i < length; i++) {
if (charAt(i) == delim) {
if (start == i) {
res.add(EMPTY_STRING);
} else {
res.add(new AsciiString(value, start + arrayOffset(), i - start, false));
}
start = i + 1;
}
}
if (start == 0) { // If no delimiter was found in the value
res.add(this);
} else {
if (start != length) {
// Add the last element if it's not empty.
res.add(new AsciiString(value, start + arrayOffset(), length - start, false));
} else {
// Truncate trailing empty elements.
for (int i = res.size() - 1; i >= 0; i--) {
if (res.get(i).isEmpty()) {
res.remove(i);
} else {
break;
}
}
}
}
return res.toArray(new AsciiString[0]);
}
/**
* {@inheritDoc}
*
* Provides a case-insensitive hash code for Ascii like byte strings.
*/
@Override
public int hashCode() {
int h = hash;
if (h == 0) {
h = PlatformDependent.hashCodeAscii(value, offset, length);
hash = h;
}
return h;
}
@Override
public boolean equals(Object obj) {
if (obj == null || obj.getClass() != AsciiString.class) {
return false;
}
if (this == obj) {
return true;
}
AsciiString other = (AsciiString) obj;
return length() == other.length() &&
hashCode() == other.hashCode() &&
PlatformDependent.equals(array(), arrayOffset(), other.array(), other.arrayOffset(), length());
}
/**
* Translates the entire byte string to a {@link String}.
* @see #toString(int)
*/
@Override
public String toString() {
String cache = string;
if (cache == null) {
cache = toString(0);
string = cache;
}
return cache;
}
/**
* Translates the entire byte string to a {@link String} using the {@code charset} encoding.
* @see #toString(int, int)
*/
public String toString(int start) {
return toString(start, length());
}
/**
* Translates the [{@code start}, {@code end}) range of this byte string to a {@link String}.
*/
public String toString(int start, int end) {
int length = end - start;
if (length == 0) {
return "";
}
if (isOutOfBounds(start, length, length())) {
throw new IndexOutOfBoundsException("expected: " + "0 <= start(" + start + ") <= srcIdx + length("
+ length + ") <= srcLen(" + length() + ')');
}
@SuppressWarnings("deprecation")
final String str = new String(value, 0, start + offset, length);
return str;
}
public boolean parseBoolean() {
return length >= 1 && value[offset] != 0;
}
public char parseChar() {
return parseChar(0);
}
public char parseChar(int start) {
if (start + 1 >= length()) {
throw new IndexOutOfBoundsException("2 bytes required to convert to character. index " +
start + " would go out of bounds.");
}
final int startWithOffset = start + offset;
return (char) ((b2c(value[startWithOffset]) << 8) | b2c(value[startWithOffset + 1]));
}
public short parseShort() {
return parseShort(0, length(), 10);
}
public short parseShort(int radix) {
return parseShort(0, length(), radix);
}
public short parseShort(int start, int end) {
return parseShort(start, end, 10);
}
public short parseShort(int start, int end, int radix) {
int intValue = parseInt(start, end, radix);
short result = (short) intValue;
if (result != intValue) {
throw new NumberFormatException(subSequence(start, end, false).toString());
}
return result;
}
public int parseInt() {
return parseInt(0, length(), 10);
}
public int parseInt(int radix) {
return parseInt(0, length(), radix);
}
public int parseInt(int start, int end) {
return parseInt(start, end, 10);
}
public int parseInt(int start, int end, int radix) {
if (radix < Character.MIN_RADIX || radix > Character.MAX_RADIX) {
throw new NumberFormatException();
}
if (start == end) {
throw new NumberFormatException();
}
int i = start;
boolean negative = byteAt(i) == '-';
if (negative && ++i == end) {
throw new NumberFormatException(subSequence(start, end, false).toString());
}
return parseInt(i, end, radix, negative);
}
private int parseInt(int start, int end, int radix, boolean negative) {
int max = Integer.MIN_VALUE / radix;
int result = 0;
int currOffset = start;
while (currOffset < end) {
