A Real Python Project Experience

import pandas as pd
import numpy as np
import jieba
from wordcloud import WordCloud
import matplotlib.pyplot as plt
# get_ipython().run_line_magic('matplotlib', 'inline')

#!/usr/bin/env python
# -*- coding: utf-8 -*-
# @Time    : 2021年06月05日
# @File    : demo4.py
import requests
from fake_useragent import UserAgent
from lxml import etree
import time

# Random request header
ua = UserAgent()

# Construct request. You need to change it on the webpage. If you can't request, refresh the webpage and deal with the captcha.
headers = {    'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.9',    'Cookie': '__mta=244176442.1622872454168.1622876903037.1622877097390.7; uuid_n_v=v1; uuid=6FFF6D30C5C211EB8D61CF53B1EFE83FE91D3C40EE5240DCBA0A422050B1E8C0; _csrf=bff9b813020b795594ff3b2ea3c1be6295b7453d19ecd72f8beb9700c679dfb4; Hm_lvt_703e94591e87be68cc8da0da7cbd0be2=1622872443; _lxsdk_cuid=1770e9ed136c8-048c356e76a22b-7d677965-1fa400-1770e9ed136c8; _lxsdk=6FFF6D30C5C211EB8D61CF53B1EFE83FE91D3C40EE5240DCBA0A422050B1E8C0; ci=59; recentCis=59; __mta=51142166.1622872443578.1622872443578.1622876719906.2; Hm_lpvt_703e94591e87be68cc8da0da7cbd0be2=1622877097; _lxsdk_s=179dafd56bf-06d-403-d81%7C%7C12',    'User-Agent': str(ua.random)}

def RequestsTools(url):    '''    Crawler request tool function    :param url: request address    :return: HTML object for xpath extraction    '''    response = requests.get(url, headers=headers).content.decode('utf-8')    html = etree.HTML(response)    return html

def Index(page):    '''    Homepage function    :param page: page number    :return:    '''    url = 'https://maoyan.com/board/4?offset={}'.format(page)    html = RequestsTools(url)    # Detail page address suffix    urls_text = html.xpath('//a[@class=image-link]/@href')    # Rating    pingfen1 = html.xpath('//i[@class=integer]/text()')    pingfen2 = html.xpath('//i[@class=fraction]/text()')
    for i, p1, p2 in zip(urls_text, pingfen1, pingfen2):        pingfen = p1 + p2        urs = 'https://maoyan.com' + i        # Request too frequently        time.sleep(2)        Details(urs, pingfen)

def Details(url, pingfen):    html = RequestsTools(url)    dianyan = html.xpath('//h1[@class=name]/text()') # Movie name    leixing = html.xpath('//li[@class=ellipsis]/a/text()') # Genre    diqu = html.xpath('/html/body/div[3]/div/div[2]/div[1]/ul/li[2]/text()') # Read total    timedata = html.xpath('/html/body/div[3]/div/div[2]/div[1]/ul/li[3]/text()') # Time    for d, l, b, t in zip(dianyan, leixing, diqu, timedata):        countyr = b.replace('\n', '').split('/')[0] # Location        shichang = b.replace('\n', '').split('/')[1] # Duration        f = open('猫眼.csv', 'a')        f.write('{}, {}, {}, {}, {}, {}, {}
'.format(d, pingfen, url, l, countyr, shichang, t))        print(d, pingfen, url, l, countyr, shichang, t )

for page in range(0, 11):    page *= 10    Index(page)

#!/usr/bin/env python
# coding: utf-8
# Load commonly used data analysis libraries
import pandas as pd
import numpy as np
import jieba
from wordcloud import WordCloud
import matplotlib.pyplot as plt
# get_ipython().run_line_magic('matplotlib', 'inline')

# In[3]:

path='./maoyan.csv'
df=pd.read_csv(path,sep=',',encoding='utf-8',index_col=False)
df.drop(df.columns[0],axis=1,inplace=True)
df.dropna(inplace=True)
df.drop_duplicates(inplace=True)
df.head(10)



# Check the structure of the data
df.info()
print(df.columns)


# In[11]:

# Year and number of movies released. The number of releases after 2018 is uncertain, so we exclude it for now.
fig,ax=plt.subplots(figsize=(9,6),dpi=70)
df[df[u'上映时间']<2018][u'上映时间'].value_counts().sort_index().plot(kind='line',ax=ax)
ax.set_xlabel(u'Year')
ax.set_ylabel(u'Number of Releases')
ax.set_title(u'Release Year and Number of Movies Released')


# Based on the above figure, create a relationship chart between release year, number of releases, and ratings.
# However, due to the small amount of data before 1980, the ratings are inaccurate, so we focus on the analysis area from 1980 to 2017.
x=df[df[u'上映时间']<2018][u'上映时间'].value_counts().sort_index().index
y=df[df[u'上映时间']<2018][u'上映时间'].value_counts().sort_index().values
y2=df[df[u'上映时间']<2018].sort_values(by=u'上映时间').groupby(u'上映时间').mean()[u'评分'].values
fig,ax=plt.subplots(figsize=(10,5),dpi=70)
ax.plot(x,y,label=u'Number of Releases')
ax.set_xlim(1980,2017)
ax.set_xlabel(u'Release Year')
ax.set_ylabel(u'Number of Releases')
ax.set_title(u'Release Year, Number of Releases, and Average Ratings')
ax2=ax.twinx()
ax2.plot(x,y2,c='y',ls='--',label=u'Ratings')
ax.legend(loc=1)
ax2.legend(loc=2)

# Solve Chinese garbled characters and the issue of negative values on the axes.
plt.rcParams['font.sans-serif'] =['Microsoft YaHei']
plt.rcParams['axes.unicode_minus'] = False

# In[12]:

# World, Release Year, and Average Ratings
fig,ax=plt.subplots(figsize=(10,7),dpi=60)
df[df[u'评分']>0].groupby(u'上映时间').mean()[u'评分'].plot(kind='line',ax=ax)
ax.set_ylabel(u'Ratings')
ax.set_title(u'World, Release Year, and Average Ratings')

# In[13]:

# Number of movies by genre in the world.
# Split the genres into the smallest units and then count.
types=[]
for tp in df[u'类型']:
    ls=tp.split(',')
    for x in ls:
        types.append(x)

tp_df=pd.DataFrame({u'类型':types})
fig,ax=plt.subplots(figsize=(9,6),dpi=60)
tp_df[u'类型'].value_counts().plot(kind='bar',ax=ax)
ax.set_xlabel(u'Genre')
ax.set_ylabel(u'Number')
ax.set_title(u'World, Genre, and Number')

# In[14]:

# Distribution of movie duration and ratings.
# Note: Some movies have not been rated, so those need to be filtered out.
x=df[df[u'评分']>0].sort_values(by=u'时长(min)')[u'时长(min)'].values
y=df[df[u'评分']>0].sort_values(by=u'时长(min)')[u'评分'].values
fig,ax=plt.subplots(figsize=(9,6),dpi=70)
ax.scatter(x,y,alpha=0.6,marker='o')
ax.set_xlabel(u'Duration (min)')
ax.set_ylabel(u'Number')
ax.set_title(u'Distribution of Movie Duration and Ratings')

# It can be seen that the ratings...

# Data filtering for China
i=0
c0=[]
c1=[]
c2=[]
c3=[]
c4=[]
c5=[]
c6=[]
c7=[]
for x in df[u'地区']:
    if u'中国大陆' in x:
        c0.append(df.iat[i, 0])
        c1.append(df.iat[i, 1])
        c2.append(df.iat[i, 2])
        c3.append(df.iat[i, 3])
        c4.append(df.iat[i, 4])
        c5.append(df.iat[i, 5])
        c6.append(df.iat[i, 6])
        c7.append(df.iat[i, 7])
    i=i+1
china_df=pd.DataFrame({u'电影':c0, u'评分':c1,u'链接':c2, u'类型':c3,u'地区':c4, u'上映地点':c5,u'时长(min)':c6,u'上映时间':c7})

