Java Code Examples for org.apache.flink.types.StringValue#length()

/**
 * Converts the given <code>StringValue</code> into a lower case variant.
 *
 * @param string The string to convert to lower case.
 */
public static void toLowerCase(StringValue string) {
	final char[] chars = string.getCharArray();
	final int len = string.length();

	for (int i = 0; i < len; i++) {
		chars[i] = Character.toLowerCase(chars[i]);
	}
}
 

/**
 * Replaces all non-word characters in a string by a given character. The only
 * characters not replaced are the characters that qualify as word characters
 * or digit characters with respect to {@link Character#isLetter(char)} or
 * {@link Character#isDigit(char)}, as well as the underscore character.
 *
 * <p>This operation is intended to simplify strings for counting distinct words.
 *
 * @param string The string value to have the non-word characters replaced.
 * @param replacement The character to use as the replacement.
 */
public static void replaceNonWordChars(StringValue string, char replacement) {
	final char[] chars = string.getCharArray();
	final int len = string.length();

	for (int i = 0; i < len; i++) {
		final char c = chars[i];
		if (!(Character.isLetter(c) || Character.isDigit(c) || c == '_')) {
			chars[i] = replacement;
		}
	}
}
 

/**
 * Converts the given <code>StringValue</code> into a lower case variant.
 *
 * @param string The string to convert to lower case.
 */
public static void toLowerCase(StringValue string) {
	final char[] chars = string.getCharArray();
	final int len = string.length();

	for (int i = 0; i < len; i++) {
		chars[i] = Character.toLowerCase(chars[i]);
	}
}
 

/**
 * Replaces all non-word characters in a string by a given character. The only
 * characters not replaced are the characters that qualify as word characters
 * or digit characters with respect to {@link Character#isLetter(char)} or
 * {@link Character#isDigit(char)}, as well as the underscore character.
 *
 * <p>This operation is intended to simplify strings for counting distinct words.
 *
 * @param string The string value to have the non-word characters replaced.
 * @param replacement The character to use as the replacement.
 */
public static void replaceNonWordChars(StringValue string, char replacement) {
	final char[] chars = string.getCharArray();
	final int len = string.length();

	for (int i = 0; i < len; i++) {
		final char c = chars[i];
		if (!(Character.isLetter(c) || Character.isDigit(c) || c == '_')) {
			chars[i] = replacement;
		}
	}
}
 

/**
 * Converts the given <code>StringValue</code> into a lower case variant.
 *
 * @param string The string to convert to lower case.
 */
public static void toLowerCase(StringValue string) {
	final char[] chars = string.getCharArray();
	final int len = string.length();

	for (int i = 0; i < len; i++) {
		chars[i] = Character.toLowerCase(chars[i]);
	}
}
 

/**
 * Replaces all non-word characters in a string by a given character. The only
 * characters not replaced are the characters that qualify as word characters
 * or digit characters with respect to {@link Character#isLetter(char)} or
 * {@link Character#isDigit(char)}, as well as the underscore character.
 *
 * <p>This operation is intended to simplify strings for counting distinct words.
 *
 * @param string The string value to have the non-word characters replaced.
 * @param replacement The character to use as the replacement.
 */
public static void replaceNonWordChars(StringValue string, char replacement) {
	final char[] chars = string.getCharArray();
	final int len = string.length();

	for (int i = 0; i < len; i++) {
		final char c = chars[i];
		if (!(Character.isLetter(c) || Character.isDigit(c) || c == '_')) {
			chars[i] = replacement;
		}
	}
}
 

/**
 * Sets the string to be tokenized and resets the state of the tokenizer.
 *
 * @param string The string value to be tokenized.
 */
public void setStringToTokenize(StringValue string) {
	this.toTokenize = string;
	this.pos = 0;
	this.limit = string.length();
}
 

@Override
public boolean add(StringValue value) {
	if (isBounded && position >= boundedLength) {
		return false;
	}

	// up to five bytes storing length
	if (position + 5 > data.length) {
		ensureCapacity(position + 5);
	}

