pdf read

Author: Decryptu

package-lock.json

@@ -33,7 +33,8 @@
         "discord.js": "^14.16.3",
         "dotenv": "^16.4.5",
         "form-data": "^4.0.1",
-        "openai": "^4.73.0"
+        "openai": "^4.73.0",
+        "pdf.js-extract": "^0.2.1"
     },
     "engines": {
         "node": ">=18.0.0"

package.json

@@ -5,6 +5,7 @@ import {
   AI_NAME,
 } from "../config.js";
 import transcribeVoiceMessage from "./transcribeVoiceMessage.js";
+import extractFileContent from "./extractFileContent.js";
 
 async function buildConversationLog(message, client) {
   const conversationLog = [
@@ -20,15 +21,11 @@ async function buildConversationLog(message, client) {
   const reversedMessages = Array.from(prevMessages.values()).reverse();
 
   for (const msg of reversedMessages) {
-    if (
-      msg.content.startsWith("!") ||
-      (msg.author.bot && msg.author.id !== client.user.id)
-    ) {
+    if (msg.content.startsWith("!") || (msg.author.bot && msg.author.id !== client.user.id)) {
       continue;
     }
 
     const role = msg.author.id === client.user.id ? "assistant" : "user";
-
     conversationLog.push({
       role: role,
       content: msg.content,
@@ -50,25 +47,41 @@ async function buildConversationLog(message, client) {
 
   if (message.attachments.size > 0) {
     console.log(`Message contains ${message.attachments.size} attachments.`);
-    const imageAttachments = Array.from(message.attachments.values())
-      .filter((attachment) => !attachment.name.endsWith(".ogg"))
-      .map((attachment) => {
-        console.log(`Image attachment URL: ${attachment.url}`);
-        return {
-          type: "image_url",
-          image_url: { url: attachment.url },
-        };
-      });
+    const attachmentPromises = Array.from(message.attachments.values()).map(async (attachment) => {
+      if (attachment.name.endsWith('.ogg')) return null;
+      
+      // Handle PDFs and text files
+      if (attachment.name.endsWith('.pdf') || attachment.name.endsWith('.txt')) {
+        try {
+          const extractedText = await extractFileContent(attachment);
+          return {
+            type: "text",
+            text: `Content from ${attachment.name}:\n${extractedText}`,
+          };
+        } catch (error) {
+          console.error(`Error extracting content from ${attachment.name}:`, error);
+          return null;
+        }
+      }
+      
+      // Handle images
+      return {
+        type: "image_url",
+        image_url: { url: attachment.url },
+      };
+    });
+
+    const processedAttachments = (await Promise.all(attachmentPromises)).filter(Boolean);
 
-    if (imageAttachments.length > 0) {
+    if (processedAttachments.length > 0) {
       conversationLog.push({
         role: "user",
         content: [
           {
             type: "text",
             text: message.content,
           },
-          ...imageAttachments,
+          ...processedAttachments,
         ],
       });
     }
@@ -105,4 +118,4 @@ async function handleVoiceMessage(message) {
   }
 }
 
-export default buildConversationLog;
+export default buildConversationLog;

utils/buildConversationLog.js

@@ -0,0 +1,27 @@
+import axios from 'axios';
+import { PDFExtract } from 'pdf.js-extract';
+
+async function extractFileContent(attachment) {
+  const response = await axios.get(attachment.url, { responseType: 'arraybuffer' });
+  const buffer = Buffer.from(response.data);
+
+  if (attachment.name.endsWith('.pdf')) {
+    try {
+      const pdfExtract = new PDFExtract();
+      const data = await pdfExtract.extractBuffer(buffer);
+      const text = data.pages
+        .map(page => page.content.map(item => item.str).join(' '))
+        .join('\n\n');
+      return text;
+    } catch (error) {
+      console.error('Error parsing PDF:', error);
+      throw new Error('Failed to parse PDF file');
+    }
+  } else if (attachment.name.endsWith('.txt')) {
+    return buffer.toString('utf-8');
+  }
+
+  throw new Error('Unsupported file type');
+}
+
+export default extractFileContent;