import glob
from bs4 import BeautifulSoup
from janome.tokenizer import Tokenizer
t = Tokenizer()

result_list = []
for file_path in glob.glob('panpiano/panpiano*.htm'):
    with open(file_path) as _f:
        html = _f.read()
        soup = BeautifulSoup(html ,'lxml')
        contents = soup.find_all('div',id='contents')
        for content in contents:
                ytds = content.find_all('ytd-comment-thread-renderer')
                for ytd in ytds:
                    span1 = ytd.find('span',class_="style-scope ytd-comment-renderer")
                    name = span1.text.replace('\n','').replace(' ','') if span1 is not None else ''
                    span2 = ytd.find('span',id="vote-count-middle")
                    high = span2.text.replace('\n','').replace(' ','') if span2 is not None else '0'
                    high = int(high)
                    come1 = ytd.find('yt-formatted-string', id="text")
                    rep = come1.text.replace(' 件の返信を表示','').replace(' ','') if come1 is not None else ''
                    come2 = ytd.find('div',id="content", class_="style-scope ytd-expander")
                    com = come2.text.replace('\n','').replace(' ','') if come2 is not None else ''
                    tmp = [name,high,rep,com]
                    result_list.append(tmp)

calc_dict = {}
for i,row in enumerate(result_list):
    for token in t.tokenize(row[3]):
        p = token.part_of_speech
        if not p.startswith('記号') and not 'サ変接続' in p:
            if any([p.startswith('名詞'),p.startswith('動詞'),'形容詞' in p]):
                #print(token)
                key = token.surface
                if key in calc_dict:
                    calc_dict[key] += 1
                else:
                    calc_dict[key] = 1

for i,(k,v) in enumerate(sorted(calc_dict.items(),reverse=True,key=lambda x:x[1])):
    print(v,k)
    if i>30:
        break

122 し
115 ピアノ
97 胸
69 見
68 の
67 ん
67 乳
63 おっぱい
63 てる
51 いい
51 人
48 れ
47 動画
44 欄
42 曲
39 て
38 き
38 さ
37 目
36 い
35 コメ
33 ある
33 ラピュタ
32 好き
32 私
27 する
27 さん
26 いる
26 素晴らしい
26 空
25 これ
24 音色

for row in sorted(result_list,key=lambda x: x[1],reverse=True):
    print(row[1],'\t',row[3])
    if row[1]<100: 
        break

タグ ：

#Python

#プログラミング

#PanPiano

#おっぱい

#オッパイ