config.java

HTML解析器是一个Java库

	 * The three types of character reference are:
	 * <ul>
	 *  <li>{@linkplain CharacterEntityReference Character entity reference}
	 *  <li><a href="NumericCharacterReference.html#DecimalCharacterReference">Decimal character reference</a>
	 *  <li><a href="NumericCharacterReference.html#HexadecimalCharacterReference">Hexadecimal character reference</a>
	 * </ul>
	 * <p>
	 * The two types of contexts used in this library are:
	 * <ul>
	 *  <li>Inside an attribute value
	 *  <li>Outside an attribute value
	 * </ul>
	 */ 
	static class UnterminatedCharacterReferenceSettings {
		// use volatile fields to make them thread safe
		public volatile int characterEntityReferenceMaxCodePoint;
		public volatile int decimalCharacterReferenceMaxCodePoint;
		public volatile int hexadecimalCharacterReferenceMaxCodePoint;

		public static UnterminatedCharacterReferenceSettings ACCEPT_ALL=new UnterminatedCharacterReferenceSettings(CompatibilityMode.CODE_POINTS_ALL,CompatibilityMode.CODE_POINTS_ALL,CompatibilityMode.CODE_POINTS_ALL);

		public UnterminatedCharacterReferenceSettings() {
			this(CompatibilityMode.CODE_POINTS_NONE,CompatibilityMode.CODE_POINTS_NONE,CompatibilityMode.CODE_POINTS_NONE);
		}

		public UnterminatedCharacterReferenceSettings(final int characterEntityReferenceMaxCodePoint, final int decimalCharacterReferenceMaxCodePoint, final int hexadecimalCharacterReferenceMaxCodePoint) {
			this.characterEntityReferenceMaxCodePoint=characterEntityReferenceMaxCodePoint;
			this.decimalCharacterReferenceMaxCodePoint=decimalCharacterReferenceMaxCodePoint;
			this.hexadecimalCharacterReferenceMaxCodePoint=hexadecimalCharacterReferenceMaxCodePoint;
		}

		public String toString() {
			return Config.NewLine+"    Character entity reference: "+getDescription(characterEntityReferenceMaxCodePoint)
						+Config.NewLine+"    Decimal character reference: "+getDescription(decimalCharacterReferenceMaxCodePoint)
						+Config.NewLine+"    Haxadecimal character reference: "+getDescription(hexadecimalCharacterReferenceMaxCodePoint);
		}

		private String getDescription(final int codePoint) {
			if (codePoint==CompatibilityMode.CODE_POINTS_NONE) return "None";
			if (codePoint==CompatibilityMode.CODE_POINTS_ALL) return "All";
			return "0x"+Integer.toString(codePoint,16);
		}
	}

	/**
	 * Represents a set of configuration parameters that relate to
	 * <a target="_blank" href="http://www.w3.org/TR/html401/conform.html#didx-user_agent">user agent</a> compatibility issues.
	 * <p>
	 * The predefined compatibility modes {@link #IE}, {@link #MOZILLA}, {@link #OPERA} and {@link #XHTML} provide an easy means of
	 * ensuring the library interprets the markup in a way consistent with some of the most commonly used browsers,
	 * at least in relation to the behaviour described by the properties in this class.
	 * <p>
	 * The properties of any <code>CompatibilityMode</code> object can be modified individually, including those in
	 * the predefined instances as well as newly constructed instances.
	 * Take note however that modifying the properties of the predefined instances has a global affect.
	 * <p>
	 * The currently active compatibility mode is stored in the static {@link Config#CurrentCompatibilityMode} property.
	 * <p>
	 */
	public static final class CompatibilityMode {
		private String name;
		private volatile boolean formFieldNameCaseInsensitive;
		volatile UnterminatedCharacterReferenceSettings unterminatedCharacterReferenceSettingsInsideAttributeValue;
		volatile UnterminatedCharacterReferenceSettings unterminatedCharacterReferenceSettingsOutsideAttributeValue;

		/**
		 * Indicates the recognition of all unicode code points.
		 * <p>
		 * This value is used in properties which specify a maximum unicode code point to be recognised by the parser.
		 *
		 * @see #getUnterminatedCharacterEntityReferenceMaxCodePoint(boolean insideAttributeValue)
		 * @see #getUnterminatedDecimalCharacterReferenceMaxCodePoint(boolean insideAttributeValue)
		 * @see #getUnterminatedHexadecimalCharacterReferenceMaxCodePoint(boolean insideAttributeValue)
		 */
		public static final int CODE_POINTS_ALL=Character.MAX_CODE_POINT; // 0x10FFFF (decimal 1114111)

