diff options
author | Julian <Julianteule@gmail.com> | 2017-02-01 10:38:02 +0100 |
---|---|---|
committer | Julian <Julianteule@gmail.com> | 2017-02-01 10:38:02 +0100 |
commit | 665aaa60c60e7cfb556b629edd74075c993d5bdf (patch) | |
tree | befdcc80ae71176184e8c1add1aef4afe61691cf /Scripts/#wiki2P.py# | |
parent | 2eb28948556f464980cf4296c09d044c21bc84de (diff) | |
parent | ebed9571e58a560bca3d38c2a8ee644866e156ae (diff) |
coolsad
Diffstat (limited to 'Scripts/#wiki2P.py#')
-rwxr-xr-x | Scripts/#wiki2P.py# | 26 |
1 files changed, 26 insertions, 0 deletions
diff --git a/Scripts/#wiki2P.py# b/Scripts/#wiki2P.py# new file mode 100755 index 0000000..2fb1a10 --- /dev/null +++ b/Scripts/#wiki2P.py# @@ -0,0 +1,26 @@ +import requests, bs4, re + +# res = requests.get("https://en.wikipedia.org/wiki/Special:Random") +res = requests.get("https://en.wikipedia.org/wiki/Linux") + + +soup = bs4.BeautifulSoup(res.text, "html.parser") + + +element = soup.select("#mw-content-text a[title]") + + +pattern = re.compile("^\/.*") + + +for i in element: + if "Edit section" not in i["title"] and pattern.match(i["href"]): + if "div" not in str(i.parent) and "th" not in str(i.parent) and "td" not in str(i.parent): + try: + i["class"] + except KeyError: + print(i) + break + +print("LOL") + |