/*
 * @(#)View.java	1.70 04/05/18
 *
 * Copyright 2004 Sun Microsystems, Inc. All rights reserved.
 * SUN PROPRIETARY/CONFIDENTIAL. Use is subject to license terms.
 */
package javax.swing.text;

import java.awt.*;
import javax.swing.SwingConstants;
import javax.swing.event.*;

/**
 * <p>
 * A very important part of the text package is the <code>View</code> class.
 * As the name suggests it represents a view of the text model,
 * or a piece of the text model. 
 * It is this class that is responsible for the look of the text component.  
 * The view is not intended to be some completely new thing that one must
 * learn, but rather is much like a lightweight component.
 * In fact, the original <code>View</code> implementation was a
 * lightweight component.   There were several reasons why
 * the <code>Component</code> implementation was abandoned in favor of
 * an alternative.
 * <ol>
 * <li>
 * <p>
 * There was barely time to get the lightweight component support in the 
 * 1.1 version of the JDK.  There simply wasn't time to lighten up the 
 * component further to where it would need to be to be used for text purposes.
 * The additions made to <code>JComponent</code> increased the memory
 * consumption, and as it currently stands its much too heavy for
 * representing text.
 * </p>
 * <li>
 * <p>
 * The layout semantics aren't quite right for text,
 * and changing the current layout 
 * semantics of component might break existing applications.
 * </p>
 * <li>
 * <p>
 * The component api uses integers, but in 1.2 one can use floating point
 * device independent coordinates.  An api that works in both 1.1 and 1.2
 * would be convenient for minimizing transition difficulties.  The
 * <code>View</code> class uses the <code>Shape</code> interface and
 * float arguments to enable View implementations for the Java 2 platform
 * v1.2 and later while still functioning in the older 1.1 JDK.
 * </p>
 * </ol>
 * <p>
By default, a view is very light.  It contains a reference to the parent 
view from which it can fetch many things without holding state, and it 
contains a reference to a portion of the model (</code>Element</code>).
A view does not 
have to exactly represent an element in the model, that is simply a typical 
and therefore convenient mapping.  A view can alternatively maintain a couple 
of Position objects to maintain its location in the model (i.e. represent 
a fragment of an element).  This is typically the result of formatting where 
views have been broken down into pieces.  The convenience of a substantial 
relationship to the element makes it easier to build factories to produce the 
views, and makes it easier  to keep track of the view pieces as the model is 
changed and the view must be changed to reflect the model.  Simple views 
therefore represent an Element directly and complex views do not.
<p>
A view has the following responsibilities:
  <dl>

    <dt><b>Participate in layout.</b>
    <dd>
    <p>The view has a <code>setSize</code> method which is like
    <code>doLayout</code> and <code>setSize</code> in <code>Component</code> combined.
    The view has a <code>preferenceChanged</code> method which is 
    like <code>invalidate</code> in <code>Component</code> except that one can
    invalidate just one axis 
    and the child requesting the change is identified.
    <p>A View expresses the size that it would like to be in terms of three
    values, a minimum, a preferred, and a maximum span.  Layout in a view is
    can be done independently upon each axis.  For a properly functioning View
    implementation, the minimum span will be <= the preferred span which in turn
    will be <= the maximum span.
    </p>
    <p align=center><img src="doc-files/View-flexibility.jpg" 
                     alt="The above text describes this graphic.">
    <p>The minimum set of methods for layout are:
    <ul>
    <li><a href="#getMinimumSpan(int)">getMinimumSpan</a>
    <li><a href="#getPreferredSpan(int)">getPreferredSpan</a>
    <li><a href="#getMaximumSpan(int)">getMaximumSpan</a>
    <li><a href="#getAlignment(int)">getAlignment</a>
    <li><a href="#preferenceChanged(javax.swing.text.View, boolean, boolean)">preferenceChanged</a>
    <li><a href="#setSize(float, float)">setSize</a>
    </ul>
  
  <p>The <code>setSize</code> method should be prepared to be called a number of times
    (i.e. It may be called even if the size didn't change). 
    The <code>setSize</code> method
    is generally called to make sure the View layout is complete prior to trying
    to perform an operation on it that requires an up-to-date layout.  A view's
    size should <em>always</em> be set to a value within the minimum and maximum
    span specified by that view.  Additionally, the view must always call the
    <code>preferenceChanged</code> method on the parent if it has changed the
    values for the
    layout it would like, and expects the parent to honor.  The parent View is
    not required to recognize a change until the <code>preferenceChanged</code>
    has been sent.
    This allows parent View implementations to cache the child requirements if
    desired.  The calling sequence looks something like the following:
    </p>
    <p align=center>
      <img src="doc-files/View-layout.jpg" 
       alt="Sample calling sequence between parent view and child view: 
       setSize, getMinimum, getPreferred, getMaximum, getAlignment, setSize">
    <p>The exact calling sequence is up to the layout functionality of
    the parent view (if the view has any children).  The view may collect
    the preferences of the children prior to determining what it will give 
    each child, or it might iteratively update the children one at a time.
    </p>

    <dt><b>Render a portion of the model.</b>
    <dd>
    <p>This is done in the paint method, which is pretty much like a component 
    paint method.  Views are expected to potentially populate a fairly large 
    tree.  A <code>View</code> has the following semantics for rendering:
    </p>
    <ul>
    <li>The view gets its allocation from the parent at paint time, so it 
    must be prepared to redo layout if the allocated area is different from 
    what it is prepared to deal with.
    <li>The coordinate system is the same as the hosting <code>Component</code>
    (i.e. the <code>Component</code> returned by the
    <a href="#getContainer">getContainer</a> method).
    This means a child view lives in the same coordinate system as the parent
    view unless the parent has explicitly changed the coordinate system.
    To schedule itself to be repainted a view can call repaint on the hosting
    <code>Component</code>.
    <li>The default is to <em>not clip</em> the children.  It is more efficient
    to allow a view to clip only if it really feels it needs clipping.
    <li>The <code>Graphics</code> object given is not initialized in any way. 
    A view should set any settings needed.
    <li>A <code>View</code> is inherently transparent.  While a view may render into its
    entire allocation, typically a view does not.  Rendering is performed by
    tranversing down the tree of <code>View</code> implementations. 
    Each <code>View</code> is responsible
    for rendering its children.  This behavior is depended upon for thread
    safety.  While view implementations do not necessarily have to be implemented
    with thread safety in mind, other view implementations that do make use of
    concurrency can depend upon a tree traversal to guarantee thread safety.
    <li>The order of views relative to the model is up to the implementation.
    Although child views will typically be arranged in the same order that they
    occur in the model, they may be visually arranged in an entirely different 
    order.  View implementations may have Z-Order associated with them if the
    children are overlapping.
    </ul>
    <p>The methods for rendering are:
    <ul>
    <li><a href="#paint(java.awt.Graphics, java.awt.Shape)">paint</a>
    </ul>
    <p>

    <dt><b>Translate between the model and view coordinate systems.</b>
    <dd>
    <p>Because the view objects are produced from a factory and therefore cannot 
    necessarily be counted upon to be in a particular pattern, one must be able 
    to perform translation to properly locate spatial representation of the model.  
    The methods for doing this are:
    <ul>
    <li><a href="#modelToView(int, javax.swing.text.Position.Bias, int, javax.swing.text.Position.Bias, java.awt.Shape)">modelToView</a>
    <li><a href="#viewToModel(float, float, java.awt.Shape, javax.swing.text.Position.Bias[])">viewToModel</a>
    <li><a href="#getDocument()">getDocument</a>
    <li><a href="#getElement()">getElement</a>
    <li><a href="#getStartOffset()">getStartOffset</a>
    <li><a href="#getEndOffset()">getEndOffset</a>
    </ul>
    <p>The layout must be valid prior to attempting to make the translation.
    The translation is not valid, and must not be attempted while changes
    are being broadcasted from the model via a <code>DocumentEvent</code>.  
    </p>

    <dt><b>Respond to changes from the model.</b>
    <dd>
    <p>If the overall view is represented by many pieces (which is the best situation 
    if one want to be able to change the view and write the least amount of new code), 
    it would be impractical to have a huge number of <code>DocumentListener</code>s.
    If each 
    view listened to the model, only a few would actually be interested in the 
    changes broadcasted at any given time.   Since the model has no knowledge of 
    views, it has no way to filter the broadcast of change information.  The view 
    hierarchy itself is instead responsible for propagating the change information.  
    At any level in the view hierarchy, that view knows enough about its children to 
    best distribute the change information further.   Changes are therefore broadcasted 
    starting from the root of the view hierarchy.
    The methods for doing this are:
    <ul>
    <li><a href="#insertUpdate">insertUpdate</a>
    <li><a href="#removeUpdate">removeUpdate</a>
    <li><a href="#changedUpdate">changedUpdate</a>
    </ul>    
    <p>
</dl>
 *
 * @author  Timothy Prinzing
 * @version 1.70 05/18/04
 */
public abstract class View implements SwingConstants {

    /**
     * Creates a new <code>View</code> object.
     *
     * @param elem the <code>Element</code> to represent
     */
    public View(Element elem) {
	this.elem = elem;
    }

    /**
     * Returns the parent of the view.
     *
     * @return the parent, or <code>null</code> if none exists
     */
    public View getParent() {
	return parent;
    }

    /**
     *  Returns a boolean that indicates whether
     *  the view is visible or not.  By default
     *  all views are visible.
     *
     *  @return always returns true
     */
    public boolean isVisible() {
	return true;
    }

	
    /**
     * Determines the preferred span for this view along an
     * axis.
     *
     * @param axis may be either <code>View.X_AXIS</code> or
     *		<code>View.Y_AXIS</code>
     * @return   the span the view would like to be rendered into.
     *           Typically the view is told to render into the span
     *           that is returned, although there is no guarantee.  
     *           The parent may choose to resize or break the view
     * @see View#getPreferredSpan
     */
    public abstract float getPreferredSpan(int axis);

    /**
     * Determines the minimum span for this view along an
     * axis.
     *
     * @param axis may be either <code>View.X_AXIS</code> or
     *		<code>View.Y_AXIS</code>
     * @return  the minimum span the view can be rendered into
     * @see View#getPreferredSpan
     */
    public float getMinimumSpan(int axis) {
	int w = getResizeWeight(axis);
	if (w == 0) {
	    // can't resize
	    return getPreferredSpan(axis);
	}
	return 0;
    }

    /**
     * Determines the maximum span for this view along an
     * axis.
     *
     * @param axis may be either <code>View.X_AXIS</code> or
     *		<code>View.Y_AXIS</code>
     * @return  the maximum span the view can be rendered into
     * @see View#getPreferredSpan
     */
    public float getMaximumSpan(int axis) {
	int w = getResizeWeight(axis);
	if (w == 0) {
	    // can't resize
	    return getPreferredSpan(axis);
	}
	return Integer.MAX_VALUE;
    }
	
    /**
     * Child views can call this on the parent to indicate that
     * the preference has changed and should be reconsidered
     * for layout.  By default this just propagates upward to 
     * the next parent.  The root view will call 
     * <code>revalidate</code> on the associated text component.
     *
     * @param child the child view
     * @param width true if the width preference has changed
     * @param height true if the height preference has changed
     * @see javax.swing.JComponent#revalidate
     */
    public void preferenceChanged(View child, boolean width, boolean height) {
	View parent = getParent();
	if (parent != null) {
	    parent.preferenceChanged(this, width, height);
	}
    }

    /**
     * Determines the desired alignment for this view along an
     * axis.  The desired alignment is returned.  This should be
     * a value >= 0.0 and <= 1.0, where 0 indicates alignment at
     * the origin and 1.0 indicates alignment to the full span
     * away from the origin.  An alignment of 0.5 would be the
     * center of the view.
     *
     * @param axis may be either <code>View.X_AXIS</code> or
     *		<code>View.Y_AXIS</code>
     * @return the value 0.5
     */
    public float getAlignment(int axis) {
	return 0.5f;
    }

    /**
     * Renders using the given rendering surface and area on that
     * surface.  The view may need to do layout and create child
     * views to enable itself to render into the given allocation.
     *
     * @param g the rendering surface to use
     * @param allocation the allocated region to render into
     * @see View#paint
     */
    public abstract void paint(Graphics g, Shape allocation);

    /**
     * Establishes the parent view for this view.  This is
     * guaranteed to be called before any other methods if the
     * parent view is functioning properly.  This is also
     * the last method called, since it is called to indicate
     * the view has been removed from the hierarchy as 
     * well. When this method is called to set the parent to
     * null, this method does the same for each of its children,
     * propogating the notification that they have been
     * disconnected from the view tree. If this is
     * reimplemented, <code>super.setParent()</code> should
     * be called.
     *
     * @param parent the new parent, or <code>null</code> if the view is
     * 		being removed from a parent
     */
    public void setParent(View parent) {
        // if the parent is null then propogate down the view tree
        if (parent == null) {
            for (int i = 0; i < getViewCount(); i++) {
		if (getView(i).getParent() == this) {
		    // in FlowView.java view might be referenced 
		    // from two super-views as a child. see logicalView
		    getView(i).setParent(null);
		}
            }
        }
        this.parent = parent;
    }

    /** 
     * Returns the number of views in this view.  Since
     * the default is to not be a composite view this
     * returns 0.
     *
     * @return the number of views >= 0
     * @see View#getViewCount
     */
    public int getViewCount() {
	return 0;
    }

    /** 
     * Gets the <i>n</i>th child view.  Since there are no
     * children by default, this returns <code>null</code>.
     *
     * @param n the number of the view to get, >= 0 && < getViewCount()
     * @return the view
     */
    public View getView(int n) {
	return null;
    }


    /**
     * Removes all of the children.  This is a convenience
     * call to <code>replace</code>.
     *
     * @since 1.3
     */
    public void removeAll() {
	replace(0, getViewCount(), null);
    }

    /**
     * Removes one of the children at the given position.
     * This is a convenience call to <code>replace</code>.
     * @since 1.3
     */
    public void remove(int i) {
	replace(i, 1, null);
    }

    /**
     * Inserts a single child view.  This is a convenience 
     * call to <code>replace</code>.
     *
     * @param offs the offset of the view to insert before >= 0
     * @param v the view
     * @see #replace
     * @since 1.3
     */
    public void insert(int offs, View v) {
	View[] one = new View[1];
	one[0] = v;
	replace(offs, 0, one);
    }

    /**
     * Appends a single child view.  This is a convenience 
     * call to <code>replace</code>.
     *
     * @param v the view
     * @see #replace
     * @since 1.3
     */
    public void append(View v) {
	View[] one = new View[1];
	one[0] = v;
	replace(getViewCount(), 0, one);
    }

    /**
     * Replaces child views.  If there are no views to remove
     * this acts as an insert.  If there are no views to
     * add this acts as a remove.  Views being removed will
     * have the parent set to <code>null</code>, and the internal reference
     * to them removed so that they can be garbage collected.
     * This is implemented to do nothing, because by default
     * a view has no children.
     *
     * @param offset the starting index into the child views to insert
     *   the new views.  This should be a value >= 0 and <= getViewCount
     * @param length the number of existing child views to remove
     *   This should be a value >= 0 and <= (getViewCount() - offset).
     * @param views the child views to add.  This value can be
     *   <code>null</code> to indicate no children are being added
     *   (useful to remove).
     * @since 1.3
     */
    public void replace(int offset, int length, View[] views) {
    }

    /**
     * Returns the child view index representing the given position in
     * the model.  By default a view has no children so this is implemented
     * to return -1 to indicate there is no valid child index for any
     * position.
     *
     * @param pos the position >= 0
     * @return  index of the view representing the given position, or 
     *   -1 if no view represents that position
     * @since 1.3
     */
    public int getViewIndex(int pos, Position.Bias b) {
	return -1;
    }
    
    /**
     * Fetches the allocation for the given child view. 
     * This enables finding out where various views
     * are located, without assuming how the views store
     * their location.  This returns <code>null</code> since the
     * default is to not have any child views.
     *
     * @param index the index of the child, >= 0 && <
     *		<code>getViewCount()</code>
     * @param a  the allocation to this view
     * @return the allocation to the child
     */
    public Shape getChildAllocation(int index, Shape a) {
	return null;
    }

