import requests def fresh(w): p1=w.rfind("<") p2=w.rfind(">",p1) sp="" if p1>-1: sp=w[p1:p2+1] k=w.replace(sp,"") return k def clean_str(ww): for i in range(100): ww=fresh(ww) return ww r = requests.get('http://otokonomesuochi.blog.fc2.com/blog-entry-303.html') r.encoding = r.apparent_encoding m=r.text sp="entry-content clearfix" p1=m.index(sp) p2=m.index("</div>",p1) m1=m[p1:p2] m1=m1.replace("<br />","sigmapin") m1=clean_str(m1) print(m1)
