1 /* |
|
2 * Copyright (c) 2007, 2011, Oracle and/or its affiliates. All rights reserved. |
|
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. |
|
4 * |
|
5 * This code is free software; you can redistribute it and/or modify it |
|
6 * under the terms of the GNU General Public License version 2 only, as |
|
7 * published by the Free Software Foundation. Oracle designates this |
|
8 * particular file as subject to the "Classpath" exception as provided |
|
9 * by Oracle in the LICENSE file that accompanied this code. |
|
10 * |
|
11 * This code is distributed in the hope that it will be useful, but WITHOUT |
|
12 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or |
|
13 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License |
|
14 * version 2 for more details (a copy is included in the LICENSE file that |
|
15 * accompanied this code). |
|
16 * |
|
17 * You should have received a copy of the GNU General Public License version |
|
18 * 2 along with this work; if not, write to the Free Software Foundation, |
|
19 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. |
|
20 * |
|
21 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA |
|
22 * or visit www.oracle.com if you need additional information or have any |
|
23 * questions. |
|
24 */ |
|
25 |
|
26 package sun.java2d.pisces; |
|
27 |
|
28 import java.util.Arrays; |
|
29 import static java.lang.Math.PI; |
|
30 import static java.lang.Math.cos; |
|
31 import static java.lang.Math.sqrt; |
|
32 import static java.lang.Math.cbrt; |
|
33 import static java.lang.Math.acos; |
|
34 |
|
35 |
|
36 final class Helpers { |
|
37 private Helpers() { |
|
38 throw new Error("This is a non instantiable class"); |
|
39 } |
|
40 |
|
41 static boolean within(final float x, final float y, final float err) { |
|
42 final float d = y - x; |
|
43 return (d <= err && d >= -err); |
|
44 } |
|
45 |
|
46 static boolean within(final double x, final double y, final double err) { |
|
47 final double d = y - x; |
|
48 return (d <= err && d >= -err); |
|
49 } |
|
50 |
|
51 static int quadraticRoots(final float a, final float b, |
|
52 final float c, float[] zeroes, final int off) |
|
53 { |
|
54 int ret = off; |
|
55 float t; |
|
56 if (a != 0f) { |
|
57 final float dis = b*b - 4*a*c; |
|
58 if (dis > 0) { |
|
59 final float sqrtDis = (float)Math.sqrt(dis); |
|
60 // depending on the sign of b we use a slightly different |
|
61 // algorithm than the traditional one to find one of the roots |
|
62 // so we can avoid adding numbers of different signs (which |
|
63 // might result in loss of precision). |
|
64 if (b >= 0) { |
|
65 zeroes[ret++] = (2 * c) / (-b - sqrtDis); |
|
66 zeroes[ret++] = (-b - sqrtDis) / (2 * a); |
|
67 } else { |
|
68 zeroes[ret++] = (-b + sqrtDis) / (2 * a); |
|
69 zeroes[ret++] = (2 * c) / (-b + sqrtDis); |
|
70 } |
|
71 } else if (dis == 0f) { |
|
72 t = (-b) / (2 * a); |
|
73 zeroes[ret++] = t; |
|
74 } |
|
75 } else { |
|
76 if (b != 0f) { |
|
77 t = (-c) / b; |
|
78 zeroes[ret++] = t; |
|
79 } |
|
80 } |
|
81 return ret - off; |
|
82 } |
|
83 |
|
84 // find the roots of g(t) = d*t^3 + a*t^2 + b*t + c in [A,B) |
|
85 static int cubicRootsInAB(float d, float a, float b, float c, |
