From f51cf2025e8f7e73193c57c0072d91c4570defb1 Mon Sep 17 00:00:00 2001
From: Vanhala Antti <vanhala1@niksula.hut.fi>
Date: Fri, 30 May 2014 17:34:00 +0300
Subject: [PATCH] Pushing nodes to a database and generating graph from it

---
 .gitignore                 |   3 +-
 mapper/conf_sh.example.py  |  19 ---
 mapper/makeGraph.py        | 231 -------------------------------------
 mapper/start-mappers.sh    |  56 ---------
 scripts/sendGraph.py       |  20 ++--
 web/database.py            |  80 +++++++++++++
 web/database.sql           |  31 +++++
 web/graph.py               |  29 +++++
 web/graphData.py           |  28 ++++-
 web/graphPlotter.py        |  74 ++++++++++++
 web/updateGraph.py         |  26 +++++
 web/web.py                 |   6 +-
 web/web_config.example.cfg |   8 ++
 13 files changed, 293 insertions(+), 318 deletions(-)
 delete mode 100644 mapper/conf_sh.example.py
 delete mode 100755 mapper/makeGraph.py
 delete mode 100755 mapper/start-mappers.sh
 create mode 100644 web/database.py
 create mode 100644 web/database.sql
 create mode 100644 web/graph.py
 create mode 100644 web/graphPlotter.py
 create mode 100755 web/updateGraph.py
 create mode 100644 web/web_config.example.cfg

diff --git a/.gitignore b/.gitignore
index 6ef403c..054c358 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,6 +1,5 @@
 .cache
 *.pyc
 graph.json
-mapper/mapper-confs/*
-conf_sh.py
+web_config.cfg
 lighttp.conf
diff --git a/mapper/conf_sh.example.py b/mapper/conf_sh.example.py
deleted file mode 100644
index 0c98ee0..0000000
--- a/mapper/conf_sh.example.py
+++ /dev/null
@@ -1,19 +0,0 @@
-#
-# This file should be runnable by bash and python!
-#
-
-cjdns_path="/home/user/cjdns"
-graph_output="../web/static/graph.json"
-num_of_nodes=30
-
-# Where mapper nodes connect to
-peer_ip="127.0.0.1"
-peer_port="33333"
-peer_pw="mapper-peers-hunter2qwertyuiopoiuytrewq"
-peer_pk="osufn28fjjduan29dajsdnasiqlqn8ahasoasa.k"
-
-# Admin RPC of mapper nodes
-rpc_bind="127.0.0.1"
-rpc_connect="127.0.0.1"
-rpc_pw="Kjs8HuaKu2afdw"
-rpc_firstport=11244
diff --git a/mapper/makeGraph.py b/mapper/makeGraph.py
deleted file mode 100755
index b7fb26a..0000000
--- a/mapper/makeGraph.py
+++ /dev/null
@@ -1,231 +0,0 @@
-#!/usr/bin/env python
-
-import conf_sh as conf
-import sys
-sys.path.append(conf.cjdns_path + '/contrib/python/cjdnsadmin/')
-import adminTools as admin
-from collections import deque
-import pygraphviz as pgv
-import json
-import time
-import httplib2
-import traceback
-
-
-class Node:
-	def __init__(self, ip):
-		self.ip = ip
-		self.version = -1
-		self.label = ip[-4:]
-
-	def __lt__(self, b):
-		return self.ip < b.ip
-
-class Edge:
-	def __init__(self, a, b):
-		self.a, self.b = sorted([a, b])
-
-	def is_in(self, edges):
-		for e in edges:
-			if e.a.ip == self.a.ip and e.b.ip == self.b.ip:
-					return True
-		return False
-
-
-
-def get_network_from_cjdns(ip, port, password):
-	nodes = dict()
-	edges = []
-
-	cjdns = admin.connect(ip, port, password)
-	me = admin.whoami(cjdns)
-	my_ip = me['IP']
-	nodes[my_ip] = Node(my_ip)
-
-	nodes_to_check = deque()
-	nodes_to_check.append(my_ip)
-
-	while len(nodes_to_check) != 0:
-		current_ip = nodes_to_check.popleft()
-		resp = cjdns.NodeStore_nodeForAddr(current_ip)
-
-		if not 'result' in resp or not 'linkCount' in resp['result']:
-			continue
-
-		result = resp['result']
-		link_count = result['linkCount']
-		
-		if 'protocolVersion' in result:
-			nodes[current_ip].version = result['protocolVersion']
-
-
-		for i in range(0, link_count):
-			result = cjdns.NodeStore_getLink(current_ip, i)['result']
-			if not 'child' in result:
-				continue
-
-			child_ip = result['child']
-
-			# Add links with one hop only
-			if result['isOneHop'] != 1:
-				continue
-
-			# Add node
-			if not child_ip in nodes:
-				nodes[child_ip] = Node(child_ip)
-				nodes_to_check.append(child_ip)
-
-			# Add edge
-			e = Edge(nodes[current_ip], nodes[child_ip])
-			if not e.is_in(edges):
-				edges.append(e)
-
-	return (nodes, edges)
-
-
-def get_full_network():
-	all_nodes = dict()
-	all_edges = []
-
-	for i in range(0, conf.num_of_nodes):
-		port = conf.rpc_firstport + i
-
-		print '[%d/%d] Connecting to %s:%d...' % (i + 1, conf.num_of_nodes, conf.rpc_connect, port),
-		sys.stdout.flush()
-
-		try:
-			nodes, edges = get_network_from_cjdns(conf.rpc_connect, port, conf.rpc_pw)
-		except Exception as ex:
-			print 'Fail! Node unresponsive!'
-			continue
-
-		print '%d nodes, %d edges' % (len(nodes), len(edges))
-
-		for ip, n in nodes.iteritems():
-			all_nodes[ip] = n
-
-		for e in edges:
-			if not e.is_in(all_edges):
-				all_edges.append(e)
-
-	return (all_nodes, all_edges)
-
-
-
-
-
-def download_names_from_nameinfo():
-	page = 'http://[fc5d:baa5:61fc:6ffd:9554:67f0:e290:7535]/nodes/list.json'
-	print 'Downloading names from Mikey\'s nodelist...',
-	sys.stdout.flush()
-	
-	ip_dict = dict()
-	http = httplib2.Http('.cache', timeout=15.0)
-	r, content = http.request(page, 'GET')
-	name_and_ip = json.loads(content)['nodes']
-
-	for node in name_and_ip:
-		ip_dict[node['ip']] = node['name']
-
-	print 'Done!'
-	return ip_dict
-
-
-
-def set_node_names(nodes):
-	try:
-		ip_dict = download_names_from_nameinfo()
-	except Exception as ex:
-		print 'Fail!'
-		# TODO use cache
