在Python 2 + GTK中检测/删除不成对的代理字符

Question

In Python 2.7 I can successfully convert the Unicode string "abc\?xyz" to UTF-8 (result is "abc\\xed\\xb0\\xb4xyz" ). But when I pass the UTF-8 string to eg. pango_parse_markup() or g_convert_with_fallback() , I get errors like "Invalid byte sequence in conversion input". Apparently the GTK/Pango functions detect the "unpaired surrogate" in the string and (correctly?) reject it.

Python 3 doesn't even allow conversion of the Unicode string to UTF-8 (error: "'utf-8' codec can't encode character '\?' in position 3: surrogates not allowed"), but I can run "abc\?xyz".encode("utf8", "replace") to get a valid UTF8 string with the lone surrogate replaced by some other character. That's fine for me, but I need a solution for Python 2.

So the question is: in Python 2.7, how can I convert that Unicode string to UTF-8 while replacing the lone surrogate with some replacement character like U+FFFD? Preferably only standard Python functions and GTK/GLib/G... functions should be used.

Btw. Iconv can convert the string to UTF8 but simply removes the bad character instead of replacing it with U+FFFD.

Answer 1

You can do the replacements yourself before encoding:

import re

lone = re.compile(
    ur'''(?x)            # verbose expression (allows comments)
    (                    # begin group
    [\ud800-\udbff]      #   match leading surrogate
    (?![\udc00-\udfff])  #   but only if not followed by trailing surrogate
    )                    # end group
    |                    #  OR
    (                    # begin group
    (?<![\ud800-\udbff]) #   if not preceded by leading surrogate
    [\udc00-\udfff]      #   match trailing surrogate
    )                    # end group
    ''')

u = u'abc\ud834\ud82a\udfcdxyz'
print repr(u)
b = lone.sub(ur'\ufffd',u).encode('utf8')
print repr(b)
print repr(b.decode('utf8'))

Output:

u'abc\ud834\U0001abcdxyz'
'abc\xef\xbf\xbd\xf0\x9a\xaf\x8dxyz'
u'abc\ufffd\U0001abcdxyz'

Answer 2

Here's what fixed this issue for me:

invalid_string.encode('utf16').decode('utf16', 'replace')

My understanding is that surrogate pairs are part of UTF-16, and that's why encoding/decoding with UTF-8 doesn't do anything.