geminispace.info

gemini search engine
git clone https://git.clttr.info/geminispace.info.git
Log (Feed) | Files | Refs (Tags) | README | LICENSE

add_is_cross_host_like.py (1001B)

      1 from gus import constants
      2 from gus.lib.db_model import init_db, Link, Page
      3 from gus.lib.gemini import GeminiResource, GeminiRobotFileParser
      4 
      5 def main():
      6     db = init_db(f"index.new/{constants.DB_FILENAME}")
      7     PageFrom = Page.alias()
      8     PageTo = Page.alias()
      9     link_query = (Link
     10                    .select(Link, PageFrom, PageTo)
     11                    .join(PageFrom, on=(Link.from_page_id == PageFrom.id))
     12                    .join(PageTo, on=(Link.to_page_id == PageTo.id)))
     13     for link in link_query.iterator():
     14         from_resource = GeminiResource(link.from_page.fetchable_url)
     15         to_resource = GeminiResource(link.to_page.fetchable_url)
     16         is_cross_host_like = Link.get_is_cross_host_like(from_resource, to_resource)
     17         link.is_cross_host_like = is_cross_host_like
     18         link.save()
     19         print("[{}] {} -> {}".format("T" if is_cross_host_like else "F", from_resource.fetchable_url, to_resource.fetchable_url))
     20     print("\nDone!")
     21 
     22 
     23 if __name__ == "__main__":
     24     main()