Merge pull request #4 from Arceliar/crawler

v0.3 changes

README.md

@@ -4,7 +4,7 @@ Forked from the code for http://www.fc00.org (http://h.fc00.org on Hyperboria).
 
 ## Sending your view of the network
 
-This code reads a map of known nodes from `y.yakamo.org:3000/current` (reachable over yggdrasil). In order to display an accurate map of the network, we need your help. If you run a yggdrasil node, plase send your network view using the [send-view.py](https://github.com/yakamok/Niflheim-api/blob/master/send-view.py) script.
+This code reads a map of known nodes from `y.yakamo.org:3000/current` (reachable over yggdrasil). You may alternatively generate your own view of the network by running [a crawler script](scripts/crawl-dht.py), but this may take some time (figuring out how to run it and use the results is left as an exercise to the user).
 
 ## Web server
 ```bash

scripts/crawl-dht.py

@@ -0,0 +1,75 @@
+import json
+import socket
+import sys
+import time
+
+#gives the option to get data from an external server instead and send that
+#if no options given it will default to localhost instead
+if len(sys.argv) == 3:
+  host_port = (sys.argv[1], int(sys.argv[2]))
+else:
+  host_port = ('localhost', 9001)
+
+def getDHTPingRequest(key, coords, target=None):
+  if target:
+    return '{{"keepalive":true, "request":"dhtPing", "box_pub_key":"{}", "coords":"{}", "target":"{}"}}'.format(key, coords, target)
+  else:
+    return '{{"keepalive":true, "request":"dhtPing", "box_pub_key":"{}", "coords":"{}"}}'.format(key, coords)
+
+def doRequest(req):
+  try:
+    ygg = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+    ygg.connect(host_port)
+    ygg.send(req)
+    data = json.loads(ygg.recv(1024*15))
+    return data
+  except:
+    return None
+
+visited = dict() # Add nodes after a successful lookup response
+rumored = dict() # Add rumors about nodes to ping
+timedout = dict()
+def handleResponse(address, info, data):
+  global visited
+  global rumored
+  global timedout
+  timedout[str(address)] = {'box_pub_key':str(info['box_pub_key']), 'coords':str(info['coords'])}
+  if not data: return
+  if 'response' not in data: return
+  if 'nodes' not in data['response']: return
+  for addr,rumor in data['response']['nodes'].iteritems():
+    if addr in visited: continue
+    rumored[addr] = rumor
+  if address not in visited:
+    # TODO? remove this, it's debug output that happens to be in the same format as yakamo's "current" json file
+    now = time.time()
+    visited[str(address)] = {'box_pub_key':str(info['box_pub_key']), 'coords':str(info['coords']), 'time':now}
+    if address in timedout: del timedout[address]
+    if len(visited) > 1: sys.stdout.write(",\n")
+    sys.stdout.write('"{}": ["{}", {}]'.format(address, info['coords'], int(now)))
+    sys.stdout.flush()
+# End handleResponse
+
+# Get self info
+selfInfo = doRequest('{"keepalive":true, "request":"getSelf"}')
+
+# Initialize dicts of visited/rumored nodes
+for k,v in selfInfo['response']['self'].iteritems(): rumored[k] = v
+
+# Loop over rumored nodes and ping them, adding to visited if they respond
+print '{"yggnodes": {'
+while len(rumored) > 0:
+  for k,v in rumored.iteritems():
+    handleResponse(k, v, doRequest(getDHTPingRequest(v['box_pub_key'], v['coords'])))
+    # These next two are imperfect workarounds to deal with old kad nodes
+    handleResponse(k, v, doRequest(getDHTPingRequest(v['box_pub_key'], v['coords'], '0'*128)))
+    handleResponse(k, v, doRequest(getDHTPingRequest(v['box_pub_key'], v['coords'], 'f'*128)))
+    break
+  del rumored[k]
+print '\n}}'
+#End
+
+# TODO do something with the results
+
+#print visited
+#print timedout

web/graphPlotter.py

@@ -8,10 +8,10 @@ def position_nodes(nodes, edges):
     G = pgv.AGraph(strict=True, directed=False, size='10!')
 
     for n in nodes.values():
-        G.add_node(n.coords, label=n.ip, version=n.version)
+        G.add_node(n.ip, label=n.label, coords=n.coords)
 
     for e in edges:
-        G.add_edge(e.a.coords, e.b.coords, len=1.0)
+        G.add_edge(e.a.ip, e.b.ip, len=1.0)
 
     G.layout(prog='neato', args='-Gepsilon=0.0001 -Gmaxiter=100000')
 
@@ -66,7 +66,7 @@ def get_graph_json(G):
             'id': n.name,
             'label': name if name else n.attr['label'],
             'name': name,
-            'version': n.attr['version'],
+            'coords': n.attr['coords'],
             'x': float(pos[0]),
             'y': float(pos[1]),
             'color': _gradient_color(neighbor_ratio, [(100, 100, 100), (0, 0, 0)]),

web/static/network.js

@@ -174,7 +174,7 @@ function showNodeInfo(node) {
         '<h2>' + node.label + '</h2>' +
         '<span class="tt">' + node.id + '</span><br>' +
         '<br>' +
-        '<strong>Version:</strong> ' + node.version + '<br>' +
+        '<strong>Coords:</strong> ' + node.coords + '<br>' +
         '<strong>Peers:</strong> ' + node.peers.length + '<br>' +
         '<strong>Centrality:</strong> ' + node.centrality + '<br>' +
         '<table>' +

web/templates/about.html

@@ -8,7 +8,7 @@
 			<br>
 
 			<h3>Network map</h3>
-			<p>The network page has a map of Yggdrasil's spanning tree as it is now. The map is not complete since it is hard/impossible to get a full picture of the network, and it only includes the minimum subset of links needed to construct the spanning tree. The known nodes and tree coordinates are taken from <a href="http://y.yakamo.org:3000/">Yakamo's API</a>. You can submit your node's view of the network by periodically running <a href="https://github.com/yakamok/Niflheim-api/blob/master/send-view.py">send-view.py</a>.</p>
+			<p>The network page has a map of Yggdrasil's spanning tree as it is now. The map is not complete since it is hard/impossible to get a full picture of the network, and it only includes the minimum subset of links needed to construct the spanning tree. The known nodes and tree coordinates are taken from <a href="http://y.yakamo.org:3000/">Yakamo's API</a>.</p>
 			<!--
 			<h3>Node names</h3>
 			<p>For now, node names are assigned manually. You can submit Pull Request to <a href="https://github.com/zielmicha/nodedb">nodedb</a> if you want to have your node named. 

web/updateGraph.py

@@ -4,7 +4,7 @@ from database import NodeDB
 import graphPlotter
 
 import urllib, json
-url = "http://y.yakamo.org:3000/current"
+url = "current" #alternatively "http://y.yakamo.org:3000/current"
 
 # nodes indexed by coords
 class NodeInfo:
@@ -40,6 +40,7 @@ def generate_graph(time_limit=60*60*3):
     nodes = dict()
     def addAncestors(info):
       parent = NodeInfo("?", info.getParent())
+      parent.label = "{} {}".format(parent.ip, parent.coords)
       nodes[parent.coords] = parent
       if parent.coords != parent.getParent(): addAncestors(parent)