Making PolarRoute-server compatible with latest PolarRoute and MeshiPhi versions

[thomaszwagerman]

Specifically updating how evaluate_route works with PolarRoute changes to route_calc api:

# From
calc_route = route_calc(route_file.name, mesh_file.name)

# To
calc_route = route_calc(df, from_wp, to_wp, mesh.json, "fuel")

Also removing all TempFile related code, as now longer required for new API.

[davidwyld]

I appreciate this is still in draft, but looks good so far, I'm very pleased to finally be getting rid of the need to deal with tempfiles now we can use the python interface more directly.

Feel free to make any other improvements to this function you feel are necessary.

I'm sure you will, but don't forget to restrict the polar-route version in the pyproject.toml from this point.

[github-actions]

Coverage Report

File	Stmts	Miss	Branch	BrPart	Cover	Missing
polarrouteserver
__init__.py	3	0	0	0	100%
_version.py	11	0	0	0	100%
celery.py	15	3	0	0	0	80%
polarrouteserver/route_api
__init__.py	4	0	0	0	100%
models.py	74	2	0	0	0	97%
responses.py	33	0	2	0	100%
serializers.py	120	5	36	4	4	94%
tasks.py	138	21	32	7	7	84%
utils.py	114	10	46	4	4	91%
views.py	277	17	52	4	4	94%
polarrouteserver/route_api/management
__init__.py	0	0	0	0	100%
polarrouteserver/route_api/management/commands
__init__.py	0	0	0	0	100%
ensure_adminuser.py	22	22	6	0	0	0%
insert_mesh.py	26	26	8	0	0	0%
polarrouteserver/utils
__init__.py	0	0	0	0	100%
loggers.py	10	10	0	0	0	0%
TOTAL	847	116	182	19	86%

[thomaszwagerman]

Satisfied this successfully runs and returns routes

[davidwyld]

As I've said, this is a huge improvement, thanks for doing this.

I wonder if we should add to the testing of this functionality in test_utils? I'm happy to write a test if you don't have time.

Don't forget to update the changelog in this PR too.

[thomaszwagerman]

I've added route_type parameter as a positional argument and updated the route evaluation API endpoint + schema.

Opened an issue in PolarRoute for related issues within route_calc

[davidwyld]

Worth discussing whether we want to expose route_type in the API?

Want me to implement a test for evaluate_route? Or we can just open an issue since there isn't one to begin with so I don't want it to hold back this branch.

[davidwyld]

I'm not sure if it's possible for the submitted route to have waypoint names, we may want to preserve those in the future, but lets leave this as is for now, they don't really matter for SIIS at least.

[thomaszwagerman]

TZ review geojson schema and safety of waypoint name here

[davidwyld]

It would be much better IMO if route_calc took a geojson/dict as input rather than a dataframe, but whatever, this is fine.
It would be nice to preserve waypoint names from the geojson if it has them

[thomaszwagerman]

waypoint names preserved with e6a2b24, agree it would be lot less hassle if route_calc took a GeoJSON though - although that is a PolarRoute change. Might investigate.

[thomaszwagerman]

see bas-logist/PolarRoute#355 - I think worth holding off on this until that is reviewed, as well as possibly implementing bas-logist/PolarRoute#350 as it touches this