    /**
     * Provides a way to determine the next visually represented model 
     * location at which one might place a caret.
     * Some views may not be visible,
     * they might not be in the same order found in the model, or they just
     * might not allow access to some of the locations in the model.
     *
     * @param pos the position to convert >= 0
     * @param a the allocated region in which to render
     * @param direction the direction from the current position that can
     *  be thought of as the arrow keys typically found on a keyboard.
     *  This will be one of the following values:
     * <ul>
     * <li>SwingConstants.WEST
     * <li>SwingConstants.EAST
     * <li>SwingConstants.NORTH
     * <li>SwingConstants.SOUTH
     * </ul>
     * @return the location within the model that best represents the next
     *  location visual position
     * @exception BadLocationException
     * @exception IllegalArgumentException if <code>direction</code>
     *		doesn't have one of the legal values above
     */
    public int getNextVisualPositionFrom(int pos, Position.Bias b, Shape a, 
					 int direction, Position.Bias[] biasRet) 
      throws BadLocationException {

	biasRet[0] = Position.Bias.Forward;
	switch (direction) {
	case NORTH:
	case SOUTH:
	{
	    if (pos == -1) {
		pos = (direction == NORTH) ? Math.max(0, getEndOffset() - 1) :
		    getStartOffset();
		break;
	    }
	    JTextComponent target = (JTextComponent) getContainer();
	    Caret c = (target != null) ? target.getCaret() : null;
	    // YECK! Ideally, the x location from the magic caret position
	    // would be passed in.
	    Point mcp;
	    if (c != null) {
		mcp = c.getMagicCaretPosition();
	    }
	    else {
		mcp = null;
	    }
	    int x;
	    if (mcp == null) {
		Rectangle loc = target.modelToView(pos);
		x = (loc == null) ? 0 : loc.x;
	    }
	    else {
		x = mcp.x;
	    }
	    if (direction == NORTH) {
		pos = Utilities.getPositionAbove(target, pos, x);
	    }
	    else {
		pos = Utilities.getPositionBelow(target, pos, x);
	    }
	}
	    break;
	case WEST:
	    if(pos == -1) {
		pos = Math.max(0, getEndOffset() - 1);
	    }
	    else {
		pos = Math.max(0, pos - 1);
	    }
	    break;
	case EAST:
	    if(pos == -1) {
		pos = getStartOffset();
	    }
	    else {
		pos = Math.min(pos + 1, getDocument().getLength());
	    }
	    break;
	default:
	    throw new IllegalArgumentException("Bad direction: " + direction);
	}
	return pos;
    }

    /**
     * Provides a mapping, for a given character,
     * from the document model coordinate space
     * to the view coordinate space.  
     *
     * @param pos the position of the desired character (>=0)
     * @param a the area of the view, which encompasses the requested character
     * @param b the bias toward the previous character or the
     *  next character represented by the offset, in case the 
     *  position is a boundary of two views; <code>b</code> will have one
     *  of these values:
     * <ul>
     * <li> <code>Position.Bias.Forward</code>
     * <li> <code>Position.Bias.Backward</code>
     * </ul>
     * @return the bounding box, in view coordinate space,
     *		of the character at the specified position
     * @exception BadLocationException  if the specified position does
     *   not represent a valid location in the associated document
     * @exception IllegalArgumentException if <code>b</code> is not one of the
     *		legal <code>Position.Bias</code> values listed above
     * @see View#viewToModel
     */
    public abstract Shape modelToView(int pos, Shape a, Position.Bias b) throws BadLocationException;

    /**
     * Provides a mapping, for a given region,
     * from the document model coordinate space
     * to the view coordinate space. The specified region is
     * created as a union of the first and last character positions.
     *
     * @param p0 the position of the first character (>=0)
     * @param b0 the bias of the first character position,
     *  toward the previous character or the
     *  next character represented by the offset, in case the 
     *  position is a boundary of two views; <code>b0</code> will have one
     *  of these values:
     * <ul>
     * <li> <code>Position.Bias.Forward</code>
     * <li> <code>Position.Bias.Backward</code>
     * </ul>
     * @param p1 the position of the last character (>=0)
     * @param b1 the bias for the second character position, defined
     *		one of the legal values shown above
     * @param a the area of the view, which encompasses the requested region
     * @return the bounding box which is a union of the region specified
     *		by the first and last character positions
     * @exception BadLocationException  if the given position does
     *   not represent a valid location in the associated document
     * @exception IllegalArgumentException if <code>b0</code> or
     *		<code>b1</code> are not one of the
     *		legal <code>Position.Bias</code> values listed above
     * @see View#viewToModel
     */
    public Shape modelToView(int p0, Position.Bias b0, int p1, Position.Bias b1, Shape a) throws BadLocationException {
	Shape s0 = modelToView(p0, a, b0);
	Shape s1;
	if (p1 == getEndOffset()) {
	    try {
		s1 = modelToView(p1, a, b1);
	    } catch (BadLocationException ble) {
		s1 = null;
	    }
	    if (s1 == null) {
		// Assume extends left to right.
		Rectangle alloc = (a instanceof Rectangle) ? (Rectangle)a :
		                  a.getBounds();
		s1 = new Rectangle(alloc.x + alloc.width - 1, alloc.y,
				   1, alloc.height);
	    }
	}
	else {
	    s1 = modelToView(p1, a, b1);
	}
	Rectangle r0 = s0.getBounds();
	Rectangle r1 = (s1 instanceof Rectangle) ? (Rectangle) s1 :
	                                           s1.getBounds();
	if (r0.y != r1.y) {
	    // If it spans lines, force it to be the width of the view.
	    Rectangle alloc = (a instanceof Rectangle) ? (Rectangle)a :
		              a.getBounds();
	    r0.x = alloc.x;
	    r0.width = alloc.width;
	}
	r0.add(r1);
	return r0;
    }