# In[16]:

# Comparing average ratings between China and the world in the time range of 1980-2017.
x1 = df[df[u'评分']>0].groupby(u'上映时间').mean()[u'评分'].index
y1 = df[df[u'评分']>0].groupby(u'上映时间').mean()[u'评分'].values    
x2 = china_df[china_df[u'评分']>0].groupby(u'上映时间').mean()[u'评分'].index
y2 = china_df[china_df[u'评分']>0].groupby(u'上映时间').mean()[u'评分'].values
fig,ax=plt.subplots(figsize=(12,9),dpi=60)
ax.plot(x1,y1,ls='-',c='DarkTurquoise',label=u'World')
ax.plot(x2,y2,ls='--',c='Gold',label=u'China')
ax.set_title(u'Average Ratings: China vs World')
ax.set_xlabel(u'Time')
ax.set_xlim(1980,2017)
ax.set_ylabel(u'Ratings')
ax.legend()

# In[17]:

# Number of movies by genre: Comparison between China and the world.
# Since genres are mixed, we need to write a function to split them first.

# In[18]:

# Function to split genres: input a Series object and return a DataFrame of split genres.
# Here, we input a Series of genres.
def Cuttig_type(typeS):    types=[]    types1=[]
    for x in typeS:        if len(x)<4:            # print x            types1.append(x)        ls=x.split(',')        for i in ls:            types.append(i)
    types.extend(types1)    df=pd.DataFrame({u'类型':types})    return pd.DataFrame(df[u'类型'].value_counts().sort_values(ascending=False))

# In[19]:

# Comparing movie types between China and the world.
df1=Cuttig_type(china_df[u'类型'])
df2=Cuttig_type(df[u'类型'])
trans=pd.concat([df1,df2],axis=1)
trans.dropna(inplace=True)
trans.columns=[u'中国',u'世界']
fig,ax=plt.subplots(figsize=(15,9),dpi=80)
trans.plot(kind='bar',ax=ax) 
fig.autofmt_xdate(rotation=30)
ax.set_title(u'Comparison of Movie Types: China vs World')
ax.set_xlabel(u'Genre')
ax.set_ylabel(u'Number of Movies')

# In[20]:

# Then there is the scatter distribution: China, World, Duration, and Rating Distribution.
y = df[df[u'评分'] > 0].sort_values(by=u'时长(min)')[u'评分'].values
x = df[df[u'评分'] > 0].sort_values(by=u'时长(min)')[u'时长(min)'].values
y2 = china_df[china_df[u'评分'] > 0].sort_values(by=u'时长(min)')[u'评分'].values
x2 = china_df[china_df[u'评分'] > 0].sort_values(by=u'时长(min)')[u'时长(min)'].values
fig, ax = plt.subplots(figsize=(10,7), dpi=80)
ax.scatter(x, y, c='DeepSkyBlue', alpha=0.6, label=u'World')
ax.scatter(x2, y2, c='Salmon', alpha=0.7, label=u'China')
ax.set_title(u'Distribution of Ratings: China vs World')
ax.set_xlabel(u'Duration (min)')
ax.set_ylabel(u'Ratings')
ax.legend(loc=4)

# In[25]:

dfs=df[(df[u'上映时间']>1980)&(df[u'上映时间']<2019)]


# for x in range(0,len(dfs)):#     print(dfs.iat[x,0],dfs.iat[x,-1])
df666 = dfs['电影'][:15]
wl = ,.join(df666.values)# Write the tokenized txt to a text file# fenciTxt  = open(fenciHou.txt,w+)# fenciTxt.writelines(wl)# fenciTxt.close()
# Set the word cloud
lwc = WordCloud(background_color='white',  # Set background color               # mask=imread('shen.jpg'),   # Set background image#                    max_words=2000,  # Set maximum number of words                   font_path='C:\Windows\Fonts\simkai.ttf', # Set to KaiTi Regular    # Set Chinese font so that the word cloud can display (the default font for word clouds is 'DroidSansMono.ttf', which does not support Chinese)               max_font_size=60, # Set maximum font size               random_state=30,  # Set how many random generation states, i.e., how many color schemes               )
myword = wc.generate(wl)  # Generate word cloud
wc.to_file('result.jpg')
# Display the word cloud image
plt.imshow(myword)
plt.axis('off')
plt.show()

# In[41]: