main.py

import asyncio
import json
import os
from datetime import UTC, datetime

import dotenv
from content_fetcher import ContentFetcher
from crawlers.hn import HackerNewsCrawler
from exporter import (
    FeedIdentity,
    items_to_json_feed,
    items_to_markdown,
    items_to_raw_json,
)
from item_store import ItemStore
from loguru import logger
from models import Item
from perspective_generator import SmolLLMPerspectiveGenerator
from transformer import Transformer

HACKER_NEWS_FEED = FeedIdentity(
    source_name="Hacker News",
    feed_title="Social Trending - Hacker News",
    home_page_url="https://news.ycombinator.com/",
    feed_url="https://github.com/RoCry/social-trending/releases/download/latest/hackernews.rss.json",
    tags=("hackernews",),
)


def llm_enabled() -> bool:
    value = os.getenv("ENABLE_LLM", "false").strip().lower()
    if value in {"", "false"}:
        return False
    if value == "true":
        return True
    raise ValueError("ENABLE_LLM must be 'true' or 'false'")


async def apply_perspectives(*, items: list[Item], enabled: bool) -> list[Item]:
    if not enabled:
        logger.info("LLM disabled; preserving cached Perspectives")
        return items

    logger.info("Generating or refreshing Perspectives")
    perspective_generator = SmolLLMPerspectiveGenerator.from_env()
    return await Transformer(perspective_generator=perspective_generator).transform(items=items)


async def main():
    _ = dotenv.load_dotenv()
    enable_llm = llm_enabled()

    # Initialize ItemStore
    db_path = "cache/social.sqlite"
    store = ItemStore(path=db_path)
    await store.init()

    # Clean up old items
    await store.cleanup()

    now = datetime.now(tz=UTC)

    # Fetch stories
    logger.info("Fetching stories from Hacker News...")
    crawler = HackerNewsCrawler(content_fetcher=ContentFetcher())
    top_n = os.getenv("HN_COUNT", 30)
    fetched = await crawler.fetch_top_stories(cache_db_path=db_path, count=int(top_n))

    # Reconcile with cached Items
    logger.info("Reconciling with cached Items...")
    items = await store.reconcile(now=now, fetched=fetched)

    # Apply Perspectives only when LLM generation is enabled
    items = await apply_perspectives(items=items, enabled=enable_llm)
    for item in items:
        await store.save(item=item)
    logger.info("Prepared {} Items", len(items))

    # Generate output files
    logger.info("Generating output files...")
    os.makedirs("cache", exist_ok=True)

    # Generate JSON Feed file
    json_feed = items_to_json_feed(items, identity=HACKER_NEWS_FEED, skip_none_perspective=True)
    with open("cache/hackernews.rss.json", "w", encoding="utf-8") as f:
        json.dump(json_feed, f, indent=2, ensure_ascii=False)
    logger.info("Generated JSON Feed file at cache/hackernews.rss.json")

    # Generate markdown file
    md_content = items_to_markdown(items)
    with open("cache/hackernews.md", "w", encoding="utf-8") as f:
        _ = f.write(md_content)
    logger.info("Generated markdown file at cache/hackernews.md")

    # Generate JSON file
    json_content = items_to_raw_json(items)
    with open("cache/hackernews.json", "w", encoding="utf-8") as f:
        json.dump(json_content, f, indent=2, ensure_ascii=False)
    logger.info("Generated JSON file at cache/hackernews.json")


if __name__ == "__main__":
    asyncio.run(main())