#1616979: Add the cp720 (Arabic DOS) encoding.

Since there is no official mapping file from unicode.org,
the codec file is generated on Windows with the new genwincodec.py script.

diff --git a/Lib/encodings/cp720.py b/Lib/encodings/cp720.py
new file mode 100644 (file)
index 0000000..c1b6a99
--- /dev/null
+++ b/Lib/encodings/cp720.py
@@ -0,0 +1,311 @@
+"""Python Character Mapping Codec cp720 generated on Windows:\r
+Vista 6.0.6002 SP2 Multiprocessor Free with the command:\r
+  python Tools/unicode/genwincodec.py 720\r
+"""#"\r
+\r
+\r
+import codecs\r
+\r
+### Codec APIs\r
+\r
+class Codec(codecs.Codec):\r
+\r
+    def encode(self,input,errors='strict'):\r
+        return codecs.charmap_encode(input,errors,encoding_table)\r
+\r
+    def decode(self,input,errors='strict'):\r
+        return codecs.charmap_decode(input,errors,decoding_table)\r
+\r
+class IncrementalEncoder(codecs.IncrementalEncoder):\r
+    def encode(self, input, final=False):\r
+        return codecs.charmap_encode(input,self.errors,encoding_table)[0]\r
+\r
+class IncrementalDecoder(codecs.IncrementalDecoder):\r
+    def decode(self, input, final=False):\r
+        return codecs.charmap_decode(input,self.errors,decoding_table)[0]\r
+\r
+class StreamWriter(Codec,codecs.StreamWriter):\r
+    pass\r
+\r
+class StreamReader(Codec,codecs.StreamReader):\r
+    pass\r
+\r
+### encodings module API\r
+\r
+def getregentry():\r
+    return codecs.CodecInfo(\r
+        name='cp720',\r
+        encode=Codec().encode,\r
+        decode=Codec().decode,\r
+        incrementalencoder=IncrementalEncoder,\r
+        incrementaldecoder=IncrementalDecoder,\r
+        streamreader=StreamReader,\r
+        streamwriter=StreamWriter,\r
+    )\r
+\r
+\r
+### Decoding Table\r
+\r
+decoding_table = (\r
+    u'\x00'     #  0x00 -> CONTROL CHARACTER\r
+    u'\x01'     #  0x01 -> CONTROL CHARACTER\r
+    u'\x02'     #  0x02 -> CONTROL CHARACTER\r
+    u'\x03'     #  0x03 -> CONTROL CHARACTER\r
+    u'\x04'     #  0x04 -> CONTROL CHARACTER\r
+    u'\x05'     #  0x05 -> CONTROL CHARACTER\r
+    u'\x06'     #  0x06 -> CONTROL CHARACTER\r
+    u'\x07'     #  0x07 -> CONTROL CHARACTER\r
+    u'\x08'     #  0x08 -> CONTROL CHARACTER\r
+    u'\t'       #  0x09 -> CONTROL CHARACTER\r
+    u'\n'       #  0x0A -> CONTROL CHARACTER\r
+    u'\x0b'     #  0x0B -> CONTROL CHARACTER\r
+    u'\x0c'     #  0x0C -> CONTROL CHARACTER\r
+    u'\r'       #  0x0D -> CONTROL CHARACTER\r
+    u'\x0e'     #  0x0E -> CONTROL CHARACTER\r
+    u'\x0f'     #  0x0F -> CONTROL CHARACTER\r
+    u'\x10'     #  0x10 -> CONTROL CHARACTER\r
+    u'\x11'     #  0x11 -> CONTROL CHARACTER\r
+    u'\x12'     #  0x12 -> CONTROL CHARACTER\r
+    u'\x13'     #  0x13 -> CONTROL CHARACTER\r
+    u'\x14'     #  0x14 -> CONTROL CHARACTER\r
+    u'\x15'     #  0x15 -> CONTROL CHARACTER\r
+    u'\x16'     #  0x16 -> CONTROL CHARACTER\r
+    u'\x17'     #  0x17 -> CONTROL CHARACTER\r
+    u'\x18'     #  0x18 -> CONTROL CHARACTER\r