    /**
     * Provides a mapping from the view coordinate space to the logical
     * coordinate space of the model.  The <code>biasReturn</code>
     * argument will be filled in to indicate that the point given is
     * closer to the next character in the model or the previous
     * character in the model.
     *
     * @param x the X coordinate >= 0
     * @param y the Y coordinate >= 0
     * @param a the allocated region in which to render
     * @return the location within the model that best represents the
     *  given point in the view >= 0.  The <code>biasReturn</code>
     *  argument will be
     * filled in to indicate that the point given is closer to the next
     * character in the model or the previous character in the model.
     */
    public abstract int viewToModel(float x, float y, Shape a, Position.Bias[] biasReturn);

    /**
     * Gives notification that something was inserted into 
     * the document in a location that this view is responsible for.  
     * To reduce the burden to subclasses, this functionality is
     * spread out into the following calls that subclasses can
     * reimplement:
     * <ol>
     * <li><a href="#updateChildren">updateChildren</a> is called
     * if there were any changes to the element this view is
     * responsible for.  If this view has child views that are
     * represent the child elements, then this method should do
     * whatever is necessary to make sure the child views correctly
     * represent the model.
     * <li><a href="#forwardUpdate">forwardUpdate</a> is called
     * to forward the DocumentEvent to the appropriate child views.
     * <li><a href="#updateLayout">updateLayout</a> is called to
     * give the view a chance to either repair its layout, to reschedule
     * layout, or do nothing.
     * </ol>
     *
     * @param e the change information from the associated document
     * @param a the current allocation of the view
     * @param f the factory to use to rebuild if the view has children
     * @see View#insertUpdate
     */
    public void insertUpdate(DocumentEvent e, Shape a, ViewFactory f) {
	if (getViewCount() > 0) {
	    Element elem = getElement();
	    DocumentEvent.ElementChange ec = e.getChange(elem);
	    if (ec != null) {
		if (! updateChildren(ec, e, f)) {
		    // don't consider the element changes they
		    // are for a view further down.
		    ec = null;
		}
	    }
	    forwardUpdate(ec, e, a, f);
	    updateLayout(ec, e, a);
	}
    }

    /**
     * Gives notification that something was removed from the document
     * in a location that this view is responsible for.
     * To reduce the burden to subclasses, this functionality is
     * spread out into the following calls that subclasses can
     * reimplement:
     * <ol>
     * <li><a href="#updateChildren">updateChildren</a> is called
     * if there were any changes to the element this view is
     * responsible for.  If this view has child views that are
     * represent the child elements, then this method should do
     * whatever is necessary to make sure the child views correctly
     * represent the model.
     * <li><a href="#forwardUpdate">forwardUpdate</a> is called
     * to forward the DocumentEvent to the appropriate child views.
     * <li><a href="#updateLayout">updateLayout</a> is called to
     * give the view a chance to either repair its layout, to reschedule
     * layout, or do nothing.
     * </ol>
     *
     * @param e the change information from the associated document
     * @param a the current allocation of the view
     * @param f the factory to use to rebuild if the view has children
     * @see View#removeUpdate
     */
    public void removeUpdate(DocumentEvent e, Shape a, ViewFactory f) {
	if (getViewCount() > 0) {
	    Element elem = getElement();
	    DocumentEvent.ElementChange ec = e.getChange(elem);
	    if (ec != null) {
		if (! updateChildren(ec, e, f)) {
		    // don't consider the element changes they
		    // are for a view further down.
		    ec = null;
		}
	    }
	    forwardUpdate(ec, e, a, f);
	    updateLayout(ec, e, a);
	}
    }

    /**
     * Gives notification from the document that attributes were changed
     * in a location that this view is responsible for.
     * To reduce the burden to subclasses, this functionality is
     * spread out into the following calls that subclasses can
     * reimplement:
     * <ol>
     * <li><a href="#updateChildren">updateChildren</a> is called
     * if there were any changes to the element this view is
     * responsible for.  If this view has child views that are
     * represent the child elements, then this method should do
     * whatever is necessary to make sure the child views correctly
     * represent the model.
     * <li><a href="#forwardUpdate">forwardUpdate</a> is called
     * to forward the DocumentEvent to the appropriate child views.
     * <li><a href="#updateLayout">updateLayout</a> is called to
     * give the view a chance to either repair its layout, to reschedule
     * layout, or do nothing.
     * </ol>
     *
     * @param e the change information from the associated document
     * @param a the current allocation of the view
     * @param f the factory to use to rebuild if the view has children
     * @see View#changedUpdate
     */
    public void changedUpdate(DocumentEvent e, Shape a, ViewFactory f) {
	if (getViewCount() > 0) {
	    Element elem = getElement();
	    DocumentEvent.ElementChange ec = e.getChange(elem);
	    if (ec != null) {
		if (! updateChildren(ec, e, f)) {
		    // don't consider the element changes they
		    // are for a view further down.
		    ec = null;
		}
	    }
	    forwardUpdate(ec, e, a, f);
	    updateLayout(ec, e, a);
	}
    }

    /**
     * Fetches the model associated with the view.
     *
     * @return the view model, <code>null</code> if none
     * @see View#getDocument
     */
    public Document getDocument() {
	return elem.getDocument();
    }

    /**
     * Fetches the portion of the model for which this view is
     * responsible.
     *
     * @return the starting offset into the model >= 0
     * @see View#getStartOffset
     */
    public int getStartOffset() {
	return elem.getStartOffset();
    }

    /**
     * Fetches the portion of the model for which this view is
     * responsible.
     *
     * @return the ending offset into the model >= 0
     * @see View#getEndOffset
     */
    public int getEndOffset() {
	return elem.getEndOffset();
    }

