Regex on unicode string

Question

I am trying to download a few hundred Korean pages like this one:

http://homeplusexpress.com/store/store_view.asp?cd_express=3

For each page, I want to use a regex to extract the "address" field, which in the above page looks like:

＊주소 : 서울시 광진구 구의1동 236-53

So I do this:

>>> import requests
>>> resp=requests.get('http://homeplusexpress.com/store/store_view.asp?cd_express=3')
>>> resp.encoding
'ISO-8859-1'
>>> # I wonder why it's ISO-8859-1, since I thought that is for Latin text (Latin-1).
>>> html = resp.text
>>> type(html)
<type 'unicode'>
>>> html
(outputs a long string that contains a lot of characters like \xc3\xb7\xaf\xbd\xba \xc0\xcd\xbd\xba\xc7\xc1\xb7\xb9\)

I then wrote a script. I set # -*- coding: utf-8 -*- on the .py file and put this:

address = re.search('주소', html)

However, re.search is returning None. I tried with and without the u prefix on the regex string.

Usually I can solve issues like this with a call to .encode or .decode but I tried a few things and am stuck. Any pointers on what I'm missing?

Answer 1

According to the tag in the html document header:

<meta http-equiv="Content-Type" content="text/html; charset=euc-kr">

the web page uses the euc-kr encoding.

I wrote this code:

# -*- coding: euc-kr -*-

import re

import requests

resp=requests.get('http://homeplusexpress.com/store/store_view.asp?cd_express=3')
html = resp.text

address = re.search('주소', html)

print address

Then I saved it in gedit using the euc-kr encoding.

I got a match.

But actually there is an even better solution! You can keep the utf-8 encoding for your files.

# -*- coding: utf-8 -*-

import re

import requests

resp=requests.get('http://homeplusexpress.com/store/store_view.asp?cd_express=3')

resp.encoding = 'euc-kr'
# we need to specify what the encoding is because the 
# requests library couldn't detect it correctly

html = resp.text
# now the html variable contains a utf-8 encoded unicode instance

print type(html)

# we use the re.search functions with unicode strings
address = re.search(u'주소', html)

print address

Answer 2

From requests documetation: When you make a request, Requests makes educated guesses about the encoding of the response based on the HTTP headers

If you check your website, we can see there is no encoding in server response:

I think the only option in this case is directly specify what encoding to use:

# -*- coding: utf-8 -*-

import requests
import re

r = requests.get('http://homeplusexpress.com/store/store_view.asp?cd_express=3')
r.encoding = 'euc-kr'
print re.search(ur'주소', r.text, re.UNICODE)