• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1""" Python Character Mapping Codec tis_620 generated from 'python-mappings/TIS-620.TXT' with gencodec.py.
2
3"""#"
4
5import codecs
6
7### Codec APIs
8
9class Codec(codecs.Codec):
10
11    def encode(self,input,errors='strict'):
12        return codecs.charmap_encode(input,errors,encoding_table)
13
14    def decode(self,input,errors='strict'):
15        return codecs.charmap_decode(input,errors,decoding_table)
16
17class IncrementalEncoder(codecs.IncrementalEncoder):
18    def encode(self, input, final=False):
19        return codecs.charmap_encode(input,self.errors,encoding_table)[0]
20
21class IncrementalDecoder(codecs.IncrementalDecoder):
22    def decode(self, input, final=False):
23        return codecs.charmap_decode(input,self.errors,decoding_table)[0]
24
25class StreamWriter(Codec,codecs.StreamWriter):
26    pass
27
28class StreamReader(Codec,codecs.StreamReader):
29    pass
30
31### encodings module API
32
33def getregentry():
34    return codecs.CodecInfo(
35        name='tis-620',
36        encode=Codec().encode,
37        decode=Codec().decode,
38        incrementalencoder=IncrementalEncoder,
39        incrementaldecoder=IncrementalDecoder,
40        streamreader=StreamReader,
41        streamwriter=StreamWriter,
42    )
43
44
45### Decoding Table
46
47decoding_table = (
48    '\x00'     #  0x00 -> NULL
49    '\x01'     #  0x01 -> START OF HEADING
50    '\x02'     #  0x02 -> START OF TEXT
51    '\x03'     #  0x03 -> END OF TEXT
52    '\x04'     #  0x04 -> END OF TRANSMISSION
53    '\x05'     #  0x05 -> ENQUIRY
54    '\x06'     #  0x06 -> ACKNOWLEDGE
55    '\x07'     #  0x07 -> BELL
56    '\x08'     #  0x08 -> BACKSPACE
57    '\t'       #  0x09 -> HORIZONTAL TABULATION
58    '\n'       #  0x0A -> LINE FEED
59    '\x0b'     #  0x0B -> VERTICAL TABULATION
60    '\x0c'     #  0x0C -> FORM FEED
61    '\r'       #  0x0D -> CARRIAGE RETURN
62    '\x0e'     #  0x0E -> SHIFT OUT
63    '\x0f'     #  0x0F -> SHIFT IN
64    '\x10'     #  0x10 -> DATA LINK ESCAPE
65    '\x11'     #  0x11 -> DEVICE CONTROL ONE
66    '\x12'     #  0x12 -> DEVICE CONTROL TWO
67    '\x13'     #  0x13 -> DEVICE CONTROL THREE
68    '\x14'     #  0x14 -> DEVICE CONTROL FOUR
69    '\x15'     #  0x15 -> NEGATIVE ACKNOWLEDGE
70    '\x16'     #  0x16 -> SYNCHRONOUS IDLE
71    '\x17'     #  0x17 -> END OF TRANSMISSION BLOCK
72    '\x18'     #  0x18 -> CANCEL
73    '\x19'     #  0x19 -> END OF MEDIUM
74    '\x1a'     #  0x1A -> SUBSTITUTE
75    '\x1b'     #  0x1B -> ESCAPE
76    '\x1c'     #  0x1C -> FILE SEPARATOR
77    '\x1d'     #  0x1D -> GROUP SEPARATOR
