gnustep/libs-back
0
0
Fork 0
mirror of https://github.com/gnustep/libs-back.git synced 2025-04-22 15:31:14 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Detect and handle large values by scaling them down in order to prevent overflow on large images.

Browse source 
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/back/trunk@17702 72102866-910b-0410-8b05-ffd578937521
This commit is contained in:
alexm 2003-09-22 11:06:19 +00:00
parent 69fb0dcf76
commit 71d48e1f2c
2 changed files with 78 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
ChangeLog
View file

@ -1,3 +1,10 @@
2003-09-22 13:03 Alexander Malmberg <alexander@malmberg.org>
* Source/art/image.m (-_image_do_rgb_transform:::): Check if the
coordinates involved are too large to handle normally. If they
are, scale down the numbers a bit (gives less accuracy, but no
overflow).
2003-09-20 Adam Fedor <fedor@gnu.org>
* Headers/xlib/XGGState.h: Add GSReadRect def.

79
Source/art/image.m
View file

@ -269,6 +269,17 @@ seem to cause edges to be off by a pixel
else
left_delta = -1;
/* printf("x/y (%i %i) (%i %i) (%i %i) (%i %i) t (%i %i) (%i %i) (%i %i) (%i %i) le=%i\n",
x[0],y[0],
x[1],y[1],
x[2],y[2],
x[3],y[3],
tx[0],ty[0],
tx[1],ty[1],
tx[2],ty[2],
tx[3],ty[3],
le);*/
/* silence the compiler */
lx = lx_frac = ldx = ldx_frac = l_de = 0;
rx = rx_frac = rdx = rdx_frac = r_de = 0;
@ -375,12 +386,18 @@ seem to cause edges to be off by a pixel
rty++, rty_frac -= r_de;
}
/* printf("cy=%i left(x=%i, tx=%i, ty=%i) right(x=%i, tx=%i, ty=%i)\n",
cy,
lx,ltx,lty,
rx,rtx,rty);*/
if (cy >= clip_y0 && rx > lx)
{
render_run_t ri;
int x0, x1, de;
int tx, ty, tx_frac, ty_frac, dtx, dty, dtx_frac, dty_frac;
int delta;
int scale_factor;
x0 = lx;
x1 = rx;
@ -388,26 +405,53 @@ seem to cause edges to be off by a pixel
tx = ltx;
ty = lty;
tx_frac = ltx_frac * de * r_de;
ty_frac = lty_frac * de * r_de;
de *= r_de * l_de;
if (r_de * l_de > 0x10000)
{
/*
If the numbers involved are really large, scale things down
a bit. This loses some accuracy, but should keep things in
range.
*/
scale_factor = 10;
tx_frac = ltx_frac * ((de * r_de) >> scale_factor);
ty_frac = lty_frac * ((de * r_de) >> scale_factor);
de *= (r_de * l_de) >> scale_factor;
}
else
{
tx_frac = ltx_frac * de * r_de;
ty_frac = lty_frac * de * r_de;
scale_factor = 0;
de *= r_de * l_de;
}
delta = (rtx * r_de + rtx_frac) * l_de
- (ltx * l_de + ltx_frac) * r_de;
delta = ((rtx * r_de + rtx_frac) >> scale_factor) * l_de
- ((ltx * l_de + ltx_frac) >> scale_factor) * r_de;
dtx = delta / de;
dtx_frac = delta % de;
delta = (rty * r_de + rty_frac) * l_de
- (lty * l_de + lty_frac) * r_de;
delta = ((rty * r_de + rty_frac) >> scale_factor) * l_de
- ((lty * l_de + lty_frac) >> scale_factor) * r_de;
dty = delta / de;
dty_frac = delta % de;
/* printf("r_de=%i l_de=%i de=%i dtx=%i dtx_frac=%i dty=%i dty_frac=%i\n",
r_de,l_de,de,dtx,dtx_frac,dty,dty_frac);
printf(" start at x(%i %i) y(%i %i)\n",
tx,tx_frac,ty,ty_frac);*/
/*
x0 x1 / x2 -> y0 y1 / y2 z
((y0 * y2 + y1) * x2 - (x0 * x2 + x1) * y2) / (x2 * y2 * z)
*/
if (x0 < clip_x0)
/*
TODO: Would like to use this code instead for speed, but
the multiplications dtx*(clip_x0-x0) may overflow. Need to figure
out a way of doing this safely, or add checks for 'large numbers'
so we can use this in most cases, at least.
*/
/* if (x0 < clip_x0)
{
tx += dtx * (clip_x0 - x0);
ty += dty * (clip_x0 - x0);
@ -422,6 +466,25 @@ seem to cause edges to be off by a pixel
while (ty_frac >= de)
ty++, ty_frac -= de;
x0 = clip_x0;
}*/
if (x0 < clip_x0)
{
tx += dtx * (clip_x0 - x0);
ty += dty * (clip_x0 - x0);
while (x0 < clip_x0)
{
tx_frac += dtx_frac;
if (tx_frac < 0)
tx--, tx_frac += de;
if (tx_frac >= de)
tx++, tx_frac -= de;
ty_frac += dty_frac;
if (ty_frac < 0)
ty--, ty_frac += de;
if (ty_frac >= de)
ty++, ty_frac -= de;
x0++;
}
}
if (x1 >= clip_x1) x1 = clip_x1 - 1; /* TODO? */