| 1 |
73
|
aaronmk
|
# String manipulation
|
| 2 |
|
|
|
| 3 |
|
|
def to_unicode(str_):
|
| 4 |
|
|
if isinstance(str_, unicode): return str_
|
| 5 |
|
|
encodings = ['utf_8', 'latin_1']
|
| 6 |
|
|
for encoding in encodings:
|
| 7 |
|
|
try: return unicode(str_, encoding)
|
| 8 |
|
|
except UnicodeDecodeError, e: pass
|
| 9 |
|
|
raise AssertionError(encoding+' is not a catch-all encoding')
|
| 10 |
340
|
aaronmk
|
|
| 11 |
|
|
def ensure_newl(str_): return str_.rstrip()+'\n'
|
| 12 |
714
|
aaronmk
|
|
| 13 |
|
|
def std_newl(str_): return str_.replace('\r\n', '\n').replace('\r', '\n')
|
| 14 |
|
|
|
| 15 |
|
|
def cleanup(str_): return std_newl(str_.strip())
|
