基于xpath选择器、PyQuery、正则表达式的格式清理工具详解

1，使用xpath清理不必要的标签元素，以及无内容标签

from lxml import etree
 
def xpath_clean(self, text: str, xpath_dict: dict) -> str:
    '''
    xpath 清除不必要的元素
    :param text: html_content
    :param xpath_dict: 清除目标xpath
    :return: string type html_content
    '''
    remove_by_xpath = xpath_dict if xpath_dict else dict()
 
    # 必然清除的项目 除非极端情况 一般这些都是要清除的
    remove_by_xpath.update({
      '_remove_2': '//iframe',
      '_remove_4': '//button',
      '_remove_5': '//form',
      '_remove_6': '//input',
      '_remove_7': '//select',
      '_remove_8': '//option',
      '_remove_9': '//textarea',
      '_remove_10': '//figure',
      '_remove_11': '//figcaption',
      '_remove_12': '//frame',
      '_remove_13': '//video',
      '_remove_14': '//script',
      '_remove_15': '//style'
    })
 
    parser = etree.HTMLParser(remove_blank_text=True, remove_comments=True)
    selector = etree.HTML(text, parser=parser)
 
    # 常规删除操作，不需要的标签删除
    for xpath in remove_by_xpath.values():
      for bad in selector.xpath(xpath):
        bad_string = etree.tostring(bad, encoding='utf-8',
                      pretty_print=True).decode()
        logger.debug(f"clean article content : {bad_string}")
        bad.getparent().remove(bad)
 
    skip_tip = "name()='img' or name()='tr' or " \
          "name()='th' or name()='tbody' or " \
          "name()='thead' or name()='table'"
    # 判断所有p标签，是否有内容存在，没有的直接删除
    for p in selector.xpath(f"//*[not({skip_tip})]"):
      # 跳过逻辑
      if p.xpath(f".//*[{skip_tip}]") or \
          bool(re.sub('\s', '', p.xpath('string(.)'))):
        continue
 
      bad_p = etree.tostring(p, encoding='utf-8',
                  pretty_print=True).decode()
      logger.debug(f"clean p tag : {bad_p}")
      p.getparent().remove(p)
 
    return etree.tostring(selector, encoding='utf-8',
               pretty_print=True).decode()

2，使用pyquery清理标签属性，并返回处理后源码和纯净文本

#!/usr/bin/env python
# -*-coding:utf-8-*-
 
from pyquery import PyQuery as pq
 
def pyquery_clean(self, text, url, pq_dict) -> object:
    '''
    pyquery 做出必要的处理，
    :param text:
    :param url:
    :param pq_dict:
    :return:
    '''
    # 删除pq表达式字典
    remove_by_pq = pq_dict if pq_dict else dict()
    # 标签属性白名单
    attr_white_list = ['rowspan', 'colspan']
    # 图片链接key
    img_key_list = ['src', 'data-echo', 'data-src', 'data-original']
    # 生成pyquery对象
    dom = pq(text)
 
    # 删除无用标签
    for bad_tag in remove_by_pq.values():
      for bad in dom(bad_tag):
        bad_string = pq(bad).html()
        logger.debug(f"clean article content : {bad_string}")
      dom.remove(bad_tag)
 
    # 标签各个属性处理
    for tag in dom('*'):
      for key, value in tag.attrib.items():
        # 跳过逻辑，保留表格的rowspan和colspan属性
        if key in attr_white_list:
          continue
        # 处理图片链接，不完整url，补充完整后替换
        if key in img_key_list:
          img_url = self.absolute_url(url, value)
          pq(tag).remove_attr(key)
          pq(tag).attr('src', img_url)
          pq(tag).attr('alt', '')
        # img标签的alt属性保留为空
        elif key == 'alt':
          pq(tag).attr(key, '')
        # 其余所有属性做删除操作
        else:
          pq(tag).remove_attr(key)
 
    return dom.text(), dom.html()

