python の lxml で html の dom を操作してみる。
実行結果
import sys
import lxml.html
# div/p0-p4
htmlstr = '<html><body><div class="a"><p>p0</p><p>p1</p><p>p2</p><p>p3</p><p>p4</p></div></body></html>'
dom = lxml.html.fromstring(htmlstr)
nodes = dom.xpath('//div[@class="a"]')
div = nodes[0]
# p0 を削除
div.remove(div[0])
print(lxml.html.tostring(dom).decode('utf-8'))
# p2 を検索
idx = div.index(div[1])
print('idx: %d' % (idx))
# p2 を更新
div[idx] = lxml.html.fromstring('<p>p2 new</p>')
print(lxml.html.tostring(dom).decode('utf-8'))
# body を出力
print(lxml.html.tostring(div.getparent()).decode('utf-8'))
# p5 を追加
p5 = lxml.html.fromstring('<p>p5</p>')
div.append(p5)
print(lxml.html.tostring(dom).decode('utf-8'))
実行結果
<html><body><div class="a"><p>p1</p><p>p2</p><p>p3</p><p>p4</p></div></body></html> idx: 1 <html><body><div class="a"><p>p1</p><p>p2 new</p><p>p3</p><p>p4</p></div></body></html> <body><div class="a"><p>p1</p><p>p2 new</p><p>p3</p><p>p4</p></div></body> <html><body><div class="a"><p>p1</p><p>p2 new</p><p>p3</p><p>p4</p><p>p5</p></div></body></html>
