summaryrefslogtreecommitdiff
path: root/src/pip/_vendor/chardet/langhebrewmodel.py
diff options
context:
space:
mode:
Diffstat (limited to 'src/pip/_vendor/chardet/langhebrewmodel.py')
-rw-r--r--src/pip/_vendor/chardet/langhebrewmodel.py533
1 files changed, 265 insertions, 268 deletions
diff --git a/src/pip/_vendor/chardet/langhebrewmodel.py b/src/pip/_vendor/chardet/langhebrewmodel.py
index 484c652a4..56d297587 100644
--- a/src/pip/_vendor/chardet/langhebrewmodel.py
+++ b/src/pip/_vendor/chardet/langhebrewmodel.py
@@ -1,9 +1,5 @@
-#!/usr/bin/env python
-# -*- coding: utf-8 -*-
-
from pip._vendor.chardet.sbcharsetprober import SingleByteCharSetModel
-
# 3: Positive
# 2: Likely
# 1: Unlikely
@@ -4115,269 +4111,270 @@ HEBREW_LANG_MODEL = {
# Character Mapping Table(s):
WINDOWS_1255_HEBREW_CHAR_TO_ORDER = {
- 0: 255, # '\x00'
- 1: 255, # '\x01'
- 2: 255, # '\x02'
- 3: 255, # '\x03'
- 4: 255, # '\x04'
- 5: 255, # '\x05'
- 6: 255, # '\x06'
- 7: 255, # '\x07'
- 8: 255, # '\x08'
- 9: 255, # '\t'
- 10: 254, # '\n'
- 11: 255, # '\x0b'
- 12: 255, # '\x0c'
- 13: 254, # '\r'
- 14: 255, # '\x0e'
- 15: 255, # '\x0f'
- 16: 255, # '\x10'
- 17: 255, # '\x11'
- 18: 255, # '\x12'
- 19: 255, # '\x13'
- 20: 255, # '\x14'
- 21: 255, # '\x15'
- 22: 255, # '\x16'
- 23: 255, # '\x17'
- 24: 255, # '\x18'
- 25: 255, # '\x19'
- 26: 255, # '\x1a'
- 27: 255, # '\x1b'
- 28: 255, # '\x1c'
- 29: 255, # '\x1d'
- 30: 255, # '\x1e'
- 31: 255, # '\x1f'
- 32: 253, # ' '
- 33: 253, # '!'
- 34: 253, # '"'
- 35: 253, # '#'
- 36: 253, # '$'
- 37: 253, # '%'
- 38: 253, # '&'
- 39: 253, # "'"
- 40: 253, # '('
- 41: 253, # ')'
- 42: 253, # '*'
- 43: 253, # '+'
- 44: 253, # ','
- 45: 253, # '-'
- 46: 253, # '.'
- 47: 253, # '/'
- 48: 252, # '0'
- 49: 252, # '1'
- 50: 252, # '2'
- 51: 252, # '3'
- 52: 252, # '4'
- 53: 252, # '5'
- 54: 252, # '6'
- 55: 252, # '7'
- 56: 252, # '8'
- 57: 252, # '9'
- 58: 253, # ':'
- 59: 253, # ';'
- 60: 253, # '<'
- 61: 253, # '='
- 62: 253, # '>'
- 63: 253, # '?'
