/***************************************************************************** * Copyright (C) The Apache Software Foundation. All rights reserved. * * ------------------------------------------------------------------------- * * This software is published under the terms of the Apache Software License * * version 1.1, a copy of which has been included with this distribution in * * the LICENSE file. * *****************************************************************************/ package com.kitfox.svg.batik; import java.awt.Color; import java.awt.Rectangle; import java.awt.RenderingHints; import java.awt.geom.AffineTransform; import java.awt.geom.NoninvertibleTransformException; import java.awt.geom.Point2D; import java.awt.geom.Rectangle2D; import java.awt.image.ColorModel; /** * Provides the actual implementation for the LinearGradientPaint This is where * the pixel processing is done. * * @author Nicholas Talian, Vincent Hardy, Jim Graham, Jerry Evans * @author <a href="mailto:vincent.hardy@eng.sun.com">Vincent Hardy</a> * @version $Id: LinearGradientPaintContext.java,v 1.2 2007/02/04 01:28:05 * kitfox Exp $ * @see java.awt.PaintContext * @see java.awt.Paint * @see java.awt.GradientPaint */ final class LinearGradientPaintContext extends MultipleGradientPaintContext { /** * The following invariants are used to process the gradient value from a * device space coordinate, (X, Y): g(X, Y) = dgdX*X + dgdY*Y + gc */ private float dgdX, dgdY, gc, pixSz; private static final int DEFAULT_IMPL = 1; private static final int ANTI_ALIAS_IMPL = 3; private int fillMethod; /** * Constructor for LinearGradientPaintContext. * * @param cm * {@link ColorModel} that receives the <code>Paint</code> data. * This is used only as a hint. * * @param deviceBounds * the device space bounding box of the graphics primitive being * rendered * * @param userBounds * the user space bounding box of the graphics primitive being * rendered * * @param t * the {@link AffineTransform} from user space into device space * (gradientTransform should be concatenated with this) * * @param hints * the hints that the context object uses to choose between * rendering alternatives * * @param start * gradient start point, in user space * * @param end * gradient end point, in user space * * @param fractions * the fractions specifying the gradient distribution * * @param colors * the gradient colors * * @param cycleMethod * either NO_CYCLE, REFLECT, or REPEAT * * @param colorSpace * which colorspace to use for interpolation, either SRGB or * LINEAR_RGB * */ public LinearGradientPaintContext(ColorModel cm, Rectangle deviceBounds, Rectangle2D userBounds, AffineTransform t, RenderingHints hints, Point2D dStart, Point2D dEnd, float[] fractions, Color[] colors, MultipleGradientPaint.CycleMethodEnum cycleMethod, MultipleGradientPaint.ColorSpaceEnum colorSpace) throws NoninvertibleTransformException { super(cm, deviceBounds, userBounds, t, hints, fractions, colors, cycleMethod, colorSpace); // Use single precision floating points Point2D.Float start = new Point2D.Float((float) dStart.getX(), (float) dStart.getY()); Point2D.Float end = new Point2D.Float((float) dEnd.getX(), (float) dEnd.getY()); // A given point in the raster should take on the same color as its // projection onto the gradient vector. // Thus, we want the projection of the current position vector // onto the gradient vector, then normalized with respect to the // length of the gradient vector, giving a value which can be mapped // into // the range 0-1. // projection = currentVector dot gradientVector / // length(gradientVector) // normalized = projection / length(gradientVector) float dx = end.x - start.x; // change in x from start to end float dy = end.y - start.y; // change in y from start to end float dSq = dx * dx + dy * dy; // total distance squared // avoid repeated calculations by doing these divides once. float constX = dx / dSq; float constY = dy / dSq; // incremental change along gradient for +x dgdX = a00 * constX + a10 * constY; // incremental change along gradient for +y dgdY = a01 * constX + a11 * constY; float dgdXAbs = Math.abs(dgdX); float dgdYAbs = Math.abs(dgdY); if (dgdXAbs > dgdYAbs) { pixSz = dgdXAbs; } else { pixSz = dgdYAbs; } // constant, incorporates the translation components from the matrix gc = (a02 - start.x) * constX + (a12 - start.y) * constY; Object colorRend = hints == null ? RenderingHints.VALUE_COLOR_RENDER_SPEED : hints.get(RenderingHints.KEY_COLOR_RENDERING); Object rend = hints == null ? RenderingHints.VALUE_RENDER_SPEED : hints.get(RenderingHints.KEY_RENDERING); fillMethod = DEFAULT_IMPL; if ((cycleMethod == MultipleGradientPaint.REPEAT) || hasDiscontinuity) { if (rend == RenderingHints.VALUE_RENDER_QUALITY) { fillMethod = ANTI_ALIAS_IMPL; } // ColorRend overrides rend. if (colorRend == RenderingHints.VALUE_COLOR_RENDER_SPEED) { fillMethod = DEFAULT_IMPL; } else if (colorRend == RenderingHints.VALUE_COLOR_RENDER_QUALITY) { fillMethod = ANTI_ALIAS_IMPL; } } } protected void fillHardNoCycle(int[] pixels, int off, int adjust, int x, int y, int w, int h) { // constant which can be pulled out of the inner loop final float initConst = (dgdX * x) + gc; for (int i = 0; i < h; i++) { // for every row // initialize current value to be start. float g = initConst + dgdY * (y + i); final int rowLimit = off + w; // end of row iteration if (dgdX == 0) { // System.out.println("In fillHard: " + g); final int val; if (g <= 0) { val = gradientUnderflow; } else if (g >= 1) { val = gradientOverflow; } else { // Could be a binary search... int gradIdx = 0; while (gradIdx < gradientsLength - 1) { if (g < fractions[gradIdx + 1]) { break; } gradIdx++; } float delta = (g - fractions[gradIdx]); float idx = ((delta * GRADIENT_SIZE_INDEX) / normalizedIntervals[gradIdx]) + 0.5f; val = gradients[gradIdx][(int) idx]; } while (off < rowLimit) { pixels[off++] = val; } } else { // System.out.println("In fillHard2: " + g); int gradSteps; int preGradSteps; final int preVal, postVal; if (dgdX >= 0) { gradSteps = (int) ((1 - g) / dgdX); preGradSteps = (int) Math.ceil((0 - g) / dgdX); preVal = gradientUnderflow; postVal = gradientOverflow; } else { // dgdX < 0 gradSteps = (int) ((0 - g) / dgdX); preGradSteps = (int) Math.ceil((1 - g) / dgdX); preVal = gradientOverflow; postVal = gradientUnderflow; } if (gradSteps > w) { gradSteps = w; } final int gradLimit = off + gradSteps; if (preGradSteps > 0) { if (preGradSteps > w) { preGradSteps = w; } final int preGradLimit = off + preGradSteps; while (off < preGradLimit) { pixels[off++] = preVal; } g += dgdX * preGradSteps; } if (dgdX > 0) { // Could be a binary search... int gradIdx = 0; while (gradIdx < gradientsLength - 1) { if (g < fractions[gradIdx + 1]) { break; } gradIdx++; } while (off < gradLimit) { float delta = (g - fractions[gradIdx]); final int[] grad = gradients[gradIdx]; int steps = (int) Math .ceil((fractions[gradIdx + 1] - g) / dgdX); int subGradLimit = off + steps; if (subGradLimit > gradLimit) { subGradLimit = gradLimit; } int idx = (int) (((delta * GRADIENT_SIZE_INDEX) / normalizedIntervals[gradIdx]) * (1 << 16)) + (1 << 15); int step = (int) (((dgdX * GRADIENT_SIZE_INDEX) / normalizedIntervals[gradIdx]) * (1 << 16)); while (off < subGradLimit) { pixels[off++] = grad[idx >> 16]; idx += step; } g += dgdX * steps; gradIdx++; } } else { // Could be a binary search... int gradIdx = gradientsLength - 1; while (gradIdx > 0) { if (g > fractions[gradIdx]) { break; } gradIdx--; } while (off < gradLimit) { float delta = (g - fractions[gradIdx]); final int[] grad = gradients[gradIdx]; int steps = (int) Math.ceil(delta / -dgdX); int subGradLimit = off + steps; if (subGradLimit > gradLimit) { subGradLimit = gradLimit; } int idx = (int) (((delta * GRADIENT_SIZE_INDEX) / normalizedIntervals[gradIdx]) * (1 << 16)) + (1 << 15); int step = (int) (((dgdX * GRADIENT_SIZE_INDEX) / normalizedIntervals[gradIdx]) * (1 << 16)); while (off < subGradLimit) { pixels[off++] = grad[idx >> 16]; idx += step; } g += dgdX * steps; gradIdx--; } } while (off < rowLimit) { pixels[off++] = postVal; } } off += adjust; // change in off from row to row } } protected void fillSimpleNoCycle(int[] pixels, int off, int adjust, int x, int y, int w, int h) { // constant which can be pulled out of the inner loop final float initConst = (dgdX * x) + gc; final float step = dgdX * fastGradientArraySize; final int fpStep = (int) (step * (1 << 16)); // fix point step final int[] grad = gradient; for (int i = 0; i < h; i++) { // for every row // initialize current value to be start. float g = initConst + dgdY * (y + i); g *= fastGradientArraySize; g += 0.5; // rounding factor... final int rowLimit = off + w; // end of row iteration if (dgdX == 0) { // System.out.println("In fillSimpleNC: " + g); final int val; if (g <= 0) { val = gradientUnderflow; } else if (g >= fastGradientArraySize) { val = gradientOverflow; } else { val = grad[(int) g]; } while (off < rowLimit) { pixels[off++] = val; } } else { // System.out.println("In fillSimpleNC2: " + g); int gradSteps; int preGradSteps; final int preVal, postVal; if (dgdX > 0) { gradSteps = (int) ((fastGradientArraySize - g) / step); preGradSteps = (int) Math.ceil(0 - g / step); preVal = gradientUnderflow; postVal = gradientOverflow; } else { // dgdX < 0 gradSteps = (int) ((0 - g) / step); preGradSteps = (int) Math .ceil((fastGradientArraySize - g) / step); preVal = gradientOverflow; postVal = gradientUnderflow; } if (gradSteps > w) { gradSteps = w; } final int gradLimit = off + gradSteps; if (preGradSteps > 0) { if (preGradSteps > w) { preGradSteps = w; } final int preGradLimit = off + preGradSteps; while (off < preGradLimit) { pixels[off++] = preVal; } g += step * preGradSteps; } int fpG = (int) (g * (1 << 16)); while (off < gradLimit) { pixels[off++] = grad[fpG >> 16]; fpG += fpStep; } while (off < rowLimit) { pixels[off++] = postVal; } } off += adjust; // change in off from row to row } } protected void fillSimpleRepeat(int[] pixels, int off, int adjust, int x, int y, int w, int h) { final float initConst = (dgdX * x) + gc; // Limit step to fractional part of // fastGradientArraySize (the non fractional part has // no affect anyways, and would mess up lots of stuff // below). float step = (dgdX - (int) dgdX) * fastGradientArraySize; // Make it a Positive step (a small negative step is // the same as a positive step slightly less than // fastGradientArraySize. if (step < 0) { step += fastGradientArraySize; } final int[] grad = gradient; for (int i = 0; i < h; i++) { // for every row // initialize current value to be start. float g = initConst + dgdY * (y + i); // now Limited between -1 and 1. g = g - (int) g; // put in the positive side. if (g < 0) { g += 1; } // scale for gradient array... g *= fastGradientArraySize; g += 0.5; // rounding factor final int rowLimit = off + w; // end of row iteration while (off < rowLimit) { int idx = (int) g; if (idx >= fastGradientArraySize) { g -= fastGradientArraySize; idx -= fastGradientArraySize; } pixels[off++] = grad[idx]; g += step; } off += adjust; // change in off from row to row } } protected void fillSimpleReflect(int[] pixels, int off, int adjust, int x, int y, int w, int h) { final float initConst = (dgdX * x) + gc; final int[] grad = gradient; for (int i = 0; i < h; i++) { // for every row // initialize current value to be start. float g = initConst + dgdY * (y + i); // now limited g to -2<->2 g = g - 2 * ((int) (g / 2.0f)); float step = dgdX; // Pull it into the positive half if (g < 0) { g = -g; // take absolute value step = -step; // Change direction.. } // Now do the same for dgdX. This is safe because // any step that is a multiple of 2.0 has no // affect, hence we can remove it which the first // part does. The second part simply adds 2.0 // (which has no affect due to the cylcle) to move // all negative step values into the positive // side. step = step - 2 * ((int) step / 2.0f); if (step < 0) { step += 2.0; } final int reflectMax = 2 * fastGradientArraySize; // Scale for gradient array. g *= fastGradientArraySize; g += 0.5; step *= fastGradientArraySize; final int rowLimit = off + w; // end of row iteration while (off < rowLimit) { int idx = (int) g; if (idx >= reflectMax) { g -= reflectMax; idx -= reflectMax; } if (idx <= fastGradientArraySize) { pixels[off++] = grad[idx]; } else { pixels[off++] = grad[reflectMax - idx]; } g += step; } off += adjust; // change in off from row to row } } /** * Return a Raster containing the colors generated for the graphics * operation. This is where the area is filled with colors distributed * linearly. * * @param x,y,w,h * The area in device space for which colors are generated. * */ @Override protected void fillRaster(int[] pixels, int off, int adjust, int x, int y, int w, int h) { // constant which can be pulled out of the inner loop final float initConst = (dgdX * x) + gc; if (fillMethod == ANTI_ALIAS_IMPL) { // initialize current value to be start. for (int i = 0; i < h; i++) { // for every row float g = initConst + dgdY * (y + i); final int rowLimit = off + w; // end of row iteration while (off < rowLimit) { // for every pixel in this row. // get the color pixels[off++] = indexGradientAntiAlias(g, pixSz); g += dgdX; // incremental change in g } off += adjust; // change in off from row to row } } else if (!isSimpleLookup) { if (cycleMethod == MultipleGradientPaint.NO_CYCLE) { fillHardNoCycle(pixels, off, adjust, x, y, w, h); } else { // initialize current value to be start. for (int i = 0; i < h; i++) { // for every row float g = initConst + dgdY * (y + i); final int rowLimit = off + w; // end of row iteration while (off < rowLimit) { // for every pixel in this row. // get the color pixels[off++] = indexIntoGradientsArrays(g); g += dgdX; // incremental change in g } off += adjust; // change in off from row to row } } } else { // Simple implementations: just scale index by array size if (cycleMethod == MultipleGradientPaint.NO_CYCLE) { fillSimpleNoCycle(pixels, off, adjust, x, y, w, h); } else if (cycleMethod == MultipleGradientPaint.REPEAT) { fillSimpleRepeat(pixels, off, adjust, x, y, w, h); } else { fillSimpleReflect(pixels, off, adjust, x, y, w, h); } } } }