oedb-backend/oedb/resources/stats.py

"""
Stats resource for the OpenEventDatabase.
"""

import subprocess
import falcon
import psycopg2.extras
from oedb.utils.db import db_connect
from oedb.utils.serialization import dumps
from oedb.utils.logging import logger

class StatsResource:
    """
    Resource for retrieving database statistics.
    Handles the /stats endpoint.
    """
    
    def on_get(self, req, resp):
        """
        Handle GET requests to the /stats endpoint.
        
        Args:
            req: The request object.
            resp: The response object.
        """
        logger.info("Processing GET request to /stats")
        
        db = db_connect()
        cur = db.cursor(cursor_factory=psycopg2.extras.DictCursor)
        
        try:
            # Estimated row count, way faster than count(*)
            cur.execute("SELECT reltuples::bigint FROM pg_class r WHERE relname = 'events';")
            count = cur.fetchone()[0]
            logger.debug(f"Estimated event count: {count}")
            
            # If the estimated count is negative, fall back to an actual count
            if count < 0:
                logger.warning(f"Estimated event count is negative ({count}), falling back to COUNT(*)")
                cur.execute("SELECT COUNT(*) FROM events;")
                count = cur.fetchone()[0]
                logger.debug(f"Actual event count: {count}")
            
            # Global info
            cur.execute("SELECT max(lastupdate) as last_updated, current_timestamp-pg_postmaster_start_time() from events;")
            pg_stats = cur.fetchone()
            last = pg_stats[0]
            pg_uptime = pg_stats[1]
            uptime = subprocess.check_output(["uptime", "-p"]).decode('utf-8')[0:-1]
            
            # Summary about last 10000 events (what, last, count, sources)
            cur.execute("SELECT row_to_json(stat) from (SELECT events_what as what, left(max(upper(events_when))::text,19) as last, count(*) as count, array_agg(distinct(regexp_replace(regexp_replace(events_tags ->> 'source','^(http://|https://)',''),'/.*',''))) as source from (select * from events order by lastupdate desc limit 10000) as last group by 1 order by 2 desc) as stat;")
            recent = cur.fetchall()
            
            resp.text = dumps(dict(events_count=count, last_updated=last, uptime=uptime, db_uptime=pg_uptime, recent=recent))
            resp.status = falcon.HTTP_200
            logger.success("Successfully processed GET request to /stats")
        except Exception as e:
            logger.error(f"Error processing GET request to /stats: {e}")
            resp.status = falcon.HTTP_500
            resp.text = dumps({"error": str(e)})
        finally:
            cur.close()
            db.close()