-		print traceback.format_exc()
-		return
-
-	for ip, node in nodes.iteritems():
-		if ip in ip_dict:
-			node.label = ip_dict[ip]
-
-
-
-
-def build_graph(nodes, edges):
-	G = pgv.AGraph(strict=True, directed=False, size='10!')
-
-	for n in nodes.values():
-		G.add_node(n.ip, label=n.label, version=n.version)
-
-	for e in edges:
-		G.add_edge(e.a.ip, e.b.ip, len=1.0)
-
-	G.layout(prog='neato', args='-Gepsilon=0.0001 -Gmaxiter=100000')
-
-	return G
-
-
-
-def gradient_color(ratio, colors):
-	jump = 1.0 / (len(colors) - 1)
-	gap_num = int(ratio / (jump + 0.0000001))
-
-	a = colors[gap_num]
-	b = colors[gap_num + 1]
-
-	ratio = (ratio - gap_num * jump) * (len(colors) - 1)
-
-	r = a[0] + (b[0] - a[0]) * ratio
-	g = a[1] + (b[1] - a[1]) * ratio
-	b = a[2] + (b[2] - a[2]) * ratio
-
-	return '#%02x%02x%02x' % (r, g, b)
-
-
-def get_graph_json(G):
-	max_neighbors = 1
-	for n in G.iternodes():
-		neighbors = len(G.neighbors(n))
-		if neighbors > max_neighbors:
-			max_neighbors = neighbors
-	print 'Max neighbors: %d' % max_neighbors
-
-	out_data = {
-		'created': int(time.time()),
-		'nodes': [],
-		'edges': []
-	}
-
-	for n in G.iternodes():
-		neighbor_ratio = len(G.neighbors(n)) / float(max_neighbors)
-		pos = n.attr['pos'].split(',', 1)
-
-		out_data['nodes'].append({
-			'id': n.name,
-			'label': n.attr['label'],
-			'version': n.attr['version'],
-			'x': float(pos[0]),
-			'y': float(pos[1]),
-			'color': gradient_color(neighbor_ratio, [(100, 100, 100), (0, 0, 0)]),
-			'size': neighbor_ratio
-		})
-
-	for e in G.iteredges():
-		out_data['edges'].append({
-			'sourceID': e[0],
-			'targetID': e[1]
-		})
-
-	return json.dumps(out_data)
-
-
-
-
-if __name__ == '__main__':
-	nodes, edges = get_full_network()
-	print 'Total:'
-	print '%d nodes, %d edges' % (len(nodes), len(edges))
-	
-	set_node_names(nodes)
-	G = build_graph(nodes, edges)
-	output = get_graph_json(G)
-
-	with open(conf.graph_output, 'w') as f:
-		f.write(output)
diff --git a/mapper/start-mappers.sh b/mapper/start-mappers.sh
deleted file mode 100755
index cb0f12c..0000000
--- a/mapper/start-mappers.sh
+++ /dev/null
@@ -1,56 +0,0 @@
-#!/bin/bash
-
-source conf_sh.py
-mkdir -p mapper-confs
-
-# Generate configurations and collect their publick keys and ports
-for i in $(seq 1 $num_of_nodes)
-do
-	file=mapper-confs/node$i.conf
-
-	$cjdns_path/cjdroute --genconf > $file
-
-	# Get connecting info
-	publicKey=$(grep -oP -m1 '(?<="publicKey": ").*(?=",)' $file)
-	connectPort=$(grep -oP -m1 '(?<="0.0.0.0:).*(?=",)' $file)
-	connectToInfo[i]='"127.0.0.1:'"$connectPort"'":{"password":"'"$rpc_pw"'","publicKey":"'"$publicKey"'"},'
-done
-
-# Modify configurations
-for i in $(seq 1 $num_of_nodes)
-do
-	echo "Starting mapper node $i/$num_of_nodes"
-
-	file=mapper-confs/node$i.conf
-	rpcport=$(($rpc_firstport + $i - 1))
-
-	# Connect to all mapper nodes except itself
-	connectInfo=""
-	for j in $(seq 1 $num_of_nodes)
-	do
-		if [[ $i != $j ]]; then
-			connectInfo+="${connectToInfo[j]}"
-		fi
-	done
-
-	# Set peer credentials
-	sed -i 's/\/\/ Add connection credentials here to join the network/'"$connectInfo"'/g' $file
-	sed -i 's/\/\/ Ask somebody who is already connected./"'"${peer_ip}"':'"${peer_port}"'":{"password":"'"${peer_pw}"'","publicKey":"'"${peer_pk}"'"}/g' $file
-
-	# Set admin rpc credentials
-	sed -i 's/127.0.0.1:11234/'"${rpc_bind}"':'"${rpcport}"'/g' $file
-	sed -i 's/"password": ".*"/"password": "'"${rpc_pw}"'"/g' $file
-
-	# Disable tun interface
-	sed -i 's/"type": "TUNInterface"/\/\/"type": "TUNInterface"/g' $file
-
-	# Start mappers
-	if [[ $* == *-d* ]]; then
-		# Log to stdout
-		sed -i 's/\/\/ "logTo":"stdout"/"logTo":"stdout"/g' $file
-
-		gdb $cjdns_path/cjdroute -ex 'set follow-fork-mode child' -ex 'run < '"${file}" -ex 'thread apply all bt' -ex 'quit' > gdb-$i.log 2>&1 &
-	else
-		$cjdns_path/cjdroute < $file
-	fi
-done
diff --git a/scripts/sendGraph.py b/scripts/sendGraph.py
index 06141ff..7fceb1a 100755
--- a/scripts/sendGraph.py
+++ b/scripts/sendGraph.py
@@ -40,13 +40,19 @@ import cjdnsadmin
 
