nnsave_backend

The backend for the NNSave app for the iNNovate hackathon

Installation

Packages

In the current state, this only runs on linux. You need to have the following installed (these are the debian package names):

python-dev
libpq-dev
postgresql-9.5
postgresql-contrib-9.5
postgresql-9.5-postgis-2.2
postgresql-9.5-postgis-scripts
postgis
binutils
libproj-dev
gdal-bin
python-gdal
python-pip

These package might not be entirely correct, and there might be some extra packages.

VirtualEnv

You also need virtualenv, which can be installed through pip

sudo pip install virtualenv

You should be able to source the existing virtualenv in venv/bin/activate, but if you need to recreate the venv, here is the packaged installed:

pip install django psycopg2 django-import-export

Postgresql setup

You will need to setup a postgres database with a user and a few options set. The database needs the postgis extension loaded. Here is what I ran to create the database:

sudo su - postgres
psql

CREATE DATABASE nnsave;
CREATE USER nnsave_user WITH PASSWORD 'nnsave_password';
ALTER ROLE nnsave_user SUPERUSER;
ALTER ROLE nnsave_user SET client_encoding TO 'utf8';
ALTER ROLE nnsave_user SET default_transaction_isolation TO 'read committed';
ALTER ROLE nnsave_user SET timezone TO 'UTC';
GRANT ALL PRIVILEGES ON DATABASE nnsave TO nnsave_user;
\q

psql nnsave -c 'CREATE EXTENSION postgis'
exit

Migrations

When first installing the server, the migrations must be applied to the database.

python manage.py migrate

Usage

To run the development server, source the virtual enviornment, then run the django dev server

source venv/bin/activate
python manage.py runserver 0.0.0.0:8000

Importing from CSV

One thing I had to do was import a list of discount locations, I did this by saving a spreadsheet as a csv. The first row of the csv should be the column headers. There should be a column for every field on the object. I ran python manage.py shell to enter the django shell then ran the following python code. This is written from memory and might not be correct. This code assumes that every row has at least a valid name, description, latitude and longitude

import csv
from nnsave_app.models import Location, Category
from django.contrib.gis.geos import Point
myfile = open('path/to/import.csv', 'r+b')
mycsv = csv.DictReader(myfile)
for row in mycsv:
     pnt = Point(row['Latitude'],row['Longitude'])
     cat, created = Category.objects.get_or_create(row['Category'].strip())
     loc, created = Location.objects.get_or_create(name=row['Name'].strip(),
                          phone=row['Phone'].strip(),
                          website=row['Website'].strip(),
                          email=row['Email'].strip(),
                          address=row['Address'].strip(),
                          loc=pnt,
                          description=row['Description'].strip(),
                          category=cat)

I had hoped to get Django import_export to work with this importing, but I couldn't figure out how to get Points with it. Maybe in the future. If import_export gets working it could be added to the admin interface to make it super simple to import locations.

It looks like I probably could have used GeoDjango's DataSource interface to do this. If you read the GeoDjango tutorial's section on "Importing Spatial Data", it tells you how to import a shp file. A shp file can be created from a csv using ogr2ogr. I think the following command will work for a file named import.csv located in /mydir:

ogr2ogr -f "ESRI Shapefile" /mydir import.csv
ogr2ogr -f "ESRI Shapefile" /mydir import.vrt

Then in the django shell:

from django.contrib.gis.utils import LayerMapping
from nnsave_app.models import Location
mapping = {} #This nees to be figured out
lm = LayerMapping(Location, 'import.shp', mapping)
lm.save(verbose=True)

Hackathon

If you are interestedin what code was written in the hackathon, everything up to commit c2c780e was done at the hackathon

legonigel/nnsave_backend