PageRenderTime 442ms CodeModel.GetById 110ms app.highlight 220ms RepoModel.GetById 102ms app.codeStats 1ms

/Lib/encodings/cp864.py

http://unladen-swallow.googlecode.com/
Python | 690 lines | 687 code | 2 blank | 1 comment | 0 complexity | f924f9672b2336896f70571faae68971 MD5 | raw file
  1""" Python Character Mapping Codec generated from 'VENDORS/MICSFT/PC/CP864.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_map)
 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_map)[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='cp864',
 36        encode=Codec().encode,
 37        decode=Codec().decode,
 38        incrementalencoder=IncrementalEncoder,
 39        incrementaldecoder=IncrementalDecoder,
 40        streamreader=StreamReader,
 41        streamwriter=StreamWriter,
 42    )
 43
 44### Decoding Map
 45
 46decoding_map = codecs.make_identity_dict(range(256))
 47decoding_map.update({
 48    0x0025: 0x066a,     #  ARABIC PERCENT SIGN
 49    0x0080: 0x00b0,     #  DEGREE SIGN
 50    0x0081: 0x00b7,     #  MIDDLE DOT
 51    0x0082: 0x2219,     #  BULLET OPERATOR
 52    0x0083: 0x221a,     #  SQUARE ROOT
 53    0x0084: 0x2592,     #  MEDIUM SHADE
 54    0x0085: 0x2500,     #  FORMS LIGHT HORIZONTAL
 55    0x0086: 0x2502,     #  FORMS LIGHT VERTICAL
 56    0x0087: 0x253c,     #  FORMS LIGHT VERTICAL AND HORIZONTAL
 57    0x0088: 0x2524,     #  FORMS LIGHT VERTICAL AND LEFT
 58    0x0089: 0x252c,     #  FORMS LIGHT DOWN AND HORIZONTAL
 59    0x008a: 0x251c,     #  FORMS LIGHT VERTICAL AND RIGHT
 60    0x008b: 0x2534,     #  FORMS LIGHT UP AND HORIZONTAL
 61    0x008c: 0x2510,     #  FORMS LIGHT DOWN AND LEFT
 62    0x008d: 0x250c,     #  FORMS LIGHT DOWN AND RIGHT
 63    0x008e: 0x2514,     #  FORMS LIGHT UP AND RIGHT
 64    0x008f: 0x2518,     #  FORMS LIGHT UP AND LEFT
 65    0x0090: 0x03b2,     #  GREEK SMALL BETA
 66    0x0091: 0x221e,     #  INFINITY
 67    0x0092: 0x03c6,     #  GREEK SMALL PHI
 68    0x0093: 0x00b1,     #  PLUS-OR-MINUS SIGN
 69    0x0094: 0x00bd,     #  FRACTION 1/2
 70    0x0095: 0x00bc,     #  FRACTION 1/4
 71    0x0096: 0x2248,     #  ALMOST EQUAL TO
 72    0x0097: 0x00ab,     #  LEFT POINTING GUILLEMET
 73    0x0098: 0x00bb,     #  RIGHT POINTING GUILLEMET
 74    0x0099: 0xfef7,     #  ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE ISOLATED FORM
 75    0x009a: 0xfef8,     #  ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE FINAL FORM
 76    0x009b: None,       #  UNDEFINED
 77    0x009c: None,       #  UNDEFINED
 78    0x009d: 0xfefb,     #  ARABIC LIGATURE LAM WITH ALEF ISOLATED FORM
 79    0x009e: 0xfefc,     #  ARABIC LIGATURE LAM WITH ALEF FINAL FORM
 80    0x009f: None,       #  UNDEFINED
 81    0x00a1: 0x00ad,     #  SOFT HYPHEN
 82    0x00a2: 0xfe82,     #  ARABIC LETTER ALEF WITH MADDA ABOVE FINAL FORM
 83    0x00a5: 0xfe84,     #  ARABIC LETTER ALEF WITH HAMZA ABOVE FINAL FORM
 84    0x00a6: None,       #  UNDEFINED
 85    0x00a7: None,       #  UNDEFINED
 86    0x00a8: 0xfe8e,     #  ARABIC LETTER ALEF FINAL FORM
 87    0x00a9: 0xfe8f,     #  ARABIC LETTER BEH ISOLATED FORM
 88    0x00aa: 0xfe95,     #  ARABIC LETTER TEH ISOLATED FORM
 89    0x00ab: 0xfe99,     #  ARABIC LETTER THEH ISOLATED FORM
 90    0x00ac: 0x060c,     #  ARABIC COMMA
 91    0x00ad: 0xfe9d,     #  ARABIC LETTER JEEM ISOLATED FORM
 92    0x00ae: 0xfea1,     #  ARABIC LETTER HAH ISOLATED FORM
 93    0x00af: 0xfea5,     #  ARABIC LETTER KHAH ISOLATED FORM
 94    0x00b0: 0x0660,     #  ARABIC-INDIC DIGIT ZERO
 95    0x00b1: 0x0661,     #  ARABIC-INDIC DIGIT ONE
 96    0x00b2: 0x0662,     #  ARABIC-INDIC DIGIT TWO
 97    0x00b3: 0x0663,     #  ARABIC-INDIC DIGIT THREE
 98    0x00b4: 0x0664,     #  ARABIC-INDIC DIGIT FOUR
 99    0x00b5: 0x0665,     #  ARABIC-INDIC DIGIT FIVE
100    0x00b6: 0x0666,     #  ARABIC-INDIC DIGIT SIX
101    0x00b7: 0x0667,     #  ARABIC-INDIC DIGIT SEVEN
102    0x00b8: 0x0668,     #  ARABIC-INDIC DIGIT EIGHT
103    0x00b9: 0x0669,     #  ARABIC-INDIC DIGIT NINE
104    0x00ba: 0xfed1,     #  ARABIC LETTER FEH ISOLATED FORM
105    0x00bb: 0x061b,     #  ARABIC SEMICOLON
106    0x00bc: 0xfeb1,     #  ARABIC LETTER SEEN ISOLATED FORM
107    0x00bd: 0xfeb5,     #  ARABIC LETTER SHEEN ISOLATED FORM
108    0x00be: 0xfeb9,     #  ARABIC LETTER SAD ISOLATED FORM
109    0x00bf: 0x061f,     #  ARABIC QUESTION MARK
110    0x00c0: 0x00a2,     #  CENT SIGN
111    0x00c1: 0xfe80,     #  ARABIC LETTER HAMZA ISOLATED FORM
