jupyrdfGH-118: Add initial example widget.

examples/GraphSearchWidget.ipynb

@@ -0,0 +1,229 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "f6f51118",
+   "metadata": {},
+   "source": [
+    "## Basic widget for searching a graph to find nodes/edges\n",
+    "\n",
+    "Opinionated search. Looking for specific keys from an RDF graph present in the converted networkx graph."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "338ea955",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import ipywidgets as W\n",
+    "import traitlets as T\n",
+    "from IPython.display import JSON, display\n",
+    "from pathlib import Path\n",
+    "from networkx import Graph as NXGraph\n",
+    "from pandas import DataFrame\n",
+    "from rdflib.graph import Graph as RDFGraph\n",
+    "from rdflib.term import URIRef"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "54b02ad2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from ipyradiant import FileManager, PathLoader\n",
+    "\n",
+    "\n",
+    "lw = FileManager(loader=PathLoader(path=\"data\"))\n",
+    "# here we hard set what we want the file to be, but ideally a user can choose a file to work with.\n",
+    "lw.loader.file_picker.value = lw.loader.file_picker.options[\"starwars.ttl\"]\n",
+    "rdf_graph = lw.graph"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "a946dabd",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# run the converter so that we can work with the networkx graph\n",
+    "from ipyradiant.rdf2nx import RDF2NX\n",
+    "\n",
+    "nx_graph = RDF2NX.convert(rdf_graph)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "ee9363c9",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def get_types(graph: NXGraph):\n",
+    "    \"\"\"Return the options for node types present in a networkx graph.\"\"\"\n",
+    "    types = set()\n",
+    "    for _, data in nx_graph.nodes(data=True):\n",
+    "        # TODO how to make this more generic, e.g. for custom type URIs\n",
+    "        rdf_type = data.get(\"rdf:type\")\n",
+    "        if not rdf_type:\n",
+    "            continue\n",
+    "        elif type(rdf_type) == URIRef:\n",
+    "            types.add(rdf_type)\n",
+    "        elif type(rdf_type) in {tuple, list}:\n",
+    "            types.update(rdf_type)\n",
+    "\n",
+    "    s_types = sorted(types)\n",
+    "    \n",
+    "    # TODO need intelligent URI parsing later on\n",
+    "    return [(f\"<< {Path(type_).name} >>\", (type_, \"node\")) for type_ in s_types]\n",
+    "\n",
+    "\n",
+    "def get_predicates(graph: NXGraph):\n",
+    "    \"\"\"Return the options for edge types present in a networkx graph.\"\"\"\n",
+    "    predicates = set()\n",
+    "    predicate_options = []\n",
+    "    for source, target, edge_data in nx_graph.edges(data=True):\n",
+    "        # TODO how to make this more generic\n",
+    "        predicate = edge_data.get(\"predicate\")\n",
+    "        if not predicate:\n",
+    "            continue\n",
+    "            \n",
+    "        # TODO should this be configurable?\n",
+    "        label = edge_data.get(\"_label\") or Path(predicate).name\n",
+    "        if predicate not in predicates:\n",
+    "            predicates.add(predicate)\n",
+    "            predicate_options.append((f\"o-- {label} --o\", (predicate, \"edge\")))\n",
+    "            \n",
+    "    return sorted(predicate_options, key=lambda x: x[0])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "7839b7bf",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "text_widget = W.Text(placeholder=\"Start typing to search....\")\n",
+    "select_options = W.Select(rows=4)\n",
+    "run_button = W.Button(description=\"Run\")\n",
+    "\n",
+    "# store on class\n",
+    "available_graph_types = get_types(nx_graph)\n",
+    "available_graph_predicates = get_predicates(nx_graph)\n",
+    "\n",
+    "\n",
+    "def return_matches(b):\n",
+    "    \"\"\"Returns nodes/edges that match the user's selection.\"\"\"\n",
+    "    # TODO get nx_graph from widget class\n",
+    "    \n",
+    "    # TODO obv this needs to be linked for real\n",
+    "    try:\n",
+    "        choice, type_ = select_options.value\n",
+    "    except TypeError:\n",
+    "        # there is no selection, so don't do anything\n",
+    "        return\n",
+    "    \n",
+    "    if type_ == \"node\":\n",
+    "        # TODO configure the return types\n",
+    "        # Challenge, node can have multiple types\n",
+    "        matches = set()\n",
+    "        for node, data in nx_graph.nodes(data=True):\n",
+    "            types = data.get(\"rdf:type\")\n",
+    "            if (\n",
+    "                (type(types) == URIRef and types == choice) or \\\n",
+    "                (type(types) in {list, tuple} and choice in types)\n",
+    "            ):\n",
+    "                matches.add(node)\n",
+    "        print(matches)\n",
+    "        \n",
+    "    elif type_ == \"edge\":\n",
+    "        matches = [\n",
+    "            (source, target) \n",
+    "            for source, target, data in nx_graph.edges(data=True) \n",
+    "            if data.get(\"predicate\") == choice\n",
+    "        ]\n",
+    "        print(matches)\n",
+    "\n",
+    "def clear_select_options():\n",
+    "    select_options.options = []\n",
+    "    select_options.value = None\n",
+    "\n",
+    "\n",
+    "def update_options(change):\n",
+    "    \"\"\"Update the Select options as the user changes the search string.\"\"\"\n",
+    "    if len(change.new) < 2:\n",
+    "        clear_select_options()\n",
+    "        return\n",
+    "    \n",
+    "    if change.old != change.new:\n",
+    "        # TODO obv this needs to be linked for real\n",
+    "        select_options.options = [\n",
+    "            option for option in [*available_graph_types, *available_graph_predicates]\n",
+    "            if change.new.lower() in option[0].lower()\n",
+    "        ]\n",
+    "\n",
+    "text_widget.observe(update_options, \"value\")\n",
+    "run_button.on_click(return_matches)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "50648f07",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# TODO gridspec layout\n",
+    "# Note: combobox does not accept tuple for display option vs value\n",
+    "W.HBox([\n",
+    "    W.VBox([\n",
+    "        text_widget,\n",
+    "        select_options\n",
+    "    ]),\n",
+    "    run_button\n",
+    "])"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "e01a4d52",
+   "metadata": {},
+   "source": [
+    "#### Questions/TODO\n",
+    "* How would we support chaining multiple selections e.g. Character - homeworld - Planet?\n",
+    "* Provide recommendations before information is types into the search bar (types, and then predicates)\n",
+    "* Add to actual graph view (cytoscape)\n",
+    "    * warn user if too many nodes/edges would be added\n",
+    "    * clear view button\n",
+    "* create actual widget class\n",
+    "* widget that displays information about the types/edges present in the networkx graph"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.10"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}