int digit = Character.digit((char) (value[currOffset++ + offset] & 0xFF), radix);
if (digit == -1) {
throw new NumberFormatException(subSequence(start, end, false).toString());
}
if (max > result) {
throw new NumberFormatException(subSequence(start, end, false).toString());
}
int next = result * radix - digit;
if (next > result) {
throw new NumberFormatException(subSequence(start, end, false).toString());
}
result = next;
}
if (!negative) {
result = -result;
if (result < 0) {
throw new NumberFormatException(subSequence(start, end, false).toString());
}
}
return result;
}
public long parseLong() {
return parseLong(0, length(), 10);
}
public long parseLong(int radix) {
return parseLong(0, length(), radix);
}
public long parseLong(int start, int end) {
return parseLong(start, end, 10);
}
public long parseLong(int start, int end, int radix) {
if (radix < Character.MIN_RADIX || radix > Character.MAX_RADIX) {
throw new NumberFormatException();
}
if (start == end) {
throw new NumberFormatException();
}
int i = start;
boolean negative = byteAt(i) == '-';
if (negative && ++i == end) {
throw new NumberFormatException(subSequence(start, end, false).toString());
}
return parseLong(i, end, radix, negative);
}
private long parseLong(int start, int end, int radix, boolean negative) {
long max = Long.MIN_VALUE / radix;
long result = 0;
int currOffset = start;
while (currOffset < end) {
int digit = Character.digit((char) (value[currOffset++ + offset] & 0xFF), radix);
if (digit == -1) {
throw new NumberFormatException(subSequence(start, end, false).toString());
}
if (max > result) {
throw new NumberFormatException(subSequence(start, end, false).toString());
}
long next = result * radix - digit;
if (next > result) {
throw new NumberFormatException(subSequence(start, end, false).toString());
}
result = next;
}
if (!negative) {
result = -result;
if (result < 0) {
throw new NumberFormatException(subSequence(start, end, false).toString());
}
}
return result;
}
public float parseFloat() {
return parseFloat(0, length());
}
public float parseFloat(int start, int end) {
return Float.parseFloat(toString(start, end));
}
public double parseDouble() {
return parseDouble(0, length());
}
public double parseDouble(int start, int end) {
return Double.parseDouble(toString(start, end));
}
public static final HashingStrategy CASE_INSENSITIVE_HASHER =
new HashingStrategy() {
@Override
public int hashCode(CharSequence o) {
return AsciiString.hashCode(o);
}
@Override
public boolean equals(CharSequence a, CharSequence b) {
return AsciiString.contentEqualsIgnoreCase(a, b);
}
};
public static final HashingStrategy CASE_SENSITIVE_HASHER =
new HashingStrategy() {
@Override
public int hashCode(CharSequence o) {
return AsciiString.hashCode(o);
}
@Override
public boolean equals(CharSequence a, CharSequence b) {
return AsciiString.contentEquals(a, b);
}
};
/**
* Returns an {@link AsciiString} containing the given character sequence. If the given string is already a
* {@link AsciiString}, just returns the same instance.
*/
public static AsciiString of(CharSequence string) {
return string instanceof AsciiString ? (AsciiString) string : new AsciiString(string);
}
/**
* Returns an {@link AsciiString} containing the given string and retains/caches the input
* string for later use in {@link #toString()}.
* Used for the constants (which already stored in the JVM's string table) and in cases
* where the guaranteed use of the {@link #toString()} method.
*/
public static AsciiString cached(String string) {
AsciiString asciiString = new AsciiString(string);
asciiString.string = string;
return asciiString;
}
/**
* Returns the case-insensitive hash code of the specified string. Note that this method uses the same hashing
* algorithm with {@link #hashCode()} so that you can put both {@link AsciiString}s and arbitrary
* {@link CharSequence}s into the same headers.