112    0x00c2: 0xfe81,     #  ARABIC LETTER ALEF WITH MADDA ABOVE ISOLATED FORM
113    0x00c3: 0xfe83,     #  ARABIC LETTER ALEF WITH HAMZA ABOVE ISOLATED FORM
114    0x00c4: 0xfe85,     #  ARABIC LETTER WAW WITH HAMZA ABOVE ISOLATED FORM
115    0x00c5: 0xfeca,     #  ARABIC LETTER AIN FINAL FORM
116    0x00c6: 0xfe8b,     #  ARABIC LETTER YEH WITH HAMZA ABOVE INITIAL FORM
117    0x00c7: 0xfe8d,     #  ARABIC LETTER ALEF ISOLATED FORM
118    0x00c8: 0xfe91,     #  ARABIC LETTER BEH INITIAL FORM
119    0x00c9: 0xfe93,     #  ARABIC LETTER TEH MARBUTA ISOLATED FORM
120    0x00ca: 0xfe97,     #  ARABIC LETTER TEH INITIAL FORM
121    0x00cb: 0xfe9b,     #  ARABIC LETTER THEH INITIAL FORM
122    0x00cc: 0xfe9f,     #  ARABIC LETTER JEEM INITIAL FORM
123    0x00cd: 0xfea3,     #  ARABIC LETTER HAH INITIAL FORM
124    0x00ce: 0xfea7,     #  ARABIC LETTER KHAH INITIAL FORM
125    0x00cf: 0xfea9,     #  ARABIC LETTER DAL ISOLATED FORM
126    0x00d0: 0xfeab,     #  ARABIC LETTER THAL ISOLATED FORM
127    0x00d1: 0xfead,     #  ARABIC LETTER REH ISOLATED FORM
128    0x00d2: 0xfeaf,     #  ARABIC LETTER ZAIN ISOLATED FORM
129    0x00d3: 0xfeb3,     #  ARABIC LETTER SEEN INITIAL FORM
130    0x00d4: 0xfeb7,     #  ARABIC LETTER SHEEN INITIAL FORM
131    0x00d5: 0xfebb,     #  ARABIC LETTER SAD INITIAL FORM
132    0x00d6: 0xfebf,     #  ARABIC LETTER DAD INITIAL FORM
133    0x00d7: 0xfec1,     #  ARABIC LETTER TAH ISOLATED FORM
134    0x00d8: 0xfec5,     #  ARABIC LETTER ZAH ISOLATED FORM
135    0x00d9: 0xfecb,     #  ARABIC LETTER AIN INITIAL FORM
136    0x00da: 0xfecf,     #  ARABIC LETTER GHAIN INITIAL FORM
137    0x00db: 0x00a6,     #  BROKEN VERTICAL BAR
138    0x00dc: 0x00ac,     #  NOT SIGN
139    0x00dd: 0x00f7,     #  DIVISION SIGN
140    0x00de: 0x00d7,     #  MULTIPLICATION SIGN
141    0x00df: 0xfec9,     #  ARABIC LETTER AIN ISOLATED FORM
142    0x00e0: 0x0640,     #  ARABIC TATWEEL
143    0x00e1: 0xfed3,     #  ARABIC LETTER FEH INITIAL FORM
144    0x00e2: 0xfed7,     #  ARABIC LETTER QAF INITIAL FORM
145    0x00e3: 0xfedb,     #  ARABIC LETTER KAF INITIAL FORM
146    0x00e4: 0xfedf,     #  ARABIC LETTER LAM INITIAL FORM
147    0x00e5: 0xfee3,     #  ARABIC LETTER MEEM INITIAL FORM
148    0x00e6: 0xfee7,     #  ARABIC LETTER NOON INITIAL FORM
149    0x00e7: 0xfeeb,     #  ARABIC LETTER HEH INITIAL FORM
150    0x00e8: 0xfeed,     #  ARABIC LETTER WAW ISOLATED FORM
151    0x00e9: 0xfeef,     #  ARABIC LETTER ALEF MAKSURA ISOLATED FORM
152    0x00ea: 0xfef3,     #  ARABIC LETTER YEH INITIAL FORM
153    0x00eb: 0xfebd,     #  ARABIC LETTER DAD ISOLATED FORM
154    0x00ec: 0xfecc,     #  ARABIC LETTER AIN MEDIAL FORM
155    0x00ed: 0xfece,     #  ARABIC LETTER GHAIN FINAL FORM
156    0x00ee: 0xfecd,     #  ARABIC LETTER GHAIN ISOLATED FORM
157    0x00ef: 0xfee1,     #  ARABIC LETTER MEEM ISOLATED FORM
158    0x00f0: 0xfe7d,     #  ARABIC SHADDA MEDIAL FORM
159    0x00f1: 0x0651,     #  ARABIC SHADDAH
160    0x00f2: 0xfee5,     #  ARABIC LETTER NOON ISOLATED FORM
161    0x00f3: 0xfee9,     #  ARABIC LETTER HEH ISOLATED FORM
162    0x00f4: 0xfeec,     #  ARABIC LETTER HEH MEDIAL FORM
163    0x00f5: 0xfef0,     #  ARABIC LETTER ALEF MAKSURA FINAL FORM
164    0x00f6: 0xfef2,     #  ARABIC LETTER YEH FINAL FORM
165    0x00f7: 0xfed0,     #  ARABIC LETTER GHAIN MEDIAL FORM
166    0x00f8: 0xfed5,     #  ARABIC LETTER QAF ISOLATED FORM
167    0x00f9: 0xfef5,     #  ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE ISOLATED FORM
168    0x00fa: 0xfef6,     #  ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE FINAL FORM
169    0x00fb: 0xfedd,     #  ARABIC LETTER LAM ISOLATED FORM
170    0x00fc: 0xfed9,     #  ARABIC LETTER KAF ISOLATED FORM
171    0x00fd: 0xfef1,     #  ARABIC LETTER YEH ISOLATED FORM
172    0x00fe: 0x25a0,     #  BLACK SQUARE
173    0x00ff: None,       #  UNDEFINED
174})
175
176### Decoding Table
177
178decoding_table = (
179    u'\x00'     #  0x0000 -> NULL
180    u'\x01'     #  0x0001 -> START OF HEADING
181    u'\x02'     #  0x0002 -> START OF TEXT
182    u'\x03'     #  0x0003 -> END OF TEXT
183    u'\x04'     #  0x0004 -> END OF TRANSMISSION
184    u'\x05'     #  0x0005 -> ENQUIRY
185    u'\x06'     #  0x0006 -> ACKNOWLEDGE
186    u'\x07'     #  0x0007 -> BELL
187    u'\x08'     #  0x0008 -> BACKSPACE
188    u'\t'       #  0x0009 -> HORIZONTAL TABULATION
189    u'\n'       #  0x000a -> LINE FEED
190    u'\x0b'     #  0x000b -> VERTICAL TABULATION
191    u'\x0c'     #  0x000c -> FORM FEED
192    u'\r'       #  0x000d -> CARRIAGE RETURN
193    u'\x0e'     #  0x000e -> SHIFT OUT
194    u'\x0f'     #  0x000f -> SHIFT IN
195    u'\x10'     #  0x0010 -> DATA LINK ESCAPE
196    u'\x11'     #  0x0011 -> DEVICE CONTROL ONE
197    u'\x12'     #  0x0012 -> DEVICE CONTROL TWO
198    u'\x13'     #  0x0013 -> DEVICE CONTROL THREE
199    u'\x14'     #  0x0014 -> DEVICE CONTROL FOUR
200    u'\x15'     #  0x0015 -> NEGATIVE ACKNOWLEDGE
201    u'\x16'     #  0x0016 -> SYNCHRONOUS IDLE
202    u'\x17'     #  0x0017 -> END OF TRANSMISSION BLOCK
203    u'\x18'     #  0x0018 -> CANCEL
204    u'\x19'     #  0x0019 -> END OF MEDIUM
205    u'\x1a'     #  0x001a -> SUBSTITUTE
206    u'\x1b'     #  0x001b -> ESCAPE
207    u'\x1c'     #  0x001c -> FILE SEPARATOR
208    u'\x1d'     #  0x001d -> GROUP SEPARATOR
209    u'\x1e'     #  0x001e -> RECORD SEPARATOR
210    u'\x1f'     #  0x001f -> UNIT SEPARATOR
211    u' '        #  0x0020 -> SPACE
212    u'!'        #  0x0021 -> EXCLAMATION MARK
213    u'"'        #  0x0022 -> QUOTATION MARK
214    u'#'        #  0x0023 -> NUMBER SIGN
215    u'$'        #  0x0024 -> DOLLAR SIGN
216    u'\u066a'   #  0x0025 -> ARABIC PERCENT SIGN
217    u'&'        #  0x0026 -> AMPERSAND
218    u"'"        #  0x0027 -> APOSTROPHE
219    u'('        #  0x0028 -> LEFT PARENTHESIS
220    u')'        #  0x0029 -> RIGHT PARENTHESIS
221    u'*'        #  0x002a -> ASTERISK
222    u'+'        #  0x002b -> PLUS SIGN
223    u','        #  0x002c -> COMMA
224    u'-'        #  0x002d -> HYPHEN-MINUS
225    u'.'        #  0x002e -> FULL STOP
226    u'/'        #  0x002f -> SOLIDUS
227    u'0'        #  0x0030 -> DIGIT ZERO
228    u'1'        #  0x0031 -> DIGIT ONE
229    u'2'        #  0x0032 -> DIGIT TWO
230    u'3'        #  0x0033 -> DIGIT THREE
231    u'4'        #  0x0034 -> DIGIT FOUR
232    u'5'        #  0x0035 -> DIGIT FIVE
233    u'6'        #  0x0036 -> DIGIT SIX
234    u'7'        #  0x0037 -> DIGIT SEVEN
235    u'8'        #  0x0038 -> DIGIT EIGHT
236    u'9'        #  0x0039 -> DIGIT NINE
237    u':'        #  0x003a -> COLON
238    u';'        #  0x003b -> SEMICOLON
239    u'<'        #  0x003c -> LESS-THAN SIGN
240    u'='        #  0x003d -> EQUALS SIGN
241    u'>'        #  0x003e -> GREATER-THAN SIGN
242    u'?'        #  0x003f -> QUESTION MARK
243    u'@'        #  0x0040 -> COMMERCIAL AT
244    u'A'        #  0x0041 -> LATIN CAPITAL LETTER A
245    u'B'        #  0x0042 -> LATIN CAPITAL LETTER B
246    u'C'        #  0x0043 -> LATIN CAPITAL LETTER C
247    u'D'        #  0x0044 -> LATIN CAPITAL LETTER D
248    u'E'        #  0x0045 -> LATIN CAPITAL LETTER E
249    u'F'        #  0x0046 -> LATIN CAPITAL LETTER F
250    u'G'        #  0x0047 -> LATIN CAPITAL LETTER G
251    u'H'        #  0x0048 -> LATIN CAPITAL LETTER H
252    u'I'        #  0x0049 -> LATIN CAPITAL LETTER I
253    u'J'        #  0x004a -> LATIN CAPITAL LETTER J
254    u'K'        #  0x004b -> LATIN CAPITAL LETTER K
255    u'L'        #  0x004c -> LATIN CAPITAL LETTER L
256    u'M'        #  0x004d -> LATIN CAPITAL LETTER M
257    u'N'        #  0x004e -> LATIN CAPITAL LETTER N
258    u'O'        #  0x004f -> LATIN CAPITAL LETTER O
259    u'P'        #  0x0050 -> LATIN CAPITAL LETTER P
260    u'Q'        #  0x0051 -> LATIN CAPITAL LETTER Q
261    u'R'        #  0x0052 -> LATIN CAPITAL LETTER R
262    u'S'        #  0x0053 -> LATIN CAPITAL LETTER S
263    u'T'        #  0x0054 -> LATIN CAPITAL LETTER T
264    u'U'        #  0x0055 -> LATIN CAPITAL LETTER U
265    u'V'        #  0x0056 -> LATIN CAPITAL LETTER V
266    u'W'        #  0x0057 -> LATIN CAPITAL LETTER W
267    u'X'        #  0x0058 -> LATIN CAPITAL LETTER X
268    u'Y'        #  0x0059 -> LATIN CAPITAL LETTER Y
269    u'Z'        #  0x005a -> LATIN CAPITAL LETTER Z
270    u'['        #  0x005b -> LEFT SQUARE BRACKET
271    u'\\'       #  0x005c -> REVERSE SOLIDUS
272    u']'        #  0x005d -> RIGHT SQUARE BRACKET
273    u'^'        #  0x005e -> CIRCUMFLEX ACCENT
274    u'_'        #  0x005f -> LOW LINE
275    u'`'        #  0x0060 -> GRAVE ACCENT
276    u'a'        #  0x0061 -> LATIN SMALL LETTER A
277    u'b'        #  0x0062 -> LATIN SMALL LETTER B
278    u'c'        #  0x0063 -> LATIN SMALL LETTER C
279    u'd'        #  0x0064 -> LATIN SMALL LETTER D
280    u'e'        #  0x0065 -> LATIN SMALL LETTER E
281    u'f'        #  0x0066 -> LATIN SMALL LETTER F
282    u'g'        #  0x0067 -> LATIN SMALL LETTER G