78    '\x1e'     #  0x1E -> RECORD SEPARATOR
79    '\x1f'     #  0x1F -> UNIT SEPARATOR
80    ' '        #  0x20 -> SPACE
81    '!'        #  0x21 -> EXCLAMATION MARK
82    '"'        #  0x22 -> QUOTATION MARK
83    '#'        #  0x23 -> NUMBER SIGN
84    '$'        #  0x24 -> DOLLAR SIGN
85    '%'        #  0x25 -> PERCENT SIGN
86    '&'        #  0x26 -> AMPERSAND
87    "'"        #  0x27 -> APOSTROPHE
88    '('        #  0x28 -> LEFT PARENTHESIS
89    ')'        #  0x29 -> RIGHT PARENTHESIS
90    '*'        #  0x2A -> ASTERISK
91    '+'        #  0x2B -> PLUS SIGN
92    ','        #  0x2C -> COMMA
93    '-'        #  0x2D -> HYPHEN-MINUS
94    '.'        #  0x2E -> FULL STOP
95    '/'        #  0x2F -> SOLIDUS
96    '0'        #  0x30 -> DIGIT ZERO
97    '1'        #  0x31 -> DIGIT ONE
98    '2'        #  0x32 -> DIGIT TWO
99    '3'        #  0x33 -> DIGIT THREE
100    '4'        #  0x34 -> DIGIT FOUR
101    '5'        #  0x35 -> DIGIT FIVE
102    '6'        #  0x36 -> DIGIT SIX
103    '7'        #  0x37 -> DIGIT SEVEN
104    '8'        #  0x38 -> DIGIT EIGHT
105    '9'        #  0x39 -> DIGIT NINE
106    ':'        #  0x3A -> COLON
107    ';'        #  0x3B -> SEMICOLON
108    '<'        #  0x3C -> LESS-THAN SIGN
109    '='        #  0x3D -> EQUALS SIGN
110    '>'        #  0x3E -> GREATER-THAN SIGN
111    '?'        #  0x3F -> QUESTION MARK
112    '@'        #  0x40 -> COMMERCIAL AT
113    'A'        #  0x41 -> LATIN CAPITAL LETTER A
114    'B'        #  0x42 -> LATIN CAPITAL LETTER B
115    'C'        #  0x43 -> LATIN CAPITAL LETTER C
116    'D'        #  0x44 -> LATIN CAPITAL LETTER D
117    'E'        #  0x45 -> LATIN CAPITAL LETTER E
118    'F'        #  0x46 -> LATIN CAPITAL LETTER F
119    'G'        #  0x47 -> LATIN CAPITAL LETTER G
120    'H'        #  0x48 -> LATIN CAPITAL LETTER H
121    'I'        #  0x49 -> LATIN CAPITAL LETTER I
122    'J'        #  0x4A -> LATIN CAPITAL LETTER J
123    'K'        #  0x4B -> LATIN CAPITAL LETTER K
124    'L'        #  0x4C -> LATIN CAPITAL LETTER L
125    'M'        #  0x4D -> LATIN CAPITAL LETTER M
126    'N'        #  0x4E -> LATIN CAPITAL LETTER N
127    'O'        #  0x4F -> LATIN CAPITAL LETTER O
128    'P'        #  0x50 -> LATIN CAPITAL LETTER P
129    'Q'        #  0x51 -> LATIN CAPITAL LETTER Q
130    'R'        #  0x52 -> LATIN CAPITAL LETTER R
131    'S'        #  0x53 -> LATIN CAPITAL LETTER S
132    'T'        #  0x54 -> LATIN CAPITAL LETTER T
133    'U'        #  0x55 -> LATIN CAPITAL LETTER U
134    'V'        #  0x56 -> LATIN CAPITAL LETTER V
135    'W'        #  0x57 -> LATIN CAPITAL LETTER W
136    'X'        #  0x58 -> LATIN CAPITAL LETTER X
137    'Y'        #  0x59 -> LATIN CAPITAL LETTER Y
138    'Z'        #  0x5A -> LATIN CAPITAL LETTER Z
139    '['        #  0x5B -> LEFT SQUARE BRACKET
140    '\\'       #  0x5C -> REVERSE SOLIDUS
141    ']'        #  0x5D -> RIGHT SQUARE BRACKET
142    '^'        #  0x5E -> CIRCUMFLEX ACCENT
143    '_'        #  0x5F -> LOW LINE
144    '`'        #  0x60 -> GRAVE ACCENT
145    'a'        #  0x61 -> LATIN SMALL LETTER A
146    'b'        #  0x62 -> LATIN SMALL LETTER B
147    'c'        #  0x63 -> LATIN SMALL LETTER C
148    'd'        #  0x64 -> LATIN SMALL LETTER D
149    'e'        #  0x65 -> LATIN SMALL LETTER E
150    'f'        #  0x66 -> LATIN SMALL LETTER F
151    'g'        #  0x67 -> LATIN SMALL LETTER G
152    'h'        #  0x68 -> LATIN SMALL LETTER H
153    'i'        #  0x69 -> LATIN SMALL LETTER I
154    'j'        #  0x6A -> LATIN SMALL LETTER J
155    'k'        #  0x6B -> LATIN SMALL LETTER K
156    'l'        #  0x6C -> LATIN SMALL LETTER L
157    'm'        #  0x6D -> LATIN SMALL LETTER M
158    'n'        #  0x6E -> LATIN SMALL LETTER N
159    'o'        #  0x6F -> LATIN SMALL LETTER O
160    'p'        #  0x70 -> LATIN SMALL LETTER P
161    'q'        #  0x71 -> LATIN SMALL LETTER Q
162    'r'        #  0x72 -> LATIN SMALL LETTER R
163    's'        #  0x73 -> LATIN SMALL LETTER S
164    't'        #  0x74 -> LATIN SMALL LETTER T
165    'u'        #  0x75 -> LATIN SMALL LETTER U
166    'v'        #  0x76 -> LATIN SMALL LETTER V
167    'w'        #  0x77 -> LATIN SMALL LETTER W
168    'x'        #  0x78 -> LATIN SMALL LETTER X
169    'y'        #  0x79 -> LATIN SMALL LETTER Y
170    'z'        #  0x7A -> LATIN SMALL LETTER Z
171    '{'        #  0x7B -> LEFT CURLY BRACKET
172    '|'        #  0x7C -> VERTICAL LINE
173    '}'        #  0x7D -> RIGHT CURLY BRACKET
174    '~'        #  0x7E -> TILDE
175    '\x7f'     #  0x7F -> DELETE
176    '\x80'     #  0x80 -> <control>
177    '\x81'     #  0x81 -> <control>
178    '\x82'     #  0x82 -> <control>
179    '\x83'     #  0x83 -> <control>
180    '\x84'     #  0x84 -> <control>
181    '\x85'     #  0x85 -> <control>
182    '\x86'     #  0x86 -> <control>
183    '\x87'     #  0x87 -> <control>
184    '\x88'     #  0x88 -> <control>
185    '\x89'     #  0x89 -> <control>
186    '\x8a'     #  0x8A -> <control>
187    '\x8b'     #  0x8B -> <control>
188    '\x8c'     #  0x8C -> <control>
189    '\x8d'     #  0x8D -> <control>
190    '\x8e'     #  0x8E -> <control>
191    '\x8f'     #  0x8F -> <control>
192    '\x90'     #  0x90 -> <control>
193    '\x91'     #  0x91 -> <control>
194    '\x92'     #  0x92 -> <control>
