Module 3, week 2, assignment 2, bidding website

Posted by redmonkey on Tue, 10 Mar 2020 11:21:39 +0100

1 problem description

Using the Scrapy framework, complete the collection of bidding information on the Internet. The collection fields are as follows:

2 Tips for problem solving

  1. Some pages of the Internet must be logged in to get response. You need to log in manually and get the Cookie value in the browser. Add the Cookie to the request header
  2. For data extraction, some regular expressions need to be customized. For example, the item number may be in the text of the detailed page, which cannot be extracted with ordinary XPath. This requires looking at several more pages, doing more tests, and analyzing the data format
  3. Data persistence can be done in the pipeline file. Take the example explained in the course, save the bidding information to MySQL database
  4. The proxy IP should be set in the download middleware. The proxy IP needs to access the interface of the third party. For details, please refer to the steps in recording and broadcasting

3 scoring criteria

  1. The crawler can collect the bidding data of the must be connected network 20 points
  2. The accuracy of data extraction, for example, 10 points of bidding number can be extracted more effectively
  3. Code notes, specification 10 points

4 key points analysis

  • register

  • cookie

5 implementation steps

  • Create a story, spider

  • spider file
import scrapy
import re
from copy import deepcopy

class TenderdataSpider(scrapy.Spider):
    name = 'tender_data'
    # Add to prevent filtering
    allowed_domains = ['', '']
    # start_urls = ['']
    # The data mode stored in the database is Dictionary: sql_data
    sql_data = dict(
        projectcode='',  # Item number
        web='Must networking',  # Information source website (e.g. must be connected to the Internet)
        keyword='',  # Keyword
        detail_url='',  # Bidding details page website
        title='',  # Third party website release title
        toptype='',  # Information type
        province='',  # Vested provinces
        product='',  # Product category
        industry='',  # Attribution industry
        tendering_manner='',  # Bidding method
        publicity_date='',  # Date of bidding announcement
        expiry_date='',  # Deadline for bidding
    # Because its submission mode is POST mode, its form data is stored according to the form data of the website
    form_data = dict(
        infoClassCodes='',  #
        rangeTyp='',  #
        projectType='bid',  # This value does not change, so it defaults directly
        fundSourceCodes='',  #
        dateType='',  #
        startDateCode='',  #
        endDateCode='',  #
        normIndustry='',  #
        normIndustryName='',  #
        zone='',  #
        zoneName='',  #
        zoneText='',  #
        key='',  # Keywords entered by router users:
        pubDateType='',  #
        pubDateBegin='',  #
        pubDateEnd='',  #
        sortMethod='timeDesc',  # This value does not change, so it defaults directly
        orgName='',  #
        currentPage='',  # Current page 2

    keyword_s = ['Router', 'Transformer']

    # Set start request
    def start_requests(self):
        # request needs to submit the form
        for keyword in self.keyword_s:
            # Because of multithreading, deep copy is needed on storage
            form_data = deepcopy(self.form_data)
            form_data['key'] = keyword
            form_data['currentPage'] = '1'
            # Set to FormRequest instead of Request because the form data needs to be submitted
            request = scrapy.FormRequest(
                # Submit the response to start_parse to get the maximum page number
            # Encapsulate the form? Data data, because start? Parse requires
            request.meta['form_data'] = form_data
            yield request

    # start_parse finds all the page numbers, encapsulates them in the request and submits them to the scheduler for processing
    # Loop passes all URLs to the scheduler once
    def start_parse(self, response):
        # Get list with maximum page number
        page_max_s = response.xpath('//form[@id="pagerSubmitForm"]/a/text()').extract()
        # re.match find number
        page_max = max([int(page_max) for page_max in page_max_s if re.match('\d+', page_max)])
        # Test use
        # page_max = 2
        # Submit the form of each page to parse ﹣ page1
        for page in range(1,page_max + 1):
            # First get the form data, and deep copy
            form_data = deepcopy(response.meta['form_data'])
            # The data submitted by the form is str:currentPage = ''
            form_data['currentPage'] = str(page)
            request = scrapy.FormRequest(
            request.meta['form_data'] = form_data
            yield request

