파충류 노트 정리

xpath('//*')

xpath('//li/a')

xpath('//a[@href="link4.html"]/../@class')

xpath('//a[@href="link4.html"]/parent::*/@class')

xpath('//li[1]/a/text()')

xpath('//li[last()]/a/text()')

xpath('//li[position()<3]/a/text()')

xpath('//li[last()-2]/a/text()')

xpath('//li[1]/ancestor::*')

xpath('//li[1]/ancestor::div')

xpath('//li[1]/attribute::*')

xpath('//li[1]/child::a[@href="link1.html"]')

xpath('//li[1]/descendant::span"]')

xpath('//li[1]/following::*[2]"]')

xpath('//li[1]/following-sibling::*"]')

xpath('//li[contains(@class,"li")]/a/text()')

xpath('//div[contains(@class,"para") and contains(@label-module,"para")]/text()')

xpath('//li[@class="item-0"]/a/text()')

xpath('//li/a/@href')

from bs4 import BeautifulSoup

soup = BeautifulSoup(html,lxml)

list(soup.a.parents)[0]

list(soup.a.parents)[0].attrs['class']

soup.find_all(name='ul')

for ul in soup.find_all(name='ul')
	for li in ul.find_all(name='li')
  	print(li.string)

soup.find_all(attrs={'id':'list-1'}

soup.find_all(class_='element')

soup.find_all(text=re.compile('link'))

soup.select('.panel .panel-heading')

for ul in soup.select('ul')
	print(ul.select('li'))

for ul in soup.select('ul')
	print(ul['id'])
  print(ul.attrs['id'])

for li in soup.select('li')
	print(li.get_text())
  print(li.string)

doc = pq(filename='demo.html')

doc('#content .list li')

items = doc('.list')
lis = items.find('li')

items = doc('.list')
lis = items.children('li')

items = doc('.list')
lis = items.parent('li')

lis = doc('li').items()
for i in lis:
  print(i)

a = doc('.item-0.active a')
print(a.attr('href'))
print(a.attr.href)
for item in a.items():
  print(item.attr('href'))

doc = pq()
a = doc('.item-0.active a')
print(a.text)
print(a.html())

doc = pq()
li = doc('.item-0.active')
li.removeClass('active')
li.addClass('active')

doc = pq()
li = doc('.item-0.active')
#attr()      ,         ,         
li.attr('name','link')
#      ,  
li.text('change item')
li.html('changed item')

doc = pq()
#       
li = doc('li:first-child')
#        
li = doc('li:last-child')
#       
li = doc('li:nth-child(2)')
#       
li = doc('li:gt(2)')
#      
li = doc('li:nth-child(2n)')
#           
li = doc('li:contains(second)')