*/
public static int hashCode(CharSequence value) {
if (value == null) {
return 0;
}
if (value instanceof AsciiString) {
return value.hashCode();
}
return PlatformDependent.hashCodeAscii(value);
}
/**
* Determine if {@code a} contains {@code b} in a case sensitive manner.
*/
public static boolean contains(CharSequence a, CharSequence b) {
return contains(a, b, DefaultCharEqualityComparator.INSTANCE);
}
/**
* Determine if {@code a} contains {@code b} in a case insensitive manner.
*/
public static boolean containsIgnoreCase(CharSequence a, CharSequence b) {
return contains(a, b, AsciiCaseInsensitiveCharEqualityComparator.INSTANCE);
}
/**
* Returns {@code true} if both {@link CharSequence}'s are equals when ignore the case. This only supports 8-bit
* ASCII.
*/
public static boolean contentEqualsIgnoreCase(CharSequence a, CharSequence b) {
if (a == null || b == null) {
return a == b;
}
if (a instanceof AsciiString) {
return ((AsciiString) a).contentEqualsIgnoreCase(b);
}
if (b instanceof AsciiString) {
return ((AsciiString) b).contentEqualsIgnoreCase(a);
}
if (a.length() != b.length()) {
return false;
}
for (int i = 0; i < a.length(); ++i) {
if (!equalsIgnoreCase(a.charAt(i), b.charAt(i))) {
return false;
}
}
return true;
}
/**
* Determine if {@code collection} contains {@code value} and using
* {@link #contentEqualsIgnoreCase(CharSequence, CharSequence)} to compare values.
* @param collection The collection to look for and equivalent element as {@code value}.
* @param value The value to look for in {@code collection}.
* @return {@code true} if {@code collection} contains {@code value} according to
* {@link #contentEqualsIgnoreCase(CharSequence, CharSequence)}. {@code false} otherwise.
* @see #contentEqualsIgnoreCase(CharSequence, CharSequence)
*/
public static boolean containsContentEqualsIgnoreCase(Collection collection, CharSequence value) {
for (CharSequence v : collection) {
if (contentEqualsIgnoreCase(value, v)) {
return true;
}
}
return false;
}
/**
* Determine if {@code a} contains all of the values in {@code b} using
* {@link #contentEqualsIgnoreCase(CharSequence, CharSequence)} to compare values.
* @param a The collection under test.
* @param b The values to test for.
* @return {@code true} if {@code a} contains all of the values in {@code b} using
* {@link #contentEqualsIgnoreCase(CharSequence, CharSequence)} to compare values. {@code false} otherwise.
* @see #contentEqualsIgnoreCase(CharSequence, CharSequence)
*/
public static boolean containsAllContentEqualsIgnoreCase(Collection a, Collection b) {
for (CharSequence v : b) {
if (!containsContentEqualsIgnoreCase(a, v)) {
return false;
}
}
return true;
}
/**
* Returns {@code true} if the content of both {@link CharSequence}'s are equals. This only supports 8-bit ASCII.