		/**
		 * Indicates the recognition of no unicode code points.
		 * <p>
		 * This value is used in properties which specify a maximum unicode code point to be recognised by the parser.
		 *
		 * @see #getUnterminatedCharacterEntityReferenceMaxCodePoint(boolean insideAttributeValue)
		 * @see #getUnterminatedDecimalCharacterReferenceMaxCodePoint(boolean insideAttributeValue)
		 * @see #getUnterminatedHexadecimalCharacterReferenceMaxCodePoint(boolean insideAttributeValue)
		 */
		public static final int CODE_POINTS_NONE=CharacterReference.INVALID_CODE_POINT;

		/**
		 * <a target="_blank" href="http://www.microsoft.com/windows/ie/">Microsoft Internet Explorer</a> compatibility mode.
		 * <p>
		 * <code>{@link #getName() Name} = IE</code><br />
		 * <code>{@link #isFormFieldNameCaseInsensitive() FormFieldNameCaseInsensitive} = true</code><br />
		 * <table cellspacing="0" cellpadding="0">
		 *  <tr><th>Recognition of unterminated character references:<th><th align="center">&nbsp; (inside attribute) &nbsp;<th align="center">&nbsp; (outside attribute) &nbsp;
		 *  <tr><td>{@link #getUnterminatedCharacterEntityReferenceMaxCodePoint(boolean) UnterminatedCharacterEntityReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">U+00FF<td align="center">U+00FF
		 *  <tr><td>{@link #getUnterminatedDecimalCharacterReferenceMaxCodePoint(boolean) UnterminatedDecimalCharacterReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">{@linkplain #CODE_POINTS_ALL All}<td align="center">{@linkplain #CODE_POINTS_ALL All}
		 *  <tr><td>{@link #getUnterminatedHexadecimalCharacterReferenceMaxCodePoint(boolean) UnterminatedHexadecimalCharacterReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">{@linkplain #CODE_POINTS_ALL All}<td align="center">{@linkplain #CODE_POINTS_NONE None}
		 * </table>		 
		 */
		public static final CompatibilityMode IE=new CompatibilityMode("IE",true,
			new UnterminatedCharacterReferenceSettings(0xFF, CODE_POINTS_ALL, CODE_POINTS_ALL), // inside attributes
			new UnterminatedCharacterReferenceSettings(0xFF, CODE_POINTS_ALL, CODE_POINTS_NONE) // outside attributes
		);

		/**
		 * <a target="_blank" href="http://www.mozilla.org/products/mozilla1.x/">Mozilla</a> / 
		 * <a target="_blank" href="http://www.mozilla.org/products/firefox/">Firefox</a> /
		 * <a target="_blank" href="http://browser.netscape.com/">Netscape</a> compatibility mode.
		 * <p>
		 * <code>{@link #getName() Name} = Mozilla</code><br />
		 * <code>{@link #isFormFieldNameCaseInsensitive() FormFieldNameCaseInsensitive} = false</code><br />
		 * <table cellspacing="0" cellpadding="0">
		 *  <tr><th>Recognition of unterminated character references:<th><th align="center">&nbsp; (inside attribute) &nbsp;<th align="center">&nbsp; (outside attribute) &nbsp;
		 *  <tr><td>{@link #getUnterminatedCharacterEntityReferenceMaxCodePoint(boolean) UnterminatedCharacterEntityReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">U+00FF<td align="center">{@linkplain #CODE_POINTS_ALL All}
		 *  <tr><td>{@link #getUnterminatedDecimalCharacterReferenceMaxCodePoint(boolean) UnterminatedDecimalCharacterReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">{@linkplain #CODE_POINTS_ALL All}<td align="center">{@linkplain #CODE_POINTS_ALL All}
		 *  <tr><td>{@link #getUnterminatedHexadecimalCharacterReferenceMaxCodePoint(boolean) UnterminatedHexadecimalCharacterReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">{@linkplain #CODE_POINTS_ALL All}<td align="center">{@linkplain #CODE_POINTS_ALL All}
		 * </table>		 
		 */
		public static final CompatibilityMode MOZILLA=new CompatibilityMode("Mozilla",false,
			new UnterminatedCharacterReferenceSettings(0xFF,            CODE_POINTS_ALL, CODE_POINTS_ALL), // inside attributes
			new UnterminatedCharacterReferenceSettings(CODE_POINTS_ALL, CODE_POINTS_ALL, CODE_POINTS_ALL) // outside attributes
		);