    /**
     * Fetches the structural portion of the subject that this
     * view is mapped to.  The view may not be responsible for the
     * entire portion of the element.
     *
     * @return the subject
     * @see View#getElement
     */
    public Element getElement() {
	return elem;
    }

    /**
     * Fetch a <code>Graphics</code> for rendering.
     * This can be used to determine
     * font characteristics, and will be different for a print view
     * than a component view.
     *
     * @return a <code>Graphics</code> object for rendering
     * @since 1.3
     */
    public Graphics getGraphics() {
	// PENDING(prinz) this is a temporary implementation
	Component c = getContainer();
	return c.getGraphics();
    }

    /**
     * Fetches the attributes to use when rendering.  By default
     * this simply returns the attributes of the associated element.
     * This method should be used rather than using the element
     * directly to obtain access to the attributes to allow
     * view-specific attributes to be mixed in or to allow the
     * view to have view-specific conversion of attributes by
     * subclasses.
     * Each view should document what attributes it recognizes
     * for the purpose of rendering or layout, and should always
     * access them through the <code>AttributeSet</code> returned
     * by this method.
     */
    public AttributeSet getAttributes() {
	return elem.getAttributes();
    }

    /**
     * Tries to break this view on the given axis.  This is
     * called by views that try to do formatting of their
     * children.  For example, a view of a paragraph will
     * typically try to place its children into row and 
     * views representing chunks of text can sometimes be 
     * broken down into smaller pieces.
     * <p>
     * This is implemented to return the view itself, which
     * represents the default behavior on not being
     * breakable.  If the view does support breaking, the
     * starting offset of the view returned should be the
     * given offset, and the end offset should be less than
     * or equal to the end offset of the view being broken.
     *
     * @param axis may be either <code>View.X_AXIS</code> or
     *		<code>View.Y_AXIS</code>
     * @param offset the location in the document model
     *   that a broken fragment would occupy >= 0.  This
     *   would be the starting offset of the fragment
     *   returned
     * @param pos the position along the axis that the
     *  broken view would occupy >= 0.  This may be useful for
     *  things like tab calculations
     * @param len specifies the distance along the axis
     *  where a potential break is desired >= 0
     * @return the fragment of the view that represents the
     *  given span, if the view can be broken.  If the view
     *  doesn't support breaking behavior, the view itself is
     *  returned.
     * @see ParagraphView
     */
    public View breakView(int axis, int offset, float pos, float len) {
	return this;
    }

    /**
     * Creates a view that represents a portion of the element.
     * This is potentially useful during formatting operations
     * for taking measurements of fragments of the view.  If 
     * the view doesn't support fragmenting (the default), it 
     * should return itself.  
     *
     * @param p0 the starting offset >= 0.  This should be a value
     *   greater or equal to the element starting offset and
     *   less than the element ending offset.
     * @param p1 the ending offset > p0.  This should be a value
     *   less than or equal to the elements end offset and
     *   greater than the elements starting offset.
     * @return the view fragment, or itself if the view doesn't
     *   support breaking into fragments
     * @see LabelView
     */
    public View createFragment(int p0, int p1) {
	return this;
    }

    /**
     * Determines how attractive a break opportunity in 
     * this view is.  This can be used for determining which
     * view is the most attractive to call <code>breakView</code>
     * on in the process of formatting.  A view that represents
     * text that has whitespace in it might be more attractive
     * than a view that has no whitespace, for example.  The
     * higher the weight, the more attractive the break.  A
     * value equal to or lower than <code>BadBreakWeight</code>
     * should not be considered for a break.  A value greater
     * than or equal to <code>ForcedBreakWeight</code> should
     * be broken.
     * <p>
     * This is implemented to provide the default behavior
     * of returning <code>BadBreakWeight</code> unless the length
     * is greater than the length of the view in which case the 
     * entire view represents the fragment.  Unless a view has
     * been written to support breaking behavior, it is not
     * attractive to try and break the view.  An example of
     * a view that does support breaking is <code>LabelView</code>.
     * An example of a view that uses break weight is 
     * <code>ParagraphView</code>.
     *
     * @param axis may be either <code>View.X_AXIS</code> or
     *		<code>View.Y_AXIS</code>
     * @param pos the potential location of the start of the 
     *   broken view >= 0.  This may be useful for calculating tab
     *   positions
     * @param len specifies the relative length from <em>pos</em>
     *   where a potential break is desired >= 0
     * @return the weight, which should be a value between
     *   ForcedBreakWeight and BadBreakWeight
     * @see LabelView
     * @see ParagraphView
     * @see #BadBreakWeight
     * @see #GoodBreakWeight
     * @see #ExcellentBreakWeight
     * @see #ForcedBreakWeight
     */
    public int getBreakWeight(int axis, float pos, float len) {
	if (len > getPreferredSpan(axis)) {
	    return GoodBreakWeight;
	}
	return BadBreakWeight;
    }

    /**
     * Determines the resizability of the view along the
     * given axis.  A value of 0 or less is not resizable.
     *
     * @param axis may be either <code>View.X_AXIS</code> or
     *		<code>View.Y_AXIS</code>
     * @return the weight
     */
    public int getResizeWeight(int axis) {
	return 0;
    }

    /**
     * Sets the size of the view.  This should cause 
     * layout of the view along the given axis, if it 
     * has any layout duties.
     *
     * @param width the width >= 0
     * @param height the height >= 0
     */
    public void setSize(float width, float height) {
    }

    /**
     * Fetches the container hosting the view.  This is useful for
     * things like scheduling a repaint, finding out the host 
     * components font, etc.  The default implementation
     * of this is to forward the query to the parent view.
     *
     * @return the container, <code>null</code> if none
     */
    public Container getContainer() {
	View v = getParent();
	return (v != null) ? v.getContainer() : null;
    }