+    u'\x19'     #  0x19 -> CONTROL CHARACTER\r
+    u'\x1a'     #  0x1A -> CONTROL CHARACTER\r
+    u'\x1b'     #  0x1B -> CONTROL CHARACTER\r
+    u'\x1c'     #  0x1C -> CONTROL CHARACTER\r
+    u'\x1d'     #  0x1D -> CONTROL CHARACTER\r
+    u'\x1e'     #  0x1E -> CONTROL CHARACTER\r
+    u'\x1f'     #  0x1F -> CONTROL CHARACTER\r
+    u' '        #  0x20 -> SPACE\r
+    u'!'        #  0x21 -> EXCLAMATION MARK\r
+    u'"'        #  0x22 -> QUOTATION MARK\r
+    u'#'        #  0x23 -> NUMBER SIGN\r
+    u'$'        #  0x24 -> DOLLAR SIGN\r
+    u'%'        #  0x25 -> PERCENT SIGN\r
+    u'&'        #  0x26 -> AMPERSAND\r
+    u"'"        #  0x27 -> APOSTROPHE\r
+    u'('        #  0x28 -> LEFT PARENTHESIS\r
+    u')'        #  0x29 -> RIGHT PARENTHESIS\r
+    u'*'        #  0x2A -> ASTERISK\r
+    u'+'        #  0x2B -> PLUS SIGN\r
+    u','        #  0x2C -> COMMA\r
+    u'-'        #  0x2D -> HYPHEN-MINUS\r
+    u'.'        #  0x2E -> FULL STOP\r
+    u'/'        #  0x2F -> SOLIDUS\r
+    u'0'        #  0x30 -> DIGIT ZERO\r
+    u'1'        #  0x31 -> DIGIT ONE\r
+    u'2'        #  0x32 -> DIGIT TWO\r
+    u'3'        #  0x33 -> DIGIT THREE\r
+    u'4'        #  0x34 -> DIGIT FOUR\r
+    u'5'        #  0x35 -> DIGIT FIVE\r
+    u'6'        #  0x36 -> DIGIT SIX\r
+    u'7'        #  0x37 -> DIGIT SEVEN\r
+    u'8'        #  0x38 -> DIGIT EIGHT\r
+    u'9'        #  0x39 -> DIGIT NINE\r
+    u':'        #  0x3A -> COLON\r
+    u';'        #  0x3B -> SEMICOLON\r
+    u'<'        #  0x3C -> LESS-THAN SIGN\r
+    u'='        #  0x3D -> EQUALS SIGN\r
+    u'>'        #  0x3E -> GREATER-THAN SIGN\r
+    u'?'        #  0x3F -> QUESTION MARK\r
+    u'@'        #  0x40 -> COMMERCIAL AT\r
+    u'A'        #  0x41 -> LATIN CAPITAL LETTER A\r
+    u'B'        #  0x42 -> LATIN CAPITAL LETTER B\r
+    u'C'        #  0x43 -> LATIN CAPITAL LETTER C\r
+    u'D'        #  0x44 -> LATIN CAPITAL LETTER D\r
+    u'E'        #  0x45 -> LATIN CAPITAL LETTER E\r
+    u'F'        #  0x46 -> LATIN CAPITAL LETTER F\r
+    u'G'        #  0x47 -> LATIN CAPITAL LETTER G\r
+    u'H'        #  0x48 -> LATIN CAPITAL LETTER H\r
+    u'I'        #  0x49 -> LATIN CAPITAL LETTER I\r
+    u'J'        #  0x4A -> LATIN CAPITAL LETTER J\r
+    u'K'        #  0x4B -> LATIN CAPITAL LETTER K\r
+    u'L'        #  0x4C -> LATIN CAPITAL LETTER L\r
+    u'M'        #  0x4D -> LATIN CAPITAL LETTER M\r
+    u'N'        #  0x4E -> LATIN CAPITAL LETTER N\r
+    u'O'        #  0x4F -> LATIN CAPITAL LETTER O\r
+    u'P'        #  0x50 -> LATIN CAPITAL LETTER P\r
+    u'Q'        #  0x51 -> LATIN CAPITAL LETTER Q\r
+    u'R'        #  0x52 -> LATIN CAPITAL LETTER R\r
+    u'S'        #  0x53 -> LATIN CAPITAL LETTER S\r
+    u'T'        #  0x54 -> LATIN CAPITAL LETTER T\r
+    u'U'        #  0x55 -> LATIN CAPITAL LETTER U\r
+    u'V'        #  0x56 -> LATIN CAPITAL LETTER V\r
+    u'W'        #  0x57 -> LATIN CAPITAL LETTER W\r
+    u'X'        #  0x58 -> LATIN CAPITAL LETTER X\r
