From 2d195c70fd5cd7fe9662e9fb6f43830f4a8306f7 Mon Sep 17 00:00:00 2001
From: Jonas Haag <jonas@lophus.org>
Date: Sun, 6 Jan 2013 22:35:04 +0100
Subject: [PATCH] Fix #43: --site-name should accept non-ASCII values
---
klaus/utils.py | 19 ++++++++++++++-----
1 file changed, 14 insertions(+), 5 deletions(-)
diff --git a/klaus/utils.py b/klaus/utils.py
index e8fc049..1311a5b 100644
--- a/klaus/utils.py
+++ b/klaus/utils.py
@@ -4,6 +4,11 @@ import re
import time
import datetime
import mimetypes
+import locale
+try:
+ import chardet
+except ImportError:
+ chardet = None
from pygments import highlight
from pygments.lexers import get_lexer_for_filename, guess_lexer, ClassNotFound
@@ -140,23 +145,27 @@ def guess_is_image(filename):
def force_unicode(s):
""" Does all kind of magic to turn `s` into unicode """
+ # It's already unicode, don't do anything:
if isinstance(s, unicode):
return s
+
+ # Try some default encodings:
try:
return s.decode('utf-8')
except UnicodeDecodeError as exc:
pass
try:
- return s.decode('iso-8859-1')
+ return s.decode(locale.getpreferredencoding())
except UnicodeDecodeError:
pass
- try:
- import chardet
+
+ if chardet is not None:
+ # Try chardet, if available
encoding = chardet.detect(s)['encoding']
if encoding is not None:
return s.decode(encoding)
- except (ImportError, UnicodeDecodeError):
- raise exc
+
+ raise exc # Give up.
def extract_author_name(email):
--
GitLab