ｇｈっじゅ - 経済統計練習帳

f=open("tv.txt","r",encoding='utf-8')
x=[]
for line in f:
    x.append(line)
f.close()
 
p1=[]
for i in range(len(x)):
    if "<td" in x[i]:p1.append(i)
 
p1.append(len(x)-1)
 
q=[]
for i in range(len(p1)-1):
    strx="" 
    for j in range(p1[i],p1[i+1]):
        strx=strx+x[j]
    q.append(strx)
 
 
a="<td rowspan=\"10\" class=\"left pl5px\">"
 
sigx=[]
bookx=[]
ensx=[]
px=[]
cpx=[]
 
for strx in q:
    strx=strx.replace(a,"")
 
    z=strx.split("<br>")
    sig=""
    book=""
    ens=""
    p=""
    cp=""
    for z1 in z:
        z1=z1.replace("etc","")
        h=0
        if "：" in z1:h=100
        if h<50:sig=sig+z1
        if "脚本 ：" in z1:book=z1
        if "演出 ：" in z1:ens=z1
        if "Ｐ ：" in z1:p=z1
        if "チーフＰ ：" in z1:cp=z1
    px.append(p)
    ensx.append(ens)
    bookx.append(book)
    cpx.append(cp)
    sigx.append(sig)
 
g=open("tv2.txt","w",encoding='utf-8')
 
for i in range(len(sigx)):
    g.write("*****")
    g.write(sigx[i])
    g.write(bookx[i])
    g.write(px[i])
    g.write(cpx[i])
    g.write(ensx[i])
g.close()