 
 def main():
+	print "Connecting to cjdns...",; sys.stdout.flush()
 	cjdns = cjdns_connect()
+	print "Done!"
+
 	success = generate_and_send_graph(cjdns)
 	sys.exit(0 if success else 1)
 
 def generate_and_send_graph(cjdns):
 	source_nodes = cjdns_get_node_store(cjdns)
+	print "Found %d source nodes." % len(source_nodes)
+
 	nodes, edges = cjdns_graph_from_nodes(cjdns, source_nodes)
+	print "Found %d nodes and %d links." % (len(nodes), len(edges))
 
 	graph_data = {
 		'nodes': [],
@@ -56,7 +62,6 @@ def generate_and_send_graph(cjdns):
 	for n in nodes.values():
 		graph_data['nodes'].append({
 			'ip':      n.ip,
-			'key':     n.key,
 			'version': n.version
 		})
 	
@@ -67,15 +72,19 @@ def generate_and_send_graph(cjdns):
 		})
 
 	json_str = json.dumps(graph_data)
-	return send_data(json_str)
+
+	print "Sending data...",; sys.stdout.flush()
+	success = send_data(json_str)
+	print ("Done!" if success else "Failed!")
+
+	return success
 
 
 
 class Node:
-	def __init__(self, ip, version=None, key=None):
+	def __init__(self, ip, version=None):
 		self.ip = ip
 		self.version = version
-		self.key = key
 
 class Edge:
 	def __init__(self, a, b):
@@ -140,9 +149,6 @@ def cjdns_graph_from_nodes(cjdns, source_nodes):
 			continue
 		res = resp['result']
 
-		if 'key' in res:
-			node.key = res['key']
-
 		if 'protocolVersion' in res:
 			node.version = res['protocolVersion']
 
