Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- # -*- coding: cp1252 -*-
- from bs4 import BeautifulSoup
- import html5lib
- import urllib
- from selenium import webdriver
- from selenium.webdriver.common.by import By
- import string
- import requests
- from requests.auth import HTTPBasicAuth
- danskAlf = list(string.ascii_uppercase)
- #danskAlf.append("Æ")
- #danskAlf.append("Ø")
- #danskAlf.append("Å")
- driver = webdriver.Firefox()
- driver.get("http://xnet.dkma.dk/indlaegsseddel/leaflets/leaflets.faces")
- for i in range(2):
- search = driver.find_element_by_partial_link_text(danskAlf[i])
- search.click()
- soup = BeautifulSoup(driver.page_source, 'html.parser')
- res = []
- for i in soup.findAll("tr", {"class" : "odd"}):
- respDict = {}
- respDict["name"] = i.text.splitlines()[0]
- respDict["producer"] = i.text.splitlines()[1]
- respDict["form"] = i.text.splitlines()[2]
- respDict["dose"] = i.text.splitlines()[3]
- res.append(respDict)
- for i in soup.findAll("tr", {"class" : "even"}):
- respDict = {}
- respDict["name"] = i.text.splitlines()[0]
- respDict["producer"] = i.text.splitlines()[1]
- respDict["form"] = i.text.splitlines()[2]
- respDict["dose"] = i.text.splitlines()[3]
- res.append(respDict)
- sendResp(res, danskAlf[i])
- def sendResp(resp, letter):
- url = "https://qa-engine.scaut.dk/integration/pharmaceutical"
- user = "scraper"
- password = "1234"
- payload = {}
- payload["Medicines with start letter \""+letter+"\""] = resp
- r = requests.post(url, json=payload, auth=HTTPBasicAuth(user, password))
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement