Program.py

import json
import os
import random
import tempfile
import time
import requests
from Time_Stamp_Extractor import TimeStampExtractor
from Frame_Extractor import FrameExtractor
from Layout_Model import LayoutModel


script_dir = os.path.dirname(os.path.abspath(__file__))

# create directories for data
lectures_data_path = os.path.join(script_dir, "lectures.json")


with open(lectures_data_path, "r", encoding="utf-8") as f:
    lectures_data = json.load(f)

videos_with_frames = []

for index, lecture in enumerate(lectures_data):

    video = random.sample(lecture["videos"], 1)[0]

    lecture_name = lecture["lecture"]
    lecture_video_name = video["video_name"]

    headers = {
        "Cookie": "JSESSIONID=node01ddxo7huujd5sufziwqk9ck2q8109666.node0",
    }

    response = requests.get(video["path"], headers=headers, stream=True)
    if response.status_code != 200:
        raise Exception(f"Failed to download video from {video['path']}, status code: {response.status_code}")

    with tempfile.NamedTemporaryFile(suffix=".mp4", delete=True) as temp_file:
        for chunk in response.iter_content(chunk_size=8192):
            if chunk:
                temp_file.write(chunk)
        temp_file.flush()
        temp_path = temp_file.name

        print("temp file")

        # extract the time stamps (frame indices) at which the slides are changing
        timeExtractor = TimeStampExtractor(lecture_name=lecture_name, video_id=video["id"], video_path=temp_file.name, sample_rate = 0.2)
        slideChanges = timeExtractor.extract_timestamps()

        print("time extractor")

        # extract the frames at the previoiusly defined indices
        frameExtractor = FrameExtractor(lecture_name=lecture_name, video_id=video["id"], video_path=temp_file.name)
        frames = frameExtractor.get_frames(slideChanges)

        time.sleep(2)

        print("frame extractor")

        random_frames = random.sample(frames, 4)

        # Add frames into the video dict (without mutating original lecture data)
        video_with_frames = dict(video)  # make a shallow copy
        video_with_frames["frames"] = random_frames
        videos_with_frames.append(video_with_frames)


    # run layout detection model and store png + json results
    layoutDetector = LayoutModel(frames_list=random_frames, lecture_name=lecture_name, video_id = video["id"])
    layoutDetector.run_all_frames()       

# save results into a combined JSON file
output_path = os.path.join(script_dir, "selected_frames.json")
with open(output_path, "w", encoding="utf-8") as f:
    json.dump(videos_with_frames, f, indent=4, ensure_ascii=False)