 3，正则表达清理空格以及换行符内容

#!/usr/bin/env python
# -*-coding:utf-8-*-
 
import re  
 
def regular_clean(self, str1: str, str2: str):
    '''
    正则表达式处理数据格式
    :param str1: content
    :param str2: html_content
    :return: 返回处理后的结果
    '''
 
    def new_line(text):
      text = re.sub('<br\s?/?>', '<br>', text)
      text = re.sub(
        '</?a>|</?em>|</?html>|</?body>|'
        '</?head>|<[a-zA-Z]{1,10}\s?/>|'
        '</?strong>|</?blockquote>|</?b>|'
        '</?span>|</?i>|</?hr>|</?font>',
        '',
        text)
      text = re.sub('\n', '', text)
      text = re.sub('<h[1-6]>', '<p>', text)
      text = re.sub('</h[1-6]>', '</p>', text)
      text = text.replace('</p>', '</p>\n').replace('<br>', '<br/>')
      return text
 
    str1, str2 = self.clean_blank(str1), self.clean_blank(str2) # TODO 处理空白行问题
 
    # TODO html_content处理 1，删除多余的无法使用的标签以及影响数据展示的标签 2，换行符问题处理以及更换
 
    str2 = new_line(text=str2)
 
    return str1, str2

结尾部分，各个方法封装类代码展示

#!/usr/bin/env python
# -*-coding:utf-8-*-
'''
author: szhan
date：2020-08-17
summery: 清理html_conent以及获取纯净数据格式
'''
 
import re
from lxml import etree
from pyquery import PyQuery as pq
from urllib.parse import urlsplit, urljoin
 
from loguru import logger
 
 
class CleanArticle:
 
  def __init__(
      self,
      text: str,
      url: str = '',
      xpath_dict: dict = None,
      pq_dict: dict = None
  ):
    self.text = text
    self.url = url
    self.xpath_dict = xpath_dict or dict()
    self.pq_dict = pq_dict or dict()
 
  @staticmethod
  def absolute_url(baseurl: str, url: str) -> str:
    '''
    补充url
    :param baseurl:scheme url
    :param url: target url
    :return: complete url
    '''
    target_url = url if urlsplit(url).scheme else urljoin(baseurl, url)
    return target_url
 
  @staticmethod
  def clean_blank(text):
    '''
    空白处理
    :param text:
    :return:
    '''
    text = text.replace('
', '').replace('\u3000', '').replace('\t', '').replace('\xa0', '')
    text = re.sub('\s{2,}', '', text)
    text = re.sub('\n{2,}', '\n', text)
    text = text.strip('\n').strip()
    return text
 
  def run(self):
    '''
    :return:处理后的content, html_content
    '''
    if (not bool(self.text)) or (not isinstance(self.text, str)):
      raise ValueError('html_content has a bad type value')
    # 首先，使用xpath去除空格，以及注释，iframe, button, form, script, style, video等标签
    text = self.xpath_clean(self.text, self.xpath_dict)
 
    # 第二步，使用pyquery处理具体细节方面
    str1, str2 = self.pyquery_clean(text, self.url, self.pq_dict)
 
    # 最终的正则处理
    content, html_content = self.regular_clean(str1, str2)
 
    return content, html_content
 
  def xpath_clean(self, text: str, xpath_dict: dict) -> str:
    '''
    xpath 清除不必要的元素
    :param text: html_content
    :param xpath_dict: 清除目标xpath
    :return: string type html_content
    '''
    remove_by_xpath = xpath_dict if xpath_dict else dict()
 
    # 必然清除的项目 除非极端情况 一般这些都是要清除的
    remove_by_xpath.update({
      '_remove_2': '//iframe',
      '_remove_4': '//button',
      '_remove_5': '//form',
      '_remove_6': '//input',
      '_remove_7': '//select',
      '_remove_8': '//option',
      '_remove_9': '//textarea',
      '_remove_10': '//figure',
      '_remove_11': '//figcaption',
      '_remove_12': '//frame',
      '_remove_13': '//video',
      '_remove_14': '//script',
      '_remove_15': '//style'
    })
 
    parser = etree.HTMLParser(remove_blank_text=True, remove_comments=True)
    selector = etree.HTML(text, parser=parser)
 
