Import XaoS r489 (trunk after version 3.5)

1 files changed, 56 insertions, 0 deletions

diff --git a/src/engine/rotated.c b/src/engine/rotated.c
new file mode 100644
index 0000000..69e24fd
--- /dev/null
+++ b/src/engine/rotated.c
@@ -0,0 +1,56 @@
+#ifndef UNSUPPORTED
+static void do_rotate(void *data, struct taskinfo *task, int r1, int r2)
+{
+    struct filter *f = (struct filter *) data;
+    struct rotatedata *s = (struct rotatedata *) f->data;
+    double xstep = (s->x2 - s->x1) * 65536 / f->image->height;
+    double ystep = (s->y2 - s->y1) * 65536 / f->image->height;
+    double x = (s->x1) * 65536, y = (s->y1) * 65536;
+
+    int ixstep = (int) ((s->xx1 - s->x1) * 65536);
+    int iystep = (int) ((s->yy1 - s->y1) * 65536);
+    int i;
+
+    if (x < 0)
+	x = 0;			/*avoid shifting problems */
+    if (y < 0)
+	y = 0;
+
+    ixstep /= f->image->width;
+    iystep /= f->image->width;
+
+    /* I do floating point dda here since I expect that registers used by dda will 
+     * not conflict with registers of integer one used by main loop so it will be 
+     * faster than dda from stack :)
+     */
+    x += r1 * xstep;
+    y += r1 * ystep;
+    for (i = r1; i < r2; i++) {
+
+	{
+	    register int ix = (int) x;
+	    register int iy = (int) y;
+	    register cpixel_t **vbuff =
+		(cpixel_t **) f->childimage->currlines;
+	    register cpixel_t *end =
+		p_add((cpixel_t *) f->image->currlines[i],
+		      f->image->width), *dest =
+		(cpixel_t *) f->image->currlines[i];
+	    register int iixstep = ixstep, iiystep = iystep;
+
+	    while (dest < end) {
+		p_copy(dest, 0, (cpixel_t *) (vbuff[iy >> 16]),
+		       (ix >> 16));
+		p_inc(dest, 1);
+		ix += iixstep;
+		iy += iiystep;
+	    }
+	}
+
+	x += xstep;
+	y += ystep;
+
+    }
+}
+#endif
+#undef do_rotate