Update app.py

app.py

@@ -7,6 +7,8 @@ import pytesseract
 from pydub import AudioSegment
 import yt_dlp
 from bs4 import BeautifulSoup
+import whisper
+import yt_dlp
 
 from openai import OpenAI
 
@@ -64,26 +66,75 @@ def execute_tool(self, tool, input_data, file_url):
     if tool == "calculator":
 
         return str(eval(input_data, {"__builtins__": {}}))
-
+    if tool == "youtube_captions":
+        return self.youtube_captions(input_data)
     if tool == "wiki_search":
 
         return self.wiki_search(input_data)
 
     return "Unknown tool"
 
-def read_youtube(url):
+def youtube_captions(self, url):
+
     try:
-        ydl_opts = {"quiet": True}
+        ydl_opts = {
+            "skip_download": True,
+            "writesubtitles": True,
+            "writeautomaticsub": True,
+        }
+
         with yt_dlp.YoutubeDL(ydl_opts) as ydl:
             info = ydl.extract_info(url, download=False)
-            return info["title"] + " " + info.get("description", "")
-    except:
-        return "YouTube access error"
+
+        captions = info.get("subtitles") or info.get("automatic_captions")
+
+        return str(captions)
+
+    except Exception as e:
+        print("YouTube error:", e)
+        return ""
 # --- Basic Agent Definition ---
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
 from openai import OpenAI
 
 class BasicAgent:
+    
+
+    def read_audio(self, file):
+    
+        try:
+            model = whisper.load_model("base")
+    
+            result = model.transcribe(file)
+    
+            return result["text"]
+    
+        except:
+            return ""
+    def execute_tool(self, tool, input_data, file_url):
+
+    if tool == "wiki_search":
+        return self.wiki_search(input_data)
+
+    if tool == "scrape_page":
+        return self.scrape_page(input_data)
+
+    if tool == "read_excel":
+        file = self.download_file(file_url)
+        return self.read_excel(file)
+
+    if tool == "read_image":
+        file = self.download_file(file_url)
+        return self.read_image(file)
+
+    if tool == "calculator":
+        try:
+            return str(eval(input_data))
+        except:
+            return "error"
+
+    return "unknown tool"
+    
     def agent_loop(self, question, file_url):
     
         memory = ""
@@ -95,11 +146,12 @@ class BasicAgent:
     
     Available tools:
     
-    1. read_excel
-    2. read_image
-    3. scrape_page
-    4. calculator
-    5. wiki_search
+    1. read_excel(file_url)
+    2. read_image(file_url)
+    3. scrape_page(url)
+    4. youtube_captions(url)
+    5. calculator(expression)
+    6. wiki_search(query)
     
     Question:
     {question}
@@ -301,7 +353,8 @@ class BasicAgent:
         if file.endswith(".png") or file.endswith(".jpg"):
 
             return self.read_image(file)
-
+        if file.endswith(".mp3"):
+            return self.read_audio(file)
         if file.endswith(".py"):
 
             with open(file) as f:
@@ -409,7 +462,11 @@ Return only the answer.
     # ------------------------------------------------
 
     def __call__(self, question, file_url=None):
-    
+        if "youtube.com" in question or "youtu.be" in question:
+            captions = self.youtube_captions(question)
+            file_content = captions
+        else:
+            file_content = self.load_file(file_url)
         print("Question:", question)
     
         answer = self.agent_loop(question, file_url)