Mercurial > repos > shellac > guppy_basecaller

annotate env/lib/python3.7/site-packages/chardet/chardistribution.py @ 3:758bc20232e8 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit 2a0fe2cc28b09e101d37293e53e82f61762262ec"
author shellac
date Thu, 14 May 2020 16:20:52 -0400
parents 26e78fe6e8c4
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1 ######################## BEGIN LICENSE BLOCK ########################
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2 # The Original Code is Mozilla Communicator client code.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
3 #
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
4 # The Initial Developer of the Original Code is
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
5 # Netscape Communications Corporation.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
6 # Portions created by the Initial Developer are Copyright (C) 1998
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
7 # the Initial Developer. All Rights Reserved.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
8 #
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
9 # Contributor(s):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
10 # Mark Pilgrim - port to Python
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
11 #
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
12 # This library is free software; you can redistribute it and/or
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
13 # modify it under the terms of the GNU Lesser General Public
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
14 # License as published by the Free Software Foundation; either
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
15 # version 2.1 of the License, or (at your option) any later version.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
16 #
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
17 # This library is distributed in the hope that it will be useful,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
18 # but WITHOUT ANY WARRANTY; without even the implied warranty of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
19 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
20 # Lesser General Public License for more details.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
21 #
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
22 # You should have received a copy of the GNU Lesser General Public
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
23 # License along with this library; if not, write to the Free Software
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
24 # Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
25 # 02110-1301 USA
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
26 ######################### END LICENSE BLOCK #########################
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
27
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
28 from .euctwfreq import (EUCTW_CHAR_TO_FREQ_ORDER, EUCTW_TABLE_SIZE,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
29 EUCTW_TYPICAL_DISTRIBUTION_RATIO)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
30 from .euckrfreq import (EUCKR_CHAR_TO_FREQ_ORDER, EUCKR_TABLE_SIZE,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
31 EUCKR_TYPICAL_DISTRIBUTION_RATIO)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
32 from .gb2312freq import (GB2312_CHAR_TO_FREQ_ORDER, GB2312_TABLE_SIZE,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
33 GB2312_TYPICAL_DISTRIBUTION_RATIO)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
34 from .big5freq import (BIG5_CHAR_TO_FREQ_ORDER, BIG5_TABLE_SIZE,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
35 BIG5_TYPICAL_DISTRIBUTION_RATIO)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
36 from .jisfreq import (JIS_CHAR_TO_FREQ_ORDER, JIS_TABLE_SIZE,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
37 JIS_TYPICAL_DISTRIBUTION_RATIO)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
38
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
39
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
40 class CharDistributionAnalysis(object):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
41 ENOUGH_DATA_THRESHOLD = 1024
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
42 SURE_YES = 0.99
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
43 SURE_NO = 0.01
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
44 MINIMUM_DATA_THRESHOLD = 3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
45
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
46 def __init__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
47 # Mapping table to get frequency order from char order (get from
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
48 # GetOrder())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
49 self._char_to_freq_order = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
50 self._table_size = None # Size of above table
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
51 # This is a constant value which varies from language to language,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
52 # used in calculating confidence. See
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
53 # http://www.mozilla.org/projects/intl/UniversalCharsetDetection.html
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
54 # for further detail.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
55 self.typical_distribution_ratio = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
56 self._done = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
57 self._total_chars = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
58 self._freq_chars = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
59 self.reset()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
60
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
61 def reset(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
62 """reset analyser, clear any state"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
63 # If this flag is set to True, detection is done and conclusion has
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
64 # been made
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
65 self._done = False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
66 self._total_chars = 0 # Total characters encountered
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
67 # The number of characters whose frequency order is less than 512
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
68 self._freq_chars = 0
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
69
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
70 def feed(self, char, char_len):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
71 """feed a character with known length"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
72 if char_len == 2:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
73 # we only care about 2-bytes character in our distribution analysis
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
74 order = self.get_order(char)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
75 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
76 order = -1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
77 if order >= 0:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
78 self._total_chars += 1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
79 # order is valid
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
80 if order < self._table_size:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
81 if 512 > self._char_to_freq_order[order]:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
82 self._freq_chars += 1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
83
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
84 def get_confidence(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
85 """return confidence based on existing data"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
86 # if we didn't receive any character in our consideration range,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
87 # return negative answer
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
88 if self._total_chars <= 0 or self._freq_chars <= self.MINIMUM_DATA_THRESHOLD:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
89 return self.SURE_NO
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
90
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
91 if self._total_chars != self._freq_chars:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
92 r = (self._freq_chars / ((self._total_chars - self._freq_chars)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
93 * self.typical_distribution_ratio))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
94 if r < self.SURE_YES:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
95 return r
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
96
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
97 # normalize confidence (we don't want to be 100% sure)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
98 return self.SURE_YES
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
99
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
100 def got_enough_data(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
101 # It is not necessary to receive all data to draw conclusion.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
102 # For charset detection, certain amount of data is enough
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
103 return self._total_chars > self.ENOUGH_DATA_THRESHOLD
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
104
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
105 def get_order(self, byte_str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
106 # We do not handle characters based on the original encoding string,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
107 # but convert this encoding string to a number, here called order.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
108 # This allows multiple encodings of a language to share one frequency
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
109 # table.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
110 return -1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
111
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
112
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
113 class EUCTWDistributionAnalysis(CharDistributionAnalysis):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
114 def __init__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
