Merge pull request #956 from scitran/check-session-length

Add integrity check for sessions that span longer than 3 hours

Author: nagem

bin/integrity_check.py

@@ -10,7 +10,8 @@
 
 # Methods should return true if all integrity checks passed
 INTEGRITY_CHECKS = {
-    "rule_alg": "Confirm alg keys in rules table can be resolved to gear in gear table"
+    "rule_alg" :        "Confirm alg keys in rules table can be resolved to gear in gear table",
+    "session_length" :   "Confirm there are no sessions whose acquisition timestamps span more than 3 hours"
 }
 
 
@@ -33,6 +34,25 @@ def rule_alg():
 
     return not errors
 
+def session_length():
+    errors = False
+
+    pipeline = [
+        {'$match': {'timestamp': {'$ne': None}}},
+        {'$group': {'_id': '$session', 'min_timestamp': { '$min': '$timestamp' }, 'max_timestamp': { '$max': '$timestamp' }}},
+        {'$project': {'_id': '$_id', 'diff': { '$subtract':  ['$max_timestamp', '$min_timestamp']}}},
+        {'$match': {'diff': {'$gt': 10800000}}}
+    ]
+
+    results = config.db.command('aggregate', 'acquisitions', pipeline=pipeline)['result']
+    if len(results) > 0:
+        errors = True
+        logging.warning('There are {} sessions that span 3 hours.'.format(len(results)))
+        for r in results:
+            logging.warning('Session {} spans {} minutes'.format(r['_id'], r['diff']/60000))
+
+    return not errors
+
 
 if __name__ == '__main__':
     try: