I think the easiest way could be using getBoundingClientRect() which returns the element's position relative to the viewport.

Based on that, we can calculate the number of pixel to scroll so that the element is in the middle of view port.

Ben: do you think this seems like a sound strategy?

http://caniuse.com/#search=scrollintoview

Sorry, clarifying question: does the scroll need to be smooth? If so, you can use jquery or manually animate window.scrollTo() as Ned suggests. If not, element.scrollIntoView() should work.

We are trying to emulate real user action, and it looks like element.scrollIntoView() just jumps to the element (based on local experimenting); so I am thinking it is probably not a good fit for this.

Ah, yep, you can do something like this:
https://github.com/GabrielDelepine/smooth-scroll
but maybe requestAnimationFrame instead of setTimeout.

Ben: we have hook in Chrome for synthesize the smooth scroll already (see https://cs.chromium.org/chromium/src/third_party/catapult/telemetry/telemetry/internal/actions/scroll.js?rcl=0&l=35)

For this bug, the problem we are trying to solve is figure out how much to scroll to get to the element. After that, we can just pass the direction & speed to existing code in https://cs.chromium.org/chromium/src/third_party/catapult/telemetry/telemetry/internal/actions/scroll.py?rcl=0&l=90

+Some other folks: does any one has bandwidth to take on implementing this? 

I have a quick quesiton about how EvaluateJavaScript in the action runner works..

On the same page I run the following command: document.querySelectorAll(".pinImageDim")[5].getBoundingClientRect()

Running from the JS console I get:
ClientRect {top: 765, right: 1102.5, bottom: 1118, left: 866.5, width: 236…}

Running from telemetry via action_runner.EvaluateJavaScript('document.querySelectorAll(".pinImageDim")[5].getBoundingClientRect()') I get
{}

Shouldn't they return the same value? 

They should return the same value. Randy, do you have a CL? I will patch yours & see what's going on.

I dont have a CL for it, I was just experimenting.

In browsing_stories.py in the _MediaBrowsingStory class I just made _DidLoadDocument the following:

  def _DidLoadDocument(self, action_runner):
    print 'Test'
    print
    i = 'document.querySelectorAll("%s")[5]' % self.ITEM_SELECTOR
    i = i + '.getBoundingClientRect()'
    print i
    print action_runner.EvaluateJavaScript(i)

    index = self.ITEM_SELECTOR_INDEX
    for _ in xrange(self.ITEMS_TO_VISIT):
      print action_runner.EvaluateJavaScript(i)
      self._NavigateToItem(action_runner, index)
      self._ViewMediaItem(action_runner)
      if self.INCREMENT_INDEX:
        index += 1


It outputs "document.querySelectorAll(".pinImageDim")[5].getBoundingClientRect()" Which I then open the JSConsole mid run(which pauses telemetry) and see that it returns something, while the print statement 'print action_runner.EvaluateJavaScript(i) just prints {}.

Did you get a chance to try and repo this Ned?

Sorry, I am quite busy with issue 653519 & can't use my desktop do anything else besides running the benchmark with many repetition. 

For fast checking, can you do:
window.alert("document.querySelectorAll(".pinImageDim")[5].getBoundingClientRect()") to see if this is a js problem or telemetry problem?

nvm, my run just finished and I found the problem:
JSON.stringify(document.querySelectorAll(".pinImageDim")[5].getBoundingClientRect()) also return "{}" in devtool.

The reason is that getBoundingClientRect() doesn't return a raw dictionary, but an actual instance of a class. Therefore you can't convert it to json string & send back to telemetry.

For your case, I would suggest doing the code of computing the distance to scroll in js land & pass it to window.__scrollAction.start (similar to the code in https://cs.chromium.org/chromium/src/third_party/catapult/telemetry/telemetry/internal/actions/scroll.py?rcl=0&l=86)

Thanks for taking this, Randy!

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/0d801d5aa8208459720512e845fe82507dc7bea3

commit 0d801d5aa8208459720512e845fe82507dc7bea3
Author: catapult-deps-roller <catapult-deps-roller@chromium.org>
Date: Fri Oct 14 20:44:05 2016

Roll src/third_party/catapult/ baf2eab47..f506b6809 (2 commits).

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/baf2eab473e1..f506b6809e99

$ git log baf2eab47..f506b6809 --date=short --no-merges --format='%ad %ae %s'
2016-10-14 rnephew Add ScrollPageToElement to action_runner.
2016-10-14 lpy Fix the usage of StackFrame in tests.

BUG= 651909 

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel
TBR=catapult-sheriff@chromium.org

Review-Url: https://codereview.chromium.org/2422723002
Cr-Commit-Position: refs/heads/master@{#425458}

[modify] https://crrev.com/0d801d5aa8208459720512e845fe82507dc7bea3/DEPS