Merge pull request #326 from Vizzuality/feat/cloud-functions/SKY30-39…

…9-terrestrial-conservation-builder

added new functionality for terrestrial cloud function analysis

cloud_functions/analysis/docker-compose.yml

@@ -25,15 +25,15 @@ services:
   skytruth-db-init:
     profiles:
       - feed_data_to_db
-    image: ghcr.io/osgeo/gdal:ubuntu-small-latest
+    image: ghcr.io/osgeo/gdal:ubuntu-small-3.9.3
     container_name: skytruth_cf_db_init
     environment:
       GRANT_SUDO: "yes"
     env_file: .env
     restart: "no"
     volumes:
       - ./init.sh:/usr/share/init.sh
-    entrypoint: bash -c "chmod +x /usr/share/init.sh && /usr/share/init.sh"
+    entrypoint: bash -c "chmod +x /usr/share/init.sh && /usr/share/init.sh eez gadm"
     network_mode: "host"
     extra_hosts:
       - docker.host:0.0.0.0

cloud_functions/analysis/init.sh

@@ -1,16 +1,47 @@
 #!/bin/sh
-
-URL='vector-data-raw/vizzuality_processed_data/analysis_data/eez_minus_mpa.zip'
+# Upload the data to the database
+# Usage: init.sh [OPTION]
+for arg in "$@"
+do
+    case $arg in
+        eez)
+            echo "uploading eez_minus_mpa"
+            URL='vector-data-raw/vizzuality_processed_data/analysis_data/eez_minus_mpa.zip'
+            TABLE_NAME='eez_minus_mpa'
+            ;;
+        gadm)
+            echo "uploading gadm_minus_pa"
+            URL='vector-data-raw/vizzuality_processed_data/analysis_data/gadm_minus_pa.zip'
+            TABLE_NAME='gadm_minus_pa'
+            ;;
+        --help)
+            echo "Usage: init.sh [OPTION]"
+            echo "Upload the data to the database"
+            echo ""
+            echo "Options:"
+            echo "  --eez    Upload eez_minus_mpa data"
+            echo "  --gadm   Upload gadm_minus_pa data"
+            echo "  --help   Display this help message"
+            exit 0
+            ;;
+        *)
+            echo "Invalid option $arg"
+            exit 1
+            ;;
+    esac
+ogr2ogr --version
 
 ogr2ogr -progress \
-  -makevalid -overwrite \
-  -nln eez_minus_mpa -nlt PROMOTE_TO_MULTI \
-  -lco GEOMETRY_NAME=the_geom \
-  -lco PRECISION=FALSE \
-  -lco SPATIAL_INDEX=GIST \
-  -lco FID=id \
-  -t_srs EPSG:4326 -a_srs EPSG:4326 \
-  -f PostgreSQL PG:"host=$POSTGRES_HOST port=$POSTGRES_PORT \
-   user=$POSTGRES_USER password=$POSTGRES_PASSWORD \
-   dbname=$POSTGRES_DB active_schema=$POSTGRES_SCHEMA" \
-   -doo "PRELUDE_STATEMENTS=CREATE SCHEMA IF NOT EXISTS $POSTGRES_SCHEMA AUTHORIZATION CURRENT_USER;" "/vsizip/vsigs/$URL";
+          -makevalid -overwrite \
+          -nln $TABLE_NAME -nlt PROMOTE_TO_MULTI \
+          -lco GEOMETRY_NAME=the_geom \
+          -lco PRECISION=FALSE \
+          -lco SPATIAL_INDEX=GIST \
+          -lco FID=id \
+          -t_srs EPSG:4326 \
+          -f PostgreSQL PG:"host=$POSTGRES_HOST port=$POSTGRES_PORT \
+          user=$POSTGRES_USER password=$POSTGRES_PASSWORD \
+          dbname=$POSTGRES_DB active_schema=$POSTGRES_SCHEMA" \
+          -doo "PRELUDE_STATEMENTS=CREATE SCHEMA IF NOT EXISTS $POSTGRES_SCHEMA AUTHORIZATION CURRENT_USER;" "/vsizip/vsigs/$URL";
+
+done

cloud_functions/analysis/main.py

@@ -3,7 +3,7 @@
 import logging
 
 from src.connect_tcp import connect_tcp_socket
-from src.analysis import get_locations_stats
+from src.analysis import get_locations_stats_terrestrial, get_locations_stats_marine
 
 logger = logging.getLogger(__name__)
 logger.addHandler(default_handler)
@@ -50,7 +50,19 @@ def index(request):
         if not geometry:
             raise ValueError("geometry is required")
 
