gemfeed/gemfeed.py

import argparse
import datetime
import glob
import os.path

from feedgen.feed import FeedGenerator

def find_files():
    files = []
    for extension in ("gmi", "gem", "gemini"):
        files.extend(glob.glob("*.{}".format(extension)))
        index = "index.{}".format(extension)
        if index in files:
            files.remove(index)
    return files

def populate_entry_from_file(filename, base_url, entry):
    url = base_url + filename
    entry.guid(url)
    entry.link(href=url)
    updated = os.path.getctime(filename)
    updated = datetime.datetime.fromtimestamp(updated, tz=datetime.timezone.utc)
    entry.updated(updated)
    with open(filename) as fp:
        for line in fp:
            if line.startswith("#"):
                while line[0] == "#":
                    line = line[1:]
                title = line.strip()
                break
        else:
            title = filename
    entry.title(title)

def main():

    # Parse arguments
    parser = argparse.ArgumentParser(description='Generate an Atom feed for Gemini content.')
    parser.add_argument('-b', '--base', dest='base_url', type=str,
            required=True, help='base URL for feed and entries')
    parser.add_argument('-o', '--output', dest='output', type=str,
            default="atom.xml", help='output filename')
    parser.add_argument('-t', '--title', dest='title', type=str,
            default="A Gemini feed", help='feed title')
    args = parser.parse_args()

    # Normalise base URL, quick and dirty
    if not args.base_url.startswith("gemini://"):
        args.base_url = "gemini://" + args.base_url
    if not args.base_url.endswith("/"):
        args.base_url = args.base_url + "/"

    # Setup feed
    feed = FeedGenerator()
    feed.id(args.base_url)
    feed.title(args.title)
    feed.link(href=args.base_url, rel='alternate')
    feed.link(href=args.base_url + "atom.xml", rel='self')

    # Add one entry per .gmi file
    files = find_files()
    latest_update = datetime.datetime.fromtimestamp(0, tz=datetime.timezone.utc)
    for filename in files:
        entry = feed.add_entry()
        populate_entry_from_file(filename, args.base_url, entry)
        print("Adding {} with title '{}'...".format(filename, entry.title()))
        if entry.updated() > latest_update:
            latest_update = entry.updated()
    feed.updated(latest_update)

    # Write file
    feed.atom_file(args.output, pretty=True)
    print("Wrote Atom feed to {}.".format(args.output))

if __name__ == "__main__":
    main()