A small reptiles, demonstrates climb a word, after deposited into txt file to

A small reptiles, demonstrates climb a word, after data cleaning, deposited into txt file to

Requests Import, Re 
from the BeautifulSoup BS4 Import 

URL = "https://trade.500.com/sfc/" 
URL2 = "https://so.gushiwen.org/shiwenv_4d3b4d132c82.aspx" 

REQ = requests.get (URL2) 
IF == 200 is req.status_code: 
    IF req.encoding == "GBK" or req.encoding == "the ISO-8859-1": 
        HTML req.content.decode = ( "GBK") 
    the else: 
        HTML = req.text 

    Soup the BeautifulSoup = (HTML, 'lxml') 

    # regular use to locate 
    # h1-h7 is a lookup tag all data 
    # soup.findAll Result = (the re.compile ( "H [1-7]")) 
    # is a lookup tag All data, plus a look for content that includes "500" character of all labels and content 
    # result2 = soup.findAll ( "a", text = re.compile (. "* (500) +. *")) 
    # Find information outside the chain, namely href = "http: // ..." or "https://..."
    Soup.findAll result3 = # ( "A", attrs = { "href": re.compile ( "^ (HTTP \ :) | ^ (HTTPS \:.) *")}) 

    # Use the navigation tree to find 
    # soup .body.children 
    # soup.body.descendants 
    . # soup.body.find ( "div") next_siblings 
    . # soup.body.find ( "div") parent 

    # get all the source code 
    # Print (Soup) 
    # Gets title: 
    title soup.findAll = ( "h1") 
    title = [x.text in for the X-title] 
    title = "" the Join (title). 
    Print (title) 
    # access to content: 
    # content = soup.body.findAll ( "div", = ID "contson4d3b4d132c82") 
    Content = soup.body.findAll ( "div", attrs = { "ID": "contson4d3b4d132c82"}) # effect a supra 
    content = [x.Content for X in text] 

    # content data cleaning: 
    Content = "" .join (Content) .strip () to spaces #
    # Content = re.sub ( "original character", "replacement character" Content) 
    # content = re.sub ( "\ (. *? \)" "" Content) #. *? Lazy match, No need to? Invincible matching 
    Print (Content) 

    # last write txt file 
    with Open (F "} {title .txt", "W", encoding = "UTF-. 8") AS F: 
        f.write (title + "\ n-" Content +) 




the else: 
    Print ( "connection is not successful, check the program and the network?")