From 4c0e849ff3a856083e5225e5665b34763f1f8888 Mon Sep 17 00:00:00 2001
From: Kevin <kew.sanchez@gmail.com>
Date: Thu, 11 Jan 2024 10:41:27 +0100
Subject: [PATCH] fix(Cost Surface): Fixes a bug in the insertion of Cost
 Surface PUs for projects without a shapefile

Also includes script to fix incorrect pu data in the DB
---
 .../planning-units/planning-units.job.ts      |  4 +-
 .../cost-surface-incorrect-pu-data-purge.py   | 85 +++++++++++++++++++
 2 files changed, 87 insertions(+), 2 deletions(-)
 create mode 100644 data/scripts/cost-surface/cost-surface-incorrect-pu-data-purge.py

diff --git a/api/apps/geoprocessing/src/modules/planning-units/planning-units.job.ts b/api/apps/geoprocessing/src/modules/planning-units/planning-units.job.ts
index da6e68425a..51d950ba6d 100644
--- a/api/apps/geoprocessing/src/modules/planning-units/planning-units.job.ts
+++ b/api/apps/geoprocessing/src/modules/planning-units/planning-units.job.ts
@@ -253,9 +253,9 @@ grid.geom
                  SELECT ppu.id, round(pug.area / 1000000) as area, $1
                     FROM projects_pu ppu
                  INNER JOIN planning_units_geom pug ON pug.id = ppu.geom_id
-                 WHERE ppu.geom_id = $2
+                 WHERE ppu.geom_id = $2 AND ppu.project_id = $3
                 `,
-                  [job.data.costSurfaceId, geometryId],
+                  [job.data.costSurfaceId, geometryId, job.data.projectId],
                 );
               });
             },
diff --git a/data/scripts/cost-surface/cost-surface-incorrect-pu-data-purge.py b/data/scripts/cost-surface/cost-surface-incorrect-pu-data-purge.py
new file mode 100644
index 0000000000..43873bcf58
--- /dev/null
+++ b/data/scripts/cost-surface/cost-surface-incorrect-pu-data-purge.py
@@ -0,0 +1,85 @@
+import psycopg2
+
+
+dry_run = False
+
+# Connect to Database 1
+api_db = psycopg2.connect(
+    host="localhost",
+    database="marxan-api",
+    user="marxan-api",
+    password="marxan-api",
+    port=3432
+)
+
+# Connect to Database 2
+geo_db = psycopg2.connect(
+    host="localhost",
+    database="marxan-geo-api",
+    user="marxan-geo-api",
+    password="marxan-geo-api",
+    port=3433
+)
+
+# Initialize the cursor for api_db
+cur_api_db = api_db.cursor()
+cur_geo_db = geo_db.cursor()
+
+try:
+
+    cur_api_db.execute("BEGIN;")
+    cur_geo_db.execute("BEGIN;")
+
+    cur_geo_db.execute(''' select cspd.cost_surface_id
+                        from cost_surface_pu_data cspd
+                        inner join projects_pu pp on pp.id = cspd.projects_pu_id
+                        group by cspd.cost_surface_id
+                        having count(distinct pp.project_id) > 1 ''')
+    incorrect_cost_surfaces = cur_geo_db.fetchall()
+
+    for cost_surface in incorrect_cost_surfaces:
+        cost_surface_id = cost_surface
+
+        cur_api_db.execute( '''select project_id from cost_surfaces where id = %s ''', (cost_surface_id),)
+        project = cur_api_db.fetchone()  # Retrieve the project id that the cost surface is associated to
+
+        if project is None:
+            #It can happen that the cost surface has been deleted, but there's dangling data because most probably
+            #the clean up process doesn't account for this incorrect pu data that has leaked into cost_surface_pu_data
+            print('Cost Surface ', cost_surface_id, ' not found on API')
+            delete_incorrect_pu_data_sql = '''
+                DELETE FROM cost_surface_pu_data cspd
+                USING projects_pu pp
+                WHERE pp.id = cspd.projects_pu_id AND cspd.cost_surface_id = %s
+                RETURNING pp.project_id
+            '''
+            cur_geo_db.execute(delete_incorrect_pu_data_sql, (cost_surface_id))
+        else:
+            project_id = project
+            delete_incorrect_pu_data_sql = '''
+                DELETE FROM cost_surface_pu_data cspd
+                USING projects_pu pp
+                WHERE pp.id = cspd.projects_pu_id AND cspd.cost_surface_id = %s  AND NOT pp.project_id = %s
+                RETURNING pp.project_id
+            '''
+            cur_geo_db.execute(delete_incorrect_pu_data_sql, (cost_surface_id, project_id))
+
+        deleted_rows = cur_geo_db.fetchall()
+
+        print("Successfully purged ", len(deleted_rows), " rows of incorrect pu data  for Cost Surface: ", cost_surface_id)
+
+
+    if dry_run is False:
+        cur_api_db.execute("COMMIT;")
+        cur_geo_db.execute("COMMIT;")
+    else:
+        cur_api_db.execute("ROLLBACK;")
+        cur_geo_db.execute("ROLLBACK;")
+
+    print("Successfully purged all incorrect Cost Surface pu data")
+
+
+except Exception as e:
+    print(f"An error occurred: {e}")
+    cur_api_db.execute("ROLLBACK;")
+    cur_geo_db.execute("ROLLBACK;")