ThirdParty/CharDet/chardet/test.py

Sat, 02 Jan 2010 15:11:35 +0000

author
Detlev Offenbach <detlev@die-offenbachs.de>
date
Sat, 02 Jan 2010 15:11:35 +0000
changeset 12
1d8dd9706f46
parent 0
de9c2efb9d02
permissions
-rw-r--r--

First commit after changing to Python 3.1.

12
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
1 import sys, glob
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
2 sys.path.insert(0, '..')
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
3 from chardet.universaldetector import UniversalDetector
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
4
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
5 count = 0
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
6 u = UniversalDetector()
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
7 for f in glob.glob(sys.argv[1]):
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
8 print(f.ljust(60), end=' ')
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
9 u.reset()
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
10 for line in open(f, 'rb'):
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
11 u.feed(line)
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
12 if u.done: break
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
13 u.close()
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
14 result = u.result
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
15 if result['encoding']:
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
16 print(result['encoding'], 'with confidence', result['confidence'])
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
17 else:
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
18 print('******** no result')
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
19 count += 1
1d8dd9706f46 First commit after changing to Python 3.1.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 0
diff changeset
20 print(count, 'tests')

eric ide

mercurial