Mercurial > repos > bcclaywell > argo_navis
annotate venv/lib/python2.7/site-packages/requests/packages/chardet/latin1prober.py @ 0:d67268158946 draft
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
author | bcclaywell |
---|---|
date | Mon, 12 Oct 2015 17:43:33 -0400 |
parents | |
children |
rev | line source |
---|---|
0
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
1 ######################## BEGIN LICENSE BLOCK ######################## |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
2 # The Original Code is Mozilla Universal charset detector code. |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
3 # |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
4 # The Initial Developer of the Original Code is |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
5 # Netscape Communications Corporation. |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
6 # Portions created by the Initial Developer are Copyright (C) 2001 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
7 # the Initial Developer. All Rights Reserved. |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
8 # |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
9 # Contributor(s): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
10 # Mark Pilgrim - port to Python |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
11 # Shy Shalom - original C code |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
12 # |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
13 # This library is free software; you can redistribute it and/or |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
14 # modify it under the terms of the GNU Lesser General Public |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
15 # License as published by the Free Software Foundation; either |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
16 # version 2.1 of the License, or (at your option) any later version. |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
17 # |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
18 # This library is distributed in the hope that it will be useful, |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
19 # but WITHOUT ANY WARRANTY; without even the implied warranty of |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
20 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
21 # Lesser General Public License for more details. |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
22 # |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
23 # You should have received a copy of the GNU Lesser General Public |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
24 # License along with this library; if not, write to the Free Software |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
25 # Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
26 # 02110-1301 USA |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
27 ######################### END LICENSE BLOCK ######################### |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
28 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
29 from .charsetprober import CharSetProber |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
30 from .constants import eNotMe |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
31 from .compat import wrap_ord |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
32 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
33 FREQ_CAT_NUM = 4 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
34 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
35 UDF = 0 # undefined |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
36 OTH = 1 # other |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
37 ASC = 2 # ascii capital letter |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
38 ASS = 3 # ascii small letter |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
39 ACV = 4 # accent capital vowel |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
40 ACO = 5 # accent capital other |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
41 ASV = 6 # accent small vowel |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
42 ASO = 7 # accent small other |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
43 CLASS_NUM = 8 # total classes |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
44 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
45 Latin1_CharToClass = ( |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
46 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # 00 - 07 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
47 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # 08 - 0F |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
48 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # 10 - 17 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
49 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # 18 - 1F |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
50 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # 20 - 27 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
51 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # 28 - 2F |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
52 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # 30 - 37 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
53 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # 38 - 3F |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
54 OTH, ASC, ASC, ASC, ASC, ASC, ASC, ASC, # 40 - 47 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
55 ASC, ASC, ASC, ASC, ASC, ASC, ASC, ASC, # 48 - 4F |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
56 ASC, ASC, ASC, ASC, ASC, ASC, ASC, ASC, # 50 - 57 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
57 ASC, ASC, ASC, OTH, OTH, OTH, OTH, OTH, # 58 - 5F |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
58 OTH, ASS, ASS, ASS, ASS, ASS, ASS, ASS, # 60 - 67 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
59 ASS, ASS, ASS, ASS, ASS, ASS, ASS, ASS, # 68 - 6F |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
60 ASS, ASS, ASS, ASS, ASS, ASS, ASS, ASS, # 70 - 77 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
61 ASS, ASS, ASS, OTH, OTH, OTH, OTH, OTH, # 78 - 7F |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
62 OTH, UDF, OTH, ASO, OTH, OTH, OTH, OTH, # 80 - 87 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
63 OTH, OTH, ACO, OTH, ACO, UDF, ACO, UDF, # 88 - 8F |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
64 UDF, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # 90 - 97 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
65 OTH, OTH, ASO, OTH, ASO, UDF, ASO, ACO, # 98 - 9F |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
66 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # A0 - A7 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
67 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # A8 - AF |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
68 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # B0 - B7 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
69 OTH, OTH, OTH, OTH, OTH, OTH, OTH, OTH, # B8 - BF |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
70 ACV, ACV, ACV, ACV, ACV, ACV, ACO, ACO, # C0 - C7 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
71 ACV, ACV, ACV, ACV, ACV, ACV, ACV, ACV, # C8 - CF |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
72 ACO, ACO, ACV, ACV, ACV, ACV, ACV, OTH, # D0 - D7 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
73 ACV, ACV, ACV, ACV, ACV, ACO, ACO, ACO, # D8 - DF |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
74 ASV, ASV, ASV, ASV, ASV, ASV, ASO, ASO, # E0 - E7 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
75 ASV, ASV, ASV, ASV, ASV, ASV, ASV, ASV, # E8 - EF |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
76 ASO, ASO, ASV, ASV, ASV, ASV, ASV, OTH, # F0 - F7 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
77 ASV, ASV, ASV, ASV, ASV, ASO, ASO, ASO, # F8 - FF |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
78 ) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
79 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
80 # 0 : illegal |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
81 # 1 : very unlikely |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
82 # 2 : normal |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
83 # 3 : very likely |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
84 Latin1ClassModel = ( |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
85 # UDF OTH ASC ASS ACV ACO ASV ASO |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
86 0, 0, 0, 0, 0, 0, 0, 0, # UDF |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
87 0, 3, 3, 3, 3, 3, 3, 3, # OTH |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
88 0, 3, 3, 3, 3, 3, 3, 3, # ASC |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
89 0, 3, 3, 3, 1, 1, 3, 3, # ASS |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
90 0, 3, 3, 3, 1, 2, 1, 2, # ACV |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
91 0, 3, 3, 3, 3, 3, 3, 3, # ACO |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
92 0, 3, 1, 3, 1, 1, 1, 3, # ASV |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
93 0, 3, 1, 3, 1, 1, 3, 3, # ASO |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
94 ) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
95 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
96 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
97 class Latin1Prober(CharSetProber): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
98 def __init__(self): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
99 CharSetProber.__init__(self) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
100 self.reset() |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
101 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
102 def reset(self): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
103 self._mLastCharClass = OTH |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
104 self._mFreqCounter = [0] * FREQ_CAT_NUM |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
105 CharSetProber.reset(self) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
106 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
107 def get_charset_name(self): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
108 return "windows-1252" |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
109 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
110 def feed(self, aBuf): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
111 aBuf = self.filter_with_english_letters(aBuf) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
112 for c in aBuf: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
113 charClass = Latin1_CharToClass[wrap_ord(c)] |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
114 freq = Latin1ClassModel[(self._mLastCharClass * CLASS_NUM) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
115 + charClass] |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
116 if freq == 0: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
117 self._mState = eNotMe |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
118 break |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
119 self._mFreqCounter[freq] += 1 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
120 self._mLastCharClass = charClass |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
121 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
122 return self.get_state() |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
123 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
124 def get_confidence(self): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
125 if self.get_state() == eNotMe: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
126 return 0.01 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
127 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
128 total = sum(self._mFreqCounter) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
129 if total < 0.01: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
130 confidence = 0.0 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
131 else: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
132 confidence = ((self._mFreqCounter[3] - self._mFreqCounter[1] * 20.0) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
133 / total) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
134 if confidence < 0.0: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
135 confidence = 0.0 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
136 # lower the confidence of latin1 so that other more accurate |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
137 # detector can take priority. |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
138 confidence = confidence * 0.73 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
139 return confidence |