Merge pull request #1487 from phac-nml/filesystem-cleanup-script-add-…

…dates

Adding start and end dates to cleanup script

CHANGELOG.md

@@ -8,6 +8,7 @@
 * [Developer]: Added override flag that determines if files should be deleted from file storage. [See PR 1486](https://github.com/phac-nml/irida/pull/1486)
 * [Developer]: Fixed flaky text in `PipelinesPhylogenomicsPageIT#testPageSetup` test. See [PR 1490](https://github.com/phac-nml/irida/pull/1492)
 * [ALL]: Added LDAP/ADLDAP support.
+* [Developer]: Added start and ends dates to filesystem clean up script. [See PR 1487](https://github.com/phac-nml/irida/pull/1487)
 
 ## [23.01.3] - 2023/05/09
 * [Developer]: Fixed issue with metadata uploader removing existing data. See [PR 1489](https://github.com/phac-nml/irida/pull/1489)

src/main/resources/scripts/sequence-files/purge_sequence_files.py

@@ -1,5 +1,6 @@
 #!/usr/bin/python
 import argparse
+import datetime
 import mysql.connector
 import os
 
@@ -17,7 +18,7 @@ def remove(path, purge):
     else:
         print(path)
 
-def list_sequence_files(host, user, password, database):
+def list_sequence_files(startDate, endDate, host, user, password, database):
     db = mysql.connector.connect(
         host=host,
         user=user,
@@ -26,7 +27,14 @@ def list_sequence_files(host, user, password, database):
     )
     cursor = db.cursor()
     # TODO: Should we double check this file doesn't exist in the actual table in case it was manually restored?
-    cursor.execute("SELECT DISTINCT file_path FROM sequence_file_AUD WHERE revtype=2")
+    if(startDate and endDate):
+      cursor.execute("SELECT DISTINCT file_path FROM sequence_file_AUD WHERE revtype=2 AND modified_date BETWEEN %s AND %s", (startDate, endDate))
+    elif(startDate):
+      cursor.execute("SELECT DISTINCT file_path FROM sequence_file_AUD WHERE revtype=2 AND modified_date >= %s", (startDate,))
+    elif(endDate):
+      cursor.execute("SELECT DISTINCT file_path FROM sequence_file_AUD WHERE revtype=2 AND modified_date <= %s", (endDate,))
+    else:
+      cursor.execute("SELECT DISTINCT file_path FROM sequence_file_AUD WHERE revtype=2")
     result = cursor.fetchall()
     cursor.close()
     db.close()
@@ -35,14 +43,16 @@ def list_sequence_files(host, user, password, database):
 def main():
     parser = argparse.ArgumentParser(description="This program lists the sequence files and folders that have been previously deleted in IRIDA.")
     parser.add_argument('--purge', help="Deletes the sequence files and folders from the filesystem.", action="store_true")
+    parser.add_argument('--startDate', type=datetime.date.fromisoformat, help="The start date in format YYYY-MM-DD (inclusive).", required=False)
+    parser.add_argument('--endDate', type=datetime.date.fromisoformat, help="The end date in format YYYY-MM-DD (inclusive).", required=False)
     parser.add_argument('--baseDirectory', default='/tmp/irida/sequence-files', help="The sequence file base directory.", required=False)
     parser.add_argument('--host', default='localhost', help="The database host name.", required=False)
     parser.add_argument('--database', default='irida_test', help="The database name.", required=False)
     parser.add_argument('--user', default='test', help="The database user name.", required=False)
     parser.add_argument('--password', default='test', help="The database password.", required=False)
 
     args = parser.parse_args()
-    rows = list_sequence_files(args.host, args.user, args.password, args.database)
+    rows = list_sequence_files(args.startDate, args.endDate, args.host, args.user, args.password, args.database)
     if rows:
         for row in rows:
             sequence_file_directory = os.path.dirname(os.path.dirname(os.path.join(args.baseDirectory, row[0])))