def setFiles(self, html, url):
self.setHTML(html)
internalList = []
for files in self.getHTML().findAll('a', href=True):
linkToFile = files['href']
if UrlUtils.externalLink(url, linkToFile) and UrlUtils.containsHTTP(linkToFile):
self._externals.append(linkToFile)
else:
if UrlUtils.containsHTTP(linkToFile) is False:
linkToFile = UrlUtils.assertSiteWithFile(url, linkToFile)
if ExtensionsFile.hasExtension(linkToFile) and not UrlUtils.externalLink(url, linkToFile):
internalList.append(linkToFile)
self.filterFiles(internalList)
def pageOrExternal(self, page, url):
if url in page:
return True
elif UrlUtils.externalLink(url, page):
self.setExternals(page)
return False