115 super(EUCTWDistributionAnalysis, self).__init__()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
116 self._char_to_freq_order = EUCTW_CHAR_TO_FREQ_ORDER
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
117 self._table_size = EUCTW_TABLE_SIZE
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
118 self.typical_distribution_ratio = EUCTW_TYPICAL_DISTRIBUTION_RATIO
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
119
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
120 def get_order(self, byte_str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
121 # for euc-TW encoding, we are interested
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
122 # first byte range: 0xc4 -- 0xfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
123 # second byte range: 0xa1 -- 0xfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
124 # no validation needed here. State machine has done that
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
125 first_char = byte_str[0]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
126 if first_char >= 0xC4:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
127 return 94 * (first_char - 0xC4) + byte_str[1] - 0xA1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
128 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
129 return -1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
130
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
131
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
132 class EUCKRDistributionAnalysis(CharDistributionAnalysis):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
133 def __init__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
134 super(EUCKRDistributionAnalysis, self).__init__()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
135 self._char_to_freq_order = EUCKR_CHAR_TO_FREQ_ORDER
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
136 self._table_size = EUCKR_TABLE_SIZE
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
137 self.typical_distribution_ratio = EUCKR_TYPICAL_DISTRIBUTION_RATIO
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
138
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
139 def get_order(self, byte_str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
140 # for euc-KR encoding, we are interested
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
141 # first byte range: 0xb0 -- 0xfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
142 # second byte range: 0xa1 -- 0xfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
143 # no validation needed here. State machine has done that
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
144 first_char = byte_str[0]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
145 if first_char >= 0xB0:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
146 return 94 * (first_char - 0xB0) + byte_str[1] - 0xA1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
147 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
148 return -1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
149
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
150
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
151 class GB2312DistributionAnalysis(CharDistributionAnalysis):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
152 def __init__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
153 super(GB2312DistributionAnalysis, self).__init__()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
154 self._char_to_freq_order = GB2312_CHAR_TO_FREQ_ORDER
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
155 self._table_size = GB2312_TABLE_SIZE
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
156 self.typical_distribution_ratio = GB2312_TYPICAL_DISTRIBUTION_RATIO
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
157
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
158 def get_order(self, byte_str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
159 # for GB2312 encoding, we are interested
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
160 # first byte range: 0xb0 -- 0xfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
161 # second byte range: 0xa1 -- 0xfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
162 # no validation needed here. State machine has done that
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
163 first_char, second_char = byte_str[0], byte_str[1]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
164 if (first_char >= 0xB0) and (second_char >= 0xA1):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
165 return 94 * (first_char - 0xB0) + second_char - 0xA1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
166 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
167 return -1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
168
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
169
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
170 class Big5DistributionAnalysis(CharDistributionAnalysis):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
171 def __init__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
172 super(Big5DistributionAnalysis, self).__init__()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
173 self._char_to_freq_order = BIG5_CHAR_TO_FREQ_ORDER
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
174 self._table_size = BIG5_TABLE_SIZE
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
175 self.typical_distribution_ratio = BIG5_TYPICAL_DISTRIBUTION_RATIO
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
176
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
177 def get_order(self, byte_str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
178 # for big5 encoding, we are interested
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
179 # first byte range: 0xa4 -- 0xfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
180 # second byte range: 0x40 -- 0x7e , 0xa1 -- 0xfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
181 # no validation needed here. State machine has done that
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
182 first_char, second_char = byte_str[0], byte_str[1]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
183 if first_char >= 0xA4:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
184 if second_char >= 0xA1:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
185 return 157 * (first_char - 0xA4) + second_char - 0xA1 + 63
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
186 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
187 return 157 * (first_char - 0xA4) + second_char - 0x40
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
188 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
189 return -1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
190
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
191
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
192 class SJISDistributionAnalysis(CharDistributionAnalysis):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
193 def __init__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
194 super(SJISDistributionAnalysis, self).__init__()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
195 self._char_to_freq_order = JIS_CHAR_TO_FREQ_ORDER
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
196 self._table_size = JIS_TABLE_SIZE
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
197 self.typical_distribution_ratio = JIS_TYPICAL_DISTRIBUTION_RATIO
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
198
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
199 def get_order(self, byte_str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
200 # for sjis encoding, we are interested
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
201 # first byte range: 0x81 -- 0x9f , 0xe0 -- 0xfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
202 # second byte range: 0x40 -- 0x7e, 0x81 -- oxfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
203 # no validation needed here. State machine has done that
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
204 first_char, second_char = byte_str[0], byte_str[1]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
205 if (first_char >= 0x81) and (first_char <= 0x9F):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
206 order = 188 * (first_char - 0x81)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
207 elif (first_char >= 0xE0) and (first_char <= 0xEF):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
208 order = 188 * (first_char - 0xE0 + 31)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
209 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
210 return -1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
211 order = order + second_char - 0x40
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
212 if second_char > 0x7F:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
213 order = -1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
214 return order
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
215
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
216
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
217 class EUCJPDistributionAnalysis(CharDistributionAnalysis):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
218 def __init__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
219 super(EUCJPDistributionAnalysis, self).__init__()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
220 self._char_to_freq_order = JIS_CHAR_TO_FREQ_ORDER
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
221 self._table_size = JIS_TABLE_SIZE
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
222 self.typical_distribution_ratio = JIS_TYPICAL_DISTRIBUTION_RATIO
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
223
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
224 def get_order(self, byte_str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
225 # for euc-JP encoding, we are interested
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
226 # first byte range: 0xa0 -- 0xfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
227 # second byte range: 0xa1 -- 0xfe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
228 # no validation needed here. State machine has done that
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
229 char = byte_str[0]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
230 if char >= 0xA0:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
231 return 94 * (char - 0xA1) + byte_str[1] - 0xa1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
232 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
233 return -1