Merge pull request #609 from mapswipe/dev

Dev - fix missing timestamps in mapping_sessions

Author: Hagellach37

mapswipe_workers/python_scripts/results_to_mapping_sessions_results.py

@@ -0,0 +1,56 @@
+import sys
+
+import pandas as pd
+
+from mapswipe_workers import auth
+from mapswipe_workers.definitions import logger
+
+
+def copy_results_batch(first_timestamp, last_timestamp):
+    logger.info(
+        f"Start process for : ms.start_time>={first_timestamp} "
+        f"and ms.start_time<{last_timestamp}"
+    )
+    p_con = auth.postgresDB()
+    query = """
+        -- create table with results for given time span
+        drop table if exists results_batch;
+        create table results_batch as
+        select ms.mapping_session_id, r.task_id, r."result"
+        from mapping_sessions ms, results r
+        where
+            ms.start_time >= %(first_timestamp)s
+            and ms.start_time < %(last_timestamp)s
+            and ms.project_id = r.project_id
+            and ms.group_id = r.group_id
+            and ms.user_id = r.user_id;
+        insert into mapping_sessions_results
+            (select r.* from results_batch r)
+        on conflict do nothing;
+    """
+    p_con.query(
+        query, {"first_timestamp": first_timestamp, "last_timestamp": last_timestamp}
+    )
+    logger.info(
+        f"Finished process for : ms.start_time >= {first_timestamp} "
+        f"and ms.start_time < {last_timestamp}"
+    )
+
+
+if __name__ == "__main__":
+    """Use this command to run in docker container.
+    docker-compose run -d mapswipe_workers_creation python3 python_scripts/results_to_mapping_sessions_results.py "2016-01-01" "2022-10-01"  # noqa
+    """
+    min_timestamp = sys.argv[1]
+    max_timestamp = sys.argv[2]
+    timestamps_list = (
+        pd.date_range(min_timestamp, max_timestamp, freq="MS")
+        .strftime("%Y-%m-%d")
+        .tolist()
+    )
+
+    for i in range(0, len(timestamps_list) - 1):
+        first_timestamp = timestamps_list[i]
+        last_timestamp = timestamps_list[i + 1]
+        copy_results_batch(first_timestamp, last_timestamp)
+        logger.info(f"progress: {i+1}/{len(timestamps_list) - 1}")

postgres/scripts/v2_to_v3/04b_fix_timestamps_for_mapping_sessions.sql

@@ -0,0 +1,28 @@
+/*
+ * This script takes the data from the 'results' table.
+ * Results submitted before 2019-09-30 11:30:02.823
+ * do NOT provide the "start_time" and "end_time" attribute,
+ * but only use a single attribute "timestamp".
+ * During the initial transfer this has not been considered.
+ */
+set search_path = 'public';
+
+insert into mapping_sessions
+(
+    select
+        project_id
+         ,group_id
+         ,user_id
+         ,nextval('mapping_sessions_mapping_session_id_seq') as mapping_session_id
+         ,Min(timestamp) - INTERVAL '2 Minutes' as start_time
+         ,Max(timestamp) as end_time
+         ,count(*) as items_count
+    from results
+    where start_time is null and end_time is null
+    group by project_id, group_id, user_id
+)
+on conflict (project_id, group_id, user_id)
+DO UPDATE SET
+    start_time = EXCLUDED.start_time,
+    end_time = EXCLUDED.end_time;
+