195    '\x93'     #  0x93 -> <control>
196    '\x94'     #  0x94 -> <control>
197    '\x95'     #  0x95 -> <control>
198    '\x96'     #  0x96 -> <control>
199    '\x97'     #  0x97 -> <control>
200    '\x98'     #  0x98 -> <control>
201    '\x99'     #  0x99 -> <control>
202    '\x9a'     #  0x9A -> <control>
203    '\x9b'     #  0x9B -> <control>
204    '\x9c'     #  0x9C -> <control>
205    '\x9d'     #  0x9D -> <control>
206    '\x9e'     #  0x9E -> <control>
207    '\x9f'     #  0x9F -> <control>
208    '\ufffe'
209    '\u0e01'   #  0xA1 -> THAI CHARACTER KO KAI
210    '\u0e02'   #  0xA2 -> THAI CHARACTER KHO KHAI
211    '\u0e03'   #  0xA3 -> THAI CHARACTER KHO KHUAT
212    '\u0e04'   #  0xA4 -> THAI CHARACTER KHO KHWAI
213    '\u0e05'   #  0xA5 -> THAI CHARACTER KHO KHON
214    '\u0e06'   #  0xA6 -> THAI CHARACTER KHO RAKHANG
215    '\u0e07'   #  0xA7 -> THAI CHARACTER NGO NGU
216    '\u0e08'   #  0xA8 -> THAI CHARACTER CHO CHAN
217    '\u0e09'   #  0xA9 -> THAI CHARACTER CHO CHING
218    '\u0e0a'   #  0xAA -> THAI CHARACTER CHO CHANG
219    '\u0e0b'   #  0xAB -> THAI CHARACTER SO SO
220    '\u0e0c'   #  0xAC -> THAI CHARACTER CHO CHOE
221    '\u0e0d'   #  0xAD -> THAI CHARACTER YO YING
222    '\u0e0e'   #  0xAE -> THAI CHARACTER DO CHADA
223    '\u0e0f'   #  0xAF -> THAI CHARACTER TO PATAK
224    '\u0e10'   #  0xB0 -> THAI CHARACTER THO THAN
225    '\u0e11'   #  0xB1 -> THAI CHARACTER THO NANGMONTHO
226    '\u0e12'   #  0xB2 -> THAI CHARACTER THO PHUTHAO
227    '\u0e13'   #  0xB3 -> THAI CHARACTER NO NEN
228    '\u0e14'   #  0xB4 -> THAI CHARACTER DO DEK
229    '\u0e15'   #  0xB5 -> THAI CHARACTER TO TAO
230    '\u0e16'   #  0xB6 -> THAI CHARACTER THO THUNG
231    '\u0e17'   #  0xB7 -> THAI CHARACTER THO THAHAN
232    '\u0e18'   #  0xB8 -> THAI CHARACTER THO THONG
233    '\u0e19'   #  0xB9 -> THAI CHARACTER NO NU
234    '\u0e1a'   #  0xBA -> THAI CHARACTER BO BAIMAI
235    '\u0e1b'   #  0xBB -> THAI CHARACTER PO PLA
236    '\u0e1c'   #  0xBC -> THAI CHARACTER PHO PHUNG
237    '\u0e1d'   #  0xBD -> THAI CHARACTER FO FA
238    '\u0e1e'   #  0xBE -> THAI CHARACTER PHO PHAN
239    '\u0e1f'   #  0xBF -> THAI CHARACTER FO FAN
240    '\u0e20'   #  0xC0 -> THAI CHARACTER PHO SAMPHAO
241    '\u0e21'   #  0xC1 -> THAI CHARACTER MO MA
242    '\u0e22'   #  0xC2 -> THAI CHARACTER YO YAK
243    '\u0e23'   #  0xC3 -> THAI CHARACTER RO RUA
244    '\u0e24'   #  0xC4 -> THAI CHARACTER RU
245    '\u0e25'   #  0xC5 -> THAI CHARACTER LO LING
246    '\u0e26'   #  0xC6 -> THAI CHARACTER LU
247    '\u0e27'   #  0xC7 -> THAI CHARACTER WO WAEN
248    '\u0e28'   #  0xC8 -> THAI CHARACTER SO SALA
