def getBillText(url): bill_page=get_contents_of_url(url) clean_bill=substitute_char(bill_page,var_re,'') clean_bill=substitute_char(clean_bill,markup_id_re,'') soup=BeautifulSoup(clean_bill) for e in soup.find_all('br'): e.extract() br = soup.new_tag('br') bill_text = soup.find_all('div','xtend') # for e in bill_text[0]('a'): # bill_text[0].a.insert_before(br) # first_link = bill_text[0].a # first_link.find_next("a") return bill_text[0]
def sendGetRequest(url): url=substitute_char(url,' ','%20') response = get_contents_of_url(url) if response: data=json.loads(response) return data else: return None
def getBillText(url): bill_page = get_contents_of_url(url) clean_bill = substitute_char(bill_page, var_re, '') clean_bill = substitute_char(clean_bill, markup_id_re, '') soup = BeautifulSoup(clean_bill) for e in soup.find_all('br'): e.extract() br = soup.new_tag('br') bill_text = soup.find_all('div', 'xtend') # for e in bill_text[0]('a'): # bill_text[0].a.insert_before(br) # first_link = bill_text[0].a # first_link.find_next("a") return bill_text[0]
def sendGetRequest(url): url = substitute_char(url, ' ', '%20') response = get_contents_of_url(url) if response: data = json.loads(response) return data else: return None
def bill_text_remove_markup(text): text=substitute_char(text,var_re,'') return text
def bill_text_remove_markup(text): text = substitute_char(text, var_re, '') return text