Update app.py

app.py

@@ -1,13 +1,15 @@
 import gradio as gr
 import asyncio
 from pathlib import Path
-import anthropic
+from google import genai
+from google.genai import types
 import os
 from dataclasses import dataclass
 from typing import Dict
 from youtube_transcript_api import YouTubeTranscriptApi
 import re
 import pandas as pd
+import assemblyai as aai
 
 # Move relevant classes and functions into app.py
 @dataclass
@@ -15,41 +17,41 @@ class ContentRequest:
     prompt_key: str
 
 class ContentGenerator:
-    def __init__(self):
+    def __init__(self,api_key):
         self.current_prompts = self._load_default_prompts()
-        self.client = anthropic.Anthropic()
-        
+        self.client = genai.Client(api_key=api_key)
+
     def _load_default_prompts(self) -> Dict[str, str]:
         """Load default prompts and examples from files and CSVs."""
-        
+
         # Load CSV examples
         try:
             timestamps_df = pd.read_csv("data/Timestamps.csv")
             titles_df = pd.read_csv("data/Titles & Thumbnails.csv")
             descriptions_df = pd.read_csv("data/Viral Episode Descriptions.csv")
             clips_df = pd.read_csv("data/Viral Twitter Clips.csv")
-            
+
             # Format timestamp examples
             timestamp_examples = "\n\n".join(timestamps_df['Timestamps'].dropna().tolist())
-            
+
             # Format title examples
             title_examples = "\n".join([
                 f'Title: "{row.Titles}"\nThumbnail: "{row.Thumbnail}"'
                 for _, row in titles_df.iterrows()
             ])
-            
+
             # Format description examples
             description_examples = "\n".join([
                 f'Tweet: "{row["Tweet Text"]}"'
                 for _, row in descriptions_df.iterrows()
             ])
-            
+
             # Format clip examples
             clip_examples = "\n\n".join([
                 f'Tweet Text: "{row["Tweet Text"]}"\nClip Transcript: "{row["Clip Transcript"]}"'
                 for _, row in clips_df.iterrows() if pd.notna(row["Tweet Text"])
             ])
-            
+
         except Exception as e:
             print(f"Warning: Error loading CSV examples: {e}")
             timestamp_examples = ""
@@ -61,7 +63,7 @@ class ContentGenerator:
         prompts = {}
         for key in ["previews", "clips", "description", "timestamps", "titles_and_thumbnails"]:
             prompt = Path(f"prompts/{key}.txt").read_text()
-            
+
             # Inject relevant examples
             if key == "timestamps":
                 prompt = prompt.replace("{timestamps_examples}", timestamp_examples)
@@ -71,31 +73,30 @@ class ContentGenerator:
                 prompt = prompt.replace("{description_examples}", description_examples)
             elif key == "clips":
                 prompt = prompt.replace("{clip_examples}", clip_examples)
-            
+
             prompts[key] = prompt
 
         return prompts
 
     async def generate_content(self, request: ContentRequest, transcript: str) -> str:
-        """Generate content using Claude asynchronously."""
+        """Generate content using Gemini asynchronously."""
         try:
             print(f"\nFull prompt for {request.prompt_key}:")
             print("=== SYSTEM PROMPT ===")
             print(self.current_prompts[request.prompt_key])
             print("=== END SYSTEM PROMPT ===\n")
-            
-            response = self.client.messages.create(
-                model="claude-3-5-sonnet-20241022",
-                max_tokens=8192,
-                system=self.current_prompts[request.prompt_key],
-                messages=[{"role": "user", "content": f"Process this transcript:\n\n{transcript}"}]
+
+            response = self.client.models.generate_content(
+                model="gemini-2.5-pro-exp-03-25",
+                config=types.GenerateContentConfig(system_instruction=self.current_prompts[request.prompt_key]),
+                contents=transcript
             )
-            
-            if response and hasattr(response, 'content'):
-                return response.content[0].text
+
+            if response and hasattr(response, 'candidates'):
+                return response.text
             else:
                 return f"Error: Unexpected response structure for {request.prompt_key}"
-                
+
         except Exception as e:
             return f"Error generating content: {str(e)}"
 
