Fix #43: --site-name should accept non-ASCII values
Jonas Haag
9 years ago
| 3 | 3 | import time |
| 4 | 4 | import datetime |
| 5 | 5 | import mimetypes |
| 6 | import locale | |
| 7 | try: | |
| 8 | import chardet | |
| 9 | except ImportError: | |
| 10 | chardet = None | |
| 6 | 11 | |
| 7 | 12 | from pygments import highlight |
| 8 | 13 | from pygments.lexers import get_lexer_for_filename, guess_lexer, ClassNotFound |
| 139 | 144 | |
| 140 | 145 | def force_unicode(s): |
| 141 | 146 | """ Does all kind of magic to turn `s` into unicode """ |
| 147 | # It's already unicode, don't do anything: | |
| 142 | 148 | if isinstance(s, unicode): |
| 143 | 149 | return s |
| 150 | ||
| 151 | # Try some default encodings: | |
| 144 | 152 | try: |
| 145 | 153 | return s.decode('utf-8') |
| 146 | 154 | except UnicodeDecodeError as exc: |
| 147 | 155 | pass |
| 148 | 156 | try: |
| 149 | return s.decode('iso-8859-1') | |
| 157 | return s.decode(locale.getpreferredencoding()) | |
| 150 | 158 | except UnicodeDecodeError: |
| 151 | 159 | pass |
| 152 | try: | |
| 153 | import chardet | |
| 160 | ||
| 161 | if chardet is not None: | |
| 162 | # Try chardet, if available | |
| 154 | 163 | encoding = chardet.detect(s)['encoding'] |
| 155 | 164 | if encoding is not None: |
| 156 | 165 | return s.decode(encoding) |
| 157 | except (ImportError, UnicodeDecodeError): | |
| 158 | raise exc | |
| 166 | ||
| 167 | raise exc # Give up. | |
| 159 | 168 | |
| 160 | 169 | |
| 161 | 170 | def extract_author_name(email): |