4byfive

[joereddington]

Currently it's a little hard to get OVF to work with other systems because of the slightly unusual design of the pagesets (the CommuniKate pagesets have the message window between two of the buttons). I want OVF to be able to cope with more 'square' approachs.

I've written some tests, and now I want to get them working.

• add Argparse, run tests, and deploy
• remove gridSize from the internal structure
• Write tests for the four corners of the top page
• add a switch so that grab text knows if it should ignore the top row
• add the code so grab_text can ignore the top row
• look into why the top row and last column is missing entirely from cough drop
• upload the OBZ to Cough Drop (again)
• Do CK12 the same way (so 4 by 3)
• send both obz files to cough drop
• modify the Javascript so that it is an option
• add to the main page as an option.

[joereddington]

19/03/18 12:28 to 13:23

Currently OVF expects everything to be square, so we have some work to do to seperate out height and width. From there we can start including irregularly shaped grids. Unhelpfully, gridSize is a global.

We also needed to upgrate the arg parsing, which has happened. I'm a bit worried about what happens when we push to server with it.

After a bit of work, we've replaced all instances of gridSize, and refactored some parts. A trial run does produce a working aid, but there are tests failing that I need to look at.

23:13:

Looking at this again.

FAILED (failures=5, errors=2)

23:31:

FAILED (failures=1, errors=6)

That's an improvement... right ?

23:38:

FAILED (failures=5, errors=2)
Our situation needs to get better.

[joereddington]

20/03/18 13:49 to 14:03:

Let's have a look...(temp.json is the file that the test suite creates for the last json comparison)

python test_grab_text.py ovfTest.test_regession_ck20v2_bg   
opendiff temp.json regression_tests_size_5/CK20V2.pptx.json 

Turns out that the only difference was that I'd added an extra term into the json (both we need both height and width now). So

cp temp.json regression_tests_size_5/CK20V2.pptx.json

resets the test to be what is *currently produced.

I've repeated the process with the other failing tests.

14:04 to 14:28:

So what errors are left?

 python test_grab_text.py ovfTest.test_create_obf_object 

is failing. I've made some changes to the test so that I can use Opendiff to see the difference.

Turns out I needed to make quite a lot of changes and now I've drilled down to find the difference - this fail was actually caused by a different alternation I'd made and for some reason I didn't pick it up. This test is failing because the OBF generation changed to include special commands and was never updated. I'm going to update it now.

14:29 to 15:03:

Next one...

 python test_grab_text.py ovfTest.test_read_CK4by5_and_check_label

Ah, the code I was actually writing...

The problem here is that I've switch row and column somewhere and it's a little unclear where.

I created a new test (test_read_CK4by5_and_check_sizes) but then found that I got the same error because I needed to setup the grid first. Eventually found a place where row and column where switched.

It's probably good coding practice if everything is col(s) and row(s) rather than x/col/width - would probably reduce confusion...

15:07:

Another test was failing because I'd put the wrong number of slides in, easy fix that. One left now.

15:17:

This is the one. test_regession_ck12, which turns out to be fine if you simply remove the exception checking code I put in for the previous test. yay! All tests passing!

20/03/18 15:19:

Now we have to make sure that 'localcreate' produces a suitable 4by5 aid for us.

15:20 to 15:54:

Here's the string we're using.

./localcreate.sh tests/CK4by5cutdown.pptx 5 4 

• I've altered localcreate so it passes both arguments in.
• I've altered grab_text so that it listens to the arguments.

I ran the new 4by5 into our normal javascript display:

The labels are where I would expect, but some are missing - also the icons have the old position. Worth looking into why that is. Let's see if it works for cough drop (although it might be faster to do a branch for the javascript)

Wait... the icons are squashed.... two rows have been pushed together, like the icon recognition is out...

Oh, I know why that is - it's because I'm still making space for the title on the example slide. I need to account for that in the settings...

Uploaded the OBZ file to coughDrop anyway. There is what it looked like:

So the next steps are:

• Write tests for the four corners of the top page
• add a switch so that grab text knows if it should ignore the top row
• add the code so grab_text can ignore the top row
• look into why the top row and last column is missing entirely from cough drop
• upload the OBZ to Cough Drop (again)

I've written the tests and found the place for the switch - it is fairly deep so we might need to either make the args global or run them deep.

The OBF also is starting to look good. Pleased with that.

16:09:

I'm working out if I should stop here. Or if I should push on and have something to show for it. Let me at least try with the CK12...

16:14:

Let's have a push.

16:23:

It feels like the 4 by 3 is going quite well, let's have a look...

16:30:

Open board format site is taking a while...

16:38:

Found the CK12 uses the old way of specifying special commands and that was translating badly to OBF - I've fixed and I'm running again.

16:52:

Had a fairly significant error that turned out to be caused by my accidentally corrupting the pptx file. Pleased that Dropbox saved me.

16:59:

I've sent the email and I'm good with this for now.

[joereddington]

01/04/18 09:16:

(review) will badger Cough Drop)