	// update local variable until serialization succeeds
	int newPosition = position;

	// write the length, variable-length encoded
	int len = value.length();

	while (len >= HIGH_BIT) {
		data[newPosition++] = (byte) (len | HIGH_BIT);
		len >>>= 7;
	}
	data[newPosition++] = (byte) len;

	// write the char data, variable-length encoded
	final char[] valueData = value.getCharArray();
	int remainingCapacity = data.length - newPosition;

	len = value.length();
	for (int i = 0; i < len; i++) {
		// up to three bytes storing length
		if (remainingCapacity < 3) {
			ensureCapacity(remainingCapacity + 3);
			remainingCapacity = data.length - newPosition;
		}

		int c = valueData[i];

		while (c >= HIGH_BIT) {
			data[newPosition++] = (byte) (c | HIGH_BIT);
			remainingCapacity--;
			c >>>= 7;
		}
		data[newPosition++] = (byte) c;
		remainingCapacity--;
	}

	length++;
	position = newPosition;

	return true;
}
 

/**
 * Sets the string to be tokenized and resets the state of the tokenizer.
 *
 * @param string The string value to be tokenized.
 */
public void setStringToTokenize(StringValue string) {
	this.toTokenize = string;
	this.pos = 0;
	this.limit = string.length();
}
 

@Override
public boolean add(StringValue value) {
	if (isBounded && position >= boundedLength) {
		return false;
	}

	// up to five bytes storing length
	if (position + 5 > data.length) {
		ensureCapacity(position + 5);
	}

	// update local variable until serialization succeeds
	int newPosition = position;

	// write the length, variable-length encoded
	int len = value.length();

	while (len >= HIGH_BIT) {
		data[newPosition++] = (byte) (len | HIGH_BIT);
		len >>>= 7;
	}
	data[newPosition++] = (byte) len;

	// write the char data, variable-length encoded
	final char[] valueData = value.getCharArray();
	int remainingCapacity = data.length - newPosition;

	len = value.length();
	for (int i = 0; i < len; i++) {
		// up to three bytes storing length
		if (remainingCapacity < 3) {
			ensureCapacity(remainingCapacity + 3);
			remainingCapacity = data.length - newPosition;
		}

		int c = valueData[i];

		while (c >= HIGH_BIT) {
			data[newPosition++] = (byte) (c | HIGH_BIT);
			remainingCapacity--;
			c >>>= 7;
		}
		data[newPosition++] = (byte) c;
		remainingCapacity--;
	}

	length++;
	position = newPosition;

	return true;
}
 

/**
 * Sets the string to be tokenized and resets the state of the tokenizer.
 *
 * @param string The string value to be tokenized.
 */
public void setStringToTokenize(StringValue string) {
	this.toTokenize = string;
	this.pos = 0;
	this.limit = string.length();
}
 

@Override
public boolean add(StringValue value) {
	if (isBounded && position >= boundedLength) {
		return false;
	}

	// up to five bytes storing length
	if (position + 5 > data.length) {
		ensureCapacity(position + 5);
	}

	// update local variable until serialization succeeds
	int newPosition = position;

	// write the length, variable-length encoded
	int len = value.length();

	while (len >= HIGH_BIT) {
		data[newPosition++] = (byte) (len | HIGH_BIT);
		len >>>= 7;
	}
	data[newPosition++] = (byte) len;

	// write the char data, variable-length encoded
	final char[] valueData = value.getCharArray();
	int remainingCapacity = data.length - newPosition;

	len = value.length();
	for (int i = 0; i < len; i++) {
		// up to three bytes storing length
		if (remainingCapacity < 3) {
			ensureCapacity(remainingCapacity + 3);
			remainingCapacity = data.length - newPosition;
		}

		int c = valueData[i];

		while (c >= HIGH_BIT) {
			data[newPosition++] = (byte) (c | HIGH_BIT);
			remainingCapacity--;
			c >>>= 7;
		}
		data[newPosition++] = (byte) c;
		remainingCapacity--;
	}

	length++;
	position = newPosition;

	return true;
}