*/
public static boolean contentEquals(CharSequence a, CharSequence b) {
if (a == null || b == null) {
return a == b;
}
if (a instanceof AsciiString) {
return ((AsciiString) a).contentEquals(b);
}
if (b instanceof AsciiString) {
return ((AsciiString) b).contentEquals(a);
}
if (a.length() != b.length()) {
return false;
}
for (int i = 0; i < a.length(); ++i) {
if (a.charAt(i) != b.charAt(i)) {
return false;
}
}
return true;
}
private static AsciiString[] toAsciiStringArray(String[] jdkResult) {
AsciiString[] res = new AsciiString[jdkResult.length];
for (int i = 0; i < jdkResult.length; i++) {
res[i] = new AsciiString(jdkResult[i]);
}
return res;
}
private interface CharEqualityComparator {
boolean equals(char a, char b);
}
private static final class DefaultCharEqualityComparator implements CharEqualityComparator {
static final DefaultCharEqualityComparator INSTANCE = new DefaultCharEqualityComparator();
private DefaultCharEqualityComparator() { }
@Override
public boolean equals(char a, char b) {
return a == b;
}
}
private static final class AsciiCaseInsensitiveCharEqualityComparator implements CharEqualityComparator {
static final AsciiCaseInsensitiveCharEqualityComparator
INSTANCE = new AsciiCaseInsensitiveCharEqualityComparator();
private AsciiCaseInsensitiveCharEqualityComparator() { }
@Override
public boolean equals(char a, char b) {
return equalsIgnoreCase(a, b);
}
}
private static final class GeneralCaseInsensitiveCharEqualityComparator implements CharEqualityComparator {
static final GeneralCaseInsensitiveCharEqualityComparator
INSTANCE = new GeneralCaseInsensitiveCharEqualityComparator();
private GeneralCaseInsensitiveCharEqualityComparator() { }
@Override
public boolean equals(char a, char b) {
//For motivation, why we need two checks, see comment in String#regionMatches
return Character.toUpperCase(a) == Character.toUpperCase(b) ||
Character.toLowerCase(a) == Character.toLowerCase(b);
}
}
private static boolean contains(CharSequence a, CharSequence b, CharEqualityComparator cmp) {
if (a == null || b == null || a.length() < b.length()) {
return false;
}
if (b.length() == 0) {
return true;
}
int bStart = 0;
for (int i = 0; i < a.length(); ++i) {
if (cmp.equals(b.charAt(bStart), a.charAt(i))) {
// If b is consumed then true.
if (++bStart == b.length()) {
return true;
}
} else if (a.length() - i < b.length()) {
// If there are not enough characters left in a for b to be contained, then false.
return false;
} else {
bStart = 0;
}
}
return false;
}
private static boolean regionMatchesCharSequences(final CharSequence cs, final int csStart,
final CharSequence string, final int start, final int length,
CharEqualityComparator charEqualityComparator) {
//general purpose implementation for CharSequences
if (csStart < 0 || length > cs.length() - csStart) {
return false;
}
if (start < 0 || length > string.length() - start) {
return false;
}
int csIndex = csStart;
int csEnd = csIndex + length;
int stringIndex = start;
while (csIndex < csEnd) {
char c1 = cs.charAt(csIndex++);
char c2 = string.charAt(stringIndex++);
if (!charEqualityComparator.equals(c1, c2)) {
return false;
}
}
return true;
}
/**
* This methods make regionMatches operation correctly for any chars in strings
* @param cs the {@code CharSequence} to be processed
* @param ignoreCase specifies if case should be ignored.
* @param csStart the starting offset in the {@code cs} CharSequence
* @param string the {@code CharSequence} to compare.
* @param start the starting offset in the specified {@code string}.
* @param length the number of characters to compare.
* @return {@code true} if the ranges of characters are equal, {@code false} otherwise.
*/
public static boolean regionMatches(final CharSequence cs, final boolean ignoreCase, final int csStart,
final CharSequence string, final int start, final int length) {
if (cs == null || string == null) {
return false;
}
if (cs instanceof String && string instanceof String) {
return ((String) cs).regionMatches(ignoreCase, csStart, (String) string, start, length);
}
if (cs instanceof AsciiString) {
return ((AsciiString) cs).regionMatches(ignoreCase, csStart, string, start, length);
}
return regionMatchesCharSequences(cs, csStart, string, start, length,
ignoreCase ? GeneralCaseInsensitiveCharEqualityComparator.INSTANCE :
DefaultCharEqualityComparator.INSTANCE);
}
/**
* This is optimized version of regionMatches for string with ASCII chars only
* @param cs the {@code CharSequence} to be processed
* @param ignoreCase specifies if case should be ignored.
* @param csStart the starting offset in the {@code cs} CharSequence
* @param string the {@code CharSequence} to compare.
