docs: use honeypot dataset in collections notebook

Author: lmeyerov

demos/more_examples/graphistry_features/collections.ipynb

@@ -22,24 +22,13 @@
    "execution_count": null,
    "outputs": [],
    "source": [
+    "from pathlib import Path\n",
     "import pandas as pd\n",
     "import graphistry\n",
     "from graphistry import collection_set, collection_intersection, n, e_forward, Chain\n",
     "\n",
-    "edges = pd.DataFrame({\n",
-    "    'src': ['acct1', 'acct2', 'acct3'],\n",
-    "    'dst': ['user1', 'user2', 'user3'],\n",
-    "    'rel': ['owns', 'owns', 'owns']\n",
-    "})\n",
-    "\n",
-    "nodes = pd.DataFrame({\n",
-    "    'id': ['acct1', 'acct2', 'acct3', 'user1', 'user2', 'user3'],\n",
-    "    'type': ['account', 'account', 'account', 'user', 'user', 'user'],\n",
-    "    'subscribed': [False, True, True, True, False, True],\n",
-    "    'status': ['purchased', 'purchased', 'new', 'new', 'purchased', 'new']\n",
-    "})\n",
-    "\n",
-    "g = graphistry.edges(edges, 'src', 'dst').nodes(nodes, 'id')\n"
+    "edges = pd.read_csv(Path('demos/data/honeypot.csv'))\n",
+    "g = graphistry.edges(edges, \"attackerIP\", \"victimIP\")\n"
    ]
   },
   {
@@ -48,24 +37,27 @@
    "execution_count": null,
    "outputs": [],
    "source": [
+    "# Use Chain to select subgraphs (nodes + edges) by edge attributes\n",
     "collections = [\n",
     "    collection_set(\n",
-    "        expr=Chain([n({'status': 'purchased'}), e_forward(), n()]),\n",
-    "        id='purchasers',\n",
-    "        name='Purchasers + Connections',\n",
+    "        expr=Chain([n(), e_forward({\"vulnName\": \"MS08067 (NetAPI)\"}), n()]),\n",
+    "        id='netapi',\n",
+    "        name='MS08067 (NetAPI)',\n",
     "        node_color='#00BFFF',\n",
     "        edge_color='#00BFFF',\n",
     "    ),\n",
     "    collection_set(\n",
-    "        expr=n({'subscribed': True}),\n",
-    "        id='subscribers',\n",
-    "        name='Subscribers',\n",
+    "        expr=Chain([n(), e_forward({\"victimPort\": 445.0}), n()]),\n",
+    "        id='port445',\n",
+    "        name='Port 445',\n",
     "        node_color='#32CD32',\n",
+    "        edge_color='#32CD32',\n",
     "    ),\n",
     "    collection_intersection(\n",
-    "        sets=['purchasers', 'subscribers'],\n",
-    "        name='High Value Subscribers',\n",
+    "        sets=['netapi', 'port445'],\n",
+    "        name='NetAPI + 445',\n",
     "        node_color='#AABBCC',\n",
+    "        edge_color='#AABBCC',\n",
     "    ),\n",
     "]\n",
     "\n",
@@ -142,16 +134,18 @@
     "```python\n",
     "collections_priority = [\n",
     "    collection_set(\n",
-    "        expr=n({\"vip\": True}),\n",
-    "        id=\"vip\",\n",
-    "        name=\"VIP\",\n",
+    "        expr=Chain([n(), e_forward({\"vulnName\": \"MS08067 (NetAPI)\"}), n()]),\n",
+    "        id=\"netapi\",\n",
+    "        name=\"MS08067 (NetAPI)\",\n",
     "        node_color=\"#FFAA00\",\n",
+    "        edge_color=\"#FFAA00\",\n",
     "    ),\n",
     "    collection_set(\n",
-    "        expr=n({\"recent\": True}),\n",
-    "        id=\"recent\",\n",
-    "        name=\"Recent\",\n",
+    "        expr=Chain([n(), e_forward({\"victimPort\": 445.0}), n()]),\n",
+    "        id=\"port445\",\n",
+    "        name=\"Port 445\",\n",
     "        node_color=\"#00BFFF\",\n",
+    "        edge_color=\"#00BFFF\",\n",
     "    ),\n",
     "]\n",
     "g.collections(collections=collections_priority)\n",