gns-science / solvis-graphql-api

solvis-graphql-api

solvis-graphql-api

GRAPHQL API for analysis of opensha modular Inversion Solutions.

Using Flask with Serverless framework to operate as a AWS Lambda API.

The API documentation is served by default from the service root.

Getting started

poetry install
npm install --save serverless
npm install --save serverless-python-requirements
npm install --save serverless-wsgi
npm install --save serverless-plugin-warmup

WSGI

sls wsgi serve

Run full stack locally

SLS_OFFLINE=1 npx serverless wsgi serve

Unit tests

poetry run pytest note that some environment variables are set in setup.cfg.

solvis-graphql-api's Issues

SOLVIS_GRAPHQL_API Feature: composite solution for IFM

SOLVIS_GRAPHQL_API: add geojson support

So we can replace the rest api and hook this into our KORORAA_APIGW

• geojson
• radius polygons
• radii lists
• location lists

S-API feature/Integrate with latest upstream libs

• nzshm-common
• solvis-store
• nzshm-model

SOLVIS_GRAPHQL_API Feature: Handle Subduction inversion solutions

Some problems in analysis to sort out ...

• surface distance only (as per crustal)
• overlaid ruptures (so)
• 3D fault surface

using SW52ZXJzaW9uU29sdXRpb246MTcwNzEuMEpyTUtZ from PROD

[ERROR] ValueError: invalid literal for int() with base 10: 'Hikurangi, Kermadec to Louisville ridge, 30km - with slip deficit smoothed near East Cape and locked near trench.'
Traceback (most recent call last):
  File "/var/task/api/process_solution.py", line 56, in handler
    process_event(evt)
  File "/var/task/api/process_solution.py", line 50, in process_event
    save_solution_fault_sections(solution_id, get_fault_section_models(solution_id,solution))
  File "/var/task/api/datastore/solvis_db.py", line 91, in save_solution_fault_sections
    for item in models:
  File "/var/task/api/datastore/solvis_db.py", line 83, in get_fault_section_models
    parent_id = int(row[11]),

Refactor Solvis API: move reuseable rupture analysis functions to main solvis lib

ref: GNS-Science/solvis#30

SOLVIS_GRAPHQL_API Feature: all ruptures

We want to get fault sections ruptures without geographic filter (i.e. no need to specify location and radius). Retain filter on magnitude and rate

Bug: number of ruptures

When filtering ruptures by location/radius we get different answers from the solvis api (as viewed on https://nshm.gns.cri.nz/RuptureMap and sovis running on it's own).

To reproduce: use the Hikurangi FaultSystemSolution

	solvis	rupture explorer
no filter	1645	1645
WLG 10km	88	0
WLG 30km	172	148

Sample code used in solivs to filter:

from pathlib import Path
from solvis import FaultSystemSolution, geometry
from nzshm_common.location.location import location_by_id

fs = "HIK"
area_filter = geometry.circle_polygon(radius_m=30_000, lon=location_by_id("WLG")['longitude'], lat=location_by_id("WLG")['latitude'])
soln_dir = Path('/home/chrisdc/tmp')
soln_path = soln_dir / f'{fs}_fault_system_solution.zip'

soln = FaultSystemSolution.from_archive(soln_path)
ruptures = soln.ruptures_with_rupture_rates
ar = soln.get_ruptures_intersecting(area_filter)
print(len(ar))

Solvis API Feature: Fault trace resolver

We need a fault trace resolver alongside fault_surfaces with same color and styling options

SOLVIS_GRAPHQL: feature sort on RuptureDetail attrributres

SOLVIS_GRAPHQL_API: FIX no ruptures match query

In Kororaa we issue the following and expect some results....

