Per our press person, we should add facts and figures to our webpage which will help demonstrate why there is a need to do this. For example, the sworn affidavit section should link to this statement source on CPD's website.

State law requires that any person making an allegation of misconduct against a Chicago Police Officer sign a sworn affidavit that certifies that the allegation is true and correct. If the person making the complaint did not actually witness the alleged conduct, they must certify that they believe that the facts in the allegation are true.

Illinois law (50 ILCS 725/3.8) requires that anyone filing a complaint against a sworn peace officer must have the complaint supported by a sworn affidavit.
Any person who knowingly files a false complaint may be subject to criminal charges or a civil suit. A person commits perjury when, under oath or affirmation, in a proceeding or in any other matter where by law such oath or affirmation is required, he makes a false statement, material to the issue or point in question, which he does not believe to be true. Perjury is a Class 3 felony.

Also, according to data from the Invisible Institute only 72% of complaints have an ID'ed officer in their older data set.

We are getting an error in the application from gunicorn when I submit the following query:
a white female commander aged 35-55. Mason and I suspect that this is because of an empty set being returned.

Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: [2016-10-04 02:52:31,264] ERROR in app: Exception on /lineup [POST]
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: Traceback (most recent call last):
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: File "/usr/local/lib/python2.7/dist-packages/flask/app.py", line 1988, in wsgi_app
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: response = self.full_dispatch_request()
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: File "/usr/local/lib/python2.7/dist-packages/flask/app.py", line 1641, in full_dispatch_request
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: rv = self.handle_user_exception(e)
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: File "/usr/local/lib/python2.7/dist-packages/flask/app.py", line 1544, in handle_user_exception
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: reraise(exc_type, exc_value, tb)
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: File "/usr/local/lib/python2.7/dist-packages/flask/app.py", line 1639, in full_dispatch_request
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: rv = self.dispatch_request()
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: File "/usr/local/lib/python2.7/dist-packages/flask/app.py", line 1625, in dispatch_request
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: return self.view_functionsrule.endpoint
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: File "/home/nginx/oovirtenv/OpenOversight/OpenOversight/app/views.py", line 59, in get_lineup
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: officer_images=officer_images)
Oct 04 02:52:31 openoversight.lucyparsonslabs.com gunicorn[1353]: UnboundLocalError: local variable 'officer_images' referenced before assignment

We should be storing our images on S3 (not in the database). Right now I have the ~10k that I have collected on my local machine

At least flask-WTF requires 0.13.1, but we should also have minimum versions in that file.

We should do a basic penetration test and internal code audit so that we don't get hacked and can harden out application. This will also become more important once we fix #5 and if we have users logging in. I can lead this project.

There are a lot of officers in CPD - which can make for slow queries as the app tries to get a photo for thousands of officers. Let's have the app show 10-50 at a time.

Once we have migrations and a local dev environment, we'll want the ability to deploy this thing in a far more straightforward manner than we currently have. I'm open to suggestions; I'm thinking either a rake task that rsyncs things or a capistrano setup, but that's because most of my deployment orchestration experience is in Ruby. For all I know, there's a better solution here, and if so, we should find it and use it.

We need a short guide on what each rank looks like on an officer's uniform. Can be done in a sidebar on the submission form template or on a separate page

Current theme is very grey. There are a few nice default bootstrap themes.

Do face detection automatically with images submitted by users and those that are scraped from social media

A user might look through the list and want to flag the N officers that look similar to the one that they interacted with. They should ideally be able to flag these officers, then return and review them to compare and see if they recognize the one.

from this repo. Face detection is fine

We can sort the returned list of officers by proximity, e.g. if the location of the incident is in district 11 we should display the officers assigned to district 11 first. Note that some officers do not have a unit assignment that corresponds to a particular part of the city as well as the fact that some officers may be assigned to one district but interacting with citizens in another district.

We need to describe the structure for the data in case anyone wants to collaborate with us.

When user submits form, generate the digital line up

We got a request for stripping of metadata from some submitted images, but really we should just be doing this for all images. Potentially the only metadata to keep would be timestamp EXIF data that has when the image was taken (something we are trying to keep track of in the database), so we should grab that and dump the rest.

Consider making a way to link directly to Invisible Institute's database of office complaints. Similar to issue #10

so that @r4v5 can vagrant the heck out of it. I will get on this Wednesday night!

We should give people a contact address and inform them we're interested to know if there is something wrong with our data. Perhaps we can find out inaccuracies and correct them

Theme in FX looks terrible

This stuff should fit on one or at most two lines in the footer of the page:

There should be some visual indication to a user after they hit go "Go!" that the gallery is loading

And see if there is a way to autofill the name and badge number. On first glance it does not look like there is an easy way to do this on IPRA's portal...

We like databases, but if we have a dev environment that isn't "work by connecting straight to prod", we're going to need to change them as we add new things. We should use something like flask-migrate to manage DB migrations as the project grows. This is separate from the snapshotting and access control configuration we'll need to do once this goes live.