|
86 float[] pts, final int off, |
|
87 final float A, final float B) |
|
88 { |
|
89 if (d == 0) { |
|
90 int num = quadraticRoots(a, b, c, pts, off); |
|
91 return filterOutNotInAB(pts, off, num, A, B) - off; |
|
92 } |
|
93 // From Graphics Gems: |
|
94 // http://tog.acm.org/resources/GraphicsGems/gems/Roots3And4.c |
|
95 // (also from awt.geom.CubicCurve2D. But here we don't need as |
|
96 // much accuracy and we don't want to create arrays so we use |
|
97 // our own customized version). |
|
98 |
|
99 /* normal form: x^3 + ax^2 + bx + c = 0 */ |
|
100 a /= d; |
|
101 b /= d; |
|
102 c /= d; |
|
103 |
|
104 // substitute x = y - A/3 to eliminate quadratic term: |
|
105 // x^3 +Px + Q = 0 |
|
106 // |
|
107 // Since we actually need P/3 and Q/2 for all of the |
|
108 // calculations that follow, we will calculate |
|
109 // p = P/3 |
|
110 // q = Q/2 |
|
111 // instead and use those values for simplicity of the code. |
|
112 double sq_A = a * a; |
|
113 double p = 1.0/3 * (-1.0/3 * sq_A + b); |
|
114 double q = 1.0/2 * (2.0/27 * a * sq_A - 1.0/3 * a * b + c); |
|
115 |
|
116 /* use Cardano's formula */ |
|
117 |
|
118 double cb_p = p * p * p; |
|
119 double D = q * q + cb_p; |
|
120 |
|
121 int num; |
|
122 if (D < 0) { |
|
123 // see: http://en.wikipedia.org/wiki/Cubic_function#Trigonometric_.28and_hyperbolic.29_method |
|
124 final double phi = 1.0/3 * acos(-q / sqrt(-cb_p)); |
|
125 final double t = 2 * sqrt(-p); |
|
126 |
|
127 pts[ off+0 ] = (float)( t * cos(phi)); |
|
128 pts[ off+1 ] = (float)(-t * cos(phi + PI / 3)); |
|
129 pts[ off+2 ] = (float)(-t * cos(phi - PI / 3)); |
|
130 num = 3; |
|
131 } else { |
|
132 final double sqrt_D = sqrt(D); |
|
133 final double u = cbrt(sqrt_D - q); |
|
134 final double v = - cbrt(sqrt_D + q); |
|
135 |
|
136 pts[ off ] = (float)(u + v); |
|
137 num = 1; |
|
138 |
|
139 if (within(D, 0, 1e-8)) { |
|
140 pts[off+1] = -(pts[off] / 2); |
|
141 num = 2; |
|
142 } |
|
143 } |
|
144 |
|
145 final float sub = 1.0f/3 * a; |
|
146 |
|
147 for (int i = 0; i < num; ++i) { |
|
148 pts[ off+i ] -= sub; |
|
149 } |
|
150 |
|
151 return filterOutNotInAB(pts, off, num, A, B) - off; |
|
152 } |
|
153 |
|
154 // These use a hardcoded factor of 2 for increasing sizes. Perhaps this |
|
155 // should be provided as an argument. |
|
156 static float[] widenArray(float[] in, final int cursize, final int numToAdd) { |
|
157 if (in.length >= cursize + numToAdd) { |
|
158 return in; |
|
159 } |
|
160 return Arrays.copyOf(in, 2 * (cursize + numToAdd)); |
|
161 } |
|
162 |
|
163 static int[] widenArray(int[] in, final int cursize, final int numToAdd) { |
|
164 if (in.length >= cursize + numToAdd) { |
|
165 return in; |
|
166 } |
|
167 return Arrays.copyOf(in, 2 * (cursize + numToAdd)); |
|
168 } |
|
169 |
|
170 static float evalCubic(final float a, final float b, |
|
171 final float c, final float d, |
|
172 final float t) |
|
173 { |
|
174 return t * (t * (t * a + b) + c) + d; |
|
175 } |
|
176 |
|
177 static float evalQuad(final float a, final float b, |
|
178 final float c, final float t) |
|
179 { |
|
180 return t * (t * a + b) + c; |
|
181 } |
|
182 |
|
183 // returns the index 1 past the last valid element remaining after filtering |