283    u'h'        #  0x0068 -> LATIN SMALL LETTER H
284    u'i'        #  0x0069 -> LATIN SMALL LETTER I
285    u'j'        #  0x006a -> LATIN SMALL LETTER J
286    u'k'        #  0x006b -> LATIN SMALL LETTER K
287    u'l'        #  0x006c -> LATIN SMALL LETTER L
288    u'm'        #  0x006d -> LATIN SMALL LETTER M
289    u'n'        #  0x006e -> LATIN SMALL LETTER N
290    u'o'        #  0x006f -> LATIN SMALL LETTER O
291    u'p'        #  0x0070 -> LATIN SMALL LETTER P
292    u'q'        #  0x0071 -> LATIN SMALL LETTER Q
293    u'r'        #  0x0072 -> LATIN SMALL LETTER R
294    u's'        #  0x0073 -> LATIN SMALL LETTER S
295    u't'        #  0x0074 -> LATIN SMALL LETTER T
296    u'u'        #  0x0075 -> LATIN SMALL LETTER U
297    u'v'        #  0x0076 -> LATIN SMALL LETTER V
298    u'w'        #  0x0077 -> LATIN SMALL LETTER W
299    u'x'        #  0x0078 -> LATIN SMALL LETTER X
300    u'y'        #  0x0079 -> LATIN SMALL LETTER Y
301    u'z'        #  0x007a -> LATIN SMALL LETTER Z
302    u'{'        #  0x007b -> LEFT CURLY BRACKET
303    u'|'        #  0x007c -> VERTICAL LINE
304    u'}'        #  0x007d -> RIGHT CURLY BRACKET
305    u'~'        #  0x007e -> TILDE
306    u'\x7f'     #  0x007f -> DELETE
307    u'\xb0'     #  0x0080 -> DEGREE SIGN
308    u'\xb7'     #  0x0081 -> MIDDLE DOT
309    u'\u2219'   #  0x0082 -> BULLET OPERATOR
310    u'\u221a'   #  0x0083 -> SQUARE ROOT
311    u'\u2592'   #  0x0084 -> MEDIUM SHADE
312    u'\u2500'   #  0x0085 -> FORMS LIGHT HORIZONTAL
313    u'\u2502'   #  0x0086 -> FORMS LIGHT VERTICAL
314    u'\u253c'   #  0x0087 -> FORMS LIGHT VERTICAL AND HORIZONTAL
315    u'\u2524'   #  0x0088 -> FORMS LIGHT VERTICAL AND LEFT
316    u'\u252c'   #  0x0089 -> FORMS LIGHT DOWN AND HORIZONTAL
317    u'\u251c'   #  0x008a -> FORMS LIGHT VERTICAL AND RIGHT
318    u'\u2534'   #  0x008b -> FORMS LIGHT UP AND HORIZONTAL
319    u'\u2510'   #  0x008c -> FORMS LIGHT DOWN AND LEFT
320    u'\u250c'   #  0x008d -> FORMS LIGHT DOWN AND RIGHT
321    u'\u2514'   #  0x008e -> FORMS LIGHT UP AND RIGHT
322    u'\u2518'   #  0x008f -> FORMS LIGHT UP AND LEFT
323    u'\u03b2'   #  0x0090 -> GREEK SMALL BETA
324    u'\u221e'   #  0x0091 -> INFINITY
325    u'\u03c6'   #  0x0092 -> GREEK SMALL PHI
326    u'\xb1'     #  0x0093 -> PLUS-OR-MINUS SIGN
327    u'\xbd'     #  0x0094 -> FRACTION 1/2
328    u'\xbc'     #  0x0095 -> FRACTION 1/4
329    u'\u2248'   #  0x0096 -> ALMOST EQUAL TO
330    u'\xab'     #  0x0097 -> LEFT POINTING GUILLEMET
331    u'\xbb'     #  0x0098 -> RIGHT POINTING GUILLEMET
332    u'\ufef7'   #  0x0099 -> ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE ISOLATED FORM
333    u'\ufef8'   #  0x009a -> ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE FINAL FORM
334    u'\ufffe'   #  0x009b -> UNDEFINED
335    u'\ufffe'   #  0x009c -> UNDEFINED
336    u'\ufefb'   #  0x009d -> ARABIC LIGATURE LAM WITH ALEF ISOLATED FORM
337    u'\ufefc'   #  0x009e -> ARABIC LIGATURE LAM WITH ALEF FINAL FORM
338    u'\ufffe'   #  0x009f -> UNDEFINED
339    u'\xa0'     #  0x00a0 -> NON-BREAKING SPACE
340    u'\xad'     #  0x00a1 -> SOFT HYPHEN
341    u'\ufe82'   #  0x00a2 -> ARABIC LETTER ALEF WITH MADDA ABOVE FINAL FORM
342    u'\xa3'     #  0x00a3 -> POUND SIGN
343    u'\xa4'     #  0x00a4 -> CURRENCY SIGN
344    u'\ufe84'   #  0x00a5 -> ARABIC LETTER ALEF WITH HAMZA ABOVE FINAL FORM
345    u'\ufffe'   #  0x00a6 -> UNDEFINED
346    u'\ufffe'   #  0x00a7 -> UNDEFINED
347    u'\ufe8e'   #  0x00a8 -> ARABIC LETTER ALEF FINAL FORM
348    u'\ufe8f'   #  0x00a9 -> ARABIC LETTER BEH ISOLATED FORM
349    u'\ufe95'   #  0x00aa -> ARABIC LETTER TEH ISOLATED FORM
350    u'\ufe99'   #  0x00ab -> ARABIC LETTER THEH ISOLATED FORM
351    u'\u060c'   #  0x00ac -> ARABIC COMMA
352    u'\ufe9d'   #  0x00ad -> ARABIC LETTER JEEM ISOLATED FORM
353    u'\ufea1'   #  0x00ae -> ARABIC LETTER HAH ISOLATED FORM
354    u'\ufea5'   #  0x00af -> ARABIC LETTER KHAH ISOLATED FORM
355    u'\u0660'   #  0x00b0 -> ARABIC-INDIC DIGIT ZERO
356    u'\u0661'   #  0x00b1 -> ARABIC-INDIC DIGIT ONE
357    u'\u0662'   #  0x00b2 -> ARABIC-INDIC DIGIT TWO
358    u'\u0663'   #  0x00b3 -> ARABIC-INDIC DIGIT THREE
359    u'\u0664'   #  0x00b4 -> ARABIC-INDIC DIGIT FOUR
360    u'\u0665'   #  0x00b5 -> ARABIC-INDIC DIGIT FIVE
361    u'\u0666'   #  0x00b6 -> ARABIC-INDIC DIGIT SIX
362    u'\u0667'   #  0x00b7 -> ARABIC-INDIC DIGIT SEVEN
363    u'\u0668'   #  0x00b8 -> ARABIC-INDIC DIGIT EIGHT
