My question: Can svg <marker> elements inherit color from the <line> they are referenced on?
The background: I have a D3 graph that has different styled lines, and I want my lines to have arrows at the end.
So at the top of my <svg> I have const defs = svg.append('defs'); and then further along I generate my defs using a generator function:
function makeDefs(defs: Selection<SVGDefsElement, unknown, null, undefined>, color: string, status: string) {
const markerSize = 3
const start = defs.append
.append('marker')
.attr('id', `arrow-start-${color}-${status}`)
.attr('viewBox', '-5 -10 20 20')
.attr('markerWidth', markerSize)
.attr('markerHeight', markerSize)
.attr('orient', 'auto-start-reverse');
start
.append('path')
.attr(
'd',
status === 'PUBLISHED' ? customPaths.arrowLarge : customPaths.arrowSmall
)
.attr('stroke', color)
.attr('fill', color);
}
And use it like so:
makeDefs(defs, 'red', 'DRAFT');
And then I add the markers to my lines with:
// d3 code to draw the lines etc
line.attr(
'marker-start',
(d) =>
`url(
#arrow-start-${d.color}-${d.status}
)`
);
This all works great, my arrows have lines. But my current setup feels burdensome and clunky. I have about 20 colors and 3 statuses. With my current setup that would be 60 different:
makeDefs(defs, 'one-of-20-colors', 'one-of-3-statues');
My understanding of markers is that they can inherit color using the currentcolor attribute. Currently my <defs> sit up top underneath my main <svg> so any color inherited is inherited directly from that top level svg which is not what I want. The issue in my case is my <line> elements are the elements who's color I want to inherit, but according to the MDN docs <line>s cannot have <defs> as children, thus leaving me with the only option, of defining all my <defs> up front all at once.
Is there a trick or some attribute I'm missing here?
Any way to pass color to my marker when doing:
line.attr(
'marker-start',
(d) =>
`url(
#arrow-start-${d.color}-${d.status}
)`
);
?
For what is is worth, I'm currently wrapping all my <line>s in <g>. I suppose I could wrap them in <svg>s instead, and apply the fill and stroke there, and then define my <defs> per svg container? I tried this briefly and swapping the <g> for an <svg> broke a lot, but I'm not even sure if it would work, or be better for that matter.
I'm trying to create an SVG element with a width defined by a percentage of the parent and a fixed value, say 50% + 20px. For normal html elements, in the CSS you can use calc(50% + 20px). Is there an equivalent way to do this for embedded SVGs? Specifically, I'm using snap.svg, though I'm not sure if this capability exists with SVGs in general.
EDIT:
Tried setting <svg> width with percentages and px, which I couldn't get to work. I tried both:
<svg width='calc(50% + 20px)'>...</svg>
<svg width='50% + 20px'>...</svg>
I also tried setting it in CSS:
svg {
width: calc(50% + 20px);
}
It should be possible with the upcoming SVG2 as width etc. become geometry properties and then you can style them with calc
I am trying to place an SVG Text-element according to the width and height of the text by getting the bounding box using the getBBox() method.
If the text is using a websafe font, this works reasonably well across different browsers, but if the text is styled using #font-face and a custom webfont, then the width of the text is returned incorrectly in Firefox (Mac) and Safari (iOS).
It works perfectly in both Safari (Mac) and Chrome (Mac).
If the gray box has the same width as the text, then it works in that browser.
Does anybody have an idea on how to get the correct width of the text bounding box in all browsers?
The browser is calculating the bounding box before it has finished loading/applying #font-face, assuming you don't need IE, you can wrap your BBox calculation function inside a document.fonts.ready promise...
document.fonts.ready.then(() => const bbox = textEl.getBBox());
Here is an example at work that exhibits the problem and the fix:
const xmlns = "http://www.w3.org/2000/svg";
const correct = document.getElementById("correct");
const incorrect = document.getElementById("incorrect");
visualizeBBox(incorrect);
document.fonts.ready.then(()=> visualizeBBox(correct));
function visualizeBBox(el){
const bbox = el.getBBox();
const rect = document.createElementNS(xmlns, "rect");
for (prop in bbox) rect.setAttribute(prop, bbox[prop]);
document.querySelector("svg").appendChild(rect);
}
svg text {
font-family: 'Diplomata SC', serif;
}
svg rect {
stroke: red;
fill: none;
}
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=Diplomata+SC&display=swap" rel="stylesheet">
<svg xmlns="https://www.w3.org/2000/svg" width="600" height="400">
<text x="0" y="40" font-size="24" id="correct">Correct dimensions</text>
<text y="100" font-size="24" id="incorrect">Incorrect dimensions</text>
<svg>
Today I ran into similar issue. Duopixel is right that getBBox() might return momental metric which may be unexpected because external font hasn't been loaded yet and some standard font is used instead.
