Mercurial > repos > bcclaywell > argo_navis

annotate venv/lib/python2.7/site-packages/requests/packages/chardet/sbcharsetprober.py @ 0:d67268158946 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
author bcclaywell
date Mon, 12 Oct 2015 17:43:33 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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 import sys
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
30 from . import constants
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
31 from .charsetprober import CharSetProber
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
32 from .compat import wrap_ord
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
33
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
34 SAMPLE_SIZE = 64
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
35 SB_ENOUGH_REL_THRESHOLD = 1024
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
36 POSITIVE_SHORTCUT_THRESHOLD = 0.95
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
37 NEGATIVE_SHORTCUT_THRESHOLD = 0.05
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
38 SYMBOL_CAT_ORDER = 250
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
39 NUMBER_OF_SEQ_CAT = 4
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
40 POSITIVE_CAT = NUMBER_OF_SEQ_CAT - 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
41 #NEGATIVE_CAT = 0
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
42
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
43
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
44 class SingleByteCharSetProber(CharSetProber):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
45 def __init__(self, model, reversed=False, nameProber=None):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
46 CharSetProber.__init__(self)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
47 self._mModel = model
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
48 # TRUE if we need to reverse every pair in the model lookup
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
49 self._mReversed = reversed
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
50 # Optional auxiliary prober for name decision
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
51 self._mNameProber = nameProber
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
52 self.reset()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
53
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
54 def reset(self):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
55 CharSetProber.reset(self)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
56 # char order of last character
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
57 self._mLastOrder = 255
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
58 self._mSeqCounters = [0] * NUMBER_OF_SEQ_CAT
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
59 self._mTotalSeqs = 0
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
60 self._mTotalChar = 0
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
61 # characters that fall in our sampling range
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
62 self._mFreqChar = 0
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
63
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
64 def get_charset_name(self):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
65 if self._mNameProber:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
66 return self._mNameProber.get_charset_name()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
67 else:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
68 return self._mModel['charsetName']
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
69
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
70 def feed(self, aBuf):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
71 if not self._mModel['keepEnglishLetter']:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
72 aBuf = self.filter_without_english_letters(aBuf)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
73 aLen = len(aBuf)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
74 if not aLen:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
75 return self.get_state()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
76 for c in aBuf:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
77 order = self._mModel['charToOrderMap'][wrap_ord(c)]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
78 if order < SYMBOL_CAT_ORDER:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
79 self._mTotalChar += 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
80 if order < SAMPLE_SIZE:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
81 self._mFreqChar += 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
82 if self._mLastOrder < SAMPLE_SIZE:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
83 self._mTotalSeqs += 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
84 if not self._mReversed:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
85 i = (self._mLastOrder * SAMPLE_SIZE) + order
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
86 model = self._mModel['precedenceMatrix'][i]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
87 else: # reverse the order of the letters in the lookup
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
88 i = (order * SAMPLE_SIZE) + self._mLastOrder
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
89 model = self._mModel['precedenceMatrix'][i]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
90 self._mSeqCounters[model] += 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
91 self._mLastOrder = order
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
92
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
93 if self.get_state() == constants.eDetecting:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
94 if self._mTotalSeqs > SB_ENOUGH_REL_THRESHOLD:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
95 cf = self.get_confidence()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
96 if cf > POSITIVE_SHORTCUT_THRESHOLD:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
97 if constants._debug:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
98 sys.stderr.write('%s confidence = %s, we have a'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
99 'winner\n' %
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
100 (self._mModel['charsetName'], cf))
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
101 self._mState = constants.eFoundIt
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
102 elif cf < NEGATIVE_SHORTCUT_THRESHOLD:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
103 if constants._debug:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
104 sys.stderr.write('%s confidence = %s, below negative'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
105 'shortcut threshhold %s\n' %
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
106 (self._mModel['charsetName'], cf,
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
107 NEGATIVE_SHORTCUT_THRESHOLD))
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
108 self._mState = constants.eNotMe
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
109
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
110 return self.get_state()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
111
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
112 def get_confidence(self):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
113 r = 0.01
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
114 if self._mTotalSeqs > 0:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
115 r = ((1.0 * self._mSeqCounters[POSITIVE_CAT]) / self._mTotalSeqs
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
116 / self._mModel['mTypicalPositiveRatio'])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
117 r = r * self._mFreqChar / self._mTotalChar
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
118 if r >= 1.0:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
119 r = 0.99
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
120 return r