From b7e43dbac0bde4aa80e0bc81a42c630b5fc152fb Mon Sep 17 00:00:00 2001
From: Raj Sharma <raj.fps2000@gmail.com>
Date: Sat, 2 Nov 2024 12:55:18 +0530
Subject: [PATCH] feat: added tool to get clustered list of notes

---
 tools/notes-vectors.ts | 29 +++++++++++++++++++++++++++++
 tools/notes.ts         | 29 ++++++++++++++++++++++++++++-
 2 files changed, 57 insertions(+), 1 deletion(-)

diff --git a/tools/notes-vectors.ts b/tools/notes-vectors.ts
index ff386eb..8877383 100644
--- a/tools/notes-vectors.ts
+++ b/tools/notes-vectors.ts
@@ -271,3 +271,32 @@ export async function initVectorStoreSync() {
 export function semantic_search_notes(query: string, limit: number) {
   return vectorStore.similaritySearch(query, limit);
 }
+
+export async function getClusteredFiles(): Promise<Record<string, string[]>> {
+  const result: Record<string, string[]> = {};
+
+  // Query to get filenames and their respective cluster assignments
+  const queryResult = await vectorStore.client?.query(
+    `SELECT ${config.columns.metadataColumnName}->>'filename' AS filename, ${config.columns.clusterColumnName} AS cluster
+     FROM ${config.tableName}`
+  );
+
+  if (!queryResult) {
+    console.log("No clustered files found in the vector store.");
+    return result;
+  }
+
+  // Group filenames by cluster
+  queryResult.rows.forEach((row) => {
+    const clusterName = `Cluster ${row.cluster}`; // Format the cluster name
+    const filename = row.filename;
+
+    if (!result[clusterName]) {
+      result[clusterName] = [];
+    }
+    result[clusterName].push(filename);
+  });
+
+  console.log("Clustered files:", result);
+  return result;
+}
diff --git a/tools/notes.ts b/tools/notes.ts
index 6be7798..86512a6 100644
--- a/tools/notes.ts
+++ b/tools/notes.ts
@@ -6,7 +6,11 @@ import Fuse from "fuse.js";
 import { ask, get_transcription } from "./ask";
 import { Message } from "../interfaces/message";
 import { memory_manager_guide, memory_manager_init } from "./memory-manager";
-import { semantic_search_notes, syncVectorStore } from "./notes-vectors";
+import {
+  getClusteredFiles,
+  semantic_search_notes,
+  syncVectorStore,
+} from "./notes-vectors";
 import { readFileSync, writeFileSync } from "fs";
 import { join } from "path";
 import { tmpdir } from "os";
@@ -104,6 +108,21 @@ async function semanticSearchNotes({
   }
 }
 
+const GetClusteredFileListParams = z.object({});
+type GetClusteredFileListParams = z.infer<typeof GetClusteredFileListParams>;
+
+export async function getClusteredFileList({}: GetClusteredFileListParams): Promise<OperationResult> {
+  try {
+    const results = await getClusteredFiles();
+    return {
+      success: true,
+      message: JSON.stringify(results, null, 2),
+    };
+  } catch (error: any) {
+    return { success: false, message: error.message };
+  }
+}
+
 // Helper function to normalize paths
 function normalizePath(path: string): string {
   if (path.startsWith("/notes/")) return path.substring(7);
@@ -615,4 +634,12 @@ You can use this to search by:
 4. Tags
 `,
   }),
+  zodFunction({
+    function: getClusteredFileList,
+    name: "getClusteredFileList",
+    schema: GetClusteredFileListParams,
+    description: `Get the list of notes files based on 4 cluster (unsupervised) semantic clustering.
+    You can use this to see how the notes are clustered based on their content.
+    `,
+  }),
 ];