check_redirects.py

   1 #! /usr/bin/env python3
   2 """Check and show URLs in the bookmarks database that have redirects
   3
   4 This file is a part of Bookmarks database and Internet robot.
   5 """
   6
   7 from __future__ import print_function
   8 import sys
   9 from bkmk_objects import make_linear
  10
  11
  12 __author__ = "Oleg Broytman <phd@phdru.name>"
  13 __copyright__ = "Copyright (C) 2023 PhiloSoft Design"
  14 __license__ = "GNU GPL"
  15
  16
  17 def report_redirect(href, moved):
  18     if log_file:
  19         log_file.write("%s\n%s\n\n" % (href, moved))
  20     else:
  21         print("URL: %s\nMoved to: %s\n" % (href, moved))
  22
  23
  24 def run():
  25     from getopt import getopt
  26     optlist, args = getopt(sys.argv[1:], "sl:")
  27
  28     report_stats = 1
  29     global log_file
  30     log_filename = None
  31
  32     for _opt, _arg in optlist:
  33         if _opt == '-s':
  34             report_stats = 0
  35         if _opt == '-l':
  36             log_filename = _arg
  37     try:
  38         del _opt, _arg
  39     except NameError:
  40         pass
  41
  42     if report_stats:
  43         print("Broytman check_redirects, Copyright (C) 2023 PhiloSoft Design")
  44
  45     if args:
  46         sys.stderr.write("check_redirects: too many arguments\n")
  47         sys.stderr.write("Usage: check_redirects [-s] [-l logfile]\n")
  48         sys.exit(1)
  49
  50     if log_filename:
  51         log_file = open(log_filename, 'wt', encoding='utf-8')
  52
  53     from storage import storage
  54     storage = storage()
  55
  56     if report_stats:
  57         sys.stdout.write("Loading %s: " % storage.filename)
  58         sys.stdout.flush()
  59
  60     root_folder = storage.load()
  61     make_linear(root_folder)
  62     objects = len(root_folder.linear)
  63
  64     if report_stats:
  65         print("Ok")
  66
  67     redirects = 0
  68     for object_no in range(objects):
  69         object = root_folder.linear[object_no]
  70
  71         if object.isBookmark:
  72             if hasattr(object, "error") or \
  73                     object.href.startswith('place:'):  # Firefox SmartBookmarks
  74                 continue
  75
  76             if hasattr(object, "moved"):
  77                 report_redirect(object.href, object.moved)
  78                 del object.moved
  79                 redirects += 1
  80
  81     if report_stats:
  82         print(objects, "objects passed")
  83         print(redirects, "redirects")
  84
  85
  86 if __name__ == '__main__':
  87     run()