Youtube feature

Youtubetranscription_summarizer.py

@@ -191,13 +191,16 @@ def download_youtube_audio_wav16k_api(
 
 
 def transcribe_faster_whisper(wav_path:str, model_name="base.en"):
-    model = WhisperModel(model_name)
-    segments, info = model.transcribe(wav_path, beam_size=1, vad_filter=True)
-    out = []
-    for s in segments:
-        out.append({"start": s.start, "end": s.end, "text": s.text})
-    #return {"language": info.language, "segments": out}
-    return {"segments": out}
+    try:
+        model = WhisperModel(model_name)
+        segments, info = model.transcribe(wav_path, beam_size=1, vad_filter=True)
+        out = []
+        for s in segments:
+            out.append({"start": s.start, "end": s.end, "text": s.text})
+        #return {"language": info.language, "segments": out}
+        return {"segments": out}
+    except Exception as e:
+        return f"Faster-Whisper transcription failed: {e}"
 
 def summarize_with_phi(transcript_segments, sysprompt, userprompt, phi_client):
     # map-reduce pseudo:

app.py

@@ -9,6 +9,11 @@ from openai import AzureOpenAI  # official OpenAI SDK, works with Azure endpoint
 import json
 import subprocess
 import Youtubetranscription_summarizer
+from app.app.Youtubeextraction import extract  # Youtube download helper functions 
+#from pydantic import BaseModel, AnyUrl # Pydantic models for request validation in yiutube extraction
+#from fastapi import FastAPI, HTTPException # FastAPI for building the API
+#app = FastAPI() ## Initialize FastAPI app for testing in local
+#from extractor.app.storage import upload_and_sign  # Youtube storage helper functions
 import re
 
 # --- LLM call (Azure OpenAI with API key) -----------------------------------
@@ -24,6 +29,7 @@ def summarize_input(audio_b64: str = None, text_input: str = None, sys_prompt: s
     deployment = os.getenv("AC_MODEL_DEPLOYMENT")
     api_version = os.getenv("AC_OPENAI_API_VERSION")
 
+
     if not endpoint or not api_key or not deployment:
         return "Server misconfiguration: required env vars missing."
     # Reset json_text for logging
@@ -84,7 +90,7 @@ def summarize_input(audio_b64: str = None, text_input: str = None, sys_prompt: s
         )
         Enddate = datetime.now()
         Callduration = Enddate - Starttime[0]
-        print(f"Azure API call with a duration of {Callduration}: prompt_length={len(user_prompt or '')}, "
+        print(f"AudioChatSummarizer API call with a duration of {Callduration}: prompt_length={len(user_prompt or '')}, "
               f"audio_size={len(audio_b64 or '')}, text_input_size={len(json_text or '')}")
         return response.choices[0].message.content
 
@@ -129,16 +135,90 @@ def download_to_temp_mp3(url: str) -> str:
                 tmp.write(chunk)
         return tmp.name
 
+# function to read files
+def file_read(filepath):
+    file_data = []
+            
+    try:
+        with open(filepath, "rb") as f:
+            file_data = f.read()
+            
+            print(f"Successfully validated {file_path} and read {len(file_data)} bytes.")
+    except Exception as e:
+                print(f"Could not read {file_path}: {e}")
+
+    return file_data
+
+###Download youtube video and extract audio using yt-dlp and ffmpeg
+#### Fixing code to resolve 404 error
+
+def fetch_audio_from_youtube(youtube_url: str) -> str:
+    """
+    Calls the extractor service and returns the signed audio URL.
+    - Tries POST /extract with youtube_url as a query param (your current server shape).
+    - Falls back to sending youtube_url in JSON body if needed.
+    - Accepts either JSON {"audio_url": "..."} or a plain string URL.
+    """
+    EXTRACT_API = os.getenv("AZURE_CONTAINER_APP_FQDN") ## Fast API endpoint for youtube extraction "https://<your-app-fqdn>/extract"
+    base = EXTRACT_API.rstrip("/")
+    endpoint = base if base.endswith("/extract") else f"{base}/extract"
+
+    payload = {"format": "wav", "sample_rate": 16000, "mono": True}
+    timeout = 90
+
+    try:
+        # 1) Preferred: youtube_url as QUERY PARAM (matches your current API)
+        r = requests.post(endpoint, params={"youtube_url": youtube_url},
+                          json=payload, timeout=timeout)
+        if r.status_code == 404 or r.status_code == 422:
+            # 2) Fallback: youtube_url in JSON body (if your API switches later)
+            body = {"youtube_url": youtube_url, **payload}
+            r = requests.post(endpoint, json=body, timeout=timeout)
+
+        if r.status_code >= 400:
+            # log details instead of raising blindly
+            print("STATUS:", r.status_code)
+            print("HEADERS:", r.headers)
+            print("BODY:", r.text[:2000])
+            r.raise_for_status()
+
+        # Response parsing: support dict or plain string
+        ctype = r.headers.get("Content-Type", "")
+        if "application/json" in ctype:
+            data = r.json()
+            # If server validates response_model to dict
+            if isinstance(data, dict) and "audio_url" in data:
+                return data["audio_url"]
+            # If server returns plain string in JSON (rare)
+            if isinstance(data, str):
+                return data
+            raise ValueError(f"Unexpected JSON shape: {data}")
+        else:
+            # Plain text URL response_model=str
+            text = r.text.strip()
+            if text.startswith("http"):
+                return text
+            raise ValueError(f"Unexpected text response: {text[:200]}")
+
+    except Exception as e:
+        msg = (f"{datetime.now()}: Error retrieving youtube wave file from Azure instance. "
+               f"url={youtube_url} endpoint={endpoint} err={e}")
+        print(msg)
+        return msg
 
+        
 def process_audio(upload_path, record_path, url, sys_prompt, user_prompt):
     tmp_to_cleanup = []
     audio_b64 = None
     text_input = None
     domaincheck = None
+    extract_input = None
+    audio_wav = None
+
     try:
         # Capture start time for logging
         Starttime = datetime.now(),
-        print(f"Azure API call starts at {datetime.now()}"),
+        print(f"AudioChatSummarizer API call starts at {datetime.now()}"),
         audio_path = None
         if upload_path:
             audio_path = upload_path
@@ -158,7 +238,15 @@ def process_audio(upload_path, record_path, url, sys_prompt, user_prompt):
                 
                 if CheckURL:
                     # Get the transcription from youtube
-                    text_input = Youtubetranscription_summarizer.main(url.strip()) # Youtube files are transcribed and summarized
+                    # text_input = Youtubetranscription_summarizer.main(url.strip()) # Youtube files are transcribed and summarized
+                    #extract_input = extract(url.strip()) # Call for local testing
+                    # Test wav file transcription using faster-whisper # Call for local testing
+                    #audio_wav = fetch_audio_from_youtube(extract_input) # Call for local testing
+                    audio_wav = fetch_audio_from_youtube(url.strip()) # Server API call
+                    #file_path = "/Users/sayedarizvi/AudioSummarizer/Data/test.wav" # Call for local testing
+                    #audio_wav = file_path # Call for local testing
+                    #text_input = Youtubetranscription_summarizer.transcribe_faster_whisper(extract_input, model_name="base.en")# Call for local testing
+                    text_input = Youtubetranscription_summarizer.transcribe_faster_whisper(audio_wav, model_name="base.en") #Call for server testing
                     tmp_to_cleanup.append(text_input)
                 else:   
                     audio_path = download_to_temp_mp3(url.strip())

requirements.txt

@@ -5,9 +5,8 @@ azure-identity==1.25.0
 azure-ai-projects==1.0.0
 numpy==1.26.4
 openai==1.107.3
-yt_dlp==2025.9.5
+yt_dlp==2025.9.23
 faster_whisper==1.2.0
 fastapi
 uvicorn[standard]==0.30.6
-azure-storage-blob==12.20.0
-pydantic==2.8.2 ###
+azure-storage-blob==12.20.0