Similarly for Mt.Cook and Te Anau

{
	"query": "query FaultModelPageSolvisQuery(\n  $solution_id: ID!\n  $location_codes: [String]\n  $radius_km: Int\n  $minimum_mag: Float\n  $maximum_mag: Float\n  $minimum_rate: Float\n  $maximum_rate: Float\n  $location_colour: String\n  $fault_colour: String\n) {\n  SOLVIS_analyse_solution(input: {solution_id: $solution_id, location_codes: $location_codes, radius_km: $radius_km, minimum_mag: $minimum_mag, maximum_mag: $maximum_mag, minimum_rate: $minimum_rate, maximum_rate: $maximum_rate, location_area_style: {stroke_color: $location_colour, stroke_width: 1, stroke_opacity: 0.5, fill_color: $location_colour, fill_opacity: 0.5}, fault_trace_style: {stroke_color: $fault_colour, stroke_width: 3, stroke_opacity: 1}}) {\n    analysis {\n      fault_sections_geojson\n      solution_id\n      location_geojson\n    }\n  }\n}\n",
	"variables": {
		"fault_colour": "red",
		"location_codes": [
			"HAW"
		],
		"location_colour": "grey",
		"maximum_mag": 10,
		"maximum_rate": 0,
		"minimum_mag": 6,
		"minimum_rate": -20,
		"radius_km": 100,
		"solution_id": "U2NhbGVkSW52ZXJzaW9uU29sdXRpb246MTIwNjcy"
	}
}

SOLVIS_GRAPHQL_API Feature: Geojson styling

Want to be able to configure the styling options so the leaflet component can use the styling supplied by the geojson instead of providing its own.
Done when:

• Can configure the stroke property of rupture results and location polygons.
• Can configure the fill property of rupture results and location polygons.

Feature/ add rutpure_set_id resolver to models returning ruptures

SOLVIS_GRAPHQL_API Feature: show fault surfaces in geojson views

We want fancy 3d pls. how do we do it?

Solvis-API Feature: Composite MFD resolver

Feature: provide subsection participation rate filter

We want to be able to filter for fault participation rate. This is the combined rate for all subsections on that "parent" fault (i.e. the highest cumulative MFD rate for all subsections of the fault)

SOLVIS_GRAPHQL_API: setup AWS deployment actions

we want github actions to automate our deployment

Chore: complete mypy typing on composite_solution/cached module

re solvis_graphql_api/composite_solution/cached.py

In preparation for potentially migrating this module (or large chunks of it) we need all it's function signatures fully typed. This will help us weed our dependencies and decide what to migrate.

Possible fishooks:

• figure out how to switch between solvis_store and direct solvis methods more elegantly. We don't want the solvis library depending on the solvis-store project
• make sure that module dependencies are compatible with solvis lib - e.g graphene dependencies not allowed, but nzshm-model is OK.

Solvis-API Feature: Composite resolver for ruptures (non-animated/simple)

Solvis API Feature Request: Disaggregations on multi-rupture map view

from Brendon Bradley ...

looking at this ‘Multi-rupture map’ tool; it seems that its not too much of an extension to be able to make a ‘disaggregation map’ tool - i.e., showing the spatial location of all the fault sources that contribute to the hazard, colored by their % contribution. This would be really useful, because the grand inversion (i.e., non-characteristic and multi-fault ruptures) means that we no longer have a disagg list that is dominated by just a handful of faults. We can see Mw-Rrup plots etc, but actually understanding specific fault sections/segments is really hard at present! Very happy to help scope this out more if there is interest

@chrisbc @chrisdicaprio and I have been discussing this one for a while. This will need more work on openquake to correlate and extract the fault source contribution. Then we need some infrastructure to store this and make it accessible online. It's definitely doable but is quite a bit of work. If we get +vote on it we'll definitely reach out for your ideas on scoping this.

feature: complete union/intersection options for parent_fault filtering and set sane defaults

suggested sane defaults are:

• location-radii: take the the intersection of sets for selected locations
• parent fault: take the the union of sets for selected parent faults
• for a combination of location + parent fault filters: take the intersection of the two sub-sets

Feature Request: Co-rupture multi-rupture

can we extend Multi-Rupture map to allow the user to specify a single fault section, and then the conditional probability of all other sections/segments rupturing, conditioned on this one rupturing, is plotted? i.e., plots like this...