checkbookmarks

check_bookmarks.py

# -*- coding: utf8 -*-
import sys
from bs4 import BeautifulSoup

reload(sys)
sys.setdefaultencoding('utf8')


f = open("/home/cain/bookmarks_3_21_14.html")
content = f.read()
f.close()

soup = BeautifulSoup(content)
i = 0
for link in soup.find_all('a'):
    i += 1
    print i
    print link.get_text() + link.get('href')

check_bookmarks.rb

require 'markio'
require 'nokogiri'
require 'open-uri'
require 'ostruct'

filename = ARGV[0] || '/home/cain/bookmarks_3_21_14.html'
bookmarks = Markio::parse(File.open(filename))
bookmarks.each do |b|
    begin
        doc = Nokogiri::HTML(open(b.href))
        if doc.css('title')[0].content != b.title
            puts doc.css('title')[0].content + " !== " + b.title
        else
            puts b.title + " found!"
        end
    rescue Exception => e
        puts b.href + " is not accessible"
    end
    sleep(1)
    # b.title           # String
    # b.href            # String with bookmark URL
    # b.folders         # Array of strings - folders (tags)
    # b.add_date        # DateTime
    # b.last_visit      # DateTime
    # b.last_modified   # DateTime
end