rssfixer.py

#!/usr/bin/env python3
# -*- coding: utf-8; mode: python -*-

"""
Fix the rss.xml files. Replace 'testserver' with the correct server name.

"""

__version__ = "1.0"
__date__ = "20180730"
__author__ = "kthoden@mpiwg-berlin.mpg.de"

import logging
import argparse
import shutil
from lxml import etree

NS_ATOM = "http://www.w3.org/2005/Atom"
NS_MAP = {"a" : NS_ATOM}

logging.basicConfig(level=logging.INFO, format=' %(asctime)s - %(levelname)s - %(message)s')

def replace_link_text(old_url):
    """Replace the link text"""

    if old_url.find("testserver/sources") == -1:
        new_link_text = old_url.replace("testserver", "www.edition-open-access.de")
    else:
        new_link_text = old_url.replace("testserver", "www.edition-open-sources.org")

    return new_link_text
# def replace_link_text ends here

def main():
    """The main bit"""

    parser = argparse.ArgumentParser()
    parser.add_argument("rssfile", help="The RSS file that is to be fixed.")
    args = parser.parse_args()

    rss_file = args.rssfile
    rss_tree = etree.parse(rss_file)

    atom_link_element = rss_tree.xpath("/rss/channel/a:link", namespaces=NS_MAP)[0]

    atom_link_url = atom_link_element.get("href")
    new_atom_link = replace_link_text(atom_link_url)
    atom_link_element.set("href", new_atom_link)

    feed_items = rss_tree.xpath("/rss/channel/item", namespaces=NS_MAP)
    logging.info("Found %d feed items.", len(feed_items))

    for book in feed_items:
        link_url = book.find("link")
        link_text = link_url.text
        new_link_text = replace_link_text(link_text)
        link_url.text = new_link_text

        guid_url = book.find("guid")
        guid_text = guid_url.text
        new_guid_text = replace_link_text(guid_text)
        guid_url.text = new_guid_text

    shutil.move(rss_file, rss_file.replace(".xml", "-old.xml"))
    rss_tree.write(rss_file, pretty_print=True, xml_declaration=True, encoding="utf-8")
    logging.info("Wrote %s.", rss_file)
# def main ends here

if __name__ == '__main__':
    main()
# finis