In Python how to encode/decode unicode characters such as ö

Question

Using Python 2.6.6 on CentOS 6.4

import json
import urllib2    

url = 'http://www.google.com.hk/complete/search?output=toolbar&hl=en&q=how%20to%20pronounce%20e'
opener = urllib2.build_opener(urllib2.HTTPCookieProcessor())
opener.addheaders = [('Accept-Charset', 'utf-8')]
response = opener.open(url)
page = response.read()
print page

Result:

...<suggestion data="how to pronounce eyjafjallaj

at which Python dies with no error message.

I think it dies because the next character is ö:

<toplevel>
<CompleteSuggestion>
<suggestion data="how to pronounce edinburgh"/>
</CompleteSuggestion>
<CompleteSuggestion>
<suggestion data="how to pronounce elle"/>
</CompleteSuggestion>
<CompleteSuggestion>
<suggestion data="how to pronounce edith"/>
</CompleteSuggestion>
<CompleteSuggestion>
<suggestion data="how to pronounce et al"/>
</CompleteSuggestion>
<CompleteSuggestion>
<suggestion data="how to pronounce eunice"/>
</CompleteSuggestion>
<CompleteSuggestion>
<suggestion data="how to pronounce english names"/>
</CompleteSuggestion>
<CompleteSuggestion>
<suggestion data="how to pronounce edamame"/>
</CompleteSuggestion>
<CompleteSuggestion>
<suggestion data="how to pronounce erudite"/>
</CompleteSuggestion>
<CompleteSuggestion>
<suggestion data="how to pronounce eyjafjallajökull"/>
</CompleteSuggestion>
<CompleteSuggestion>
<suggestion data="how to pronounce either"/>
</CompleteSuggestion>
</toplevel>

_{http://www.google.com.hk/complete/search?output=toolbar&hl=en&q=how%20to%20pronounce%20e}

This appears to be a unicode issue, I have tried encode('utf-8') and decode('utf-8') in many ways, but it still dies. Any ideas?

PS It seems I need to stay with urllib2 not urllib as urllib ignores cookies that causes other problems.

Answer 1

response.read() returns a bytestring. Python shouldn't die while printing a bytestring because no character conversion occurs, bytes are printed as is.

You could try to print Unicode instead:

text = page.decode(response.info().getparam('charset') or 'utf-8')
print text