9

i'm trying to save an image from a website using selenium server & python client. i know the image's URL, but i can't find the code to save it, either when it's the the document itself, or when it's embedded in the current browser session.

the workaround i found so far is to save the page's screenshot (there are 2 selenium methods for doing just that), but i want the original image.

i don't mind fiddling with the clicking menu options etc. but i couldn't found how.

thanks

Berry Tsakala
  • 15,313
  • 12
  • 57
  • 80
  • It 's been long since, but I am wondering, did you find a solution? – johnjohn Apr 10 '11 at 22:55
  • I have a solution for this. I extended the userscript.js of Selenium. Extended the rights of Firefox to capture screenshots in the userpref.js and used something similar to Scott Szretter to capture the screen and send the image data to the server. – powtac Mar 01 '12 at 11:51
  • Here is a trick that does this http://www.testingexcellence.com/how-to-download-and-save-images-with-selenium-rc/ – Amir Ghahrai Sep 04 '12 at 20:09

5 Answers5

5

I found code that puts an image in to a canvas, then converts it to data - which could then be base64 encoded for example. My thought was to call this using the eval command in selenium, however in my testing the toDataURL is throwing a security error 1000. Seems like it is so close to a solution if not for that error.

var data, canvas, ctx;
var img = new Image();
img = document.getElementById("yourimageID");
canvas = document.createElement('canvas');
canvas.width = img.width;
canvas.height = img.height;
ctx = canvas.getContext("2d");
ctx.drawImage(img, 0, 0);  // everything works up to here
data = canvas.toDataURL();  // this fails ***
var base64Img = data.replace(/^data:image\/(png|jpg);base64,/, "");

Doing some research I found references that it is not allowed to use toDataURL when the image is from a different domain. However, I even tried this code by saving the page, stripping everything out except the image itself and this script.

For example (index.html) :

<html><head></head><body>
<img src="local/hard/disk/img.jpg" id="yourimageID">
<script>
// script from above
</script>
</body></html>

The img.jpg and index.html are stored locally, opening the page in firefox locally, still get a security error 1000!

Scott Szretter
  • 3,938
  • 11
  • 57
  • 76
3

To do this the way you want (to actually capture the content sent down to the browser) you'd need to modify Selenium RC's proxy code (see ProxyHandler.java) and store the files locally on the disk in parallel to sending the response back to the browser.

Patrick Lightbody
  • 4,424
  • 2
  • 28
  • 38
  • That's ... interesting. But then it will save ALL files, and i'll have to guess which file belongs to which point in time, not to mention learning a bit of Java. But it's a possible solution. – Berry Tsakala Jun 15 '09 at 05:25
2

I was trying to accomplish the same task, but the images I wanted to grab were the size of my monitor (wallpaper) -- so the capture screenshot workaround didn't work for me. I figured out a way to do it...

I've got selenium set up to go to the page I want (which induces all the session goodies) Then I used a program called "Workspace Macro" to loop through the selenium tasks.

Grab it from here http://www.tethyssolutions.com/product.htm -- they have a trial version, which I think works for 30 runs or something.

So here's the progression:

  • start firefox
  • open selenium and load test case
  • start it, but quickly pause it.
  • record a macro, which pushes "step" on selenium, then goes over to the firefox window and clicks file->save page as, saves, then stop recording
  • run the macro x times...
  • profit??

Cheers

humanumbrella
  • 21
  • 3
0

I haven't used selenium, but if you know the image's URL, why not just do:

from urllib import urlretrieve

urlretrieve(url, filename)

which will save the url to the filename. more info here

Adrian Mester
  • 2,523
  • 1
  • 19
  • 23
  • 4
    because saving the image depends on the session. the reason for selenium in the first place is testing on a real environment. the server gets the url indeed, but it parses lots of environment variables related to the session before deciding which image to deliver, if any! – Berry Tsakala May 03 '09 at 11:30
  • then, maybe you can get the cookies from selenium and use urllib2 to also pass them in the request? – Adrian Mester May 03 '09 at 11:45
  • 3
    session is more than just cookies, and to mimic it in urrlib2, means to mimic Selenium RC, so why would i use selenium in the first place?? – Berry Tsakala Jun 15 '09 at 05:28
  • Not everyone uses selenium to have session-state, and images generally do not require authentication, so I think the answer will be useful to some people. – Frank Schwieterman Sep 24 '14 at 19:47
-1

How about going to the image URL and then taking a screenshot of the page? Firefox displays the image in full screen. Hope this helps..

Webber
  • 300
  • 3
  • 12