Skip to content
This repository was archived by the owner on Apr 8, 2026. It is now read-only.

ADVANCED_USAGE.md

Latest commit

 

History

History
161 lines (114 loc) · 5.19 KB

ADVANCED_USAGE.md

File metadata and controls

161 lines (114 loc) · 5.19 KB
Raw

Advanced Usage Guide

This document provides advanced usage examples and customization options for yt-digest.

Customizing the Script

You can modify the behavior of yt-digest by editing app.py directly or by using the provided functions in your own Python scripts.

Customization Options

# Adjust number of videos to process per search query (default: 2)
data = get_recent_transcripts(search_url, limit=5)

# Customize OpenAI model (default: "gpt-5-mini-2025-08-07")
newsletter = generate_newsletter_digest(data, model="gpt-4-turbo-preview")

# Customize RSS feed output file name (default: "feed.xml")
generate_rss_feed(all_summaries, output_file="my_custom_feed.xml")

Advanced Workflows

Workflow 1: Generate summaries and RSS feed

This workflow demonstrates how to search for videos, generate summaries, and create an RSS feed programmatically:

import logging
from app import get_recent_transcripts, generate_newsletter_digest, generate_rss_feed
from datetime import datetime

logging.basicConfig(level=logging.INFO)

# Search and extract transcripts using a full YouTube URL
url = "https://www.youtube.com/results?search_query=Python+tutorials&sp=EgIIAw%253D%253D"
data = get_recent_transcripts(url, limit=3)

# Generate summaries for each video
all_summaries = []
for video in data:
    summary = generate_newsletter_digest([video])
    all_summaries.append({
        "title": video["title"],
        "video_id": video["video_id"],
        "summary": summary,
        "timestamp": datetime.utcnow().strftime("%a, %d %b %Y %H:%M:%S GMT")
    })

# Generate RSS feed
generate_rss_feed(all_summaries, output_file="feed.xml")

Workflow 2: Generate digest without RSS feed

This workflow shows how to extract transcripts and generate summaries without creating an RSS feed:

from app import get_recent_transcripts, save_results_to_json, generate_newsletter_digest

# Search and extract transcripts
url = "https://www.youtube.com/results?search_query=Python+tutorials&sp=EgIIAw%253D%253D"
data = get_recent_transcripts(url, limit=3)

# Save raw data
save_results_to_json(data, "python_transcripts.json")

# Generate digest
newsletter = generate_newsletter_digest(data)

# Save to file
with open("python_digest.md", "w") as f:
    f.write(newsletter)

Function Documentation

Getting Help

For more details on individual functions and their parameters, refer to their docstrings:

python -c "from app import get_recent_transcripts; help(get_recent_transcripts)"

Core Functions

The main functions available for programmatic use include:

  • get_transcript_api(): Initialize the YouTube Transcript API with proxy configuration
  • load_queries_config(config_path): Load and validate queries from a JSON file
  • get_recent_transcripts(search_url, limit): Search and extract video transcripts
  • generate_newsletter_digest(data, model): Generate AI-powered summaries
  • generate_rss_feed(summaries, output_file): Create an RSS feed from summaries
  • save_results_to_json(data, filename): Save transcript data to a JSON file

Custom Integration Examples

Integration with Other Tools

You can integrate yt-digest with other tools and services:

Example: Email notifications

import smtplib
from email.mime.text import MIMEText
from app import get_recent_transcripts, generate_newsletter_digest

# Generate digest
url = "https://www.youtube.com/results?search_query=tech+news&sp=EgIIAw%253D%253D"
data = get_recent_transcripts(url, limit=3)
digest = generate_newsletter_digest(data)

# Send via email
msg = MIMEText(digest)
msg['Subject'] = 'Daily YouTube Digest'
msg['From'] = 'sender@example.com'
msg['To'] = 'recipient@example.com'

# Configure your SMTP server
s = smtplib.SMTP('localhost')
s.send_message(msg)
s.quit()

Example: Slack notifications

import requests
from app import get_recent_transcripts, generate_newsletter_digest

# Generate digest
url = "https://www.youtube.com/results?search_query=tech+news&sp=EgIIAw%253D%253D"
data = get_recent_transcripts(url, limit=3)
digest = generate_newsletter_digest(data)

# Send to Slack
webhook_url = 'https://hooks.slack.com/services/YOUR/WEBHOOK/URL'
requests.post(webhook_url, json={'text': digest})

Performance Considerations

  • Transcript Length: Transcripts are automatically truncated to 15,000 characters to manage API costs and processing time
  • Summary Length: RSS feed summaries are truncated to 10,000 characters to prevent excessive feed sizes
  • API Rate Limits: Be mindful of OpenAI API rate limits when processing large numbers of videos
  • Video Limit: By default, only 2 videos are processed per query. Adjust the limit parameter based on your needs and API constraints

Troubleshooting

Common Issues

Issue: API rate limits

  • Solution: Reduce the number of videos processed per query or implement exponential backoff

Issue: Large transcript processing

  • Solution: Transcripts are automatically truncated, but you can adjust the limit in the code if needed

Issue: Memory usage

  • Solution: Process videos in smaller batches if working with many queries

For additional help, refer to the main README or the Development Guide.