TEMP: Working data pipeline & rendering

app/collector.py

@@ -1,102 +1,63 @@
-#!/usr/bin/env python3.11
-
-import re
 import json
-import ipaddress
 import uuid
 import hashlib
 
+from datetime import datetime
+
 from .db import Database
 
-def parse_traceroute_output(output, timestamp):
-    lines = output.strip().split('\n')
-    trace = {}
-    hops = []
 
-    ip_regex = r"\((.*?)\)" # ipaddress are in ()
+def parse_traceroute_output(data: str, origin: str):
+    lines = data.strip().split("\n")
+    target = lines[0].split()[2]
 
-    target = output.strip().split('\n')[0].split()[2]
+    created = datetime.now().isoformat()
+
+    trace = {"target": target, "created": created, "origin": origin, "hops": []}
+
+    prev_latency = 0
 
     for line in lines[1:]:
-        hop = {}
         hop_info = line.split()
-        hop_number = int(hop_info[0])
-        hop_name = None
-        hop_ip = None
-        hop_latency = None
+        print("LINE:", hop_info)
+        try:
+            number, name, ip, latency, _ = hop_info
+            hop = {
+                "created": created,
+                "number": number,
+                "name": name,
+                "ip": ip.strip("()"),
+                "latency": float(latency),
+            }
+        except ValueError:
+            number, name = hop_info
+            hop = {
+                "created": created,
+                "number": number,
+                "name": name,
+                "ip": None,
+                "latency": None,
+            }
 
-        latencies = []
-
-        #print("##### "+str(hop_info))
-        count = 0
-        for part in hop_info[1:]:
-            count += 1
-
-            # source node drops or blocks icmp packages
-            # We will give funny to name to hop for not answering and move on.
-            if part == '*':
-                hop_name = 'unresponsive'
-                hop_ip = 'unresponsive'
-                break
-
-
-            # If first colum is either name or ip-address
-            if count == 1:
-                match = re.search(ip_regex, part)
-                if  match:
-                    hop_ip = part.strip('()')
-                else:
-                    hop_name = part
-
-            # Second colum is ip-address first latency reading
-            if count == 2:
-                if re.search(ip_regex, part):
-                    try:
-                        _ip = ipaddress.ip_address(part.strip('()'))
-                        hop_ip = part.strip('()')
-                    except ValueError:
-                        pass  # Ignore if it's not a valid IP address
-
-            # Rest of the input colums are either latency floats, 'ms' or
-            # reruns of the hop_name and hop_ip...
-            # We only need the latency floats anymore.
-            else:
-                try:
-                    latency = float(part)
-                    latencies.append(latency)
-                except ValueError:
-                    pass
-
-        hop_latency = sum(latencies) / len(latencies) if latencies else None
-
-        hop['hop_number'] = hop_number
-        if not hop_name == None:
-            hop['hop_name'] = hop_name
-        hop['hop_ip'] = hop_ip
-        hop['hop_latency'] = hop_latency
-        hops.append(hop)
-
-    trace['target'] = target
-    trace['timestamp'] = timestamp
-    trace['hops'] = hops
+        trace["hops"].append(hop)
 
     return trace
 
 
-def store_traceroute(node, trace):
+def store_traceroute(trace):
     db = Database()
 
-    #hops_json = json.dumps(trace['hops'])
+    # hops_json = json.dumps(trace['hops'])
 
     path_ids = {}
 
     previous_hop_ip = None
     previous_hop_latency = None
-    for hop in trace['hops']:
-        hop_number = hop['hop_number']
-        hop_name = hop.get('hop_name')
-        hop_ip = hop.get('hop_ip')
-        hop_latency = hop.get('hop_latency')
+    for hop in trace["hops"]:
+        hop_number = hop["number"]
+        hop_name = hop.get("name")
+        hop_ip = hop.get("ip")
+        hop_latency = hop.get("latency")
         link_id = None
 
         # insert links and get their id's
@@ -112,20 +73,21 @@ def store_traceroute(node, trace):
         # calculate link latency if possible and store it
         if link_id and previous_hop_latency:
             link_latency = hop_latency - previous_hop_latency
-            db.create_latency(link_id, trace['timestamp'], link_latency)
+            db.create_latency(link_id, trace["created"], link_latency)
 
     # make entry to "Paths" table
     if path_ids:
         json_path_ids = json.dumps(path_ids)
-        db.create_path(node, trace['target'], json_path_ids)
+        db.create_path(node, trace["target"], json_path_ids)
 
     db.end()
 
 
-
 def generate_node_id():
     mac = uuid.getnode()
-    mac_str = ':'.join(['{:02x}'.format((mac >> ele) & 0xff) for ele in range(0,8*6,8)][::-1])
+    mac_str = ":".join(
+        ["{:02x}".format((mac >> ele) & 0xFF) for ele in range(0, 8 * 6, 8)][::-1]
+    )
 
     # Hash the MAC address using SHA-256 to generate a unique ID
     unique_id = hashlib.sha256(mac_str.encode()).hexdigest()