BBF download checker

bbf_download_checker.py

import bs4
import re
import urllib2


class Project:
    project_page_url = None
    download_page_url = None
    title = None
    is_download_page_working = False

    def __init__(self, project_page_url):
        self.project_page_url = project_page_url
        self._get_project_title()
        self.__get_download_page_url()
        if self.download_page_url is not None:
            self.__check_download_page()

    def _get_project_title(self):
        html = urllib2.urlopen(self.project_page_url)
        soup = bs4.BeautifulSoup(html)
        self.title = soup.find("h1").string

    def __get_download_page_url(self):
        html = urllib2.urlopen(self.project_page_url)
        soup = bs4.BeautifulSoup(html)
        download_image = soup.find(title="DOWNLOAD")
        if download_image is not None:
            download_link = download_image.parent
            if download_link is not None:
                self.download_page_url = download_link["href"]

    def __check_download_page(self):
        html = urllib2.urlopen(self.download_page_url)
        soup = bs4.BeautifulSoup(html)
        accept_button = soup.find("img", src=re.compile("accetto"))
        self.is_download_page_working = accept_button is not None


class Checker:
    projects_page_url = None

    def __init__(self, projects_page_url):
        self.projects_page_url = projects_page_url
        self._check_projects()

    def _check_projects(self):
        html = urllib2.urlopen(self.projects_page_url)
        soup = bs4.BeautifulSoup(html)
        download_snippets = soup.find_all(id=re.compile("^attachment"), class_="wp-caption aligncenter")
        if download_snippets is not None:
            for download_snippet in download_snippets:
                download_link = download_snippet.find("a")
                if download_link is not None:
                    project = Project(download_link["href"])
                    print("%s: %s" % (project.title, project.is_download_page_working))


if __name__ == "__main__":
    projects_page_url = "http://www.vazcomics.org/bbf/le-nostre-serie/"
    checker = Checker(projects_page_url)