|
184 static int filterOutNotInAB(float[] nums, final int off, final int len, |
|
185 final float a, final float b) |
|
186 { |
|
187 int ret = off; |
|
188 for (int i = off; i < off + len; i++) { |
|
189 if (nums[i] >= a && nums[i] < b) { |
|
190 nums[ret++] = nums[i]; |
|
191 } |
|
192 } |
|
193 return ret; |
|
194 } |
|
195 |
|
196 static float polyLineLength(float[] poly, final int off, final int nCoords) { |
|
197 assert nCoords % 2 == 0 && poly.length >= off + nCoords : ""; |
|
198 float acc = 0; |
|
199 for (int i = off + 2; i < off + nCoords; i += 2) { |
|
200 acc += linelen(poly[i], poly[i+1], poly[i-2], poly[i-1]); |
|
201 } |
|
202 return acc; |
|
203 } |
|
204 |
|
205 static float linelen(float x1, float y1, float x2, float y2) { |
|
206 final float dx = x2 - x1; |
|
207 final float dy = y2 - y1; |
|
208 return (float)Math.sqrt(dx*dx + dy*dy); |
|
209 } |
|
210 |
|
211 static void subdivide(float[] src, int srcoff, float[] left, int leftoff, |
|
212 float[] right, int rightoff, int type) |
|
213 { |
|
214 switch(type) { |
|
215 case 6: |
|
216 Helpers.subdivideQuad(src, srcoff, left, leftoff, right, rightoff); |
|
217 break; |
|
218 case 8: |
|
219 Helpers.subdivideCubic(src, srcoff, left, leftoff, right, rightoff); |
|
220 break; |
|
221 default: |
|
222 throw new InternalError("Unsupported curve type"); |
|
223 } |
|
224 } |
|
225 |
|
226 static void isort(float[] a, int off, int len) { |
|
227 for (int i = off + 1; i < off + len; i++) { |
|
228 float ai = a[i]; |
|
229 int j = i - 1; |
|
230 for (; j >= off && a[j] > ai; j--) { |
|
231 a[j+1] = a[j]; |
|
232 } |
|
233 a[j+1] = ai; |
|
234 } |
|
235 } |
|
236 |
|
237 // Most of these are copied from classes in java.awt.geom because we need |
|
238 // float versions of these functions, and Line2D, CubicCurve2D, |
|
239 // QuadCurve2D don't provide them. |
|
240 /** |
|
241 * Subdivides the cubic curve specified by the coordinates |
|
242 * stored in the {@code src} array at indices {@code srcoff} |
|
243 * through ({@code srcoff} + 7) and stores the |
|
244 * resulting two subdivided curves into the two result arrays at the |
|
245 * corresponding indices. |
|
246 * Either or both of the {@code left} and {@code right} |
|
247 * arrays may be {@code null} or a reference to the same array |
|
248 * as the {@code src} array. |
|
249 * Note that the last point in the first subdivided curve is the |
|
250 * same as the first point in the second subdivided curve. Thus, |
|
251 * it is possible to pass the same array for {@code left} |
|
252 * and {@code right} and to use offsets, such as {@code rightoff} |
|
253 * equals ({@code leftoff} + 6), in order |
|
254 * to avoid allocating extra storage for this common point. |
|
255 * @param src the array holding the coordinates for the source curve |
|
256 * @param srcoff the offset into the array of the beginning of the |
|
257 * the 6 source coordinates |
|
258 * @param left the array for storing the coordinates for the first |
|
259 * half of the subdivided curve |
|
260 * @param leftoff the offset into the array of the beginning of the |
|
261 * the 6 left coordinates |
|
