1.目标

python爬取三国演义，生成词云、图表

2.码前须知

项目目标：三国人物名称及出现次数—–数据统计分析

提出问题：哪个人物在三国演义中出现的次数最多?，我们希望通过数据分析来获得答案。

分析工具：pandas,Matplotlib

pip install bs4

pip install lxml

pip install pandas

pip install Matplotlib

bs4数据解析必备知识点：标签定位,提取标签中的数据值

1.实例化一个BeautifulSoup对象，并将页面源码数据加载到该对象中,lxml是解析器，固定的参数,下面是举例

本地html加载到该对象:

fp = open(’./test.html’,‘r’,encoding=‘utf-8’)

soup = BeautifulSoup(fp,‘lxml’)

print(soup)

互联网上获取的源码数据(常用)

page_text = response.text

soup = BeautifulSoup(page_text,‘lxml’)

2.通过调用BeautifulSoup对象中相关的属性或者方法对标签进行定位和提取

bs4具体属性的用法

1.标签 ，如< p > < a >< div >等等

soup.tagName

例如

soup.a #返回的是html第一次出现的tagName的a标签

soup.div #返回的是html第一次出现的tagName的div标签

2.查找

soup.find(‘div’) 用法相当于soup.div

属性定位, < div class=‘song’ >

soup.find(‘div’,class_=‘song’) class_需要带下划线

class_/id/attr

3.所有符合条件的标签

soup.find_all(‘a’) #返回符合所有的a标签，也可以属性定位

4.select放置选择器 类选择器. .代表的就是tang

soup.select(’.tang’)

soup.select(‘某种选择器(id,class,标签，选择器)’)

返回的是一个列表

定位到标签下面的标签 >表示标签一个层级选择器

soup.select(’.tang > ul >li >a’[0])

空格表示多级选择器

soup.select(‘tang’ > ul a’[0]) 与上述的表达式相同

常用层级选择器

5.获取标签中间的文本数据

soup.a.text/string/get_text()

区别

text/get_text():可以获取某一个标签中所有的文本内容，即使不是直系的文本

string:只可以获取直系文本

6.获取标签中的属性值

soup.a[‘href’] 相当于列表操作

3.操作流程

1.爬取数据来源： 古诗词网《三国演义》

2.编码流程：

发起请求–requests

获取响应数据—页面信息

数据解析(通过bs4) –1进行指定标签的定位;2取得标签当中的文本内容

持久化存储–保存文件

2.文本词频统计：中文分词库–jieba库，具体的解释已在代码处声明

3.生成词云：wordcloud，具体的解释已在代码处声明

4.生成柱状分析图:matplotlib，具体的解释已在代码处声明

4.完整代码

外汇经纪商对比http://www.fx61.com/brokerlist

from bs4 import BeautifulSoup

import requests

import jieba#优秀的中文分词第三方库

import wordcloud

import pandas as pd

from matplotlib import pyplot as plt

1.对首页html进行爬取

headers={

'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_0) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/69.0.3497.100 Safari/537.36',

}

fp = open('./sanguo.txt','w',encoding='utf-8')

page_text = requests.get(url=url,headers=headers).text

2.数据解析

实例化对象

soup = BeautifulSoup(page_text,'lxml')

获得li标签

li_list = soup.select('.book-mulu > ul > li')

取得li标签里的属性

for li in li_list:

通过bs4的方法直接获取a标签直系文本

title = li.a.string

对url进行拼接得到详情页的url

detail_url = 'http://www.shicimingju.com'+li.a['href']

对详情页发起请求

detail_page_text = requests.get(url=detail_url,headers=headers).text

解析详情页的标签内容,重新实例化一个详情页bs对象,lxml解析器

detail_soup = BeautifulSoup(detail_page_text,'lxml')

属性定位

div_tag = detail_soup.find('div',class_='chapter_content')

解析到了章节的内容，利用text方法获取

content = div_tag.text

持久化存储

fp.write(title+':'+content+'n')

print(title,'爬取成功!!!')

print('爬取文本成功，进行下一步，jieba分词，并生成一个sanguo.xlsx文件用于数据分析')