		/**
		 * Opera compatibility mode.
		 * <p>
		 * <code>{@link #getName() Name} = Opera</code><br />
		 * <code>{@link #isFormFieldNameCaseInsensitive() FormFieldNameCaseInsensitive} = true</code><br />
		 * <table cellspacing="0" cellpadding="0">
		 *  <tr><th>Recognition of unterminated character references:<th><th align="center">&nbsp; (inside attribute) &nbsp;<th align="center">&nbsp; (outside attribute) &nbsp;
		 *  <tr><td>{@link #getUnterminatedCharacterEntityReferenceMaxCodePoint(boolean) UnterminatedCharacterEntityReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">U+003E<td align="center">{@linkplain #CODE_POINTS_ALL All}
		 *  <tr><td>{@link #getUnterminatedDecimalCharacterReferenceMaxCodePoint(boolean) UnterminatedDecimalCharacterReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">{@linkplain #CODE_POINTS_ALL All}<td align="center">{@linkplain #CODE_POINTS_ALL All}
		 *  <tr><td>{@link #getUnterminatedHexadecimalCharacterReferenceMaxCodePoint(boolean) UnterminatedHexadecimalCharacterReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">{@linkplain #CODE_POINTS_ALL All}<td align="center">{@linkplain #CODE_POINTS_ALL All}
		 * </table>		 
		 */
		public static final CompatibilityMode OPERA=new CompatibilityMode("Opera",true,
			new UnterminatedCharacterReferenceSettings(0x3E,            CODE_POINTS_ALL, CODE_POINTS_ALL), // inside attributes
			new UnterminatedCharacterReferenceSettings(CODE_POINTS_ALL, CODE_POINTS_ALL, CODE_POINTS_ALL) // outside attributes
		);

		/**
		 * <a target="_blank" href="http://www.w3.org/TR/xhtml1/#xhtml">XHTML</a> compatibility mode.
		 * <p>
		 * <code>{@link #getName() Name} = XHTML</code><br />
		 * <code>{@link #isFormFieldNameCaseInsensitive() FormFieldNameCaseInsensitive} = false</code><br />
		 * <table cellspacing="0" cellpadding="0">
		 *  <tr><th>Recognition of unterminated character references:<th><th align="center">&nbsp; (inside attribute) &nbsp;<th align="center">&nbsp; (outside attribute) &nbsp;
		 *  <tr><td>{@link #getUnterminatedCharacterEntityReferenceMaxCodePoint(boolean) UnterminatedCharacterEntityReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">{@linkplain #CODE_POINTS_NONE None}<td align="center">{@linkplain #CODE_POINTS_NONE None}
		 *  <tr><td>{@link #getUnterminatedDecimalCharacterReferenceMaxCodePoint(boolean) UnterminatedDecimalCharacterReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">{@linkplain #CODE_POINTS_NONE None}<td align="center">{@linkplain #CODE_POINTS_NONE None}
		 *  <tr><td>{@link #getUnterminatedHexadecimalCharacterReferenceMaxCodePoint(boolean) UnterminatedHexadecimalCharacterReferenceMaxCodePoint}<td><code>&nbsp;=</code><td align="center">{@linkplain #CODE_POINTS_NONE None}<td align="center">{@linkplain #CODE_POINTS_NONE None}
		 * </table>		 
		 */
		public static final CompatibilityMode XHTML=new CompatibilityMode("XHTML");

		/**
		 * Constructs a new <code>CompatibilityMode</code> with the given {@linkplain #getName() name}.
		 * <p>
		 * All properties in the new instance are initially assigned their default values, which are the same as the strict
		 * rules of the {@link #XHTML} compatibility mode.
		 *
		 * @param name  the {@linkplain #getName() name} of the new compatibility mode
		 */
		public CompatibilityMode(final String name) {
			this(name,false,new UnterminatedCharacterReferenceSettings(),new UnterminatedCharacterReferenceSettings());
		}

		private CompatibilityMode(final String name, final boolean formFieldNameCaseInsensitive, final UnterminatedCharacterReferenceSettings unterminatedCharacterReferenceSettingsInsideAttributeValue, final UnterminatedCharacterReferenceSettings unterminatedCharacterReferenceSettingsOutsideAttributeValue) {
			this.name=name;
			this.formFieldNameCaseInsensitive=formFieldNameCaseInsensitive;
			this.unterminatedCharacterReferenceSettingsInsideAttributeValue=unterminatedCharacterReferenceSettingsInsideAttributeValue;
			this.unterminatedCharacterReferenceSettingsOutsideAttributeValue=unterminatedCharacterReferenceSettingsOutsideAttributeValue;
		}

		/**
		 * Returns the name of this compatibility mode.
		 * @return the name of this compatibility mode.
		 */
		public String getName() {
			return name;