check_urls.py

   1 #! /usr/bin/env python3
   2 """Robot interface - check URLs from bookmarks database
   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
  10
  11 __author__ = "Oleg Broytman <phd@phdru.name>"
  12 __copyright__ = "Copyright (C) 2000-2023 PhiloSoft Design"
  13 __license__ = "GNU GPL"
  14
  15
  16 def run():
  17     from getopt import getopt
  18     optlist, args = getopt(sys.argv[1:], "ise")
  19
  20     show_pbar = 1
  21     report_stats = 1
  22     only_errors = 0
  23
  24     for _opt, _arg in optlist:
  25         if _opt == '-i':
  26             show_pbar = 0
  27         if _opt == '-s':
  28             report_stats = 0
  29         if _opt == '-e':
  30             only_errors = 1
  31     try:
  32         del _opt, _arg
  33     except NameError:
  34         pass
  35
  36     if report_stats:
  37         print("Broytman check_urls, Copyright (C) 1997-2023 PhiloSoft Design")
  38
  39     if args:
  40         sys.stderr.write("check_urls: too many arguments\n")
  41         sys.stderr.write("Usage: check_urls [-ise]\n")
  42         sys.exit(1)
  43
  44     if show_pbar:
  45         show_pbar = sys.stderr.isatty()
  46
  47     if show_pbar:
  48         try:
  49             from m_lib.pbar.tty_pbar import ttyProgressBar
  50         except ImportError:
  51             show_pbar = 0
  52
  53     from m_lib.flog import makelog, openlog
  54     if only_errors:
  55         log = openlog("check.log")
  56         log("chk_urls restarted for errors")
  57         if report_stats:
  58             print("chk_urls restarted for errors")
  59     else:
  60         log = makelog("check.log")
  61         log("check_urls started")
  62         if report_stats:
  63             print("   check_urls: normal start")
  64     log.outfile.reconfigure(encoding='utf-8')
  65
  66     from storage import storage
  67     storage = storage()
  68
  69     from robots import robot
  70     robot = robot(log)
  71
  72     if report_stats:
  73         sys.stdout.write("Loading %s: " % storage.filename)
  74         sys.stdout.flush()
  75
  76     root_folder = storage.load()
  77     from bkmk_objects import make_linear, break_tree
  78     make_linear(root_folder)
  79     objects = len(root_folder.linear)
  80
  81     if report_stats:
  82         print("Ok")
  83
  84     if report_stats:
  85         if only_errors:
  86             s = "Rechecking errors: "
  87         else:
  88             s = "Checking: "
  89         sys.stdout.write(s)
  90         sys.stdout.flush()
  91
  92     if show_pbar:
  93         pbar = ttyProgressBar(0, objects)
  94
  95     urls_no = 0
  96     object_count = 0
  97     size = 0
  98
  99     checked = {}
 100     rcode = 1
 101
 102     for object_no in range(objects):
 103         if show_pbar:
 104             pbar.display(object_no+1)
 105
 106         object = root_folder.linear[object_no]
 107         object_count = object_count + 1
 108
 109         if object.isBookmark:
 110             href = object.href
 111             if (href.startswith('place:')  # Firefox SmartBookmarks
 112                     or '%s' in href):  # Bookmark with keyword
 113                 log("Skipped %s" % href)
 114                 continue
 115
 116             if only_errors:
 117                 if hasattr(object, "error"):
 118                     delattr(object, "error")
 119                 else:
 120                     continue
 121
 122             if href in checked:
 123                 log("Already checked %s" % href)
 124                 old_object = root_folder.linear[checked[href]]
 125                 for attr_name in (
 126                     "last_visit", "last_modified",
 127                     "error", "no_error", "moved", "size", "md5", "real_title",
 128                     "last_tested", "test_time", "icon", "charset",
 129                 ):
 130                     if hasattr(old_object, attr_name):
 131                         setattr(object, attr_name,
 132                                 getattr(old_object, attr_name))
 133             else:
 134                 log("Checking %s" % href)
 135                 rcode = robot.check_url(object)
 136
 137                 if rcode:
 138                     checked[href] = object_no
 139                     urls_no = urls_no + 1
 140                     try:
 141                         size = size + int(object.size)
 142                     except (AttributeError, TypeError, ValueError):
 143                         pass  # Some object does not have a size :(
 144                 else:
 145                     log("Interrupted by user (^C)")
 146                     break
 147     robot.stop()
 148
 149     if show_pbar:
 150         del pbar
 151
 152     if report_stats:
 153         print("Ok")
 154         print(object_count, "objects passed")
 155         print(urls_no, "URLs checked")
 156         print(size, "bytes eaten")
 157
 158     break_tree(root_folder.linear)
 159     storage.store(root_folder)
 160
 161     if rcode:
 162         log("check_urls finished ok")
 163     log.close()
 164
 165
 166 if __name__ == '__main__':
 167     run()