Python selenium

09 Feb 2018 on Computer Science

from selenium import webdriver

driver =  webdriver.PhantomJS()    
driver.get("https://www.weather.gov/okx/stormevents")

date = driver.find_elements_by_css_selector('#pagebody > div:nth-child(3) > div > table > tbody > tr > td ul:nth-child(6)')
lis = date[0].find_elements_by_css_selector('li')

li_wea = []
count = 0

rows=[]
for i in lis:
    li_wea.append(i.text)
    li_wea[count] = '2016 ' + li_wea[count]
    li_wea[count] = li_wea[count].split(" ")

    li_wea[count] = li_wea[count][0] + "-" + li_wea[count][1] + "-" + li_wea[count][2] + "-" + li_wea[count][3]
#     rows.append([li_wea[count][0], li_wea[count][1], li_wea[count][2] ,li_wea[count][3]])
    count += 1


#rows

new1 = pd.DataFrame(li_wea, columns=['old'])
new1['date'] = new1['old'].str.extract('(\d\d\d\d-...-\d\d)', expand=True)

new1['date'][4] = '2016-Feb-05'
new1['date'][5] = '2016-Feb-08'
new1['date'][11] = '2016-Apr-03'
new1['date'][12] = '2016-Apr-04'
new1['date'][14] = '2016-June-28'
new1['date'][15] = '2016-July-18'
new1['date'][16] = '2016-July-29'
new1['date'][17] = '2016-July-31'
new1['date'][25] = '2016-Oct-08'
new1['date'][35] = '2016-Dec-05'
new1 = new1.drop('old', axis=1)

new2 = pd.DataFrame(['2016-August-01', '2016-Dec-01'], columns=['date'])
new1 = new1.append(new2, ignore_index=True).dropna()

new1['date'] = pd.to_datetime(new1['date'])
new1

	date
0	2016-01-10
1	2016-01-13
2	2016-01-17
3	2016-01-23
4	2016-02-05
5	2016-02-08
6	2016-02-15
7	2016-02-24
8	2016-03-14
9	2016-03-21
10	2016-03-28
11	2016-04-03
12	2016-04-04
13	2016-05-30
14	2016-06-28
15	2016-07-18
16	2016-07-29
17	2016-07-31
18	2016-08-10
19	2016-08-11
20	2016-08-12
21	2016-08-13
22	2016-08-20
24	2016-09-19
25	2016-10-08
26	2016-10-22
27	2016-10-22
28	2016-10-27
29	2016-10-30
30	2016-11-11
31	2016-11-14
32	2016-11-20
33	2016-11-29
34	2016-11-30
35	2016-12-05
36	2016-12-15
37	2016-12-17
38	2016-12-18
39	2016-08-01
40	2016-12-01

---

Reference