get only groups/tasks of which a project exists and results of that project exists

Author: Matthias-Schaub

postgres/scripts/v1_to_v2/generate_copy_to_csv.py

@@ -4,24 +4,28 @@
 def get_query(project_ids):
     clause = 'WHERE project_id in (select project_id from results group by project_id)'
     clause_import = 'WHERE import_id in (select importkey as import_id from projects)'
+    clause_group = 'WHERE project_id in (select project_id from projects group by project_id) and project_id in (select project_id from results group by project_id)'
     if project_ids is None:
         pass
     elif len(project_ids) == 1:
         clause = f'{clause} AND project_id = {project_ids[0]}'
         clause_import = f', projects p {clause_import} AND p.project_id = {project_ids[0]} AND p.importkey = i.import_id'
+        clause_group = f'{clause_group} AND project_id = {project_ids[0]}'
     else:
         clause = f'{clause} AND project_id = {project_ids[0]}'
+        clause_group = f'{clause_group} AND project_id = {project_ids[0]}'
         clause_import = f', projects p {clause_import} AND p.project_id = {project_ids[0]} AND p.importkey = i.import_id'
         for project_id in project_ids[1:]:
             clause = clause + f' OR project_id = {project_id}'
+            clause_group = clause_group + f' OR project_id = {project_id}'
             clause_import = clause + f' OR p.project_id = {project_id} AND p.importkey = i.import_id'
 
     query = f'-- Export v1 MapSwipe data to csv.\n' \
             f'-- Rename attributes to conform to v2.\n' \
             f'\copy (SELECT archive, image, importkey as "import_id", isfeatured AS "is_featured", lookfor AS "look_for", name, progress, projectdetails AS "project_details", project_id, project_type, state AS "status", info AS "project_type_specifics" FROM projects {clause}) TO projects.csv WITH (FORMAT CSV, DELIMITER ",", HEADER TRUE);\n' \
             f'\copy (SELECT i.import_id, i.info FROM imports i {clause_import}) TO imports.csv WITH (FORMAT CSV, DELIMITER ",", HEADER TRUE);\n' \
-            f'\copy (SELECT project_id, group_id as "v1_group_id", count as "number_of_tasks", completedcount as "finished_count", verificationcount as "required_count", info as "project_type_specifics" FROM groups {clause} ) TO groups.csv WITH (FORMAT CSV, DELIMITER ",", HEADER TRUE);\n' \
-            f'\copy (SELECT project_id, group_id as "v1_group_id", task_id, info as "project_type_specifics" FROM tasks {clause}) TO tasks.csv WITH (FORMAT CSV, DELIMITER ",", HEADER TRUE);\n'
+            f'\copy (SELECT project_id, group_id as "v1_group_id", count as "number_of_tasks", completedcount as "finished_count", verificationcount as "required_count", info as "project_type_specifics" FROM groups {clause_group} ) TO groups.csv WITH (FORMAT CSV, DELIMITER ",", HEADER TRUE);\n' \
+            f'\copy (SELECT project_id, group_id as "v1_group_id", task_id, info as "project_type_specifics" FROM tasks {clause_group}) TO tasks.csv WITH (FORMAT CSV, DELIMITER ",", HEADER TRUE);\n'
     return query