@@ -117,7 +118,8 @@ def get_transcript(video_id: str) -> str:
 
 class TranscriptProcessor:
     def __init__(self):
-        self.generator = ContentGenerator()
+        self.generator = ContentGenerator(api_key=os.getenv("GOOGLE_API_KEY"))
+
 
     def _get_youtube_transcript(self, url: str) -> str:
         """Get transcript from YouTube URL."""
@@ -128,14 +130,14 @@ class TranscriptProcessor:
         except Exception as e:
             raise Exception(f"Error fetching YouTube transcript: {str(e)}")
 
-    async def process_transcript(self, input_text: str):
+    async def process_transcript(self, audio_file):
         """Process input and generate all content."""
+        audio_path = audio_file.name
         try:
-            transcript = (
-                self._get_youtube_transcript(input_text)
-                if any(x in input_text for x in ["youtube.com", "youtu.be"])
-                else input_text
-            )
+            aai.settings.api_key = os.getenv("ASSEMBLYAI_API_KEY")
+            config = aai.TranscriptionConfig(speaker_labels=True, language_code="en")
+            transcript_iter = aai.Transcriber().transcribe(str(audio_path), config=config)
+            transcript = transcript_iter.text
 
             # Process each type sequentially
             sections = {}
@@ -178,34 +180,36 @@ class TranscriptProcessor:
         ))
         return "Prompts updated for this session!"
 
+
+
 def create_interface():
     """Create the Gradio interface."""
     processor = TranscriptProcessor()
-    
-    with gr.Blocks(title="Podcast Content Generator") as app:
+
+    with gr.Blocks(title="Gemini Podcast Content Generator") as app:
         gr.Markdown(
             """
-            # Podcast Content Generator
-            Generate preview clips, timestamps, descriptions and more from podcast transcripts or YouTube videos.
-            
-            Simply paste a YouTube URL or raw transcript text to get started!
+            # Gemini Podcast Content Generator
+            Generate preview clips, timestamps, descriptions and more from podcast transcripts using Gemini.
+
+            Upload an audio file to get started!
             """
         )
-        
+
         with gr.Tab("Generate Content"):
-            input_text = gr.Textbox(
-                label="Input", 
-                placeholder="YouTube URL or transcript text...",
-                lines=10
+            input_audio = gr.File(
+                label="Upload Audio File",
+                file_count="single",
+                file_types=["audio"]
             )
-            submit_btn = gr.Button("Generate Content")
-            
+            submit_btn = gr.Button("Generate Content with Gemini")
+
             output = gr.Markdown()  # Single markdown output
 
             async def process_wrapper(text):
                 print("Process wrapper started")
                 print(f"Input text: {text[:100]}...")
-                
+
                 try:
                     result = await processor.process_transcript(text)
                     print("Process completed, got results")
@@ -216,7 +220,7 @@ def create_interface():
 
             submit_btn.click(
                 fn=process_wrapper,
-                inputs=input_text,
+                inputs=input_audio,
                 outputs=output,
                 queue=True
             )
@@ -224,10 +228,10 @@ def create_interface():
         with gr.Tab("Customize Prompts"):
             gr.Markdown(
                 """
-                ## Customize Generation Prompts
+                ## Customize Generation Prompts for Gemini
                 Here you can experiment with different prompts during your session.
                 Changes will remain active until you reload the page.
-                
+
                 Tip: Copy your preferred prompts somewhere safe if you want to reuse them later!
                 """
             )
@@ -240,7 +244,7 @@ def create_interface():
                 )
                 for key in [
                     "previews",
-                    "clips", 
+                    "clips",
                     "description",
                     "timestamps",
                     "titles_and_thumbnails"
@@ -257,7 +261,7 @@ def create_interface():
                 )
 
             # Reset button
-            reset_btn = gr.Button("Reset to Default Prompts")
+            reset_btn = gr.Button("Reset to Default Gemini Prompts")
             reset_btn.click(
                 fn=lambda: (
                     processor.update_prompts(*processor.generator.current_prompts.values()),
@@ -269,4 +273,4 @@ def create_interface():
     return app
 
 if __name__ == "__main__":
-    create_interface().launch() 
+    create_interface().launch()