diff --git a/web/database.py b/web/database.py
new file mode 100644
index 0000000..5c9111e
--- /dev/null
+++ b/web/database.py
@@ -0,0 +1,80 @@
+import MySQLdb as mdb
+from graph import Node, Edge
+import time
+
+
+class NodeDB:
+	def __init__(self, config):
+		self.con = mdb.connect(
+			config['MYSQL_DATABASE_HOST'],
+			config['MYSQL_DATABASE_USER'],
+			config['MYSQL_DATABASE_PASSWORD'],
+			config['MYSQL_DATABASE_DB'])
+		self.cur = self.con.cursor()
+
+	def __enter__(self):
+		return self
+
+	def __exit__(self, type, value, traceback):
+		self.con.commit()
+		self.con.close()
+
+
+
+	def insert_node(self, node):
+		now = int(time.time())
+		self.cur.execute('''
+			INSERT INTO nodes (ip, name, version, first_seen, last_seen)
+			VALUES (%s, %s, %s, %s, %s)
+			ON DUPLICATE KEY
+			UPDATE name = %s, version = %s, last_seen = %s''', (
+			node.ip, node.label, node.version, now, now,
+			node.label, node.version, now))
+
+	def insert_edge(self, edge):
+		now = int(time.time())
+		self.cur.execute('''
+			INSERT INTO edges (a, b, first_seen, last_seen)
+			VALUES (%s, %s, %s, %s)
+			ON DUPLICATE KEY
+			UPDATE last_seen = %s''', (
+			edge.a.ip, edge.b.ip, now, now,
+			now))
+
+	def insert_graph(self, nodes, edges):
+		for n in nodes.itervalues():
+			self.insert_node(n)
+
+		for e in edges:
+			self.insert_edge(e)
+
+
+
+	def get_nodes(self, time_limit):
+		since = int(time.time() - time_limit)
+		cur = self.con.cursor(mdb.cursors.DictCursor)
+		cur.execute("SELECT ip, version, name FROM nodes WHERE last_seen > %s", (since,))
+		db_nodes = cur.fetchall()
+
+		nodes = dict()
+		for n in db_nodes:
+			nodes[n['ip']] = Node(n['ip'], n['version'], n['name'])
+
+		return nodes
+
+	def get_edges(self, nodes, time_limit):
+		since = int(time.time() - time_limit)
+		cur = self.con.cursor(mdb.cursors.DictCursor)
+		cur.execute("SELECT a, b FROM edges WHERE last_seen > %s", (since,))
+		db_edges = cur.fetchall()
+
+		edges = []
+		for e in db_edges:
+			edges.append(Edge(nodes[e['a']], nodes[e['b']]))
+
+		return edges
+
+	def get_graph(self, time_limit):
+		nodes = self.get_nodes(time_limit)
+		edges = self.get_edges(nodes, time_limit)
+		return (nodes, edges)
diff --git a/web/database.sql b/web/database.sql
new file mode 100644
index 0000000..dbe7dbd
--- /dev/null
+++ b/web/database.sql
@@ -0,0 +1,31 @@
+CREATE DATABASE  IF NOT EXISTS `fc00`;
+USE `fc00`;
+
+
+--
+-- Table structure for table `edges`
+--
+
+DROP TABLE IF EXISTS `edges`;
+CREATE TABLE `edges` (
+  `a` varchar(39) NOT NULL,
+  `b` varchar(39) NOT NULL,
+  `first_seen` int(11) NOT NULL,
+  `last_seen` int(11) NOT NULL,
+  PRIMARY KEY (`a`,`b`)
+) ENGINE=InnoDB DEFAULT CHARSET=utf8;
+
+
+--
+-- Table structure for table `nodes`
+--
+
+DROP TABLE IF EXISTS `nodes`;
+CREATE TABLE `nodes` (
+  `ip` varchar(39) NOT NULL,
+  `name` varchar(64) DEFAULT NULL,
+  `version` int(11) DEFAULT NULL,
+  `first_seen` int(11) NOT NULL,
+  `last_seen` int(11) NOT NULL,
+  PRIMARY KEY (`ip`)
+) ENGINE=InnoDB DEFAULT CHARSET=utf8;
diff --git a/web/graph.py b/web/graph.py
new file mode 100644
index 0000000..ff9e470
--- /dev/null
+++ b/web/graph.py
@@ -0,0 +1,29 @@
+class Node:
+	def __init__(self, ip, version=None, label=None):
+		self.ip = ip
+		self.version = version
+		self.label = ip[-4:] if label == None else label
+
+	def __lt__(self, b):
+		return self.ip < b.ip
+
+	def __repr__(self):
+		return 'Node(ip="%s", version=%s, label="%s")' % (
+			self.ip,
+			self.version,
+			self.label)
+
+class Edge:
+	def __init__(self, a, b):
+		self.a, self.b = sorted([a, b])
+
+	def is_in(self, edges):
+		for e in edges:
+			if e.a.ip == self.a.ip and e.b.ip == self.b.ip:
+					return True
+		return False
+
+	def __repr__(self):
+		return 'Edge(a.ip="%s", b.ip="%s")' % (
+			self.a.ip,
+			self.b.ip)
diff --git a/web/graphData.py b/web/graphData.py
index 2e5d0a8..28c3dc0 100644
--- a/web/graphData.py
+++ b/web/graphData.py
@@ -1,9 +1,35 @@
 import json
+from database import NodeDB
+from graph import Node, Edge
 
-def insert_graph_data(json_str):
+def insert_graph_data(config, json_str):
 	try:
 		graph_data = json.loads(json_str)
 	except ValueError:
 		return False
 
+	nodes = dict()
+	edges = []
+
+	if not 'nodes' in graph_data or not 'edges' in graph_data:
+		return False
+	
+
+	try:
+		for n in graph_data['nodes']:
+			node = Node(n['ip'], version=n['version'])
+			nodes[n['ip']] = node
+
+		for e in graph_data['edges']:
+			edge = Edge(nodes[e['a']], nodes[e['b']])
+			edges.append(edge)
+
+	except TypeError:
+		return False
+
+	print "Received %d nodes and %d links." % (len(nodes), len(edges))
+
+	with NodeDB(config) as db:
+		db.insert_graph(nodes, edges)
+
 	return True
diff --git a/web/graphPlotter.py b/web/graphPlotter.py
new file mode 100644
index 0000000..017b153
--- /dev/null
+++ b/web/graphPlotter.py
@@ -0,0 +1,74 @@
+import pygraphviz as pgv
+import time
+import json
+
+
+def position_nodes(nodes, edges):
+	G = pgv.AGraph(strict=True, directed=False, size='10!')
+
+	for n in nodes.values():
+		G.add_node(n.ip, label=n.label, version=n.version)
+
+	for e in edges:
+		G.add_edge(e.a.ip, e.b.ip, len=1.0)
+
+	G.layout(prog='neato', args='-Gepsilon=0.0001 -Gmaxiter=100000')
+
+	return G
+
+
+
+def get_graph_json(G):
+	max_neighbors = 1
+	for n in G.iternodes():
+		neighbors = len(G.neighbors(n))
+		if neighbors > max_neighbors:
+			max_neighbors = neighbors
+	print 'Max neighbors: %d' % max_neighbors
+
+	out_data = {
+		'created': int(time.time()),
+		'nodes': [],
+		'edges': []
+	}
+
+	for n in G.iternodes():
+		neighbor_ratio = len(G.neighbors(n)) / float(max_neighbors)
+		pos = n.attr['pos'].split(',', 1)
+
+		out_data['nodes'].append({
+			'id': n.name,
+			'label': n.attr['label'],
+			'version': n.attr['version'],
+			'x': float(pos[0]),
+			'y': float(pos[1]),
+			'color': _gradient_color(neighbor_ratio, [(100, 100, 100), (0, 0, 0)]),
+			'size': neighbor_ratio
+		})
+
+	for e in G.iteredges():
+		out_data['edges'].append({
+			'sourceID': e[0],
+			'targetID': e[1]
+		})
+
+	return json.dumps(out_data)
+
+
+
+
+
+def _gradient_color(ratio, colors):
+	jump = 1.0 / (len(colors) - 1)
+	gap_num = int(ratio / (jump + 0.0000001))
+
+	a = colors[gap_num]
+	b = colors[gap_num + 1]
+
+	ratio = (ratio - gap_num * jump) * (len(colors) - 1)
+
+	r = a[0] + (b[0] - a[0]) * ratio
+	g = a[1] + (b[1] - a[1]) * ratio
+	b = a[2] + (b[2] - a[2]) * ratio
+
+	return '#%02x%02x%02x' % (r, g, b)
\ No newline at end of file
diff --git a/web/updateGraph.py b/web/updateGraph.py
new file mode 100755
index 0000000..b06d679
--- /dev/null
+++ b/web/updateGraph.py
@@ -0,0 +1,26 @@
+#!/usr/bin/env python
+from flask import Config
+from database import NodeDB
+import graphPlotter
+
+
+def generate_graph(time_limit=60*60*3):
+	nodes, edges = load_graph_from_db(time_limit)
+
+	graph = graphPlotter.position_nodes(nodes, edges)
+	json = graphPlotter.get_graph_json(graph)
+
+	with open('static/graph.json', 'w') as f:
+		f.write(json)
+
+
+def load_graph_from_db(time_limit):
+	config = Config('./')
+	config.from_pyfile('web_config.cfg')
+
+	with NodeDB(config) as db:
+		return db.get_graph(time_limit)
+
+
+if __name__ == '__main__':
+	generate_graph()
diff --git a/web/web.py b/web/web.py
index 1c5e4b9..33a4f99 100644
--- a/web/web.py
+++ b/web/web.py
@@ -2,7 +2,7 @@ from flask import Flask, render_template, request
 from graphData import insert_graph_data
 
 app = Flask(__name__)
-app.debug = False
+app.config.from_pyfile('web_config.cfg')
 
 
 @app.context_processor
@@ -21,8 +21,10 @@ def page_about():
 
 @app.route('/sendGraph', methods=['POST'])
 def page_sendGraph():
+	print "Receiving graph from %s" % (request.remote_addr)
+	
 	data = request.form['data']
-	ret = insert_graph_data(data)
+	ret = insert_graph_data(app.config, data)
 	if ret:
 		return 'OK'
 	else:
diff --git a/web/web_config.example.cfg b/web/web_config.example.cfg
new file mode 100644
index 0000000..50aacc0
--- /dev/null
+++ b/web/web_config.example.cfg
@@ -0,0 +1,8 @@
+DEBUG = False
+
+MYSQL_DATABASE_HOST     = 'localhost'
+MYSQL_DATABASE_PORT     = 3306
+MYSQL_DATABASE_USER     = 'fc00'
+MYSQL_DATABASE_PASSWORD = 'hunter2'
+MYSQL_DATABASE_DB       = 'fc00'
+MYSQL_DATABASE_CHARSET  = 'utf8'