1 1.英文歌词 2 # -*- coding:UTF-8 -*- 3 # -*- author:肥仔丫丫 -*- 4 5 news ='''I'm a big big girl in a big big world 6 It's not a big big thing if you leave me 7 but I do do feel that I do do will miss you much 8 miss you much… 9 I can see the first leaf faalling it's all yellow and nice 10 It's so very cold outside like the way I'm feeling inside 11 12 Outside it's now raining and tears are falling from my eyes 13 why did it have to happen why did all have to end 14 15 I have your arms around me warm like fire 16 but when I open my eyes 17 you're gone… 18 I'm a big big girl in a big big world 19 It's not a big big thing if you leave me 20 but I do feel that I will miss you much 21 miss you much…''' 22 #去掉不需要的单词 23 exclude={'do','i','a','s','m','that','in','to'} 24 25 seq=''',./'?;:..."''' 26 for c in seq: 27 news =news.replace(c,' ') 28 29 wordList=news.lower().split() 30 31 '''方法1：利用get方法来遍历扫描''' 32 ''' 33 word_dict={} 34 for w in wordList: 35 word_dict[w]=word_dict.get(w, 0)+1 36 ''' 37 38 39 # 方法2，利用统计列表Count的方法 40 word_dict={} 41 #消除不要的单词 42 wordSet=set(wordList)-exclude 43 for w in wordSet: 44 word_dict[w]=wordList.count(w) 45 dictList =list(word_dict.items()) 46 dictList.sort(key=lambda x:x[1],reverse=True) 47 for i in range(20): 48 print (dictList[i]) 49 50 51 ''' 52 53 for w in word_dict: 54 print (w, word_dict[w]) 55 ''' 56 57 '''方法3：''' 58 ''' 59 word_dict={} 60 for w in wordList: 61 if w not in word_dict: 62 word_dict[w]=1 63 else: 64 word_dict[w]=word_dict[w]+1 65 print(w, word_dict[w]) 66 ''' 67 #print sorted(news.split()) 68 69 70 ''' 71 for w in wordList: 72 73 if w not in word_dict: 74 word_dict[w]=1 75 else: 76 word_dict[w]=word_dict[w]+1 77 ''' 78 79 2.中文 80 str5='''''' 81 jieba.cut(str) 82 for i in list(jieba.cut(str)): 83 ... print(i) 84 85 广州 86 商学院 87 前身 88 是 89 广东省 90 第一所 91 独立 92 学院 93 - 94 - 95 华南 96 师范大学 97 增城 98 学院 99 。 100 学院 101 由康大 102 公司 103 与 104 " 105 211 106 工程 107 " 108 重点 109 建设 110 大学 111 - 112 - 113 华南 114 师范大学 115 联合 116 创办 117 于 118 1998 119 年 120 。 121 2014 122 年经 123 教育
