Aigle API

Deploy

Set up docker

1. Create local volume to persist db data:

docker volume create aigle_data

2. Create .env and .env.compose from templates
3. Build and run docker containers:

docker build -f Dockerfile -t aigle_api_app_container .
docker-compose --env-file .env -f docker-compose.yml up --force-recreate -d db app

Django

Add an app

python manage.py startapp my_app

Authentication

Authentication in this project is managed with djoser You can create a user by running command:

python manage.py create_super_admin --email [email protected] --password mypassword

Development

Set-up

This project is meant to be used with Python 3.12.3.

1. Create a virtual environment and activate it (here an example with venv)

python3 -m venv venv
source venv/bin/activate

2. Install dependencies

pip3 install -r requirements.txt

3. Create .env and .env.compose file and replace values

cp .env.template .env
cp .env.compose.template .env.compose

4. Run local server

source .env && source venv/bin/activate && make start

During the development, a graphic interface is provided by Django to test the API: make GET, POST,... requests easily. It is accessible by default on http://127.0.0.1:8000/

I recommend to use an extension like Requestly to add the token generated in the header and access to protected routes.

Commands

Some commands need to be run to seed data necessary for the app to works well.

# import all collectivites
python manage.py import_georegion
python manage.py import_geodepartment
python manage.py import_geocommune

# import hérault
python manage.py import_georegion --insee-codes 76
python manage.py import_geodepartment --insee-codes 34
python manage.py import_geocommune

# insert tiles: for montpellier and its surroundings
python manage.py create_tile --x-min 265750 --x-max 268364 --y-min 190647 --y-max 192325

# import parcels
python manage.py import_parcels

Useful SQL queries

Custom zones

Custom zones are big objects so we need to pre-compute and store in database associated zones for each detections. To do so, there is a command BUT this command is not really performant when updating a lot of objects. So here is a SQL request to create links between detections and custom zones:

Query

insert
	into
	core_detectionobject_geo_custom_zones(
	detectionobject_id,
	geocustomzone_id
)
select
	distinct
	dobj.id as detectionobject_id,
	{custom_zone_id} as geocustomzone_id
from
	core_detectionobject dobj
join core_detection detec on
	detec.detection_object_id = dobj.id
where
	ST_Intersects(
		detec.geometry,
		(
		select
			geozone.geometry
		from
			core_geozone geozone
		where
			id = {custom_zone_id}
		)
	)
on conflict do nothing;

Data from previous AIGLE version (SIA database)

This is the query to get data from previous AIGLE version (SIA database):

Query

select
    rel.id,
    rel.polygon as "geometry",
    case
        when (rel.dessine_interface) then 1
        when score is null then 1
        else rel.score
    end as score,
    null as "address",
    ann_t.name_n as "object_type",
    case
        when (rel.dessine_interface) then 'INTERFACE_DRAWN'
        else 'ANALYSIS'
    end
    as "detection_source",
    case
        when rel.signale_terrain then 'CONTROLLED_FIELD'
        when rel.control_status_id = 1 then 'NOT_CONTROLLED'
        when rel.control_status_id = 2 then 'SIGNALED_COMMUNE'
        when rel.control_status_id = 3 then 'SIGNALED_COLLECTIVITY'
        when rel.control_status_id = 4 then 'CONTROLLED_FIELD'
        when rel.control_status_id = 5 then 'REHABILITATED'
        when rel.control_status_id = 6 then 'VERBALIZED'
    end
    as "detection_control_status",
    case
        when rel.validation = 0 then 'INVALIDATED'
        when rel.vrai_legitime
        and rel.vrai_positif
        and not rel.faux_positif then 'LEGITIMATE'
        when not rel.vrai_legitime
        and rel.vrai_positif
        and not rel.faux_positif then 'SUSPECT'
        when not rel.vrai_legitime
        and not rel.vrai_positif
        and rel.faux_positif then 'INVALIDATED'
        when not rel.vrai_legitime
        and not rel.vrai_positif
        and not rel.faux_positif then 'DETECTED_NOT_VERIFIED'
        else 'DETECTED_NOT_VERIFIED'
    end
    as "detection_validation_status",
    case
        when rel.prescrit_manuel then 'PRESCRIBED'
        else 'NOT_PRESCRIBED'
    end
    as "detection_prescription_status",
    rel.validation is not null as "user_reviewed",
    null as tile_x,
    null as tile_y,
    case 
        when tiles.dataset_id = 7 then 'sia_2012'
        when tiles.dataset_id = 4 then 'sia_2015'
        when tiles.dataset_id = 5 then 'sia_2018'
        when tiles.dataset_id = 8 then 'sia_2021'
    end as "batch_id"
from
    relevant_detections rel
join
    annotation_types ann_t on
    (
        case
        when rel.validation is null
            or rel.validation = 0 then rel.type_id
            else rel.validation
        end
    ) = ann_t.id
join
    tiles on
    tiles.id = rel.tile_id
where 
    tiles.dataset_id in (4, 5, 7, 8)
order by
    score desc

Remove whole detetctions from specific batch

Query

delete from core_detection where batch_id = 'sia_2021';

delete
from
	core_detectiondata
where id in (
	select
		core_detectiondata.id
	from
		core_detectiondata
	left join core_detection on
		core_detectiondata.id = core_detection.detection_data_id
	where
		core_detection.detection_data_id is null
);

delete from core_detectionobject_geo_custom_zones where detectionobject_id in (
	select
		obj.id
	from
		core_detectionobject as obj
	left join core_detection as det on
		obj.id = det.detection_object_id
	where
		det.detection_object_id is null
);

delete
from
	core_detectionobject
where id in (
	select
		obj.id
	from
		core_detectionobject as obj
	left join core_detection as det on
		obj.id = det.detection_object_id
	where
		det.detection_object_id is null
);

Emails

To send emails locally, you'll need to install local certificates, here is how to do it in MacOS