    /**
     * Fetches the <code>ViewFactory</code> implementation that is feeding
     * the view hierarchy.  Normally the views are given this
     * as an argument to updates from the model when they
     * are most likely to need the factory, but this
     * method serves to provide it at other times.
     *
     * @return the factory, <code>null</code> if none
     */
    public ViewFactory getViewFactory() {
	View v = getParent();
	return (v != null) ? v.getViewFactory() : null;
    }

    /**
     * Returns the tooltip text at the specified location. The default
     * implementation returns the value from the child View identified by
     * the passed in location.
     *
     * @since 1.4
     * @see JTextComponent#getToolTipText
     */
    public String getToolTipText(float x, float y, Shape allocation) {
	int viewIndex = getViewIndex(x, y, allocation);
        if (viewIndex >= 0) {
            allocation = getChildAllocation(viewIndex, allocation);
            Rectangle rect = (allocation instanceof Rectangle) ?
                             (Rectangle)allocation : allocation.getBounds();
            if (rect.contains(x, y)) {
                return getView(viewIndex).getToolTipText(x, y, allocation);
            }
        }
        return null;
    }

    /**
     * Returns the child view index representing the given position in
     * the view. This iterates over all the children returning the
     * first with a bounds that contains <code>x</code>, <code>y</code>.
     *
     * @param x the x coordinate
     * @param y the y coordinate
     * @param allocation current allocation of the View.
     * @return  index of the view representing the given location, or 
     *   -1 if no view represents that position
     * @since 1.4
     */
    public int getViewIndex(float x, float y, Shape allocation) {
	for (int counter = getViewCount() - 1; counter >= 0; counter--) {
	    Shape childAllocation = getChildAllocation(counter, allocation);

            if (childAllocation != null) {
                Rectangle rect = (childAllocation instanceof Rectangle) ?
		         (Rectangle)childAllocation : allocation.getBounds();

                if (rect.contains(x, y)) {
                    return counter;
                }
            }
	}
	return -1;
    }

    /**
     * Updates the child views in response to receiving notification
     * that the model changed, and there is change record for the 
     * element this view is responsible for.  This is implemented
     * to assume the child views are directly responsible for the
     * child elements of the element this view represents.  The
     * <code>ViewFactory</code> is used to create child views for each element
     * specified as added in the <code>ElementChange</code>, starting at the
     * index specified in the given <code>ElementChange</code>.  The number of
     * child views representing the removed elements specified are 
     * removed.
     * 
     * @param ec the change information for the element this view
     *  is responsible for.  This should not be <code>null</code> if 
     *  this method gets called
     * @param e the change information from the associated document
     * @param f the factory to use to build child views
     * @return whether or not the child views represent the
     *  child elements of the element this view is responsible
     *  for.  Some views create children that represent a portion 
     *  of the element they are responsible for, and should return
     *  false.  This information is used to determine if views 
     *  in the range of the added elements should be forwarded to
     *  or not
     * @see #insertUpdate
     * @see #removeUpdate
     * @see #changedUpdate
     * @since 1.3
     */
    protected boolean updateChildren(DocumentEvent.ElementChange ec, 
					 DocumentEvent e, ViewFactory f) {
	Element[] removedElems = ec.getChildrenRemoved();
	Element[] addedElems = ec.getChildrenAdded();
	View[] added = null;
	if (addedElems != null) {
	    added = new View[addedElems.length];
	    for (int i = 0; i < addedElems.length; i++) {
		added[i] = f.create(addedElems[i]);
	    }
	}
	int nremoved = 0;
	int index = ec.getIndex();
	if (removedElems != null) {
	    nremoved = removedElems.length;
	}
	replace(index, nremoved, added);
	return true;
    }

    /**
     * Forwards the given <code>DocumentEvent</code> to the child views
     * that need to be notified of the change to the model.
     * If there were changes to the element this view is 
     * responsible for, that should be considered when 
     * forwarding (i.e. new child views should not get
     * notified).
     *
     * @param ec changes to the element this view is responsible
     *  for (may be <code>null</code> if there were no changes).
     * @param e the change information from the associated document
     * @param a the current allocation of the view
     * @param f the factory to use to rebuild if the view has children
     * @see #insertUpdate
     * @see #removeUpdate
     * @see #changedUpdate     
     * @since 1.3
     */
    protected void forwardUpdate(DocumentEvent.ElementChange ec, 
				      DocumentEvent e, Shape a, ViewFactory f) {
	Element elem = getElement();
	int pos = e.getOffset();
	int index0 = getViewIndex(pos, Position.Bias.Forward);
	if (index0 == -1 && e.getType() == DocumentEvent.EventType.REMOVE &&
	    pos >= getEndOffset()) {
	    // Event beyond our offsets. We may have represented this, that is
	    // the remove may have removed one of our child Elements that
	    // represented this, so, we should foward to last element.
	    index0 = getViewCount() - 1;
	}
	int index1 = index0;
	View v = (index0 >= 0) ? getView(index0) : null;
	if (v != null) {
	    if ((v.getStartOffset() == pos) && (pos > 0)) {
		// If v is at a boundary, forward the event to the previous
		// view too.
		index0 = Math.max(index0 - 1, 0);
	    }
	}
	if (e.getType() != DocumentEvent.EventType.REMOVE) {
	    index1 = getViewIndex(pos + e.getLength(), Position.Bias.Forward);
	    if (index1 < 0) {
		index1 = getViewCount() - 1;
	    }
	}
	int hole0 = index1 + 1;
	int hole1 = hole0;
	Element[] addedElems = (ec != null) ? ec.getChildrenAdded() : null;
	if ((addedElems != null) && (addedElems.length > 0)) {
	    hole0 = ec.getIndex();
	    hole1 = hole0 + addedElems.length - 1;
	}

