Ahoy StackOverflow!
I've run into the following problem: on IE7 only, if I call drawImage() after drawing anything to the canvas, excanvas does not draw the image at the desired x,y coordinates. I have looked on the excanvas project page/ google group, and found that there are known issues with the drawImage() function. ( for example: http://code.google.com/p/explorercanvas/issues/detail?id=104&q=IE7 )
I have tried restoring the identity matrix, as suggested here: Excanvas vml positioning issue , though it appears to have had no effect.
Attached is a simple html document demonstrating this issue:
<!DOCTYPE HTML>
<html>
<head>
<script type="text/javascript" src="excanvas.js"></script>
<script type="text/javascript">
window.onload = function(){
var icon, ctx;
icon = new Image();
icon.width = "20";
icon.height = "20";
icon.src = "http://www.shangalulu.com/resources/images/icons/small/30.png";
icon.onload = function(){
var ctx;
ctx = document.getElementById('the_canvas').getContext('2d');
ctx.beginPath();
ctx.moveTo(0,0);
ctx.lineTo(500,0);
ctx.lineTo(500,500);
ctx.lineTo(0,500);
ctx.lineTo(0,0);
ctx.moveTo(200,200);
ctx.lineTo(300,200);
ctx.lineTo(300,300);
ctx.lineTo(200,300);
ctx.lineTo(200,200);
ctx.moveTo(0,0);
ctx.lineTo(500,500);
ctx.moveTo(0,500);
ctx.lineTo(500,0);
ctx.stroke();
ctx.drawImage(icon, 190, 190, 20, 20);
ctx.drawImage(icon, 240, 240, 20, 20);
ctx.drawImage(icon, 290, 290, 20, 20);
return;
}
}
</script>
</head>
<body>
<canvas id="the_canvas" width="500" height="500"></canvas>
</body>
</html>
If you need the excanvas library to run this, you can grab it from here: http://code.google.com/p/explorercanvas/
The above script should do the following:
What I would like to know is: Is there a known patch/workaround to this problem?
I've been pulling my hair on that one too...
I was only able to find a workaround, which works fine as long as you do not apply transformations before drawing your image. It implies that you replace the function drawImage in excanvas.js with that one:
contextPrototype.drawImage = function(image, var_args) {
var dx, dy, dw, dh, sx, sy, sw, sh;
// to find the original width we overide the width and height
var oldRuntimeWidth = image.runtimeStyle.width;
var oldRuntimeHeight = image.runtimeStyle.height;
image.runtimeStyle.width = 'auto';
image.runtimeStyle.height = 'auto';
// get the original size
var w = image.width;
var h = image.height;
// and remove overides
image.runtimeStyle.width = oldRuntimeWidth;
image.runtimeStyle.height = oldRuntimeHeight;
if (arguments.length == 3) {
dx = arguments[1];
dy = arguments[2];
sx = sy = 0;
sw = dw = w;
sh = dh = h;
} else if (arguments.length == 5) {
dx = arguments[1];
dy = arguments[2];
dw = arguments[3];
dh = arguments[4];
sx = sy = 0;
sw = w;
sh = h;
} else if (arguments.length == 9) {
sx = arguments[1];
sy = arguments[2];
sw = arguments[3];
sh = arguments[4];
dx = arguments[5];
dy = arguments[6];
dw = arguments[7];
dh = arguments[8];
} else {
throw Error('Invalid number of arguments');
}
/////////////////////////// MODIFIED BIT ///////////////////////////
var vmlStr = [];
vmlStr.push('<g_vml_:image src="', image.src, '"',
' style="position:absolute;',
' top:', dy, 'px;',
' left:', dx, 'px;',
' width:', dw, 'px;',
' height:', dh, 'px;"',
' cropleft="', sx / w, '"',
' croptop="', sy / h, '"',
' cropright="', (w - sx - sw) / w, '"',
' cropbottom="', (h - sy - sh) / h, '"',
' />');
/////////////////////////// END OF MODIFIED BIT ///////////////////////////
this.element_.insertAdjacentHTML('BeforeEnd',
vmlStr.join(''));
};
What it does is simply create a VML image directly inside the pseudo canvas, instead of wrapping it inside a group as it was originally done. For some reason, the original code was causing a left and top padding to appear around the image, and can't figure out why. Until we can understand the reason for that, we won't be able to find a proper fix...