https://github.com/nasumilu/ogc_rest_api
https://github.com/nasumilu/ogc_rest_api
Last synced: 5 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/nasumilu/ogc_rest_api
- Owner: nasumilu
- License: apache-2.0
- Created: 2023-09-27T21:50:51.000Z (almost 2 years ago)
- Default Branch: main
- Last Pushed: 2023-09-28T05:45:51.000Z (almost 2 years ago)
- Last Synced: 2025-01-12T15:08:52.557Z (6 months ago)
- Language: Python
- Size: 135 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Prototype Project
This project is a rough prototype which provides the scaffolding for an OGC Feature API v1.0
utilizing the FastAPI framework.
## How it works
To configure a __Feature__ class utilized the decorator provided in the `app.collection` module and extent
the abstract `FeatureDataProvider` class.
### Example
```python
@feature_collection(collection_id='tc-location',
title='Tropical Cyclone Locations',
description='The location points for tropical cyclones',
spatial_extent=(-180, -68.5, 180, 83.01000),
temporal_extent=(
datetime.fromisoformat('1842-10-25 03:00:00.000000'),
datetime.fromisoformat('2023-06-16 12:00:00.000000')
))
class TCLocation(FeatureDataProvider):
def _map_feature_to_geojson(self, feature: Row) -> Dict[str, Any]:
# there are tools for this but currently this a prototype project so some janky stuff for now
# todo: Model features rather then brute forcing the issue.
geom = from_wkt(feature[4])
return {
'type': 'Feature',
'geometry': {
'type': 'Point', # not very dynamic?
'coordinates': get_coordinates(geom).tolist()
},
'properties': {
'id': feature[0],
'sid': feature[1],
'name': feature[2],
'iso_time': feature[3]
}
}
def _sql_for_items_query(self) -> TextClause:
return text("""
select id,
sid,
name,
iso_time,
st_astext(geom) as geom
from storm_location
limit 25
""")
def _sql_for_item_query(self) -> TextClause:
return text("""
select id,
sid,
name,
iso_time,
st_astext(geom) geom
from storm_location
where sid = :feature_id
""")
```
The above example you can see that the `@feature_collection` decorator provides the necessary metadata
to describe the collection. Then the `_sql_for_items_query` and `_sql_for_item_query` method uses
to provide the `TextClause` to retrieve the data from the database. Then finally each of the `Row` are converted to a
dictionary using the `_map_feature_to_geojson`.
In short, the decorator is the configuration and the class is the adapter to retrieve the data. Given enough time each
of the `FeatureDataProvider` should have an ORM which would provide the necessary data conversion to/from the database.
In fact, the `FeatureDataProvider` has generic largely unused at this point to type the modeled results.
## What needs to be done
- Provide a mechanism to allow multiple datasource, rather than one global datasource
- Pagination for the /collections/{collection_id}/items endpoint
- Model the results for the `@feature_collection` decorated classes to convert the data from the database into a
value GeoJSON response
- ~~Configuration to define modules with `@feature_collections` so they are discovered and configured when the app starts~~
- Finish unit tests
- Document the code
- And much more...
> The apps module `__init__.py` imports the `app.features` module if additional features are added that module to should
> be added to the root or its `__init__.py` to preload (discover) the decorated `FeatureDataProvider` classes.
### Example Output
#### Landing Page (/)
```json
{
"title": "OGC Feature API",
"description": "Simple one-off OGC Feature API Server",
"links": [
{
"href": "http://127.0.0.1:8000/docs",
"rel": "service-doc",
"type": "text/html",
"title": "The Swagger UI for this API"
},
{
"href": "http://127.0.0.1:8000/redoc",
"rel": "alternate",
"type": "text/html",
"title": "The ReDoc UI for this API"
},
{
"href": "http://127.0.0.1:8000/openapi.json",
"rel": "service-desc",
"type": "application/json",
"title": "The OpenAPI JSON definition"
},
{
"href": "http://127.0.0.1:8000/conformance",
"rel": "conformance",
"type": "application/json"
},
{
"href": "http://127.0.0.1:8000/collections",
"rel": "data"
},
{
"href": "http://127.0.0.1:8000/",
"rel": "self",
"type": "application/json"
}
]
}
```
The Swagger UI generated by FastAPI's app, route, and model decorators (some inferred) found using the `rel="alternative"`
link.
The ReDoc UI generated by FastAPI's app, route, and model decorates (some inferred). Since the OGC responses are nearly
like Collections+JSON HATEOAS format they provide nice discoverable linked data.
#### Conformance (/conformance)
> **IMPORTANT**: This isn't true or at least the current state does not conform to these standards. Once the app is
> finished and the unit testing is completed per the OGC Feature API requirements; This will be so...
```json
{
"conformsTo": [
"http://www.opengis.net/spec/ogcapi-features-1/1.0/req/core",
"http://www.opengis.net/spec/ogcapi-features-1/1.0/conf/geojson",
"http://www.opengis.net/spec/ogcapi-features-1/1.0/conf/html",
"http://www.opengis.net/spec/ogcapi-features-1/1.0/conf/oas30"
]
}
```
#### Collections (/collections)
> Much more testing is needed for the extent, as you see it is keyed wrong, one to many extents. That is where those
> unit test come in handy. That bug would never get by, with proper unit testing.
```json
{
"links": [
{
"href": "http://127.0.0.1:8000/collections/tc-location",
"rel": "collection",
"title": ""
},
{
"href": "http://127.0.0.1:8000/collections/tc-track",
"rel": "collection",
"title": ""
}
],
"collections": [
{
"id": "tc-location",
"title": "Tropical Cyclone Locations",
"description": "The location points for tropical cyclones",
"extent": {
"extent": {
"spatial": [
[
-180,
-68.5,
180,
83.01
]
],
"temporal": [
[
"1842-10-25T03:00:00",
"2023-06-16T12:00:00"
]
]
}
},
"itemType": "feature",
"links": [
{
"href": "http://127.0.0.1:8000/collections/tc-location",
"rel": "self",
"type": "application/json",
"title": "Tropical Cyclone Locations"
},
{
"href": "http://127.0.0.1:8000/collections/tc-location/items",
"rel": "items",
"type": "application/json",
"title": "Tropical Cyclone Locations Items"
}
]
},
{
"id": "tc-track",
"title": "Tropical Cyclone Tracks",
"description": "The storm track (connecting the dots)",
"extent": {
"extent": {
"spatial": [
[
-180,
-68.5,
180,
83.01
]
],
"temporal": [
[
"1842-10-25T03:00:00",
"2023-06-16T12:00:00"
]
]
}
},
"itemType": "feature",
"links": [
{
"href": "http://127.0.0.1:8000/collections/tc-track",
"rel": "self",
"type": "application/json",
"title": "Tropical Cyclone Tracks"
},
{
"href": "http://127.0.0.1:8000/collections/tc-track/items",
"rel": "items",
"type": "application/json",
"title": "Tropical Cyclone Tracks Items"
}
]
}
]
}
```
#### Collection Items (/collections/{collection_id}/items)
> Not valid GeoJSON still need a lot of work. Example is only a subset of 4 features for brevity.
```json
{
"type": "FeatureCollection",
"features": [
{
"type": "Feature",
"geometry": {
"type": "Point",
"coordinates": [
[
80.30000305175781,
10.899999618530273
]
]
},
"properties": {
"id": 1,
"sid": "1842298N11080",
"name": "NOT_NAMED",
"iso_time": "1842-10-25T03:00:00"
}
},
{
"type": "Feature",
"geometry": {
"type": "Point",
"coordinates": [
[
79.83000183105469,
10.869999885559082
]
]
},
"properties": {
"id": 2,
"sid": "1842298N11080",
"name": "NOT_NAMED",
"iso_time": "1842-10-25T06:00:00"
}
},
{
"type": "Feature",
"geometry": {
"type": "Point",
"coordinates": [
[
79.3499984741211,
10.84000015258789
]
]
},
"properties": {
"id": 3,
"sid": "1842298N11080",
"name": "NOT_NAMED",
"iso_time": "1842-10-25T09:00:00"
}
},
{
"type": "Feature",
"geometry": {
"type": "Point",
"coordinates": [
[
78.87999725341797,
10.819999694824219
]
]
},
"properties": {
"id": 4,
"sid": "1842298N11080",
"name": "NOT_NAMED",
"iso_time": "1842-10-25T12:00:00"
}
}
]
}
```
#### Collection Item Endpoint (/collections/{collection_id}/items/{feature_id})
```json
{
"type": "Feature",
"geometry": {
"type": "Point",
"coordinates": [
[
80.30000305175781,
10.899999618530273
]
]
},
"properties": {
"id": 1,
"sid": "1842298N11080",
"name": "NOT_NAMED",
"iso_time": "1842-10-25T03:00:00"
}
}
```
## Docker
> **NOTE**: This is a BYBD (Bring Your Own Database) at least for now. If time permits I'll include the instructions to build the
> database in a container.
```shell
$ docker-compose up --build
```
or if you prefer podman (as I do)
```shell
$ podman-compose up --build
```