249    '\u0e29'   #  0xC9 -> THAI CHARACTER SO RUSI
250    '\u0e2a'   #  0xCA -> THAI CHARACTER SO SUA
251    '\u0e2b'   #  0xCB -> THAI CHARACTER HO HIP
252    '\u0e2c'   #  0xCC -> THAI CHARACTER LO CHULA
253    '\u0e2d'   #  0xCD -> THAI CHARACTER O ANG
254    '\u0e2e'   #  0xCE -> THAI CHARACTER HO NOKHUK
255    '\u0e2f'   #  0xCF -> THAI CHARACTER PAIYANNOI
256    '\u0e30'   #  0xD0 -> THAI CHARACTER SARA A
257    '\u0e31'   #  0xD1 -> THAI CHARACTER MAI HAN-AKAT
258    '\u0e32'   #  0xD2 -> THAI CHARACTER SARA AA
259    '\u0e33'   #  0xD3 -> THAI CHARACTER SARA AM
260    '\u0e34'   #  0xD4 -> THAI CHARACTER SARA I
261    '\u0e35'   #  0xD5 -> THAI CHARACTER SARA II
262    '\u0e36'   #  0xD6 -> THAI CHARACTER SARA UE
263    '\u0e37'   #  0xD7 -> THAI CHARACTER SARA UEE
264    '\u0e38'   #  0xD8 -> THAI CHARACTER SARA U
265    '\u0e39'   #  0xD9 -> THAI CHARACTER SARA UU
266    '\u0e3a'   #  0xDA -> THAI CHARACTER PHINTHU
267    '\ufffe'
268    '\ufffe'
269    '\ufffe'
270    '\ufffe'
271    '\u0e3f'   #  0xDF -> THAI CURRENCY SYMBOL BAHT
272    '\u0e40'   #  0xE0 -> THAI CHARACTER SARA E
273    '\u0e41'   #  0xE1 -> THAI CHARACTER SARA AE
274    '\u0e42'   #  0xE2 -> THAI CHARACTER SARA O
275    '\u0e43'   #  0xE3 -> THAI CHARACTER SARA AI MAIMUAN
276    '\u0e44'   #  0xE4 -> THAI CHARACTER SARA AI MAIMALAI
277    '\u0e45'   #  0xE5 -> THAI CHARACTER LAKKHANGYAO
278    '\u0e46'   #  0xE6 -> THAI CHARACTER MAIYAMOK
279    '\u0e47'   #  0xE7 -> THAI CHARACTER MAITAIKHU
280    '\u0e48'   #  0xE8 -> THAI CHARACTER MAI EK
281    '\u0e49'   #  0xE9 -> THAI CHARACTER MAI THO
282    '\u0e4a'   #  0xEA -> THAI CHARACTER MAI TRI
283    '\u0e4b'   #  0xEB -> THAI CHARACTER MAI CHATTAWA
284    '\u0e4c'   #  0xEC -> THAI CHARACTER THANTHAKHAT
285    '\u0e4d'   #  0xED -> THAI CHARACTER NIKHAHIT
286    '\u0e4e'   #  0xEE -> THAI CHARACTER YAMAKKAN
287    '\u0e4f'   #  0xEF -> THAI CHARACTER FONGMAN
288    '\u0e50'   #  0xF0 -> THAI DIGIT ZERO
289    '\u0e51'   #  0xF1 -> THAI DIGIT ONE
290    '\u0e52'   #  0xF2 -> THAI DIGIT TWO
291    '\u0e53'   #  0xF3 -> THAI DIGIT THREE
292    '\u0e54'   #  0xF4 -> THAI DIGIT FOUR
293    '\u0e55'   #  0xF5 -> THAI DIGIT FIVE
294    '\u0e56'   #  0xF6 -> THAI DIGIT SIX
295    '\u0e57'   #  0xF7 -> THAI DIGIT SEVEN
296    '\u0e58'   #  0xF8 -> THAI DIGIT EIGHT
297    '\u0e59'   #  0xF9 -> THAI DIGIT NINE
298    '\u0e5a'   #  0xFA -> THAI CHARACTER ANGKHANKHU
299    '\u0e5b'   #  0xFB -> THAI CHARACTER KHOMUT
300    '\ufffe'
301    '\ufffe'
302    '\ufffe'
303    '\ufffe'
304)
305
306### Encoding table
307encoding_table=codecs.charmap_build(decoding_table)
308