Currently its failing to start but if I run gunicorn app:app manually it functions properly.

We need to get rid of non-officer faces in the images to protect the privacy of citizens. We should incorporate redaction into our tagging app by enabling people to flag citizen faces that should be blocked out. Once a sufficient number of tags from trustworthy users for a given face are present we can blank out that area in the image (this is to prevent officers from going in and blanking out officer faces). We'll preserve the raw hashes of images generated upon upload in #5.

We've been writing code to scrape social media, but we should definitely automate this more:

• Each scraper should run every day and automatically upload images into the raw_images table for volunteers to tag
• Each scraper should get only the images that have been put up since the last time we scraped
• Each scraper should be putting the timestamps into the database. We want to keep track of how old each photograph is if possible (currently, all photographs in the database do have timestamps tracking when they were inserted).

We need some integration tests to check the end-to-end functionality of our webapp. Should probably use selenium for this

Describe our privacy policy (we take your privacy very seriously!) in the webapp.

To enable users to SAFELY upload certain file formats. These uploads should go to S3 and then a row should be inserted into officers.raw_images.

Suggestion by one of our lawyers: we could use the Citizen's police data project to sort the returned lineup by number of complaints. Those officers that have a history of misconduct are likely to reoffend and we might make it easier for people to find the offending officers by putting them at the top. This approach is similar to search engines putting the most popular sites near the top of the search results list.

There are a lot of high-quality photos with very visible names and badge numbers are on Flickr.
Some groups worth initial scrape

We want an application that displays submitted images that have not been tagged yet, which one can grab by select * from officers.raw_images where is_tagged = 'f'.

Then this application would allow people to click on a face and drag a box to select the cutout of the face - this is what will be stored in face_position in table faces and displayed in the digital lineup. Then next to each face, there should be two fields: one for name and one for badge number which the user should fill in one or both of. The user should be able to tag a single image multiple times.

Upon submission of the form, if there is one officer that matches the name or badge number then we add a row in table faces linking the officer_id with the img_id at face_position. If there is more than one officer (e.g. in the case of a surname that is very common or a badge number that has been reused) or if there are zero officers (e.g. in the case of a new officer that does not appear in our roster yet) we will need to manually review it.

Set up travis and coveralls to run our tests. needs to be done once we make the repo public so we can get that free service

Current templates render poorly on mobile - make much more friendly for mobile interface (e.g. the navbar should collapse, etc.)

We're looking for additional photographs of officers in public places for the database. It turns out there are lots of public cameras we might be able to use. Someone should try to FOIA for POD camera data from times and locations there are likely to be police officers in public spaces (e.g. during the day). Resolution might be an issue, but it's worth a shot, and they do have zoom capability so there might be some rubies and gems in there. Let me know if you want to do this - I have a list of the POD camera locations so you can pick a good one.

We want to take in information about where incidents occur so that eventually we can use officer allocation information to identify officers. For now as a placeholder, there are latitude and longitude fields in the 'Find an Officer Form' which are just decimal fields. We should replace these two fields with an inset map: users can just drop a pin on this inset map and then the lat/long of that position should be sent to the backend.

Per our press person, we need to add links that point to our SM and more text about LPL.

According to OpenOversight/database/readme.md the database can be created by running the script at OpenOversight/createdb.py. This script doesn't appear to exist at this location. I've run the one at OpenOversight/OpenOversight/createdb.py but it appears to expect an environmental variable PGPASS. I'm assuming there should be a initial script that reads the .pgpass file and sets the variable so it feels like a file may be missing at OpenOversight/createdb.py. Please advise. Thanks.

We should start trying to keep track of when photos were taken.

I tagged a bunch of photos and manually made cutouts of the officer faces. In the future we should implement face_position to make cutouts on the fly of the faces.

The most recent "Date of Employment" in our current roster is "2015-12-14 00:00:00". Time to update! More broadly, this is something that we will need to do periodically to keep data fresh. Also, we might want to consider adding photographs and A&A sheets to the list of things to FOIA for this project (so that we have information about where an officer is assigned in the city - this is necessary for #12).

It would be good for people with some information to be able to search quickly and confirm "officer friendly"'s face matches their recollection.

If we want to develop with any amount of rigor, we should have a dev environment. This is probably (because I'm doing it, and I stick with the devils I know) going to be a vagrant box provisioned via puppet, with a helluva lot of rake magic. I'm also going to try to make this match the production schema with dummy values, but an actual migrations management system is also work we should do (in a separate issue).

We should avoid nasty bugs by writing an initial test suite once we have vagrant setting up our local postgres database for testing

For now in our queries we assume that the incident in question is recent - we should also include the incident date in the web form. Then we should select the right row in the assignments table such that we can select officers based on their rank and assignment at the time of the incident.

currently if you run echo "localhost:5432:openoversight-dev:openoversight:terriblepassword" > ~/.pgpass more than once, it will throw a "too many keys" value because it appends it to the pgpass file. I manually edited it and removed all but one lines and it worked fine after a soure ~/.bashrc so I know its the >> append redirect.