364    u'\u0669'   #  0x00b9 -> ARABIC-INDIC DIGIT NINE
365    u'\ufed1'   #  0x00ba -> ARABIC LETTER FEH ISOLATED FORM
366    u'\u061b'   #  0x00bb -> ARABIC SEMICOLON
367    u'\ufeb1'   #  0x00bc -> ARABIC LETTER SEEN ISOLATED FORM
368    u'\ufeb5'   #  0x00bd -> ARABIC LETTER SHEEN ISOLATED FORM
369    u'\ufeb9'   #  0x00be -> ARABIC LETTER SAD ISOLATED FORM
370    u'\u061f'   #  0x00bf -> ARABIC QUESTION MARK
371    u'\xa2'     #  0x00c0 -> CENT SIGN
372    u'\ufe80'   #  0x00c1 -> ARABIC LETTER HAMZA ISOLATED FORM
373    u'\ufe81'   #  0x00c2 -> ARABIC LETTER ALEF WITH MADDA ABOVE ISOLATED FORM
374    u'\ufe83'   #  0x00c3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE ISOLATED FORM
375    u'\ufe85'   #  0x00c4 -> ARABIC LETTER WAW WITH HAMZA ABOVE ISOLATED FORM
376    u'\ufeca'   #  0x00c5 -> ARABIC LETTER AIN FINAL FORM
377    u'\ufe8b'   #  0x00c6 -> ARABIC LETTER YEH WITH HAMZA ABOVE INITIAL FORM
378    u'\ufe8d'   #  0x00c7 -> ARABIC LETTER ALEF ISOLATED FORM
379    u'\ufe91'   #  0x00c8 -> ARABIC LETTER BEH INITIAL FORM
380    u'\ufe93'   #  0x00c9 -> ARABIC LETTER TEH MARBUTA ISOLATED FORM
381    u'\ufe97'   #  0x00ca -> ARABIC LETTER TEH INITIAL FORM
382    u'\ufe9b'   #  0x00cb -> ARABIC LETTER THEH INITIAL FORM
383    u'\ufe9f'   #  0x00cc -> ARABIC LETTER JEEM INITIAL FORM
384    u'\ufea3'   #  0x00cd -> ARABIC LETTER HAH INITIAL FORM
385    u'\ufea7'   #  0x00ce -> ARABIC LETTER KHAH INITIAL FORM
386    u'\ufea9'   #  0x00cf -> ARABIC LETTER DAL ISOLATED FORM
387    u'\ufeab'   #  0x00d0 -> ARABIC LETTER THAL ISOLATED FORM
388    u'\ufead'   #  0x00d1 -> ARABIC LETTER REH ISOLATED FORM
389    u'\ufeaf'   #  0x00d2 -> ARABIC LETTER ZAIN ISOLATED FORM
390    u'\ufeb3'   #  0x00d3 -> ARABIC LETTER SEEN INITIAL FORM
391    u'\ufeb7'   #  0x00d4 -> ARABIC LETTER SHEEN INITIAL FORM
392    u'\ufebb'   #  0x00d5 -> ARABIC LETTER SAD INITIAL FORM
393    u'\ufebf'   #  0x00d6 -> ARABIC LETTER DAD INITIAL FORM
394    u'\ufec1'   #  0x00d7 -> ARABIC LETTER TAH ISOLATED FORM
395    u'\ufec5'   #  0x00d8 -> ARABIC LETTER ZAH ISOLATED FORM
396    u'\ufecb'   #  0x00d9 -> ARABIC LETTER AIN INITIAL FORM
397    u'\ufecf'   #  0x00da -> ARABIC LETTER GHAIN INITIAL FORM
398    u'\xa6'     #  0x00db -> BROKEN VERTICAL BAR
399    u'\xac'     #  0x00dc -> NOT SIGN
400    u'\xf7'     #  0x00dd -> DIVISION SIGN
401    u'\xd7'     #  0x00de -> MULTIPLICATION SIGN
402    u'\ufec9'   #  0x00df -> ARABIC LETTER AIN ISOLATED FORM
403    u'\u0640'   #  0x00e0 -> ARABIC TATWEEL
404    u'\ufed3'   #  0x00e1 -> ARABIC LETTER FEH INITIAL FORM
405    u'\ufed7'   #  0x00e2 -> ARABIC LETTER QAF INITIAL FORM
406    u'\ufedb'   #  0x00e3 -> ARABIC LETTER KAF INITIAL FORM
407    u'\ufedf'   #  0x00e4 -> ARABIC LETTER LAM INITIAL FORM
408    u'\ufee3'   #  0x00e5 -> ARABIC LETTER MEEM INITIAL FORM
409    u'\ufee7'   #  0x00e6 -> ARABIC LETTER NOON INITIAL FORM
410    u'\ufeeb'   #  0x00e7 -> ARABIC LETTER HEH INITIAL FORM
411    u'\ufeed'   #  0x00e8 -> ARABIC LETTER WAW ISOLATED FORM
412    u'\ufeef'   #  0x00e9 -> ARABIC LETTER ALEF MAKSURA ISOLATED FORM
413    u'\ufef3'   #  0x00ea -> ARABIC LETTER YEH INITIAL FORM
414    u'\ufebd'   #  0x00eb -> ARABIC LETTER DAD ISOLATED FORM
415    u'\ufecc'   #  0x00ec -> ARABIC LETTER AIN MEDIAL FORM
416    u'\ufece'   #  0x00ed -> ARABIC LETTER GHAIN FINAL FORM
417    u'\ufecd'   #  0x00ee -> ARABIC LETTER GHAIN ISOLATED FORM
418    u'\ufee1'   #  0x00ef -> ARABIC LETTER MEEM ISOLATED FORM
419    u'\ufe7d'   #  0x00f0 -> ARABIC SHADDA MEDIAL FORM
420    u'\u0651'   #  0x00f1 -> ARABIC SHADDAH
421    u'\ufee5'   #  0x00f2 -> ARABIC LETTER NOON ISOLATED FORM
422    u'\ufee9'   #  0x00f3 -> ARABIC LETTER HEH ISOLATED FORM
423    u'\ufeec'   #  0x00f4 -> ARABIC LETTER HEH MEDIAL FORM
424    u'\ufef0'   #  0x00f5 -> ARABIC LETTER ALEF MAKSURA FINAL FORM
425    u'\ufef2'   #  0x00f6 -> ARABIC LETTER YEH FINAL FORM
426    u'\ufed0'   #  0x00f7 -> ARABIC LETTER GHAIN MEDIAL FORM
427    u'\ufed5'   #  0x00f8 -> ARABIC LETTER QAF ISOLATED FORM
428    u'\ufef5'   #  0x00f9 -> ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE ISOLATED FORM
429    u'\ufef6'   #  0x00fa -> ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE FINAL FORM
430    u'\ufedd'   #  0x00fb -> ARABIC LETTER LAM ISOLATED FORM
431    u'\ufed9'   #  0x00fc -> ARABIC LETTER KAF ISOLATED FORM
432    u'\ufef1'   #  0x00fd -> ARABIC LETTER YEH ISOLATED FORM
