Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- # -*- coding: utf-8 -*-
- """
- Created on Mon Nov 23 23:36:47 2015
- @author: User
- """
- import urllib2
- from bs4 import BeautifulSoup
- from time import sleep
- def words(needle):
- url = 'http://classic.jisho.org/words?jap=*' + needle + '*&eng=&dict=edict&common=on'
- f = urllib2.urlopen(url)
- data = f.read()
- #print data
- soup = BeautifulSoup(data,'html.parser')
- table = soup.find(id='word_result')
- words = []
- if table == None:
- print needle + " yield 0 results"
- return []
- for tr in table.find_all('tr'):
- c1 = tr.find(class_='kanji_column')
- c2 = tr.find(class_='kana_column')
- c3 = tr.find(class_='meanings_column')
- if c1 != None and c2 != None and c3 != None:
- words.append(c1.getText().strip() + "\t" + c2.getText().strip() + "\t" + c3.getText().strip())
- return words
- f = open('D:/w/jisho/sixth_grade.txt','w')
- for k in ['並','乱','乳','亡','仁','供','俳','値','傷','優','党','冊','処','刻','割','創','劇','勤','危','卵','厳','収','后','否','吸','呼','善','困','垂','城','域','奏','奮','姿','存','孝','宅','宇','宗','宙','宝','宣','密','寸','専','射','将','尊','就','尺','届','展','層','己','巻','幕','干','幼','庁','座','延','律','従','忘','忠','憲','我','批','担','拝','拡','捨','探','推','揮','操','敬','映','晩','暖','暮','朗','机','枚','染','株','棒','模','権','樹','欲','段','沿','泉','洗','派','済','源','潮','激','灰','熟','片','班','異','疑','痛','皇','盛','盟','看','砂','磁','私','秘','穀','穴','窓','筋','策','簡','糖','系','紅','納','純','絹','縦','縮','署','翌','聖','肺','背','胸','脳','腹','臓','臨','至','若','著','蒸','蔵','蚕','衆','裁','装','裏','補','視','覧','討','訪','訳','詞','誌','認','誕','誠','誤','論','諸','警','貴','賃','遺','郵','郷','針','鋼','閉','閣','降','陛','除','障','難','革','頂','骨']:
- w = words(k)
- f.write(k + "\n" + "\n".join(w).encode('utf8') + "\n\n")
- sleep(5)
- f.close()
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement