check_urls.py

   1 #! /usr/bin/env python
   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 os
   9 import sys
  10
  11
  12 __author__ = "Oleg Broytman <phd@phdru.name>"
  13 __copyright__ = "Copyright (C) 2000-2023 PhiloSoft Design"
  14 __license__ = "GNU GPL"
  15
  16
  17 def run():
  18     from getopt import getopt
  19     optlist, args = getopt(sys.argv[1:], "ise")
  20
  21     show_pbar = 1
  22     report_stats = 1
  23     only_errors = 0
  24
  25     for _opt, _arg in optlist:
  26         if _opt == '-i':
  27             show_pbar = 0
  28         if _opt == '-s':
  29             report_stats = 0
  30         if _opt == '-e':
  31             only_errors = 1
  32     try:
  33         del _opt, _arg
  34     except NameError:
  35         pass
  36
  37     if report_stats:
  38         print("Broytman check_urls, Copyright (C) 1997-2023 PhiloSoft Design")
  39
  40     if args:
  41         sys.stderr.write("check_urls: too many arguments\n")
  42         sys.stderr.write("Usage: check_urls [-ise]\n")
  43         sys.exit(1)
  44
  45     if show_pbar:
  46         show_pbar = sys.stderr.isatty()
  47
  48     if show_pbar:
  49         try:
  50             from m_lib.pbar.tty_pbar import ttyProgressBar
  51         except ImportError:
  52             show_pbar = 0
  53
  54     from m_lib.flog import makelog, openlog
  55     if only_errors:
  56         log = openlog("check.log")
  57         log("chk_urls restarted for errors")
  58         if report_stats:
  59             print("chk_urls restarted for errors")
  60     else:
  61         log = makelog("check.log")
  62         log("check_urls started")
  63         if report_stats:
  64             print("   check_urls: normal start")
  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 checked.has_key(href):
 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, getattr(old_object, attr_name))
 132             else:
 133                 log("Checking %s" % href)
 134                 rcode = robot.check_url(object)
 135
 136                 if rcode:
 137                     checked[href] = object_no
 138                     urls_no = urls_no + 1
 139                     try:
 140                         size = size + int(object.size)
 141                     except (AttributeError, TypeError, ValueError):
 142                         pass  # Some object does not have a size :(
 143                 else:
 144                     log("Interrupted by user (^C)")
 145                     break
 146     robot.stop()
 147
 148     if show_pbar:
 149         del pbar
 150
 151     if report_stats:
 152         print("Ok")
 153         print(object_count, "objects passed")
 154         print(urls_no, "URLs checked")
 155         print(size, "bytes eaten")
 156
 157     break_tree(root_folder.linear)
 158     storage.store(root_folder)
 159
 160     if rcode:
 161         log("check_urls finished ok")
 162     log.close()
 163
 164
 165 if __name__ == '__main__':
 166     run()