Default screen height should be the documentHeight

[staylos92]

The default document height for screenshots is set to 1500px, this causes issues with some pages which are longer than this. Current workaround is to set something like:

screen_widths:

• 320x5000
• 600x5000
• 1024x5000

in order to get the full page in the screenshot. This obviously isn't ideal if we end up with a very long page.

[wildlyinaccurate]

We could quite easily set the viewport height to the full document height, but I wonder what effect that would have when diffing images with different heights?

Feel free to experiment -- you should be able to do this by changing the following line in javascript/_getDimensions.js from

'viewportHeight': parseInt(dimensions[2] || 1500)

to

'viewportHeight': parseInt(dimensions[2] || document.body.offsetHeight)

[ChrisBAshton]

Hmm, this isn't an easy fix. These dimensions are passed to Phantom/Casper before it loads the page - so the height is unreliable. Not to mention that the height might change at runtime, or after you do something with the before_capture hook.

I think what you'll want to do is put this in your YAML config:

before_capture: javascript/resizeHeight.js

And make a resizeHeight.js file something along the lines of:

module.exports = function (phantom) {
    phantom.viewportSize.height = document.body.offsetHeight;
    phantom.open(phantom.page.url);
}

[dsingleton]

I've been running into the same issue, wanting to get the the full page height, while using wraith with history/latest. We've used wraith in the past and the output was full height. I'm not sure if thats a change in wraiths behaviour, or perhaps in an un-versioned dependency on phantomjs behaving differently?

Currently, i'm trying to put together a wraith config for a new app, using the latest version of wraith, and a modern config (which is 👍 better). The behaviour we're after is full page height capture and image diffs.

I tried some variations of the phantom.viewportSize.height suggestion, but unfortunately couldn't get it to alter the capture height - it also looks like crop_images would override that before the diff is generated anyway :/

It sounds like this isn't supported right now, and not straight forward to add? If I can find the time i'd be interested in looking into this in more detail, but i'd be good to get a steer first 😄

(edit: sorry about the commit reference spam, there was some excessive rebasing)

[Gribnif]

I, too, consider this a huge failing of version 3.x, as capturing with an arbitrarily large height of, say, 10000px has a major performance impact. A job that used to take about 2-3 minutes with version 2.x now takes over 10 minutes to perform the same analysis.

Why was this changed from version 2? Can the old behavior be restored?

[ChrisBAshton]

I hear you - it would definitely be great to fix this.

Reopening the issue by popular demand, will try to look at this more closely in the coming weeks. In the meantime, pull requests are welcome!

Wraith v3.1.1 has more resilient image cropping, so should cope with dynamic heights much better, so hopefully this issue can be fixed more easily now.