The problem in WebKit (tested in Chrome 24.0.1312.52 and 26.0.1389.0 canary) is that the browser defers external font loading until it is first effectively used anywhere on the page. So even if you wait for onreadystatechange to become "complete" you are not guaranteed to have font metrics ready when calling getBBox() - you may still be the first one rendering a text styled with external font, inserting it into the document and immediately calling getBBox() on it (my case).
My workaround instead of calling mySVGInitCode() directly I do:
$("body").append(
$("<div/>")
.attr("class", "force-external-font-loading")
.attr("style", "font-family: \"xkcd\";visibility:hidden;position:absolute")
.text("x")
);
setTimeout(function(){ mySVGInitCode() }, 100); // 100ms is just arbitrary waiting time which should be sufficient for fetching the external font on a fast network, anyone has a better solution?
As you can see I dynamically insert absolutely positioned styled piece of text to force external font loading (visibility:hidden is important here instead of display:none). Then I wait some time before I execute my SVG code which could potentially render something and then immediately ask for metrics.
I get an XML or JSON with paths only, and I need to recreate the SVG image.
I create an empty
<svg xmlns='http://www.w3.org/2000/svg' version='1.1'></svg>,
I add a <g transform="scale(1 -1)" fill='#aaa' stroke='black' stroke-width='5' ></g> in it, and then in this element I add all of the paths in it (e.g. <path d= ... />).
In the end I get a SVG image, but because I haven't set the viewBox attribute in the SVG element the image isn't properly displayed - when I open it in browser, a part of it is displayed full size.
Can the viewBox be calculated from the values from the paths?
Thank you!
Similar to Martin Spa's answer, but a better way to do get the max viewport area is using the getBBox function:
var clientrect = path.getBBox();
var viewBox = clientrect.x+' '+clientrect.y+' '+clientrect.width+' '+clientrect.height;
You can then set the viewbox to these co-ordinates.
n.b. i think you can change the viewbox of an svg after it's rendered so you may have to re-render the svg.
OK so I solved it the following way:
removed all letters from the paths string and made an array out of it with
var values = pathValue.split('L').join(' ').split('M').join(' ').split('z').join('').split(' ');
found max and min from those values:
var max = Math.max.apply( Math, values );
var min = Math.min.apply( Math, values );
set the viewBox:
viewBox = max min max max
This worked in my case excellent. Hope that it will be helpful to someone else too.
For fun I am trying to see how far I can get at implementing an SVG browser client for a RIA I'm messing around with in my spare time.
But have hit what appears to be a HUGE stumbling block. There is no word wrap!!
Does anyone know of any work around (I'm thinking some kind of JavaScript or special tag I don't know)?
If not I'm either going to have to go the xhtml route and start sticking HTML elements in my SVG (ouch), or just come back again in ten years when SVG 1.2 is ready.
This SVG stuff is baffling, isn't it ?
Thankfully, you can achieve some good results, but it takes more work than using the HTML 5 .
Here's a screenshot of my ASP.Net / SVG app, featuring a bit of "faked" word wrapping.
The following function will create an SVG text element for you, broken into tspan pieces, where each line is no longer than 20 characters in length.
<text x="600" y="400" font-size="12" fill="#FFFFFF" text-anchor="middle">
<tspan x="600" y="400">Here a realy long </tspan>
<tspan x="600" y="416">title which needs </tspan>
<tspan x="600" y="432">wrapping </tspan>
</text>
It's not perfect, but it's simple, fast, and the users will never know the difference.
My createSVGtext() JavaScript function takes three parameters: an x-position, y-position and the text to be displayed. The font, maximum-chars-per-line and text color are all hardcoded in my function, but this can be easily changed.
