opennorth / represent-boundaries Goto Github PK

• Add to http://poplus.org/components/current/
• represent.poplus.org landing page, and docs landing page
• Compare with MapIt generations
• Add examples to /import step
• Add examples to /manage step
• Copy API docs to represent.poplus.org/docs/api/
• Configure DNS
• Push represent-boundaries
• Update component_url
• Update footers on *.popit.org
• More images: images from represent.opennorth.ca in left column, images above each step, screenshots of websites, screenshot of demo

e.g. format=popolo

See http://www.popoloproject.com/specs/area.html and popolo-project/popolo-spec#59

Add a configuration variable to set the default format (classic by default).

Current quirks:

• If a shapefile happens to have _cleaned_ in its name (and if Represent Boundaries didn't create that file), it will not be loaded.
• If a ZIP file contains ZIP files, loadshapefiles won't decompress them, but it can handle a directory that contains ZIP files.
• If a ZIP file contains multiple shapefiles, loadshapefiles can't handle them, but it can handle a directory that contains multiple shapefiles.
• loadshapefiles will not recurse a directory looking for shapefiles, but it will recurse zipped directories.

As written, if extract_shapefile_from_zip were made to extract multiple shapefiles, and if two shapefiles in different parts of the zipped directory tree had the same name, it would overwrite those shapefiles. To avoid this, we need to recreate the zipped directory tree.

We should update create_data_sources and extract_shapefile_from_zip to do everything recursively and not have these quirks.

Hi @jpmckinney - I hope all is well!

It appears the service has been down (looking at client logs) today. I sent an email to peck@ and to represent@ but if there are DNS issues those may not work either. So I thought I'd try to ping you on GitHub.

• apibrowser.html
• API error messages (BadRequest, HttpResponseForbidden)
• management command errors (ValueError), help text, log messages, print statements
• model help text

Each boundaryset loaded should be a transaction, so we don't have half-finished imports.

For example, a boundary detail should perhaps contain the URL of its shape resource.

And perhaps also the URL for the resource describing its representatives? In other words, this would have to be something external apps could hook in to.

I've noticed that http://represent.poplus.org/ is down right now. Has it been moved somewhere else, and/or is it coming back?

The loader will walk the data/shapefiles/ directory tree and discover all files matching /.+definitions?.py$/

Those files will use a format like:

import boundaryservice

boundaryservice.register('Federal Electoral Districts',
    singular='Federal Electoral District',
    file='.', # relative to the path of the definitions file
    # etc.
)

Like open-contracting/software-development-handbook#101

The default behaviour of the loadshapefiles command should be to only load shapefiles where no corresponding BoundarySet already exists.

It seems like running python manage.py loadshapefiles --reload does not delete shapefiles that have previously been loaded. It does of course add new ones. I noticed that data/shapefiles has a remove_set.py file but it doesn't run successfully either.

Not sure if this is intentional/still needs to be built out, but wanted to point it out in case.

Once the experimental branch has settled a bit, documentation needs to be updated to reflect changes.

When I install represent-boundaries without specifying a Django version, I'm on 2.2.0, which does not support python manage.py syncdb --noinput. What is the most recent version of Django that is compatible with this app?

Add ?format=apibrowser to a URL to see pretty-printed JSON with docs.

I'm realizing that when upgrading to recent Heroku stacks (ex heroku-22), Heroku will complain about the Python version I've been using to run our instance of the boundary service (python-3.7.5).

I tried running an upgrade locally, only to realize that upgrading causes problems with Django, where it runs from django.utils.encoding import python_2_unicode_compatible. At least according to Stack Overflow, it might be possible to either remove that line entirely, or to run pip install six and from six import python_2_unicode_compatible.

I haven't been able test this yet, I'm trying to figure out if I can revert my setup to deploy anything, but I'll test it if I can figure that part out.

• If a definition file has no shapefiles, the boundary set is nonetheless created.

• eliminate dumps/loads by using less strict assertion
• Python 3 failures
• Django 1.7 failures (app registry not ready)
• boundary list filtering
• loadshapefiles (can still do more)
• compute_intersections (skipped)
• autodiscover
• pep8

http://poplus.org/components/examples/

• Remove throttling
• Add start_date and end_date to BoundarySet
• Tag and release

An at filter to exclude boundaries whose boundary set's date range doesn't overlap with the given time.