262 * @param right the array for storing the coordinates for the second |
|
263 * half of the subdivided curve |
|
264 * @param rightoff the offset into the array of the beginning of the |
|
265 * the 6 right coordinates |
|
266 * @since 1.7 |
|
267 */ |
|
268 static void subdivideCubic(float src[], int srcoff, |
|
269 float left[], int leftoff, |
|
270 float right[], int rightoff) |
|
271 { |
|
272 float x1 = src[srcoff + 0]; |
|
273 float y1 = src[srcoff + 1]; |
|
274 float ctrlx1 = src[srcoff + 2]; |
|
275 float ctrly1 = src[srcoff + 3]; |
|
276 float ctrlx2 = src[srcoff + 4]; |
|
277 float ctrly2 = src[srcoff + 5]; |
|
278 float x2 = src[srcoff + 6]; |
|
279 float y2 = src[srcoff + 7]; |
|
280 if (left != null) { |
|
281 left[leftoff + 0] = x1; |
|
282 left[leftoff + 1] = y1; |
|
283 } |
|
284 if (right != null) { |
|
285 right[rightoff + 6] = x2; |
|
286 right[rightoff + 7] = y2; |
|
287 } |
|
288 x1 = (x1 + ctrlx1) / 2.0f; |
|
289 y1 = (y1 + ctrly1) / 2.0f; |
|
290 x2 = (x2 + ctrlx2) / 2.0f; |
|
291 y2 = (y2 + ctrly2) / 2.0f; |
|
292 float centerx = (ctrlx1 + ctrlx2) / 2.0f; |
|
293 float centery = (ctrly1 + ctrly2) / 2.0f; |
|
294 ctrlx1 = (x1 + centerx) / 2.0f; |
|
295 ctrly1 = (y1 + centery) / 2.0f; |
|
296 ctrlx2 = (x2 + centerx) / 2.0f; |
|
297 ctrly2 = (y2 + centery) / 2.0f; |
|
298 centerx = (ctrlx1 + ctrlx2) / 2.0f; |
|
299 centery = (ctrly1 + ctrly2) / 2.0f; |
|
300 if (left != null) { |
|
301 left[leftoff + 2] = x1; |
|
302 left[leftoff + 3] = y1; |
|
303 left[leftoff + 4] = ctrlx1; |
|
304 left[leftoff + 5] = ctrly1; |
|
305 left[leftoff + 6] = centerx; |
|
306 left[leftoff + 7] = centery; |
|
307 } |
|
308 if (right != null) { |
|
309 right[rightoff + 0] = centerx; |
|
310 right[rightoff + 1] = centery; |
|
311 right[rightoff + 2] = ctrlx2; |
|
312 right[rightoff + 3] = ctrly2; |
|
313 right[rightoff + 4] = x2; |
|
314 right[rightoff + 5] = y2; |
|
315 } |
|
316 } |
|
317 |
|
318 |
|
319 static void subdivideCubicAt(float t, float src[], int srcoff, |
|
320 float left[], int leftoff, |
|
321 float right[], int rightoff) |
|
322 { |
|
323 float x1 = src[srcoff + 0]; |
|
324 float y1 = src[srcoff + 1]; |
|
325 float ctrlx1 = src[srcoff + 2]; |
|
326 float ctrly1 = src[srcoff + 3]; |
|
327 float ctrlx2 = src[srcoff + 4]; |
|
328 float ctrly2 = src[srcoff + 5]; |
|
329 float x2 = src[srcoff + 6]; |
|
330 float y2 = src[srcoff + 7]; |
|
331 if (left != null) { |
|
332 left[leftoff + 0] = x1; |
|
333 left[leftoff + 1] = y1; |
|
334 } |
|
335 if (right != null) { |
|
336 right[rightoff + 6] = x2; |
|
337 right[rightoff + 7] = y2; |
|
338 } |
|
339 x1 = x1 + t * (ctrlx1 - x1); |
|
340 y1 = y1 + t * (ctrly1 - y1); |
|
341 x2 = ctrlx2 + t * (x2 - ctrlx2); |
|
342 y2 = ctrly2 + t * (y2 - ctrly2); |
|
343 float centerx = ctrlx1 + t * (ctrlx2 - ctrlx1); |
|
344 float centery = ctrly1 + t * (ctrly2 - ctrly1); |
|
345 ctrlx1 = x1 + t * (centerx - x1); |
|
346 ctrly1 = y1 + t * (centery - y1); |
|
347 ctrlx2 = centerx + t * (x2 - centerx); |
|
348 ctrly2 = centery + t * (y2 - centery); |
|
349 centerx = ctrlx1 + t * (ctrlx2 - ctrlx1); |
|
350 centery = ctrly1 + t * (ctrly2 - ctrly1); |
|
351 if (left != null) { |
|
352 left[leftoff + 2] = x1; |
|
