From 4817424c1aea5a41e6c8913278bdd9caf7e82659 Mon Sep 17 00:00:00 2001
From: sherifButt <90522472+sherifButt@users.noreply.github.com>
Date: Fri, 8 Mar 2024 07:03:03 +0000
Subject: [PATCH] added extractMetaData function to chat utils

---
 .../src/components/WorkspaceChat/index.jsx    |  1 +
 frontend/src/utils/chat/extractMetaData.js    | 53 +++++++++++++++++++
 2 files changed, 54 insertions(+)
 create mode 100644 frontend/src/utils/chat/extractMetaData.js

diff --git a/frontend/src/components/WorkspaceChat/index.jsx b/frontend/src/components/WorkspaceChat/index.jsx
index d2d81c72..38c4dbde 100644
--- a/frontend/src/components/WorkspaceChat/index.jsx
+++ b/frontend/src/components/WorkspaceChat/index.jsx
@@ -5,6 +5,7 @@ import ChatContainer from "./ChatContainer";
 import paths from "@/utils/paths";
 import ModalWrapper from "../ModalWrapper";
 import { useParams } from "react-router-dom";
+import { extractMetaData } from "@/utils/chat/extractMetaData";
 
 export default function WorkspaceChat({ loading, workspace }) {
   const { threadSlug = null } = useParams();
diff --git a/frontend/src/utils/chat/extractMetaData.js b/frontend/src/utils/chat/extractMetaData.js
new file mode 100644
index 00000000..2b7849f5
--- /dev/null
+++ b/frontend/src/utils/chat/extractMetaData.js
@@ -0,0 +1,53 @@
+/**
+ * Extracts JSON objects and arrays from a text response from OpenAI API call and returns the remaining text with the extracted JSON strings removed.
+ * @param {String} textResponse  The text response from OpenAI API call 
+ * @returns  {Object} { remainingText, meta: extractedObjects }  remainingText is the textResponse with the extracted JSON strings removed, and meta is an array of the extracted JSON objects/arrays
+ */
+
+export const extractMetaData = (textResponse) => {
+    let remainingText = textResponse;
+    let inString = false;
+    let char, prevChar;
+    let braceCount = 0;
+    let bracketCount = 0;
+    let startIndex = null;
+    let extractedObjects = {}; // Keep as an object as requested
+
+    for (let i = 0; i < textResponse.length; i++) {
+        char = textResponse[i];
+        if (char === '"' && prevChar !== '\\') inString = !inString;
+        if (inString) continue;
+
+        if (char === '{' || char === '[') {
+            if (braceCount === 0 && bracketCount === 0) startIndex = i;
+            if (char === '{') braceCount++;
+            if (char === '[') bracketCount++;
+        } else if (char === '}' || char === ']') {
+            if (char === '}') braceCount--;
+            if (char === ']') bracketCount--;
+
+            if (braceCount === 0 && bracketCount === 0 && startIndex !== null) {
+                let json = textResponse.substring(startIndex, i + 1);
+
+                try {
+                    let parsedJson = JSON.parse(json);
+                    for (let key in parsedJson) {
+                        if (parsedJson.hasOwnProperty(key)) {
+                            extractedObjects[key] = parsedJson[key];
+                        }
+                    }
+                } catch (error) {
+                    console.error("Error parsing JSON:", error, 'in JSON:', json);
+                }
+                startIndex = null;
+            }
+        }
+        prevChar = char;
+    }
+
+    // Remove any json objects from the text tat starts with ```json nad ends with ```
+    const jsonRegex = /```json[\s\S]*?```/g;
+    remainingText = remainingText.replace(jsonRegex, '');
+
+    return { remainingText, metaData: extractedObjects };
+}
\ No newline at end of file