+    u'Y'        #  0x59 -> LATIN CAPITAL LETTER Y\r
+    u'Z'        #  0x5A -> LATIN CAPITAL LETTER Z\r
+    u'['        #  0x5B -> LEFT SQUARE BRACKET\r
+    u'\\'       #  0x5C -> REVERSE SOLIDUS\r
+    u']'        #  0x5D -> RIGHT SQUARE BRACKET\r
+    u'^'        #  0x5E -> CIRCUMFLEX ACCENT\r
+    u'_'        #  0x5F -> LOW LINE\r
+    u'`'        #  0x60 -> GRAVE ACCENT\r
+    u'a'        #  0x61 -> LATIN SMALL LETTER A\r
+    u'b'        #  0x62 -> LATIN SMALL LETTER B\r
+    u'c'        #  0x63 -> LATIN SMALL LETTER C\r
+    u'd'        #  0x64 -> LATIN SMALL LETTER D\r
+    u'e'        #  0x65 -> LATIN SMALL LETTER E\r
+    u'f'        #  0x66 -> LATIN SMALL LETTER F\r
+    u'g'        #  0x67 -> LATIN SMALL LETTER G\r
+    u'h'        #  0x68 -> LATIN SMALL LETTER H\r
+    u'i'        #  0x69 -> LATIN SMALL LETTER I\r
+    u'j'        #  0x6A -> LATIN SMALL LETTER J\r
+    u'k'        #  0x6B -> LATIN SMALL LETTER K\r
+    u'l'        #  0x6C -> LATIN SMALL LETTER L\r
+    u'm'        #  0x6D -> LATIN SMALL LETTER M\r
+    u'n'        #  0x6E -> LATIN SMALL LETTER N\r
+    u'o'        #  0x6F -> LATIN SMALL LETTER O\r
+    u'p'        #  0x70 -> LATIN SMALL LETTER P\r
+    u'q'        #  0x71 -> LATIN SMALL LETTER Q\r
+    u'r'        #  0x72 -> LATIN SMALL LETTER R\r
+    u's'        #  0x73 -> LATIN SMALL LETTER S\r
+    u't'        #  0x74 -> LATIN SMALL LETTER T\r
+    u'u'        #  0x75 -> LATIN SMALL LETTER U\r
+    u'v'        #  0x76 -> LATIN SMALL LETTER V\r
+    u'w'        #  0x77 -> LATIN SMALL LETTER W\r
+    u'x'        #  0x78 -> LATIN SMALL LETTER X\r
+    u'y'        #  0x79 -> LATIN SMALL LETTER Y\r
+    u'z'        #  0x7A -> LATIN SMALL LETTER Z\r
+    u'{'        #  0x7B -> LEFT CURLY BRACKET\r
+    u'|'        #  0x7C -> VERTICAL LINE\r
+    u'}'        #  0x7D -> RIGHT CURLY BRACKET\r
+    u'~'        #  0x7E -> TILDE\r
+    u'\x7f'     #  0x7F -> CONTROL CHARACTER\r
+    u'\x80'\r
+    u'\x81'\r
+    u'\xe9'     #  0x82 -> LATIN SMALL LETTER E WITH ACUTE\r
+    u'\xe2'     #  0x83 -> LATIN SMALL LETTER A WITH CIRCUMFLEX\r
+    u'\x84'\r
+    u'\xe0'     #  0x85 -> LATIN SMALL LETTER A WITH GRAVE\r
+    u'\x86'\r
+    u'\xe7'     #  0x87 -> LATIN SMALL LETTER C WITH CEDILLA\r
+    u'\xea'     #  0x88 -> LATIN SMALL LETTER E WITH CIRCUMFLEX\r
+    u'\xeb'     #  0x89 -> LATIN SMALL LETTER E WITH DIAERESIS\r
+    u'\xe8'     #  0x8A -> LATIN SMALL LETTER E WITH GRAVE\r
+    u'\xef'     #  0x8B -> LATIN SMALL LETTER I WITH DIAERESIS\r
+    u'\xee'     #  0x8C -> LATIN SMALL LETTER I WITH CIRCUMFLEX\r
+    u'\x8d'\r
+    u'\x8e'\r
+    u'\x8f'\r
+    u'\x90'\r
+    u'\u0651'   #  0x91 -> ARABIC SHADDA\r
+    u'\u0652'   #  0x92 -> ARABIC SUKUN\r
+    u'\xf4'     #  0x93 -> LATIN SMALL LETTER O WITH CIRCUMFLEX\r
+    u'\xa4'     #  0x94 -> CURRENCY SIGN\r
+    u'\u0640'   #  0x95 -> ARABIC TATWEEL\r
+    u'\xfb'     #  0x96 -> LATIN SMALL LETTER U WITH CIRCUMFLEX\r
