small fixes, script to collect statistics from db

compose.yml

@@ -43,6 +43,8 @@ services:
       - POSTGRES_USER=$PGUSER
       - POSTGRES_DB=$PGDATABASE
       - POSTGRES_PASSWORD_FILE=/run/secrets/postgres_db_pass
+    ports:
+      - 5432:5432
     healthcheck:
       test: pg_isready -U postgres
       start_period: 5s

postgres/initdb/03-initdb_fill_edge_uuid.sql

@@ -34,4 +34,5 @@ with new_edges as(
     where type='UnnamedPipeObject'
 )
 insert into edge (origin, destination, type)
-select origin, destination, type::EDGE_TYPE from new_edges;
+select origin::UUID, destination::UUID, type::EDGE_TYPE from new_edges
+where destination is not null;

postgres/queries.sql

@@ -66,7 +66,7 @@ PREPARE shortest_path (int, int, int) AS
 
 PREPARE two_hop_old (INTEGER) AS
   WITH hop1 AS (
-    SELECT origin, destination
+    SELECT DISTINCT origin, destination
     FROM edge
     WHERE origin = $1
     OR destination = $1
@@ -103,3 +103,13 @@ PREPARE two_hop (INTEGER) AS
   SELECT destination
   FROM double_edge
   JOIN hop1 ON vertex = origin AND destination <> $1;
+
+PREPARE neighborhood (INTEGER) AS
+  WITH double_edge AS NOT MATERIALIZED (
+    SELECT origin, destination FROM edge
+    UNION ALL
+    SELECT destination, origin FROM edge
+  )
+  SELECT destination AS vertex
+  FROM double_edge
+  WHERE origin = $1 AND destination <> $1;
\ No newline at end of file

stats.py

+import csv
+from statistics import fmean, mode, pvariance
+from os import environ
+from random import sample
+from typing import Any
+import psycopg
+
+max_vert = 46761949
+
+neighbors = """
+explain (analyze, format json) (
+WITH double_edge AS NOT MATERIALIZED (
+    SELECT origin, destination FROM edge
+    UNION ALL
+    SELECT destination, origin FROM edge
+)
+SELECT destination AS vertex
+FROM double_edge
+WHERE origin = %(id)s AND destination <> %(id)s
+);
+"""
+
+two_hop = """
+explain (analyze, format json) (
+WITH double_edge AS NOT MATERIALIZED (
+    SELECT origin, destination FROM edge
+    UNION ALL
+    SELECT destination, origin FROM edge
+), hop1 AS (
+    SELECT destination AS vertex
+    FROM double_edge
+    WHERE origin = %(id)s AND destination <> %(id)s AND destination <> 5
+)
+SELECT vertex
+FROM hop1
+UNION
+SELECT destination
+FROM double_edge
+JOIN hop1 ON vertex = origin AND destination <> %(id)s
+);
+"""
+
+environ["PGHOST"] = "localhost"
+environ["PGPASSFILE"] = ".pgpass"
+
+with psycopg.connect() as conn, conn.cursor() as cur:
+    selected = (
+        {"id": id}
+        for id in {
+            22280515,
+            29465322,
+            29545621,
+            29701639,
+            30584080,
+            31295043,
+            35584630,
+            35723190,
+            36003677,
+        }
+    )
+    stats = []
+    for id in selected:
+        result: Any = cur.execute(two_hop, id, prepare=True).fetchone()
+        rows = result[0][0]["Plan"]["Actual Rows"]
+        time = result[0][0]["Plan"]["Actual Total Time"]
+        stats.append((id["id"], int(rows), float(time)))
+
+stats.sort(key=lambda tpl: tpl[1])
+with open("result/two_hop_stats.csv", mode="x") as file:
+    writer = csv.writer(file)
+    writer.writerow(("VERTEX", "RESULT_SIZE", "QUERY_LATENCY"))
+    writer.writerows(stats)