To display the right-hand label shown in the screenshot above, you would call the function using:
var svgText = createSVGtext("Here a realy long title which needs wrapping", 600, 400);
$('svg').append(svgText);
And here's the JavaScript function:
function createSVGtext(caption, x, y) {
// This function attempts to create a new svg "text" element, chopping
// it up into "tspan" pieces, if the caption is too long
//
var svgText = document.createElementNS('http://www.w3.org/2000/svg', 'text');
svgText.setAttributeNS(null, 'x', x);
svgText.setAttributeNS(null, 'y', y);
svgText.setAttributeNS(null, 'font-size', 12);
svgText.setAttributeNS(null, 'fill', '#FFFFFF'); // White text
svgText.setAttributeNS(null, 'text-anchor', 'middle'); // Center the text
// The following two variables should really be passed as parameters
var MAXIMUM_CHARS_PER_LINE = 20;
var LINE_HEIGHT = 16;
var words = caption.split(" ");
var line = "";
for (var n = 0; n < words.length; n++) {
var testLine = line + words[n] + " ";
if (testLine.length > MAXIMUM_CHARS_PER_LINE)
{
// Add a new <tspan> element
var svgTSpan = document.createElementNS('http://www.w3.org/2000/svg', 'tspan');
svgTSpan.setAttributeNS(null, 'x', x);
svgTSpan.setAttributeNS(null, 'y', y);
var tSpanTextNode = document.createTextNode(line);
svgTSpan.appendChild(tSpanTextNode);
svgText.appendChild(svgTSpan);
line = words[n] + " ";
y += LINE_HEIGHT;
}
else {
line = testLine;
}
}
var svgTSpan = document.createElementNS('http://www.w3.org/2000/svg', 'tspan');
svgTSpan.setAttributeNS(null, 'x', x);
svgTSpan.setAttributeNS(null, 'y', y);
var tSpanTextNode = document.createTextNode(line);
svgTSpan.appendChild(tSpanTextNode);
svgText.appendChild(svgTSpan);
return svgText;
}
The logic for word-wrapping is based on this HTML5 Canvas tutorial
I hope you find this useful !
Mike
http://www.MikesKnowledgeBase.com
UPDATE
One thing I forgot to mention.
That "Workflow diagram" screen that I've shown above was originally just written using an HTML 5 canvas. It worked beautifully, the icons could be dragged, popup menus could appear when you clicked on them, and even IE8 seemed happy with it.
But I found that if the diagram became "too big" (eg 4000 x 4000 pixels), then the would fail to initialise in all browsers, nothing would appear - but - as far as the JavaScript code was concerned, everything was working fine.
So, even with error-checking, my diagram was appearing blank, and I was unable to detect when this showstopper problem was occurring.
var canvasSupported = !!c.getContext;
if (!canvasSupported) {
// The user's browser doesn't support HTML 5 <Canvas> controls.
prompt("Workflow", "Your browser doesn't support drawing on HTML 5 canvases.");
return;
}
var context = c.getContext("2d");
if (context == null) {
// The user's browser doesn't support HTML 5 <Canvas> controls.
prompt("Workflow", "The canvas isn't drawable.");
return;
}
// With larger diagrams, the error-checking above failed to notice that
// the canvas wasn't being drawn.
So, this is why I've had to rewrite the JavaScript code to use SVG instead. It just seems to cope better with larger diagrams.
There is also foreignObject tag. Then you can embed HTML in SVG which gives the greatest flexibility. HTML is great for document layout and has been hacked to no end to support application layout, drawing, and everything us developers want. But it's strength is word wrapping and document layout. Let HTML do what it does best, and let SVG do what it does best.
http://www.w3.org/TR/SVG/extend.html
This works for most browsers FireFox, Opera, Webkit, except IE (as of IE11). :-( Story of the web ain't it?
SVGT 1.2 introduces the textArea element http://www.w3.org/TR/SVGTiny12/text.html#TextInAnArea , but it is only experimentally supported by Opera 10 at the moment. I don't know if other browsers will ever plan on implementing it, though I hope they will.
Per this document, it appears that tspan can give the illusion of word wrap:
The tspan tag is identical to the text tag but can be nested inside text tags and inside itself. Coupled with the 'dy' attribute this allows the illusion of word wrap in SVG 1.1. Note that 'dy' is relative to the last glyph (character) drawn.
The svg.js library has a svg.textflow.js plugin. It's not ultra fast but it does the trick. It even stores overflowing text in a data attribute so you can use it to create continuously flowing columns. Here the text flow example page.
An alternative method is to use Andreas Neuman's text box object.
These days, flowPara can do word wrapping, but I have yet to find a browser that supports it properly.
I've been looking for a solution about word wrapping in svg so many hours (or many days).
If you can in your app, edit your code to put some tspan, or any other method, go in it.
Text wrapping will be implement in the 1.2 version but except opera, no browser fully implement it yet (4 years, the specification are on the W3 ...).
Because I had to use some alignment settings, i couldn't use any of the code that many forum can provide (no foreign object, no carto script or anything).
If I post this message, it's just in order to be usefull to some other people when googling word wrapping svg because this post on the top result and in many case, this post doesn't help.
Here is a cool, easy and light solution :
http://dev.w3.org/SVG/profiles/1.1F2/test/svg/text-dom-01-f.svg