* @param start the starting offset in the specified {@code string}.
* @param length the number of characters to compare.
* @return {@code true} if the ranges of characters are equal, {@code false} otherwise.
*/
public static boolean regionMatchesAscii(final CharSequence cs, final boolean ignoreCase, final int csStart,
final CharSequence string, final int start, final int length) {
if (cs == null || string == null) {
return false;
}
if (!ignoreCase && cs instanceof String && string instanceof String) {
//we don't call regionMatches from String for ignoreCase==true. It's a general purpose method,
//which make complex comparison in case of ignoreCase==true, which is useless for ASCII-only strings.
//To avoid applying this complex ignore-case comparison, we will use regionMatchesCharSequences
return ((String) cs).regionMatches(false, csStart, (String) string, start, length);
}
if (cs instanceof AsciiString) {
return ((AsciiString) cs).regionMatches(ignoreCase, csStart, string, start, length);
}
return regionMatchesCharSequences(cs, csStart, string, start, length,
ignoreCase ? AsciiCaseInsensitiveCharEqualityComparator.INSTANCE :
DefaultCharEqualityComparator.INSTANCE);
}
/**
* Case in-sensitive find of the first index within a CharSequence
* from the specified position.
*
* A {@code null} CharSequence will return {@code -1}.
* A negative start position is treated as zero.
* An empty ("") search CharSequence always matches.
* A start position greater than the string length only matches
* an empty search CharSequence.
*
*
* AsciiString.indexOfIgnoreCase(null, *, *) = -1
* AsciiString.indexOfIgnoreCase(*, null, *) = -1
* AsciiString.indexOfIgnoreCase("", "", 0) = 0
* AsciiString.indexOfIgnoreCase("aabaabaa", "A", 0) = 0
* AsciiString.indexOfIgnoreCase("aabaabaa", "B", 0) = 2
* AsciiString.indexOfIgnoreCase("aabaabaa", "AB", 0) = 1
* AsciiString.indexOfIgnoreCase("aabaabaa", "B", 3) = 5
* AsciiString.indexOfIgnoreCase("aabaabaa", "B", 9) = -1
* AsciiString.indexOfIgnoreCase("aabaabaa", "B", -1) = 2
* AsciiString.indexOfIgnoreCase("aabaabaa", "", 2) = 2
* AsciiString.indexOfIgnoreCase("abc", "", 9) = -1
*
*
* @param str the CharSequence to check, may be null
* @param searchStr the CharSequence to find, may be null
* @param startPos the start position, negative treated as zero
* @return the first index of the search CharSequence (always ≥ startPos),
* -1 if no match or {@code null} string input
*/
public static int indexOfIgnoreCase(final CharSequence str, final CharSequence searchStr, int startPos) {
if (str == null || searchStr == null) {
return INDEX_NOT_FOUND;
}
if (startPos < 0) {
startPos = 0;
}
int searchStrLen = searchStr.length();
final int endLimit = str.length() - searchStrLen + 1;
if (startPos > endLimit) {
return INDEX_NOT_FOUND;
}
if (searchStrLen == 0) {
return startPos;
}
for (int i = startPos; i < endLimit; i++) {
if (regionMatches(str, true, i, searchStr, 0, searchStrLen)) {
return i;
}
}
return INDEX_NOT_FOUND;
}
/**
* Case in-sensitive find of the first index within a CharSequence
* from the specified position. This method optimized and works correctly for ASCII CharSequences only
*
* A {@code null} CharSequence will return {@code -1}.
* A negative start position is treated as zero.
* An empty ("") search CharSequence always matches.
* A start position greater than the string length only matches
* an empty search CharSequence.