433    u'\u25a0'   #  0x00fe -> BLACK SQUARE
434    u'\ufffe'   #  0x00ff -> UNDEFINED
435)
436
437### Encoding Map
438
439encoding_map = {
440    0x0000: 0x0000,     #  NULL
441    0x0001: 0x0001,     #  START OF HEADING
442    0x0002: 0x0002,     #  START OF TEXT
443    0x0003: 0x0003,     #  END OF TEXT
444    0x0004: 0x0004,     #  END OF TRANSMISSION
445    0x0005: 0x0005,     #  ENQUIRY
446    0x0006: 0x0006,     #  ACKNOWLEDGE
447    0x0007: 0x0007,     #  BELL
448    0x0008: 0x0008,     #  BACKSPACE
449    0x0009: 0x0009,     #  HORIZONTAL TABULATION
450    0x000a: 0x000a,     #  LINE FEED
451    0x000b: 0x000b,     #  VERTICAL TABULATION
452    0x000c: 0x000c,     #  FORM FEED
453    0x000d: 0x000d,     #  CARRIAGE RETURN
454    0x000e: 0x000e,     #  SHIFT OUT
455    0x000f: 0x000f,     #  SHIFT IN
456    0x0010: 0x0010,     #  DATA LINK ESCAPE
457    0x0011: 0x0011,     #  DEVICE CONTROL ONE
458    0x0012: 0x0012,     #  DEVICE CONTROL TWO
459    0x0013: 0x0013,     #  DEVICE CONTROL THREE
460    0x0014: 0x0014,     #  DEVICE CONTROL FOUR
461    0x0015: 0x0015,     #  NEGATIVE ACKNOWLEDGE
462    0x0016: 0x0016,     #  SYNCHRONOUS IDLE
463    0x0017: 0x0017,     #  END OF TRANSMISSION BLOCK
464    0x0018: 0x0018,     #  CANCEL
465    0x0019: 0x0019,     #  END OF MEDIUM
466    0x001a: 0x001a,     #  SUBSTITUTE
467    0x001b: 0x001b,     #  ESCAPE
468    0x001c: 0x001c,     #  FILE SEPARATOR
469    0x001d: 0x001d,     #  GROUP SEPARATOR
470    0x001e: 0x001e,     #  RECORD SEPARATOR
471    0x001f: 0x001f,     #  UNIT SEPARATOR
472    0x0020: 0x0020,     #  SPACE
473    0x0021: 0x0021,     #  EXCLAMATION MARK
474    0x0022: 0x0022,     #  QUOTATION MARK
475    0x0023: 0x0023,     #  NUMBER SIGN
476    0x0024: 0x0024,     #  DOLLAR SIGN
477    0x0026: 0x0026,     #  AMPERSAND
478    0x0027: 0x0027,     #  APOSTROPHE
479    0x0028: 0x0028,     #  LEFT PARENTHESIS
480    0x0029: 0x0029,     #  RIGHT PARENTHESIS
481    0x002a: 0x002a,     #  ASTERISK
482    0x002b: 0x002b,     #  PLUS SIGN
483    0x002c: 0x002c,     #  COMMA
484    0x002d: 0x002d,     #  HYPHEN-MINUS
485    0x002e: 0x002e,     #  FULL STOP
486    0x002f: 0x002f,     #  SOLIDUS
487    0x0030: 0x0030,     #  DIGIT ZERO
488    0x0031: 0x0031,     #  DIGIT ONE
489    0x0032: 0x0032,     #  DIGIT TWO
490    0x0033: 0x0033,     #  DIGIT THREE
491    0x0034: 0x0034,     #  DIGIT FOUR
492    0x0035: 0x0035,     #  DIGIT FIVE
493    0x0036: 0x0036,     #  DIGIT SIX
494    0x0037: 0x0037,     #  DIGIT SEVEN
495    0x0038: 0x0038,     #  DIGIT EIGHT
496    0x0039: 0x0039,     #  DIGIT NINE
497    0x003a: 0x003a,     #  COLON
498    0x003b: 0x003b,     #  SEMICOLON
499    0x003c: 0x003c,     #  LESS-THAN SIGN
500    0x003d: 0x003d,     #  EQUALS SIGN
501    0x003e: 0x003e,     #  GREATER-THAN SIGN
502    0x003f: 0x003f,     #  QUESTION MARK
503    0x0040: 0x0040,     #  COMMERCIAL AT
504    0x0041: 0x0041,     #  LATIN CAPITAL LETTER A
505    0x0042: 0x0042,     #  LATIN CAPITAL LETTER B
506    0x0043: 0x0043,     #  LATIN CAPITAL LETTER C
507    0x0044: 0x0044,     #  LATIN CAPITAL LETTER D
508    0x0045: 0x0045,     #  LATIN CAPITAL LETTER E
509    0x0046: 0x0046,     #  LATIN CAPITAL LETTER F
510    0x0047: 0x0047,     #  LATIN CAPITAL LETTER G
511    0x0048: 0x0048,     #  LATIN CAPITAL LETTER H
512    0x0049: 0x0049,     #  LATIN CAPITAL LETTER I
513    0x004a: 0x004a,     #  LATIN CAPITAL LETTER J
514    0x004b: 0x004b,     #  LATIN CAPITAL LETTER K
515    0x004c: 0x004c,     #  LATIN CAPITAL LETTER L
516    0x004d: 0x004d,     #  LATIN CAPITAL LETTER M
517    0x004e: 0x004e,     #  LATIN CAPITAL LETTER N
518    0x004f: 0x004f,     #  LATIN CAPITAL LETTER O
519    0x0050: 0x0050,     #  LATIN CAPITAL LETTER P
520    0x0051: 0x0051,     #  LATIN CAPITAL LETTER Q
521    0x0052: 0x0052,     #  LATIN CAPITAL LETTER R
522    0x0053: 0x0053,     #  LATIN CAPITAL LETTER S
523    0x0054: 0x0054,     #  LATIN CAPITAL LETTER T
524    0x0055: 0x0055,     #  LATIN CAPITAL LETTER U
525    0x0056: 0x0056,     #  LATIN CAPITAL LETTER V
526    0x0057: 0x0057,     #  LATIN CAPITAL LETTER W
527    0x0058: 0x0058,     #  LATIN CAPITAL LETTER X
528    0x0059: 0x0059,     #  LATIN CAPITAL LETTER Y
529    0x005a: 0x005a,     #  LATIN CAPITAL LETTER Z
530    0x005b: 0x005b,     #  LEFT SQUARE BRACKET
531    0x005c: 0x005c,     #  REVERSE SOLIDUS
532    0x005d: 0x005d,     #  RIGHT SQUARE BRACKET
533    0x005e: 0x005e,     #  CIRCUMFLEX ACCENT
