From 0e0cba842648e097ff4ff7cc61b8ebc7380382a5 Mon Sep 17 00:00:00 2001
From: kalzu rekku <kalzu@tower.kotiverkko2>
Date: Sun, 26 May 2024 19:10:16 +0300
Subject: [PATCH] New parser and storage functions to support the new database
 schema.

---
 traceroute_collector.py | 118 ++++++++++++++++++++--------------------
 1 file changed, 60 insertions(+), 58 deletions(-)

diff --git a/traceroute_collector.py b/traceroute_collector.py
index 475108f..c8550c2 100755
--- a/traceroute_collector.py
+++ b/traceroute_collector.py
@@ -4,6 +4,7 @@
 import subprocess
 import sqlite3
 import re
+import json
 from datetime import datetime
 
 def run_traceroute(host):
@@ -11,36 +12,25 @@ def run_traceroute(host):
     return result.stdout.decode()
 
 def parse_traceroute_output(output):
-    hops = []
-    lines = output.strip().split('\n')[1:]  # Skip the first line (traceroute to ...)
-    previous_ip = None
-    for line in lines:
-        parts = line.split()
-        hop_number = int(parts[0])
-        ip_address = parts[1]
-        latencies = []
-        for part in parts[2:]:
-            if re.match(r'^[0-9+\s+ms$]', part):
-                latency_str = re.sub(r'[^0-9.]', '', part)
-                if latency_str and not latency_str == '':
-                    try:
-                        print(part)
-                        print(latency_str)
-                        latencies.append(float(latency_str))
-                    except ValueError:
-                        print(f"Could not convert '{latency_str}' to float.")
 
-        avg_latency = sum(latencies) / len(latencies) if latencies else None
-        timestamp = datetime.now().isoformat()
-        if previous_ip:
-            hops.append({
-                'hop_number': hop_number,
-                'source_ip': previous_ip,
-                'destination_ip': ip_address,
-                'latency': avg_latency,
-                'timestamp': timestamp,
-            })
-        previous_ip = ip_address
+    lines = output.strip().split('\n')
+    hops = []
+    for line in lines[1:]:
+        hop_info = line.split()
+        hop_number = int(hop_info[0])
+        hop_ips = []
+        hop_latencies = []
+        for part in hop_info[1:]:
+            if re.match(r'\d+\.\d+\.\d+\.\d+', part):  # Match IP address
+                hop_ips.append(part)
+            elif re.match(r'\d+(\.\d+)? ms', part):  # Match latency
+                hop_latencies.append(float(part.replace(' ms', '')))
+        
+        # If multiple IPs are present, we consider the first as primary
+        primary_ip = hop_ips[0] if hop_ips else None
+        avg_latency = sum(hop_latencies) / len(hop_latencies) if hop_latencies else None
+        hops.append((primary_ip, avg_latency))
+
     return hops
 
 def create_tables(databasefile):
@@ -100,32 +90,40 @@ def create_tables(databasefile):
     conn.commit()
     conn.close()
 
-def store_traceroute(hops):
-    conn = sqlite3.connect('traceroute.db')
+def store_traceroute(db_file, start_ip, end_ip, hops):
+    conn = sqlite3.connect(db_file)
     cursor = conn.cursor()
-    
+
+    # Insert links and get their IDs
+    link_ids = []
     for hop in hops:
-        # Insert or ignore link into links table
-        cursor.execute('''
-            INSERT OR IGNORE INTO links (source_ip, destination_ip)
-            VALUES (?, ?)
-        ''', (hop['source_ip'], hop['destination_ip']))
-        
-        # Retrieve the link_id
-        cursor.execute('''
-            SELECT id FROM links WHERE source_ip = ? AND destination_ip = ?
-        ''', (hop['source_ip'], hop['destination_ip']))
+        source_ip = start_ip if not link_ids else hops[len(link_ids)-1][0]
+        destination_ip = hop[0]
+        latency = hop[1]
+
+        cursor.execute("""
+            INSERT OR IGNORE INTO Links (source_ip, destination_ip) VALUES (?, ?)
+        """, (source_ip, destination_ip))
+        cursor.execute("""
+            SELECT id FROM Links WHERE source_ip = ? AND destination_ip = ?
+        """, (source_ip, destination_ip))
         link_id = cursor.fetchone()[0]
-        
-        # Insert latency data into link_latency table
-        cursor.execute('''
-            INSERT INTO link_latency (link_id, latency, timestamp)
-            VALUES (?, ?, ?)
-        ''', (link_id, hop['latency'], hop['timestamp']))
-    
+        link_ids.append(link_id)
+
+        cursor.execute("""
+            INSERT INTO Latency (link_id, latency_ms) VALUES (?, ?)
+        """, (link_id, latency))
+
+    # Insert path
+    hops_json = json.dumps(link_ids)
+    cursor.execute("""
+        INSERT INTO Paths (start_ip, end_ip, hops_json) VALUES (?, ?, ?)
+    """, (start_ip, end_ip, hops_json))
+
     conn.commit()
     conn.close()
 
+
 def retrieve_traceroute():
     conn = sqlite3.connect('traceroute.db')
     cursor = conn.cursor()
@@ -139,17 +137,21 @@ def retrieve_traceroute():
     conn.close()
     return rows
 
-# Usage
 if __name__ == '__main__':
-    create_tables()
-    traceroute_output = run_traceroute('vi.fi')
+
+    databasefile="./traceroute.db"
+    create_tables(databasefile)
+
+
+    target='vi.fi'
+    traceroute_output = run_traceroute(target)
     hops = parse_traceroute_output(traceroute_output)
-#    for hop in hops:
-#        print(hop)
-    store_traceroute(hops)
-    stored_hops = retrieve_traceroute()
-    for hop in stored_hops:
-        print(f"Link: {hop[0]} -> {hop[1]}, Latency: {hop[2]} ms, Timestamp: {hop[3]}")
+    if hops:
+        start_ip = hops[0][0]
+        store_traceroute(databasefile, start_ip, target, hops)
+#    stored_hops = retrieve_traceroute()
+#    for hop in stored_hops:
+#        print(f"Link: {hop[0]} -> {hop[1]}, Latency: {hop[2]} ms, Timestamp: {hop[3]}")
 
 
 exit(0)