SHOW:
|
|
- or go back to the newest paste.
| 1 | import urllib2 | |
| 2 | import re | |
| 3 | url = "http://services.runescape.com/m=news/behind-the-scenes-video-30-player-owned-ports-" | |
| 4 | print "Printing stuffs..." | |
| 5 | a = urllib2.urlopen(url).read() | |
| 6 | - | b = a[a.index('<div class=\"Content\">'):]
|
| 6 | + | a = a[a.index('<div class=\"Content\">')+21:]
|
| 7 | - | c = b[:b.index('</div>')]
|
| 7 | + | a = a[:a.index('</div>')]
|
| 8 | - | cc = c.decode("windows-1252")
|
| 8 | + | a = a.decode("windows-1252")
|
| 9 | - | html = cc.replace("<p>", "\n\n")
|
| 9 | + | a = a.replace('\n','')
|
| 10 | - | html = re.sub(r"</?b>", "\'\'\'", html, flags = "g") |
| 10 | + | html = a.replace("</p>", "\n\n").replace('<p>','')
|
| 11 | - | html = re.sub(r"</?i>", "\'\'", html, flags = "g") |
| 11 | + | html = re.sub(r"</?b>", "'''", html) |
| 12 | - | html = re.sub(r"</?li>", "\n", html, flags = "g") |
| 12 | + | html = re.sub(r"</?i>", "''", html) |
| 13 | - | html = re.sub(r"</?ul>", "\n", html, flags = "g") |
| 13 | + | html = html.replace('<li>','*').replace('</li>','\n')
|
| 14 | - | html = re.sub(r'<a.*?href="(.*?)".*?>(.*?)<\/a>', '[$1, $2]', html, flags = "g") |
| 14 | + | html = re.sub(r"</?ul>", "\n", html) |
| 15 | - | html = re.sub(r'<a.*?href="?([^ "]*)"?.*?>(.*?)<\/a>', "[$1, $2]", html, flags = "g") |
| 15 | + | html = re.sub(r'<a.*?href="?([^ "]*)"?.*?>(.*?)<\/a>', r"[\1 \2]", html) |
| 16 | - | html = re.sub(r'<hr.*?>', "\n----\n", html, flags = "g") |
| 16 | + | html = re.sub(r'<hr.*?>', "\n----\n", html) |
| 17 | - | html = re.sub(r'<iframe.*?embed\/(.*?)".*?<\/iframe>', '<youtube>$1</youtube>', html, flags = "g") |
| 17 | + | html = re.sub(r'<iframe.*?embed\/(.*?)".*?<\/iframe>', r'<youtube>\1</youtube>\n\n', html) |
| 18 | html = re.sub(r'\n{4,}','\n\n',html)
| |
| 19 | print html |