534    0x005f: 0x005f,     #  LOW LINE
535    0x0060: 0x0060,     #  GRAVE ACCENT
536    0x0061: 0x0061,     #  LATIN SMALL LETTER A
537    0x0062: 0x0062,     #  LATIN SMALL LETTER B
538    0x0063: 0x0063,     #  LATIN SMALL LETTER C
539    0x0064: 0x0064,     #  LATIN SMALL LETTER D
540    0x0065: 0x0065,     #  LATIN SMALL LETTER E
541    0x0066: 0x0066,     #  LATIN SMALL LETTER F
542    0x0067: 0x0067,     #  LATIN SMALL LETTER G
543    0x0068: 0x0068,     #  LATIN SMALL LETTER H
544    0x0069: 0x0069,     #  LATIN SMALL LETTER I
545    0x006a: 0x006a,     #  LATIN SMALL LETTER J
546    0x006b: 0x006b,     #  LATIN SMALL LETTER K
547    0x006c: 0x006c,     #  LATIN SMALL LETTER L
548    0x006d: 0x006d,     #  LATIN SMALL LETTER M
549    0x006e: 0x006e,     #  LATIN SMALL LETTER N
550    0x006f: 0x006f,     #  LATIN SMALL LETTER O
551    0x0070: 0x0070,     #  LATIN SMALL LETTER P
552    0x0071: 0x0071,     #  LATIN SMALL LETTER Q
553    0x0072: 0x0072,     #  LATIN SMALL LETTER R
554    0x0073: 0x0073,     #  LATIN SMALL LETTER S
555    0x0074: 0x0074,     #  LATIN SMALL LETTER T
556    0x0075: 0x0075,     #  LATIN SMALL LETTER U
557    0x0076: 0x0076,     #  LATIN SMALL LETTER V
558    0x0077: 0x0077,     #  LATIN SMALL LETTER W
559    0x0078: 0x0078,     #  LATIN SMALL LETTER X
560    0x0079: 0x0079,     #  LATIN SMALL LETTER Y
561    0x007a: 0x007a,     #  LATIN SMALL LETTER Z
562    0x007b: 0x007b,     #  LEFT CURLY BRACKET
563    0x007c: 0x007c,     #  VERTICAL LINE
564    0x007d: 0x007d,     #  RIGHT CURLY BRACKET
565    0x007e: 0x007e,     #  TILDE
566    0x007f: 0x007f,     #  DELETE
567    0x00a0: 0x00a0,     #  NON-BREAKING SPACE
568    0x00a2: 0x00c0,     #  CENT SIGN
569    0x00a3: 0x00a3,     #  POUND SIGN
570    0x00a4: 0x00a4,     #  CURRENCY SIGN
571    0x00a6: 0x00db,     #  BROKEN VERTICAL BAR
572    0x00ab: 0x0097,     #  LEFT POINTING GUILLEMET
573    0x00ac: 0x00dc,     #  NOT SIGN
574    0x00ad: 0x00a1,     #  SOFT HYPHEN
575    0x00b0: 0x0080,     #  DEGREE SIGN
576    0x00b1: 0x0093,     #  PLUS-OR-MINUS SIGN
577    0x00b7: 0x0081,     #  MIDDLE DOT
578    0x00bb: 0x0098,     #  RIGHT POINTING GUILLEMET
579    0x00bc: 0x0095,     #  FRACTION 1/4
580    0x00bd: 0x0094,     #  FRACTION 1/2
581    0x00d7: 0x00de,     #  MULTIPLICATION SIGN
582    0x00f7: 0x00dd,     #  DIVISION SIGN
583    0x03b2: 0x0090,     #  GREEK SMALL BETA
584    0x03c6: 0x0092,     #  GREEK SMALL PHI
585    0x060c: 0x00ac,     #  ARABIC COMMA
586    0x061b: 0x00bb,     #  ARABIC SEMICOLON
587    0x061f: 0x00bf,     #  ARABIC QUESTION MARK
588    0x0640: 0x00e0,     #  ARABIC TATWEEL
589    0x0651: 0x00f1,     #  ARABIC SHADDAH
590    0x0660: 0x00b0,     #  ARABIC-INDIC DIGIT ZERO
591    0x0661: 0x00b1,     #  ARABIC-INDIC DIGIT ONE
592    0x0662: 0x00b2,     #  ARABIC-INDIC DIGIT TWO
593    0x0663: 0x00b3,     #  ARABIC-INDIC DIGIT THREE
594    0x0664: 0x00b4,     #  ARABIC-INDIC DIGIT FOUR
595    0x0665: 0x00b5,     #  ARABIC-INDIC DIGIT FIVE
596    0x0666: 0x00b6,     #  ARABIC-INDIC DIGIT SIX
597    0x0667: 0x00b7,     #  ARABIC-INDIC DIGIT SEVEN
598    0x0668: 0x00b8,     #  ARABIC-INDIC DIGIT EIGHT
599    0x0669: 0x00b9,     #  ARABIC-INDIC DIGIT NINE
600    0x066a: 0x0025,     #  ARABIC PERCENT SIGN
601    0x2219: 0x0082,     #  BULLET OPERATOR
602    0x221a: 0x0083,     #  SQUARE ROOT
603    0x221e: 0x0091,     #  INFINITY
604    0x2248: 0x0096,     #  ALMOST EQUAL TO
605    0x2500: 0x0085,     #  FORMS LIGHT HORIZONTAL
606    0x2502: 0x0086,     #  FORMS LIGHT VERTICAL
607    0x250c: 0x008d,     #  FORMS LIGHT DOWN AND RIGHT
608    0x2510: 0x008c,     #  FORMS LIGHT DOWN AND LEFT
609    0x2514: 0x008e,     #  FORMS LIGHT UP AND RIGHT
610    0x2518: 0x008f,     #  FORMS LIGHT UP AND LEFT
611    0x251c: 0x008a,     #  FORMS LIGHT VERTICAL AND RIGHT
612    0x2524: 0x0088,     #  FORMS LIGHT VERTICAL AND LEFT
613    0x252c: 0x0089,     #  FORMS LIGHT DOWN AND HORIZONTAL
614    0x2534: 0x008b,     #  FORMS LIGHT UP AND HORIZONTAL
615    0x253c: 0x0087,     #  FORMS LIGHT VERTICAL AND HORIZONTAL
616    0x2592: 0x0084,     #  MEDIUM SHADE
617    0x25a0: 0x00fe,     #  BLACK SQUARE
618    0xfe7d: 0x00f0,     #  ARABIC SHADDA MEDIAL FORM
619    0xfe80: 0x00c1,     #  ARABIC LETTER HAMZA ISOLATED FORM
620    0xfe81: 0x00c2,     #  ARABIC LETTER ALEF WITH MADDA ABOVE ISOLATED FORM
621    0xfe82: 0x00a2,     #  ARABIC LETTER ALEF WITH MADDA ABOVE FINAL FORM
