Das deutsche Python-Forum

Code: Alles auswählen

	def _build_session(self):
		"""
		typical mechanize browser settings,
		using shared proxy vom proxy module
		"""
		cj = cookielib.LWPCookieJar()
		used_proxy = self._proxy.get_shared_proxy()
		self._logger.debug("used proxy: %s" % used_proxy)
		self._browser = mechanize.Browser()
		self._browser.addheaders = [('User-agent', self._get_random_useragent())]
		self._browser.set_cookiejar(cj)
		self._browser.set_proxies({"http": used_proxy})
		self._browser.set_handle_equiv(True)
		self._browser.set_handle_redirect(True)
		self._browser.set_handle_referer(True)
		self._browser.set_handle_robots(False)
		self._browser.set_handle_refresh(mechanize._http.HTTPRefreshProcessor(), max_time=1)

Code: Alles auswählen

# python has no javascript, so we need the noscript challenge
			p = re.compile('src="(http://api.solvemedia.com/papi/challenge.noscript[^\s]+)"')
			captcha_url = p.findall(response.read())[0]
			html_content = self._browser.open_novisit(captcha_url).read()
			if 'client error 117B' in html_content:
				# solve media checks for same IP/user agent in following 30 seconds
				self._proxy.trash_shared_proxy()
				self._build_session()
				return self.register()
			# looking for captcha img src
			pattern = re.compile('<img src="(/papi/media\?c=[^\s]+)"')
			captcha_urls = pattern.findall(html_content)
			if captcha_urls:
				# open_novisit to prevent deleting/refreshing of captcha
				captcha_url = captcha_urls[0]
				captcha = self._browser.open_novisit('http://api.solvemedia.com' + captcha_url).read()
				self._logger.debug('http://api.solvemedia.com' + captcha_url)
				open('captcha.gif', 'wb').write(captcha)

				cw = CaptchaWindow()
				if self.ocr_captcha:
					cs = captchaOCR.CaptchaSolver()
					cs.load_captcha('captcha')
					cs.smoothe_captcha()
					cs.resize_captcha()
					cs.convert_captcha_BMP()
					text, confidence = cs.solve_captcha()
					cs.clean()
					cw.solved_captcha = text
				else:
					cw.display_captcha('captcha.gif')