    # Parse [page1] extracts the list of the xpath of the first page, and saves some SQL [u data data data]
    #  detail_url = '',      # Bidding details page website
    #  title='',             # Third party website title
    #  toptype='',           # Information type
    #  province='',          # Vested provinces
    #  product='',           # Product category
    #  tendering_manner='',  # Bidding method
    #  publicity_date='',    # Date of bidding announcement
    #  expiry_date='',       # Deadline for bidding
    def parse_page1(self, response):
        form_data = response.meta['form_data']
        # xpath all page div s, which is a list
        div_x_s = response.xpath('//div[contains(@class,"abstract-box")]')
        for div_x in div_x_s:
            # class attribute is referenced for the first time, so self is used
            sql_data = deepcopy(self.sql_data)
            # sql_data['web '] =' must be connected to the Internet ', which has been defaulted in the class attribute
            sql_data['detail_url'] = div_x.xpath('./div[1]/a/@href').extract_first()
            sql_data['toptype'] = div_x.xpath('./div[1]/i[1]/text()').extract_first()
            sql_data['title'] = div_x.xpath('./div[1]/a/text()').extract_first()
            sql_data['province'] = div_x.xpath('./div[2]/div[2]/p[2]/span[2]/text()').extract_first()
            sql_data['product'] = div_x.xpath('./div[2]/div[1]/p[2]/span[2]/text()').extract_first()
            sql_data['tendering_manner'] = div_x.xpath('./div[2]/div[1]/p[1]/span[2]/text()').extract_first()
            sql_data['publicity_date'] = div_x.xpath('./div[1]/i[2]/text()').extract_first()
            # Remove 'release date:'
            sql_data['publicity_date'] = re.sub('[^0-9\-]', '', sql_data['publicity_date'])
            sql_data['expiry_date'] = div_x.xpath('./div[2]/div[2]/p[1]/span[2]/text()').extract_first()
            if sql_data['expiry_date']:
                sql_data['expiry_date'] = re.sub('[0-9]{2}[:][0-9]{2}[:][0-9]{2}', '', sql_data['expiry_date'])
                sql_data['expiry_date'] = ""
            sql_data['keyword'] = form_data.get('key')
            # print( sql_data['detail_url'],sql_data['toptype'],sql_data['title'])
            # Because page2 is a get request, FormRequest is not required
            request = scrapy.Request(
            # Encapsulate sql_data in meta and transfer values
            request.meta['sql_data'] = sql_data
            # print(sql_data)
            yield request

    # Page 2 deals with other parts of SQL data
    # projectcode='',  # Item number
    # industry='',  # Attribution industry
    def parse_page2(self, response):
        sql_data = response.meta['sql_data']
        # print(sql_data)
        sql_data['projectcode'] = response.xpath(
        # Use regular to find item number in page
        if not sql_data['projectcode']:
            projectcode_find = re.findall(
                '(Item encoding|Item label|Purchase document No|Tender number|Project bidding No|Item number|Bidding Document No|Bidding Document No)[: :]{0,1}\s{0,2}\n*(</span\s*>)*\n*(<span.*?>)*\n*(<u*?>)*\n*([a-zA-Z0-9\-_\[\]]{1,100})',
            if projectcode_find:
                sql_data['projectcode'] = projectcode_find[0][4] if projectcode_find else ""
        sql_data['industry'] = response.xpath(
        # When SQL ﹣ data is a dictionary, the summary engine automatically judges to pass the SQL ﹣ data to the pipeline
        yield sql_data

  • middleware

  • Connect to database and store data

160 original articles published, praised 37, visited 50000+
Private letter follow

Topics: SQL Database Attribute MySQL