class Part(object):
def __init__(self, url):
self.url = url
self.partPage = requests.get(url)
self.html = BeautifulSoup(self.partPage.text, "lxml")
self.standard_number = self.find_standard_number()
self.mainIndex = self.html.find("div", class_="main")
self.setupPart()
#print(self.standard_number)
def table_of_contents(self):
return self.table_of_contents;
def subparts(self):
return self.subpartArray;
def setupPart(self):
lineItems = self.mainIndex.find('ul').find_all('li')
for li in lineItems:
if self.find_toc(li):
print('found toc')
continue
elif self.find_subparts(li):
print('foundsubpart')
continue
else:
print('foundsubpart def')
self.find_subpartdefinition(li)
def find_standard_number(self):
return self.html.find("div", class_="blackTen").find('strong').text;
def find_toc(self, li):
if 'Table' in li.find_all('a')[1]['title']:
self.table_of_contents = ToC(li.find_all('a')[1]['href'])
return True
return False
def find_subparts(self, li):
self.subparts = []
if 'Subpart' in li.find_all('a')[0]['name']:
self.subpart = SubPart(li.find_all('a')[1]['href'])
self.subparts.append(self.subpart)
return True
return False
def find_subpartdefinition(self, li):
for li in self.mainIndex.find('ul').find_all('li'):
self.subpart.append_subpartdef(li.find_all('a')[1]['href'])
def find_subparts(self, li):
self.subparts = []
if 'Subpart' in li.find_all('a')[0]['name']:
self.subpart = SubPart(li.find_all('a')[1]['href'])
self.subparts.append(self.subpart)
return True
return False