	// forward to any view not in the forwarding hole 
	// formed by added elements (i.e. they will be updated
	// by initialization.
	index0 = Math.max(index0, 0);
	for (int i = index0; i <= index1; i++) {
	    if (! ((i >= hole0) && (i <= hole1))) {
		v = getView(i);
		if (v != null) {
		    Shape childAlloc = getChildAllocation(i, a);
		    forwardUpdateToView(v, e, childAlloc, f);
		}
	    }
	}
    }

    /**
     * Forwards the <code>DocumentEvent</code> to the give child view.  This
     * simply messages the view with a call to <code>insertUpdate</code>, 
     * <code>removeUpdate</code>, or <code>changedUpdate</code> depending
     * upon the type of the event.  This is called by
     * <a href="#forwardUpdate">forwardUpdate</a> to forward 
     * the event to children that need it.
     *
     * @param v the child view to forward the event to
     * @param e the change information from the associated document
     * @param a the current allocation of the view
     * @param f the factory to use to rebuild if the view has children
     * @see #forwardUpdate
     * @since 1.3
     */
    protected void forwardUpdateToView(View v, DocumentEvent e, 
					   Shape a, ViewFactory f) {
	DocumentEvent.EventType type = e.getType();
	if (type == DocumentEvent.EventType.INSERT) {
	    v.insertUpdate(e, a, f);
	} else if (type == DocumentEvent.EventType.REMOVE) {
	    v.removeUpdate(e, a, f);
	} else {
	    v.changedUpdate(e, a, f);
	}
    }

    /**
     * Updates the layout in response to receiving notification of
     * change from the model.  This is implemented to call
     * <code>preferenceChanged</code> to reschedule a new layout
     * if the <code>ElementChange</code> record is not <code>null</code>.
     *
     * @param ec changes to the element this view is responsible
     *  for (may be <code>null</code> if there were no changes)
     * @param e the change information from the associated document
     * @param a the current allocation of the view
     * @see #insertUpdate
     * @see #removeUpdate
     * @see #changedUpdate     
     * @since 1.3
     */
    protected void updateLayout(DocumentEvent.ElementChange ec, 
				    DocumentEvent e, Shape a) {
	if ((ec != null) && (a != null)) {
	    // should damage more intelligently
	    preferenceChanged(null, true, true);
	    Container host = getContainer();
	    if (host != null) {
		host.repaint();
	    }
	}
    }

    /**
     * The weight to indicate a view is a bad break
     * opportunity for the purpose of formatting.  This
     * value indicates that no attempt should be made to
     * break the view into fragments as the view has 
     * not been written to support fragmenting.
     *
     * @see #getBreakWeight
     * @see #GoodBreakWeight
     * @see #ExcellentBreakWeight
     * @see #ForcedBreakWeight
     */
    public static final int BadBreakWeight = 0;

    /**
     * The weight to indicate a view supports breaking,
     * but better opportunities probably exist.
     * 
     * @see #getBreakWeight
     * @see #BadBreakWeight
     * @see #ExcellentBreakWeight
     * @see #ForcedBreakWeight
     */
    public static final int GoodBreakWeight = 1000;

    /**
     * The weight to indicate a view supports breaking,
     * and this represents a very attractive place to
     * break.
     *
     * @see #getBreakWeight
     * @see #BadBreakWeight
     * @see #GoodBreakWeight
     * @see #ForcedBreakWeight
     */
    public static final int ExcellentBreakWeight = 2000;

    /**
     * The weight to indicate a view supports breaking,
     * and must be broken to be represented properly 
     * when placed in a view that formats its children
     * by breaking them.
     *
     * @see #getBreakWeight
     * @see #BadBreakWeight
     * @see #GoodBreakWeight
     * @see #ExcellentBreakWeight
     */
    public static final int ForcedBreakWeight = 3000;

    /**
     * Axis for format/break operations.
     */
    public static final int X_AXIS = HORIZONTAL;

    /**
     * Axis for format/break operations.
     */
    public static final int Y_AXIS = VERTICAL;

    /**
     * Provides a mapping from the document model coordinate space
     * to the coordinate space of the view mapped to it. This is 
     * implemented to default the bias to <code>Position.Bias.Forward</code>
     * which was previously implied.
     *
     * @param pos the position to convert >= 0
     * @param a the allocated region in which to render
     * @return the bounding box of the given position is returned
     * @exception BadLocationException  if the given position does
     *   not represent a valid location in the associated document
     * @see View#modelToView
     * @deprecated
     */
    @Deprecated
    public Shape modelToView(int pos, Shape a) throws BadLocationException {
	return modelToView(pos, a, Position.Bias.Forward);
    }


    /**
     * Provides a mapping from the view coordinate space to the logical
     * coordinate space of the model.
     *
     * @param x the X coordinate >= 0
     * @param y the Y coordinate >= 0
     * @param a the allocated region in which to render
     * @return the location within the model that best represents the
     *  given point in the view >= 0
     * @see View#viewToModel
     * @deprecated
     */
    @Deprecated
    public int viewToModel(float x, float y, Shape a) {
	sharedBiasReturn[0] = Position.Bias.Forward;
	return viewToModel(x, y, a, sharedBiasReturn);
    }

    // static argument available for viewToModel calls since only
    // one thread at a time may call this method.
    static final Position.Bias[] sharedBiasReturn = new Position.Bias[1];

    private View parent;
    private Element elem;

};