353 left[leftoff + 3] = y1; |
|
354 left[leftoff + 4] = ctrlx1; |
|
355 left[leftoff + 5] = ctrly1; |
|
356 left[leftoff + 6] = centerx; |
|
357 left[leftoff + 7] = centery; |
|
358 } |
|
359 if (right != null) { |
|
360 right[rightoff + 0] = centerx; |
|
361 right[rightoff + 1] = centery; |
|
362 right[rightoff + 2] = ctrlx2; |
|
363 right[rightoff + 3] = ctrly2; |
|
364 right[rightoff + 4] = x2; |
|
365 right[rightoff + 5] = y2; |
|
366 } |
|
367 } |
|
368 |
|
369 static void subdivideQuad(float src[], int srcoff, |
|
370 float left[], int leftoff, |
|
371 float right[], int rightoff) |
|
372 { |
|
373 float x1 = src[srcoff + 0]; |
|
374 float y1 = src[srcoff + 1]; |
|
375 float ctrlx = src[srcoff + 2]; |
|
376 float ctrly = src[srcoff + 3]; |
|
377 float x2 = src[srcoff + 4]; |
|
378 float y2 = src[srcoff + 5]; |
|
379 if (left != null) { |
|
380 left[leftoff + 0] = x1; |
|
381 left[leftoff + 1] = y1; |
|
382 } |
|
383 if (right != null) { |
|
384 right[rightoff + 4] = x2; |
|
385 right[rightoff + 5] = y2; |
|
386 } |
|
387 x1 = (x1 + ctrlx) / 2.0f; |
|
388 y1 = (y1 + ctrly) / 2.0f; |
|
389 x2 = (x2 + ctrlx) / 2.0f; |
|
390 y2 = (y2 + ctrly) / 2.0f; |
|
391 ctrlx = (x1 + x2) / 2.0f; |
|
392 ctrly = (y1 + y2) / 2.0f; |
|
393 if (left != null) { |
|
394 left[leftoff + 2] = x1; |
|
395 left[leftoff + 3] = y1; |
|
396 left[leftoff + 4] = ctrlx; |
|
397 left[leftoff + 5] = ctrly; |
|
398 } |
|
399 if (right != null) { |
|
400 right[rightoff + 0] = ctrlx; |
|
401 right[rightoff + 1] = ctrly; |
|
402 right[rightoff + 2] = x2; |
|
403 right[rightoff + 3] = y2; |
|
404 } |
|
405 } |
|
406 |
|
407 static void subdivideQuadAt(float t, float src[], int srcoff, |
|
408 float left[], int leftoff, |
|
409 float right[], int rightoff) |
|
410 { |
|
411 float x1 = src[srcoff + 0]; |
|
412 float y1 = src[srcoff + 1]; |
|
413 float ctrlx = src[srcoff + 2]; |
|
414 float ctrly = src[srcoff + 3]; |
|
415 float x2 = src[srcoff + 4]; |
|
416 float y2 = src[srcoff + 5]; |
|
417 if (left != null) { |
|
418 left[leftoff + 0] = x1; |
|
419 left[leftoff + 1] = y1; |
|
420 } |
|
421 if (right != null) { |
|
422 right[rightoff + 4] = x2; |
|
423 right[rightoff + 5] = y2; |
|
424 } |
|
425 x1 = x1 + t * (ctrlx - x1); |
|
426 y1 = y1 + t * (ctrly - y1); |
|
427 x2 = ctrlx + t * (x2 - ctrlx); |
|
428 y2 = ctrly + t * (y2 - ctrly); |
|
429 ctrlx = x1 + t * (x2 - x1); |
|
430 ctrly = y1 + t * (y2 - y1); |
|
431 if (left != null) { |
|
432 left[leftoff + 2] = x1; |
|
433 left[leftoff + 3] = y1; |
|
434 left[leftoff + 4] = ctrlx; |
|
435 left[leftoff + 5] = ctrly; |
|
436 } |
|
437 if (right != null) { |
|
438 right[rightoff + 0] = ctrlx; |
|
439 right[rightoff + 1] = ctrly; |
|
440 right[rightoff + 2] = x2; |
|
441 right[rightoff + 3] = y2; |
|
442 } |
|
443 } |
|
444 |
|
445 static void subdivideAt(float t, float src[], int srcoff, |
|
446 float left[], int leftoff, |
|
447 float right[], int rightoff, int size) |
|
448 { |
|
449 switch(size) { |
|
450 case 8: |
|
451 subdivideCubicAt(t, src, srcoff, left, leftoff, right, rightoff); |
|
452 break; |
|
453 case 6: |
|
454 subdivideQuadAt(t, src, srcoff, left, leftoff, right, rightoff); |
|
455 break; |
|
456 } |
|
457 } |
|
458 } |
|