622    0xfe83: 0x00c3,     #  ARABIC LETTER ALEF WITH HAMZA ABOVE ISOLATED FORM
623    0xfe84: 0x00a5,     #  ARABIC LETTER ALEF WITH HAMZA ABOVE FINAL FORM
624    0xfe85: 0x00c4,     #  ARABIC LETTER WAW WITH HAMZA ABOVE ISOLATED FORM
625    0xfe8b: 0x00c6,     #  ARABIC LETTER YEH WITH HAMZA ABOVE INITIAL FORM
626    0xfe8d: 0x00c7,     #  ARABIC LETTER ALEF ISOLATED FORM
627    0xfe8e: 0x00a8,     #  ARABIC LETTER ALEF FINAL FORM
628    0xfe8f: 0x00a9,     #  ARABIC LETTER BEH ISOLATED FORM
629    0xfe91: 0x00c8,     #  ARABIC LETTER BEH INITIAL FORM
630    0xfe93: 0x00c9,     #  ARABIC LETTER TEH MARBUTA ISOLATED FORM
631    0xfe95: 0x00aa,     #  ARABIC LETTER TEH ISOLATED FORM
632    0xfe97: 0x00ca,     #  ARABIC LETTER TEH INITIAL FORM
633    0xfe99: 0x00ab,     #  ARABIC LETTER THEH ISOLATED FORM
634    0xfe9b: 0x00cb,     #  ARABIC LETTER THEH INITIAL FORM
635    0xfe9d: 0x00ad,     #  ARABIC LETTER JEEM ISOLATED FORM
636    0xfe9f: 0x00cc,     #  ARABIC LETTER JEEM INITIAL FORM
637    0xfea1: 0x00ae,     #  ARABIC LETTER HAH ISOLATED FORM
638    0xfea3: 0x00cd,     #  ARABIC LETTER HAH INITIAL FORM
639    0xfea5: 0x00af,     #  ARABIC LETTER KHAH ISOLATED FORM
640    0xfea7: 0x00ce,     #  ARABIC LETTER KHAH INITIAL FORM
641    0xfea9: 0x00cf,     #  ARABIC LETTER DAL ISOLATED FORM
642    0xfeab: 0x00d0,     #  ARABIC LETTER THAL ISOLATED FORM
643    0xfead: 0x00d1,     #  ARABIC LETTER REH ISOLATED FORM
644    0xfeaf: 0x00d2,     #  ARABIC LETTER ZAIN ISOLATED FORM
645    0xfeb1: 0x00bc,     #  ARABIC LETTER SEEN ISOLATED FORM
646    0xfeb3: 0x00d3,     #  ARABIC LETTER SEEN INITIAL FORM
647    0xfeb5: 0x00bd,     #  ARABIC LETTER SHEEN ISOLATED FORM
648    0xfeb7: 0x00d4,     #  ARABIC LETTER SHEEN INITIAL FORM
649    0xfeb9: 0x00be,     #  ARABIC LETTER SAD ISOLATED FORM
650    0xfebb: 0x00d5,     #  ARABIC LETTER SAD INITIAL FORM
651    0xfebd: 0x00eb,     #  ARABIC LETTER DAD ISOLATED FORM
652    0xfebf: 0x00d6,     #  ARABIC LETTER DAD INITIAL FORM
653    0xfec1: 0x00d7,     #  ARABIC LETTER TAH ISOLATED FORM
654    0xfec5: 0x00d8,     #  ARABIC LETTER ZAH ISOLATED FORM
655    0xfec9: 0x00df,     #  ARABIC LETTER AIN ISOLATED FORM
656    0xfeca: 0x00c5,     #  ARABIC LETTER AIN FINAL FORM
657    0xfecb: 0x00d9,     #  ARABIC LETTER AIN INITIAL FORM
658    0xfecc: 0x00ec,     #  ARABIC LETTER AIN MEDIAL FORM
659    0xfecd: 0x00ee,     #  ARABIC LETTER GHAIN ISOLATED FORM
660    0xfece: 0x00ed,     #  ARABIC LETTER GHAIN FINAL FORM
661    0xfecf: 0x00da,     #  ARABIC LETTER GHAIN INITIAL FORM
662    0xfed0: 0x00f7,     #  ARABIC LETTER GHAIN MEDIAL FORM
663    0xfed1: 0x00ba,     #  ARABIC LETTER FEH ISOLATED FORM
664    0xfed3: 0x00e1,     #  ARABIC LETTER FEH INITIAL FORM
665    0xfed5: 0x00f8,     #  ARABIC LETTER QAF ISOLATED FORM
666    0xfed7: 0x00e2,     #  ARABIC LETTER QAF INITIAL FORM
667    0xfed9: 0x00fc,     #  ARABIC LETTER KAF ISOLATED FORM
668    0xfedb: 0x00e3,     #  ARABIC LETTER KAF INITIAL FORM
669    0xfedd: 0x00fb,     #  ARABIC LETTER LAM ISOLATED FORM
670    0xfedf: 0x00e4,     #  ARABIC LETTER LAM INITIAL FORM
671    0xfee1: 0x00ef,     #  ARABIC LETTER MEEM ISOLATED FORM
672    0xfee3: 0x00e5,     #  ARABIC LETTER MEEM INITIAL FORM
673    0xfee5: 0x00f2,     #  ARABIC LETTER NOON ISOLATED FORM
674    0xfee7: 0x00e6,     #  ARABIC LETTER NOON INITIAL FORM
675    0xfee9: 0x00f3,     #  ARABIC LETTER HEH ISOLATED FORM
676    0xfeeb: 0x00e7,     #  ARABIC LETTER HEH INITIAL FORM
677    0xfeec: 0x00f4,     #  ARABIC LETTER HEH MEDIAL FORM
678    0xfeed: 0x00e8,     #  ARABIC LETTER WAW ISOLATED FORM
679    0xfeef: 0x00e9,     #  ARABIC LETTER ALEF MAKSURA ISOLATED FORM
680    0xfef0: 0x00f5,     #  ARABIC LETTER ALEF MAKSURA FINAL FORM
681    0xfef1: 0x00fd,     #  ARABIC LETTER YEH ISOLATED FORM
682    0xfef2: 0x00f6,     #  ARABIC LETTER YEH FINAL FORM
683    0xfef3: 0x00ea,     #  ARABIC LETTER YEH INITIAL FORM
684    0xfef5: 0x00f9,     #  ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE ISOLATED FORM
685    0xfef6: 0x00fa,     #  ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE FINAL FORM
686    0xfef7: 0x0099,     #  ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE ISOLATED FORM
687    0xfef8: 0x009a,     #  ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE FINAL FORM
688    0xfefb: 0x009d,     #  ARABIC LIGATURE LAM WITH ALEF ISOLATED FORM
689    0xfefc: 0x009e,     #  ARABIC LIGATURE LAM WITH ALEF FINAL FORM
690}