+    u'\xf9'     #  0x97 -> LATIN SMALL LETTER U WITH GRAVE\r
+    u'\u0621'   #  0x98 -> ARABIC LETTER HAMZA\r
+    u'\u0622'   #  0x99 -> ARABIC LETTER ALEF WITH MADDA ABOVE\r
+    u'\u0623'   #  0x9A -> ARABIC LETTER ALEF WITH HAMZA ABOVE\r
+    u'\u0624'   #  0x9B -> ARABIC LETTER WAW WITH HAMZA ABOVE\r
+    u'\xa3'     #  0x9C -> POUND SIGN\r
+    u'\u0625'   #  0x9D -> ARABIC LETTER ALEF WITH HAMZA BELOW\r
+    u'\u0626'   #  0x9E -> ARABIC LETTER YEH WITH HAMZA ABOVE\r
+    u'\u0627'   #  0x9F -> ARABIC LETTER ALEF\r
+    u'\u0628'   #  0xA0 -> ARABIC LETTER BEH\r
+    u'\u0629'   #  0xA1 -> ARABIC LETTER TEH MARBUTA\r
+    u'\u062a'   #  0xA2 -> ARABIC LETTER TEH\r
+    u'\u062b'   #  0xA3 -> ARABIC LETTER THEH\r
+    u'\u062c'   #  0xA4 -> ARABIC LETTER JEEM\r
+    u'\u062d'   #  0xA5 -> ARABIC LETTER HAH\r
+    u'\u062e'   #  0xA6 -> ARABIC LETTER KHAH\r
+    u'\u062f'   #  0xA7 -> ARABIC LETTER DAL\r
+    u'\u0630'   #  0xA8 -> ARABIC LETTER THAL\r
+    u'\u0631'   #  0xA9 -> ARABIC LETTER REH\r
+    u'\u0632'   #  0xAA -> ARABIC LETTER ZAIN\r
+    u'\u0633'   #  0xAB -> ARABIC LETTER SEEN\r
+    u'\u0634'   #  0xAC -> ARABIC LETTER SHEEN\r
+    u'\u0635'   #  0xAD -> ARABIC LETTER SAD\r
+    u'\xab'     #  0xAE -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK\r
+    u'\xbb'     #  0xAF -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK\r
+    u'\u2591'   #  0xB0 -> LIGHT SHADE\r
+    u'\u2592'   #  0xB1 -> MEDIUM SHADE\r
+    u'\u2593'   #  0xB2 -> DARK SHADE\r
+    u'\u2502'   #  0xB3 -> BOX DRAWINGS LIGHT VERTICAL\r
+    u'\u2524'   #  0xB4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT\r
+    u'\u2561'   #  0xB5 -> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE\r
+    u'\u2562'   #  0xB6 -> BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE\r
+    u'\u2556'   #  0xB7 -> BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE\r
+    u'\u2555'   #  0xB8 -> BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE\r
+    u'\u2563'   #  0xB9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT\r
+    u'\u2551'   #  0xBA -> BOX DRAWINGS DOUBLE VERTICAL\r
+    u'\u2557'   #  0xBB -> BOX DRAWINGS DOUBLE DOWN AND LEFT\r
+    u'\u255d'   #  0xBC -> BOX DRAWINGS DOUBLE UP AND LEFT\r
+    u'\u255c'   #  0xBD -> BOX DRAWINGS UP DOUBLE AND LEFT SINGLE\r
+    u'\u255b'   #  0xBE -> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE\r
+    u'\u2510'   #  0xBF -> BOX DRAWINGS LIGHT DOWN AND LEFT\r
+    u'\u2514'   #  0xC0 -> BOX DRAWINGS LIGHT UP AND RIGHT\r
+    u'\u2534'   #  0xC1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL\r
+    u'\u252c'   #  0xC2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL\r
+    u'\u251c'   #  0xC3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT\r
+    u'\u2500'   #  0xC4 -> BOX DRAWINGS LIGHT HORIZONTAL\r
+    u'\u253c'   #  0xC5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL\r
+    u'\u255e'   #  0xC6 -> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE\r
+    u'\u255f'   #  0xC7 -> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE\r
+    u'\u255a'   #  0xC8 -> BOX DRAWINGS DOUBLE UP AND RIGHT\r
+    u'\u2554'   #  0xC9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT\r
+    u'\u2569'   #  0xCA -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL\r
+    u'\u2566'   #  0xCB -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL\r
+    u'\u2560'   #  0xCC -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT\r
+    u'\u2550'   #  0xCD -> BOX DRAWINGS DOUBLE HORIZONTAL\r
+    u'\u256c'   #  0xCE -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL\r
+    u'\u2567'   #  0xCF -> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE\r
+    u'\u2568'   #  0xD0 -> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE\r
+    u'\u2564'   #  0xD1 -> BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE\r
+    u'\u2565'   #  0xD2 -> BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE\r
+    u'\u2559'   #  0xD3 -> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE\r
+    u'\u2558'   #  0xD4 -> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE\r
+    u'\u2552'   #  0xD5 -> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE\r
+    u'\u2553'   #  0xD6 -> BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE\r
+    u'\u256b'   #  0xD7 -> BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE\r
+    u'\u256a'   #  0xD8 -> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE\r
+    u'\u2518'   #  0xD9 -> BOX DRAWINGS LIGHT UP AND LEFT\r
+    u'\u250c'   #  0xDA -> BOX DRAWINGS LIGHT DOWN AND RIGHT\r
+    u'\u2588'   #  0xDB -> FULL BLOCK\r
+    u'\u2584'   #  0xDC -> LOWER HALF BLOCK\r
+    u'\u258c'   #  0xDD -> LEFT HALF BLOCK\r
+    u'\u2590'   #  0xDE -> RIGHT HALF BLOCK\r
+    u'\u2580'   #  0xDF -> UPPER HALF BLOCK\r
+    u'\u0636'   #  0xE0 -> ARABIC LETTER DAD\r
+    u'\u0637'   #  0xE1 -> ARABIC LETTER TAH\r
+    u'\u0638'   #  0xE2 -> ARABIC LETTER ZAH\r
+    u'\u0639'   #  0xE3 -> ARABIC LETTER AIN\r
+    u'\u063a'   #  0xE4 -> ARABIC LETTER GHAIN\r
+    u'\u0641'   #  0xE5 -> ARABIC LETTER FEH\r
+    u'\xb5'     #  0xE6 -> MICRO SIGN\r
+    u'\u0642'   #  0xE7 -> ARABIC LETTER QAF\r
+    u'\u0643'   #  0xE8 -> ARABIC LETTER KAF\r
+    u'\u0644'   #  0xE9 -> ARABIC LETTER LAM\r
+    u'\u0645'   #  0xEA -> ARABIC LETTER MEEM\r
+    u'\u0646'   #  0xEB -> ARABIC LETTER NOON\r
+    u'\u0647'   #  0xEC -> ARABIC LETTER HEH\r
+    u'\u0648'   #  0xED -> ARABIC LETTER WAW\r
+    u'\u0649'   #  0xEE -> ARABIC LETTER ALEF MAKSURA\r
+    u'\u064a'   #  0xEF -> ARABIC LETTER YEH\r
+    u'\u2261'   #  0xF0 -> IDENTICAL TO\r
+    u'\u064b'   #  0xF1 -> ARABIC FATHATAN\r
+    u'\u064c'   #  0xF2 -> ARABIC DAMMATAN\r
+    u'\u064d'   #  0xF3 -> ARABIC KASRATAN\r
+    u'\u064e'   #  0xF4 -> ARABIC FATHA\r
+    u'\u064f'   #  0xF5 -> ARABIC DAMMA\r
+    u'\u0650'   #  0xF6 -> ARABIC KASRA\r
+    u'\u2248'   #  0xF7 -> ALMOST EQUAL TO\r