排除一些不是人名，但是出现次数比较靠前的单词

excludes = {"将军","却说","荆州","二人","不可","不能","如此","商议","如何","主公","军士", "左右","军马","引兵","次日","大喜","天下","东吴","于是","今日","不敢","魏兵", "陛下","一人","都督","人马","不知","汉中","只见","众将","后主","蜀兵","上马", "大叫","太守","此人","夫人","先主","后人","背后","城中","天子","一面","何不", "大军","忽报","先生","百姓","何故","然后","先锋","不如","赶来","原来","令人", "江东","下马","喊声","正是","徐州","忽然","因此","成都","不见","未知","大败", "大事","之后","一军","引军","起兵","军中","接应","进兵","大惊","可以","以为", "大怒","不得","心中","下文","一声","追赶","粮草","曹兵","一齐","分解","回报", "分付","只得","出马","三千","大将","许都","随后","报知","前面","之兵","且说", "众官","洛阳","领兵","何人","星夜","精兵","城上","之计","不肯","相见","其言", "一日","而行","文武","襄阳","准备","若何","出战","亲自","必有","此事","军师", "之中","伏兵","祁山","乘势","忽见","大笑","樊城","兄弟","首级","立于","西川","朝廷","三军","大王","传令","当先","五百","一彪","坚守","此时","之间","投降","五千","埋伏","长安","三路","遣使","英雄"}

打开爬取下来的文件，并设置编码格式

txt = open("sanguo.txt", "r", encoding='utf-8').read()

精确模式，把文本精确的切分开，不存在冗余单词，返回列表类型

words = jieba.lcut(txt)

构造一个字典，来表达单词和出现频率的对应关系

counts = {}

逐一从words中取出每一个元素

for word in words:

已经有这个键的话就把相应的值加1，没有的话就取值为0，再加1

if len(word) == 1:

continue

elif word == "诸葛亮" or word == "孔明曰":

rword = "孔明"

elif word == "关公" or word == "云长":

rword = "关羽"

elif word == "玄德" or word == "玄德曰":

rword = "刘备"

elif word == "孟德" or word == "丞相":

rword = "曹操"

else:

rword = word

如果在里面返回他的次数，如果不在则添加到字典里面并加一

counts[rword] = counts.get(rword,0) + 1

删除停用词

for word in excludes:

del counts[word]

排序，变成list类型，并使用sort方法

items = list(counts.items())

对一个列表按照键值对的2个元素的第二个元素进行排序

Ture从大到小，结果保存在items中，第一个元素就是出现次数最多的元素

items.sort(key=lambda x:x[1], reverse=True)

将前十个单词以及出现的次数打印出来

name=[]

times=[]

for i in range(30):

word, count = items[i]

print ("{0:<10}{1:>5}".format(word, count))

name.append(word)

times.append(count)

print(name)

print(times)

创建索引

id=[]

for i in range(1,31):

id.append(i)

数据帧，相当于Excel中的一个工作表

df = pd.DataFrame({

'id':id,

'name':name,

'times':times,

})

自定义索引，不然pandas会使用默认的索引，这会导致生成的工作表

也会存在这些索引，默认从0开始

df = df.set_index('id')

print(df)

df.to_excel('sanguo.xlsx')

print("DONE!")

print('生成文件成功，进行下一步，生成词云')

词云部分

w=wordcloud.WordCloud(

font_path="C:\Windows\Fonts\simhei.ttf", #设置字体

background_color="white", #设置词云背景颜色

max_words=1000, #词云允许最大词汇数

max_font_size=100, #最大字体大小

random_state=50 #配色方案的种数

)

txt=" ".join(name)

w.generate(txt)

w.to_file("ciyun.png")

print("done!")

print("词云生成并保存成功!!!,进行下一步生成柱状图")

dirpath = 'sanguo.xlsx'

data = pd.read_excel(dirpath,index_col='id',sheet_name='Sheet1')#指定id列为索引

print(data.head())#到此数据正常

print('OK!，到此数据正常')

柱状图部分

直接使用plt.bar() 绘制柱状图,颜色紫罗兰

plt.bar(data.name,data.times,color="#87CEFA")

添加中文字体支持

plt.rcParams['font.sans-serif'] = ['SimHei']

plt.rcParams['axes.unicode_minus'] = False

设置标题，x轴，y轴，fontsize设置字号

plt.title('三国人物名字前三十名出现的次数',fontsize=16)

plt.xlabel('人名')

plt.ylabel('统计次数')

因为X轴字体太长，利用rotation将其旋转90度

plt.xticks(data.name,rotation='90')

紧凑型布局，x轴太长为了显示全

plt.tight_layout()

imgname = 'sanguo.jpg'#设置图片保存的位置

plt.savefig(imgname)#保存图片

plt.show()

print('柱状图生成完毕!!!')

print('所有程序执行完成')