    # 常规删除操作，不需要的标签删除
    for xpath in remove_by_xpath.values():
      for bad in selector.xpath(xpath):
        bad_string = etree.tostring(bad, encoding='utf-8',
                      pretty_print=True).decode()
        logger.debug(f"clean article content : {bad_string}")
        bad.getparent().remove(bad)
 
    skip_tip = "name()='img' or name()='tr' or " \
          "name()='th' or name()='tbody' or " \
          "name()='thead' or name()='table'"
    # 判断所有p标签，是否有内容存在，没有的直接删除
    for p in selector.xpath(f"//*[not({skip_tip})]"):
      # 跳过逻辑
      if p.xpath(f".//*[{skip_tip}]") or \
          bool(re.sub('\s', '', p.xpath('string(.)'))):
        continue
 
      bad_p = etree.tostring(p, encoding='utf-8',
                  pretty_print=True).decode()
      logger.debug(f"clean p tag : {bad_p}")
      p.getparent().remove(p)
 
    return etree.tostring(selector, encoding='utf-8',
               pretty_print=True).decode()
 
  def pyquery_clean(self, text, url, pq_dict) -> object:
    '''
    pyquery 做出必要的处理，
    :param text:
    :param url:
    :param pq_dict:
    :return:
    '''
    # 删除pq表达式字典
    remove_by_pq = pq_dict if pq_dict else dict()
    # 标签属性白名单
    attr_white_list = ['rowspan', 'colspan']
    # 图片链接key
    img_key_list = ['src', 'data-echo', 'data-src', 'data-original']
    # 生成pyquery对象
    dom = pq(text)
 
    # 删除无用标签
    for bad_tag in remove_by_pq.values():
      for bad in dom(bad_tag):
        bad_string = pq(bad).html()
        logger.debug(f"clean article content : {bad_string}")
      dom.remove(bad_tag)
 
    # 标签各个属性处理
    for tag in dom('*'):
      for key, value in tag.attrib.items():
        # 跳过逻辑，保留表格的rowspan和colspan属性
        if key in attr_white_list:
          continue
        # 处理图片链接，不完整url，补充完整后替换
        if key in img_key_list:
          img_url = self.absolute_url(url, value)
          pq(tag).remove_attr(key)
          pq(tag).attr('src', img_url)
          pq(tag).attr('alt', '')
        # img标签的alt属性保留为空
        elif key == 'alt':
          pq(tag).attr(key, '')
        # 其余所有属性做删除操作
        else:
          pq(tag).remove_attr(key)
 
    return dom.text(), dom.html()
 
  def regular_clean(self, str1: str, str2: str):
    '''
    正则表达式处理数据格式
    :param str1: content
    :param str2: html_content
    :return: 返回处理后的结果
    '''
 
    def new_line(text):
      text = re.sub('<br\s?/?>', '<br>', text)
      text = re.sub(
        '</?a>|</?em>|</?html>|</?body>|'
        '</?head>|<[a-zA-Z]{1,10}\s?/>|'
        '</?strong>|</?blockquote>|</?b>|'
        '</?span>|</?i>|</?hr>|</?font>',
        '',
        text)
      text = re.sub('\n', '', text)
      text = re.sub('<h[1-6]>', '<p>', text)
      text = re.sub('</h[1-6]>', '</p>', text)
      text = text.replace('</p>', '</p>\n').replace('<br>', '<br/>')
      return text
 
    str1, str2 = self.clean_blank(str1), self.clean_blank(str2) # TODO 处理空白行问题
 
    # TODO html_content处理 1，删除多余的无法使用的标签以及影响数据展示的标签 2，换行符问题处理以及更换
 
    str2 = new_line(text=str2)
 
    return str1, str2
 
if __name__ == '__main__':
  with open('html_content.html', 'r', encoding='utf-8') as f:
    lines = f.readlines()
    html = ''
    for line in lines:
      html += line
  ca = CleanArticle(text=html)
  _, html_content = ca.run()
  print(html_content)

总结