-        return (get_locations_stats(db, geometry), 200, headers)
+        funct = {
+            "marine": get_locations_stats_marine,
+            "terrestrial": get_locations_stats_terrestrial,
+        }
+
+        environment = ({**request.args, **request.get_json()}).get(
+            "environment", "marine"
+        )
+        if environment not in funct.keys():
+            raise ValueError("environment must be one of `marine` or `terrestrial`")
+
+        return (funct[environment](db, geometry), 200, headers)
+
     except ValueError as e:
         logger.exception(str(e))
         return {"error": str(e)}, 400, headers

cloud_functions/analysis/src/analysis.py

@@ -12,8 +12,10 @@ def get_geojson(geojson: JSON) -> dict:
     else:
         return geojson
 
+### Marine
 
-def serialize_response(data: dict) -> dict:
+
+def serialize_response_marine(data: dict) -> dict:
     """Converts the data from the database
     into a Dict {locations_area:{"code":<location_iso>, "protected_area": <area>, "area":<location_marine_area>}, "total_area":<total_area>} response
     """
@@ -48,7 +50,9 @@ def serialize_response(data: dict) -> dict:
     return result
 
 
-def get_locations_stats(db: sqlalchemy.engine.base.Engine, geojson: JSON) -> dict:
+def get_locations_stats_marine(
+    db: sqlalchemy.engine.base.Engine, geojson: JSON
+) -> dict:
     with db.connect() as conn:
         stmt = sqlalchemy.text(
             """
@@ -65,4 +69,62 @@ def get_locations_stats(db: sqlalchemy.engine.base.Engine, geojson: JSON) -> dic
             stmt, parameters={"geometry": get_geojson(geojson)}
         ).all()
 
-    return serialize_response(data_response)
+    return serialize_response_marine(data_response)
+
+
+### Terrestrial
+def serialize_response_terrestrial(data: dict) -> dict:
+    """Converts the data from the database
+    into a Dict {locations_area:{"code":<location_iso>, "protected_area": <area>, "area":<location_marine_area>}, "total_area":<total_area>} response
+    """
+    if not data or len(data) == 0:
+        raise ValueError(
+            "No data found, this is likely due to a geometry that does not intersect with a Marine area."
+        )
+
+    result = {"total_area": data[0][3]}
+    sub_result = {}
+    total_protected_area = 0
+    for row in data:
+        for iso in filter(lambda item: item is not None, [row[1]]):
+            total_protected_area += row[2]
+            if iso not in sub_result:
+                sub_result[iso] = {
+                    "code": iso,
+                    "protected_area": row[2],
+                    "area": row[0],
+                }
+            else:
+                sub_result[iso]["protected_area"] += row[2]
+                sub_result[iso]["area"] += row[0]
+
+    result.update(
+        {
+            "locations_area": list(sub_result.values()),
+            "total_protected_area": total_protected_area,
+        }
+    )
+
+    return result
+
+
+def get_locations_stats_terrestrial(
+    db: sqlalchemy.engine.base.Engine, geojson: JSON
+) -> dict:
+    with db.connect() as conn:
+        stmt = sqlalchemy.text(
+            """
+        with user_data as (select ST_GeomFromGeoJSON(:geometry) as geom),
+	            user_data_stats as (select *, round((st_area(st_transform(geom,'+proj=longlat +datum=WGS84 +no_defs +type=crs', '+proj=moll +lon_0=0 +x_0=0 +y_0=0 +datum=WGS84 +units=m +no_defs +type=crs'))/1e6)) user_area_km2 from user_data), 
+	          stats as (select area_km2,gid_0, round((st_area(st_transform(st_makevalid(st_intersection(the_geom, user_data_stats.geom)),'+proj=longlat +datum=WGS84 +no_defs +type=crs', '+proj=moll +lon_0=0 +x_0=0 +y_0=0 +datum=WGS84 +units=m +no_defs +type=crs'))/1e6)) portion_area_km2,
+	          user_data_stats.user_area_km2
+	          from data.gadm_minus_pa gmp , user_data_stats
+            where st_intersects(the_geom, user_data_stats.geom))
+      select avg(area_km2) as area_km2, gid_0, sum(portion_area_km2) as portion_area_km2, avg(user_area_km2) as user_area_km2 from stats group  by gid_0
+            """
+        )
+        data_response = conn.execute(
+            stmt, parameters={"geometry": get_geojson(geojson)}
+        ).all()
+
+    return serialize_response_terrestrial(data_response)