+    u'\xb0'     #  0xF8 -> DEGREE SIGN\r
+    u'\u2219'   #  0xF9 -> BULLET OPERATOR\r
+    u'\xb7'     #  0xFA -> MIDDLE DOT\r
+    u'\u221a'   #  0xFB -> SQUARE ROOT\r
+    u'\u207f'   #  0xFC -> SUPERSCRIPT LATIN SMALL LETTER N\r
+    u'\xb2'     #  0xFD -> SUPERSCRIPT TWO\r
+    u'\u25a0'   #  0xFE -> BLACK SQUARE\r
+    u'\xa0'     #  0xFF -> NO-BREAK SPACE\r
+)\r
+\r
+### Encoding table\r
+encoding_table=codecs.charmap_build(decoding_table)\r
+\r
+\r

diff --git a/Tools/unicode/genwincodec.py b/Tools/unicode/genwincodec.py
new file mode 100644 (file)
index 0000000..32dcada
--- /dev/null
+++ b/Tools/unicode/genwincodec.py
@@ -0,0 +1,61 @@
+"""This script generates a Python codec module from a Windows Code Page.\r
+\r
+It uses the function MultiByteToWideChar to generate a decoding table.\r
+"""\r
+\r
+import ctypes\r
+from ctypes import wintypes\r
+from gencodec import codegen\r
+import unicodedata\r
+\r
+def genwinmap(codepage):\r
+    MultiByteToWideChar = ctypes.windll.kernel32.MultiByteToWideChar\r
+    MultiByteToWideChar.argtypes = [wintypes.UINT, wintypes.DWORD,\r
+                                    wintypes.LPCSTR, ctypes.c_int,\r
+                                    wintypes.LPWSTR, ctypes.c_int]\r
+    MultiByteToWideChar.restype = ctypes.c_int\r
+\r
+    enc2uni = {}\r
+\r
+    for i in range(32) + [127]:\r
+        enc2uni[i] = (i, 'CONTROL CHARACTER')\r
+\r
+    for i in range(256):\r
+        buf = ctypes.create_unicode_buffer(2)\r
+        ret = MultiByteToWideChar(\r
+            codepage, 0,\r
+            chr(i), 1,\r
+            buf, 2)\r
+        assert ret == 1, "invalid code page"\r
+        assert buf[1] == '\x00'\r
+        try:\r
+            name = unicodedata.name(buf[0])\r
+        except ValueError:\r
+            try:\r
+                name = enc2uni[i][1]\r
+            except KeyError:\r
+                name = ''\r
+\r
+        enc2uni[i] = (ord(buf[0]), name)\r
+\r
+    return enc2uni\r
+\r
+def genwincodec(codepage):\r
+    import platform\r
+    map = genwinmap(codepage)\r
+    encodingname = 'cp%d' % codepage\r
+    code = codegen("", map, encodingname)\r
+    # Replace first lines with our own docstring\r
+    code = '''\\r
+"""Python Character Mapping Codec %s generated on Windows:\r
+%s with the command:\r
+  python Tools/unicode/genwincodec.py %s\r
+"""#"\r
+''' % (encodingname, ' '.join(platform.win32_ver()), codepage\r
+      ) + code.split('"""#"', 1)[1]\r
+\r
+    print code\r
+\r
+if __name__ == '__main__':\r
+    import sys\r
+    genwincodec(int(sys.argv[1]))\r

diff --git a/Tools/unicode/genwincodecs.bat b/Tools/unicode/genwincodecs.bat
new file mode 100644 (file)
index 0000000..6a6a671
--- /dev/null
+++ b/Tools/unicode/genwincodecs.bat
@@ -0,0 +1,7 @@
+@rem Recreate some python charmap codecs from the Windows function\r
+@rem MultiByteToWideChar.\r
+\r
+@cd /d %~dp0\r
+@mkdir build\r
+@rem Arabic DOS code page\r
+c:\python26\python genwincodec.py 720 > build/cp720.py\r