*
*
* AsciiString.indexOfIgnoreCase(null, *, *) = -1
* AsciiString.indexOfIgnoreCase(*, null, *) = -1
* AsciiString.indexOfIgnoreCase("", "", 0) = 0
* AsciiString.indexOfIgnoreCase("aabaabaa", "A", 0) = 0
* AsciiString.indexOfIgnoreCase("aabaabaa", "B", 0) = 2
* AsciiString.indexOfIgnoreCase("aabaabaa", "AB", 0) = 1
* AsciiString.indexOfIgnoreCase("aabaabaa", "B", 3) = 5
* AsciiString.indexOfIgnoreCase("aabaabaa", "B", 9) = -1
* AsciiString.indexOfIgnoreCase("aabaabaa", "B", -1) = 2
* AsciiString.indexOfIgnoreCase("aabaabaa", "", 2) = 2
* AsciiString.indexOfIgnoreCase("abc", "", 9) = -1
*
*
* @param str the CharSequence to check, may be null
* @param searchStr the CharSequence to find, may be null
* @param startPos the start position, negative treated as zero
* @return the first index of the search CharSequence (always ≥ startPos),
* -1 if no match or {@code null} string input
*/
public static int indexOfIgnoreCaseAscii(final CharSequence str, final CharSequence searchStr, int startPos) {
if (str == null || searchStr == null) {
return INDEX_NOT_FOUND;
}
if (startPos < 0) {
startPos = 0;
}
int searchStrLen = searchStr.length();
final int endLimit = str.length() - searchStrLen + 1;
if (startPos > endLimit) {
return INDEX_NOT_FOUND;
}
if (searchStrLen == 0) {
return startPos;
}
for (int i = startPos; i < endLimit; i++) {
if (regionMatchesAscii(str, true, i, searchStr, 0, searchStrLen)) {
return i;
}
}
return INDEX_NOT_FOUND;
}
/**
* Finds the first index in the {@code CharSequence} that matches the
* specified character.
*
* @param cs the {@code CharSequence} to be processed, not null
* @param searchChar the char to be searched for
* @param start the start index, negative starts at the string start
* @return the index where the search char was found,
* -1 if char {@code searchChar} is not found or {@code cs == null}
*/
//-----------------------------------------------------------------------
public static int indexOf(final CharSequence cs, final char searchChar, int start) {
if (cs instanceof String) {
return ((String) cs).indexOf(searchChar, start);
} else if (cs instanceof AsciiString) {
return ((AsciiString) cs).indexOf(searchChar, start);
}
if (cs == null) {
return INDEX_NOT_FOUND;
}
final int sz = cs.length();
for (int i = start < 0 ? 0 : start; i < sz; i++) {
if (cs.charAt(i) == searchChar) {
return i;
}
}
return INDEX_NOT_FOUND;
}
private static boolean equalsIgnoreCase(byte a, byte b) {
return a == b || toLowerCase(a) == toLowerCase(b);
}
private static boolean equalsIgnoreCase(char a, char b) {
return a == b || toLowerCase(a) == toLowerCase(b);
}
private static byte toLowerCase(byte b) {
return isUpperCase(b) ? (byte) (b + 32) : b;
}
/**
* If the character is uppercase - converts the character to lowercase,
* otherwise returns the character as it is. Only for ASCII characters.
*
* @return lowercase ASCII character equivalent
*/
public static char toLowerCase(char c) {
return isUpperCase(c) ? (char) (c + 32) : c;
}
private static byte toUpperCase(byte b) {
return isLowerCase(b) ? (byte) (b - 32) : b;
}
private static boolean isLowerCase(byte value) {
return value >= 'a' && value <= 'z';
}
public static boolean isUpperCase(byte value) {
return value >= 'A' && value <= 'Z';
}
public static boolean isUpperCase(char value) {
return value >= 'A' && value <= 'Z';
}
public static byte c2b(char c) {
return (byte) ((c > MAX_CHAR_VALUE) ? '?' : c);
}
private static byte c2b0(char c) {
return (byte) c;
}
public static char b2c(byte b) {
return (char) (b & 0xFF);
}
}