如何解决解析包含西里尔符号的HTML文件的问题？
我有一个包含元素的html文件：
<html>
<body>
<span class="one">Text</span>some text</br>
<span class="two">Привет</span>Текст на русском</br>
</body>
</html>
要获取“some text”：
# -*- coding:cp1251 -*-
import lxml
from lxml import html
filename = "t.html"
fread = open(filename, 'r')
source = fread.read()
tree = html.fromstring(source)
fread.close()
tags = tree.xpath('//span[@class="one" and text()="Text"]') #This OK
print "name: ",tags[0].text
print "value: ",tags[0].tail
tags = tree.xpath('//span[@class="two" and text()="Привет"]') #This False
print "name: ",tags[0].text
print "value: ",tags[0].tail
这样显示：
name: Text
value: some text
Traceback: ... in line `tags = tree.xpath('//span[@class="two" and text()="Привет"]')`
ValueError: All strings must be XML compatible: Unicode or ASCII, no NULL bytes
怎么解决这个问题呢？
4 个回答
1
我在用lxml生成XML的时候遇到了同样的错误。在这里找到了解决办法：http://lethain.com/stripping-illegal-characters-from-xml-in-python/
我只做了：
remove_re = re.compile(u'[\x00-\x08\x0B-\x0C\x0E-\x1F\x7F]')
etree_sub_el.text = remove_re.sub('', text)
4
试试这个
tree = html.fromstring(source.decode('utf-8'))
还有这个
tags = tree.xpath('//span[@class="two" and text()="%s"]' % u'Привет' )
4
lxml
（观察到在系统编码方面，这个有点不太靠谱，显然在Windows XP上运行不正常，不过在Linux上可以。）
我通过解码源字符串让它工作了 - tree = html.fromstring(source.decode('utf-8'))：
# -*- coding:cp1251 -*-
import lxml
from lxml import html
filename = "t.html"
fread = open(filename, 'r')
source = fread.read()
tree = html.fromstring(source.decode('utf-8'))
fread.close()
tags = tree.xpath('//span[@class="one" and text()="Text"]') #This OK
print "name: ",tags[0].text
print "value: ",tags[0].tail
tags = tree.xpath('//span[@class="two" and text()="Привет"]') #This is now OK too
print "name: ",tags[0].text
print "value: ",tags[0].tail
这意味着实际的树结构都是unicode对象。如果你直接把xpath参数作为unicode传入，它会找不到任何匹配项。
BeautifulSoup
反正我更喜欢用BeautifulSoup来处理这些事情。这里是我的交互式会话；我把文件保存为cp1251编码。
>>> from BeautifulSoup import BeautifulSoup
>>> filename = '/tmp/cyrillic'
>>> fread = open(filename, 'r')
>>> source = fread.read()
>>> source # Scary
'<html>\n<body>\n<span class="one">Text</span>some text</br>\n<span class="two">\xcf\xf0\xe8\xe2\xe5\xf2</span>\xd2\xe5\xea\xf1\xf2 \xed\xe0 \xf0\xf3\xf1\xf1\xea\xee\xec</br>\n</body>\n</html>\n'
>>> source = source.decode('cp1251') # Let's try getting this right.
u'<html>\n<body>\n<span class="one">Text</span>some text</br>\n<span class="two">\u041f\u0440\u0438\u0432\u0435\u0442</span>\u0422\u0435\u043a\u0441\u0442 \u043d\u0430 \u0440\u0443\u0441\u0441\u043a\u043e\u043c</br>\n</body>\n</html>\n'
>>> soup = BeautifulSoup(source)
>>> soup # OK, that's looking right now. Note the </br> was dropped as that's bad HTML with no meaning.
<html>
<body>
<span class="one">Text</span>some text
<span class="two">Привет</span>Текст на русском
</body>
</html>
>>> soup.find('span', 'one').findNextSibling(text=True)
u'some text'
>>> soup.find('span', 'two').findNextSibling(text=True) # This looks a bit daunting ...
u'\u0422\u0435\u043a\u0441\u0442 \u043d\u0430 \u0440\u0443\u0441\u0441\u043a\u043e\u043c'
>>> print _ # ... but it's not, really. Just Unicode chars.
Текст на русском
>>> # Then you may also wish to get things by text:
>>> print soup.find(text=u'Привет').findParent().findNextSibling(text=True)
Текст на русском
>>> # You can't get things by attributes and the contained NavigableString at the same time, though. That may be a limitation.
在那之后，考虑在从文件系统读取时尝试source.decode('cp1251')而不是source.decode('utf-8')可能是值得的。这样lxml可能就能正常工作了。