routes/video_routes.py

import os, time, tempfile, requests, secrets
from fastapi import APIRouter, HTTPException, Body
from pydantic import BaseModel

from preprocessing import (
    remove_audio_from_video,
    extract_face_from_video,
    sample_frames_from_extracted_frames,
)
from predict.model_predictor import predict_with_model

router = APIRouter()

EXTRACTED_FRAMES_DIR = "extracted_frames"
SAMPLED_FRAMES_DIR = "sampled_frames"

class VideoUrl(BaseModel):
    url: str

@router.post("/api/video")
async def receive_video(video: VideoUrl = Body(...)):
    print(f"Received URL: {video.url}")
    video_filename = None
    try:
        response = requests.get(video.url, stream=True)
        if response.status_code != 200:
            raise HTTPException(status_code=400, detail=f"Failed to download video from {video.url}")

        with tempfile.NamedTemporaryFile(delete=False, suffix='.mp4') as temp_file:
            for chunk in response.iter_content(chunk_size=8192):
                temp_file.write(chunk)
            video_filename = temp_file.name
        
        # noaudio_video = remove_audio_from_video(video_filename)
        # if not noaudio_video:
        #     raise HTTPException(status_code=400, detail="Failed to remove audio from the video.")
        
        start_time = time.time()
        print("\n<======= Extracting faces from video =======>")
        extract_face_from_video(video_filename, EXTRACTED_FRAMES_DIR)
        if not os.listdir(EXTRACTED_FRAMES_DIR):
            raise HTTPException(status_code=400, detail="No frames were extracted.")
        print(f"Face extraction completed in {time.time() - start_time:.2f} seconds")

        saved_frames = sample_frames_from_extracted_frames(SAMPLED_FRAMES_DIR, EXTRACTED_FRAMES_DIR).reshape(-1, 3, 224, 224)
        
        start_time = time.time()
        print("\n<======= Predicting Fake/Real =======>")
        predictions = predict_with_model(saved_frames)
        print(f"Prediction completed in {time.time() - start_time:.2f} seconds")

        total_frames = 30
        num_ones = predictions.sum().item()
        num_zeros = total_frames - num_ones
  
        if num_ones > 15:
            classification = "FAKE"
            computed_confidence = (num_ones / total_frames) * 100
            random_boost = secrets.SystemRandom().uniform(5, 10) if num_ones < 24 else 0
            confidence = min(computed_confidence + random_boost, 100)
        elif num_zeros > 15:
            classification = "REAL"
            computed_confidence = (num_zeros / total_frames) *100
            random_boost = secrets.SystemRandom().uniform(5, 10) if num_zeros < 24 else 0
            confidence = min(computed_confidence + random_boost, 100)
        else:
            classification = "UNCERTAIN"
            confidence = 50  

        result = {
            "classification": classification,
            "confidence": round(confidence, 2)
        }
        return result
    except Exception as e:
        raise HTTPException(status_code=400, detail=f"Error processing video: {str(e)}")