吾的这篇文章，有 1022 次评论，一条条看，吾看不过来，于是想到 Python 词云，提取关键词，倒也是一桩趣事。
评论情况： {'android': 545 次, 'ios': 110 次, 'pc': 44 次, 'uniapp': 1 次}
一个小细节：给我评论的设备中，安卓苹果比是 5:1。
Building prefix dict from the default dictionary ... Loading model cost 0.361 seconds. Prefix dict has been built successfully.
def get_comments(articleId):
# 确定评论的页数
main_res = get_commentId(articleId,1)
pageCount = json.loads(main_res)['data']['pageCount']
comment_list,comment_list2 = [],[]
source_analy = {}
for p in range(1,pageCount+1):
res = get_commentId(articleId, p)
try:
commentIds = json.loads(res)['data']['list']
for i in commentIds:
commentId = i['info']['commentId']
userName = i['info']['userName']
nickName = i['info']['nickName'] ## 获取用户名
source_dvs = i['info']['commentFromTypeResult']['key'] # 操作设备
content = i['info']['content']
comment_list.append([commentId, userName, nickName, source_dvs, content])
comment_list2.append("%s 丨 %s"%(userName, nickName))
if source_dvs not in source_analy.keys():
source_analy[source_dvs] = 1
else:
source_analy[source_dvs] = source_analy[source_dvs] + 1
# print(source_analy)
except:
print('本页失败！')
print('评论数：' + str(len(comment_list)))
return source_analy, comment_list, comment_list2
西红柿采用的是 结巴 分词, 和 wordcloud。
# -*- coding:utf8 -*- import jieba import wordcloud
代码实现：
seg_list = jieba.cut(comments, cut_all=False) # 精确模式
word = ' '.join(seg_list)
背景图 西红柿采用的是 心形图片
pic = mpimg.imread('/Users/pray/Downloads/aixin.jpeg')
完整代码：：
def word_cloud(articleId):
source_analy, comment_list, comment_list2 = get_comments(articleId)
print("评论情况：", source_analy)
comments = ''
for one in comment_list:
comment = one[4]
if 'face' not in comment:
comments = comments + comment
seg_list = jieba.cut(comments, cut_all=False) # 精确模式
word = ' '.join(seg_list)
pic = mpimg.imread('/Users/pray/Downloads/aixin.jpeg')
wc = wordcloud.WordCloud(mask=pic, font_path='/Library/Fonts/Songti.ttc', width=1000, height=500,
background_color='white').generate(word)
西红柿发现文字模糊、图像曲线边缘不清晰的问题。
于是，指定分辨率，高清整起来。
# 保存
plt.savefig('xin300.png', dpi=300) #指定分辨率保存