- 64: 253, # '@'
- 65: 69, # 'A'
- 66: 91, # 'B'
- 67: 79, # 'C'
- 68: 80, # 'D'
- 69: 92, # 'E'
- 70: 89, # 'F'
- 71: 97, # 'G'
- 72: 90, # 'H'
- 73: 68, # 'I'
- 74: 111, # 'J'
- 75: 112, # 'K'
- 76: 82, # 'L'
- 77: 73, # 'M'
- 78: 95, # 'N'
- 79: 85, # 'O'
- 80: 78, # 'P'
- 81: 121, # 'Q'
- 82: 86, # 'R'
- 83: 71, # 'S'
- 84: 67, # 'T'
- 85: 102, # 'U'
- 86: 107, # 'V'
- 87: 84, # 'W'
- 88: 114, # 'X'
- 89: 103, # 'Y'
- 90: 115, # 'Z'
- 91: 253, # '['
- 92: 253, # '\\'
- 93: 253, # ']'
- 94: 253, # '^'
- 95: 253, # '_'
- 96: 253, # '`'
- 97: 50, # 'a'
- 98: 74, # 'b'
- 99: 60, # 'c'
- 100: 61, # 'd'
- 101: 42, # 'e'
- 102: 76, # 'f'
- 103: 70, # 'g'
- 104: 64, # 'h'
- 105: 53, # 'i'
- 106: 105, # 'j'
- 107: 93, # 'k'
- 108: 56, # 'l'
- 109: 65, # 'm'
- 110: 54, # 'n'
- 111: 49, # 'o'
- 112: 66, # 'p'
- 113: 110, # 'q'
- 114: 51, # 'r'
- 115: 43, # 's'
- 116: 44, # 't'
- 117: 63, # 'u'
- 118: 81, # 'v'
- 119: 77, # 'w'
- 120: 98, # 'x'
- 121: 75, # 'y'
- 122: 108, # 'z'
- 123: 253, # '{'
- 124: 253, # '|'
- 125: 253, # '}'
- 126: 253, # '~'
- 127: 253, # '\x7f'
- 128: 124, # '€'
- 129: 202, # None
- 130: 203, # '‚'
- 131: 204, # 'ƒ'
- 132: 205, # '„'
- 133: 40, # '…'
- 134: 58, # '†'
- 135: 206, # '‡'
- 136: 207, # 'ˆ'
- 137: 208, # '‰'
- 138: 209, # None
- 139: 210, # '‹'
- 140: 211, # None
- 141: 212, # None
- 142: 213, # None
- 143: 214, # None
- 144: 215, # None
- 145: 83, # '‘'
- 146: 52, # '’'
- 147: 47, # '“'
- 148: 46, # '”'
- 149: 72, # '•'
- 150: 32, # '–'
- 151: 94, # '—'
- 152: 216, # '˜'
- 153: 113, # '™'
- 154: 217, # None
- 155: 109, # '›'
- 156: 218, # None
- 157: 219, # None
- 158: 220, # None
- 159: 221, # None
- 160: 34, # '\xa0'
- 161: 116, # '¡'
- 162: 222, # '¢'
- 163: 118, # '£'
- 164: 100, # '₪'
- 165: 223, # '¥'
- 166: 224, # '¦'
- 167: 117, # '§'
- 168: 119, # '¨'
- 169: 104, # '©'
- 170: 125, # '×'
- 171: 225, # '«'
- 172: 226, # '¬'
- 173: 87, # '\xad'
- 174: 99, # '®'
- 175: 227, # '¯'
- 176: 106, # '°'
- 177: 122, # '±'
- 178: 123, # '²'
- 179: 228, # '³'
- 180: 55, # '´'
- 181: 229, # 'µ'
- 182: 230, # '¶'
- 183: 101, # '·'
- 184: 231, # '¸'
- 185: 232, # '¹'
- 186: 120, # '÷'
- 187: 233, # '»'
- 188: 48, # '¼'
- 189: 39, # '½'
- 190: 57, # '¾'
- 191: 234, # '¿'
- 192: 30, # 'ְ'
- 193: 59, # 'ֱ'
- 194: 41, # 'ֲ'
- 195: 88, # 'ֳ'
- 196: 33, # 'ִ'
- 197: 37, # 'ֵ'
- 198: 36, # 'ֶ'
- 199: 31, # 'ַ'
- 200: 29, # 'ָ'
- 201: 35, # 'ֹ'
- 202: 235, # None
- 203: 62, # 'ֻ'
- 204: 28, # 'ּ'
- 205: 236, # 'ֽ'
- 206: 126, # '־'
- 207: 237, # 'ֿ'
- 208: 238, # '׀'
- 209: 38, # 'ׁ'
- 210: 45, # 'ׂ'
- 211: 239, # '׃'
- 212: 240, # 'װ'
- 213: 241, # 'ױ'
- 214: 242, # 'ײ'
- 215: 243, # '׳'
- 216: 127, # '״'
- 217: 244, # None
- 218: 245, # None
- 219: 246, # None
- 220: 247, # None
- 221: 248, # None
- 222: 249, # None
- 223: 250, # None
- 224: 9, # 'א'
- 225: 8, # 'ב'
- 226: 20, # 'ג'
- 227: 16, # 'ד'
- 228: 3, # 'ה'
- 229: 2, # 'ו'
- 230: 24, # 'ז'
- 231: 14, # 'ח'
- 232: 22, # 'ט'
- 233: 1, # 'י'
- 234: 25, # 'ך'
- 235: 15, # 'כ'
- 236: 4, # 'ל'
- 237: 11, # 'ם'
- 238: 6, # 'מ'
- 239: 23, # 'ן'
- 240: 12, # 'נ'
- 241: 19, # 'ס'
- 242: 13, # 'ע'
- 243: 26, # 'ף'
- 244: 18, # 'פ'
- 245: 27, # 'ץ'
- 246: 21, # 'צ'
- 247: 17, # 'ק'
- 248: 7, # 'ר'
- 249: 10, # 'ש'
- 250: 5, # 'ת'
- 251: 251, # None
- 252: 252, # None
- 253: 128, # '\u200e'
- 254: 96, # '\u200f'
- 255: 253, # None
+ 0: 255, # '\x00'
+ 1: 255, # '\x01'
+ 2: 255, # '\x02'
+ 3: 255, # '\x03'
+ 4: 255, # '\x04'
+ 5: 255, # '\x05'
+ 6: 255, # '\x06'
+ 7: 255, # '\x07'
+ 8: 255, # '\x08'
+ 9: 255, # '\t'
+ 10: 254, # '\n'
+ 11: 255, # '\x0b'
+ 12: 255, # '\x0c'
+ 13: 254, # '\r'
+ 14: 255, # '\x0e'
+ 15: 255, # '\x0f'
+ 16: 255, # '\x10'
+ 17: 255, # '\x11'
+ 18: 255, # '\x12'
+ 19: 255, # '\x13'
+ 20: 255, # '\x14'
+ 21: 255, # '\x15'
+ 22: 255, # '\x16'
+ 23: 255, # '\x17'
+ 24: 255, # '\x18'
+ 25: 255, # '\x19'
+ 26: 255, # '\x1a'
+ 27: 255, # '\x1b'
+ 28: 255, # '\x1c'
+ 29: 255, # '\x1d'
+ 30: 255, # '\x1e'
+ 31: 255, # '\x1f'
+ 32: 253, # ' '
+ 33: 253, # '!'
+ 34: 253, # '"'
+ 35: 253, # '#'
+ 36: 253, # '$'
+ 37: 253, # '%'
+ 38: 253, # '&'
+ 39: 253, # "'"
+ 40: 253, # '('
+ 41: 253, # ')'
+ 42: 253, # '*'
+ 43: 253, # '+'
+ 44: 253, # ','
+ 45: 253, # '-'
+ 46: 253, # '.'
+ 47: 253, # '/'
+ 48: 252, # '0'
+ 49: 252, # '1'
+ 50: 252, # '2'
+ 51: 252, # '3'
+ 52: 252, # '4'
+ 53: 252, # '5'
+ 54: 252, # '6'
+ 55: 252, # '7'
+ 56: 252, # '8'
+ 57: 252, # '9'
+ 58: 253, # ':'
+ 59: 253, # ';'
+ 60: 253, # '<'
+ 61: 253, # '='
+ 62: 253, # '>'
+ 63: 253, # '?'
+ 64: 253, # '@'
+ 65: 69, # 'A'
+ 66: 91, # 'B'
+ 67: 79, # 'C'
+ 68: 80, # 'D'
+ 69: 92, # 'E'
+ 70: 89, # 'F'
+ 71: 97, # 'G'
+ 72: 90, # 'H'
+ 73: 68, # 'I'
+ 74: 111, # 'J'
+ 75: 112, # 'K'
+ 76: 82, # 'L'
+ 77: 73, # 'M'
+ 78: 95, # 'N'
+ 79: 85, # 'O'
+ 80: 78, # 'P'
+ 81: 121, # 'Q'
+ 82: 86, # 'R'
+ 83: 71, # 'S'
+ 84: 67, # 'T'
+ 85: 102, # 'U'
+ 86: 107, # 'V'
+ 87: 84, # 'W'
+ 88: 114, # 'X'
+ 89: 103, # 'Y'
+ 90: 115, # 'Z'
+ 91: 253, # '['
+ 92: 253, # '\\'
+ 93: 253, # ']'
+ 94: 253, # '^'
+ 95: 253, # '_'
+ 96: 253, # '`'
+ 97: 50, # 'a'
+ 98: 74, # 'b'
+ 99: 60, # 'c'
+ 100: 61, # 'd'
+ 101: 42, # 'e'
+ 102: 76, # 'f'
+ 103: 70, # 'g'
+ 104: 64, # 'h'
+ 105: 53, # 'i'
+ 106: 105, # 'j'
+ 107: 93, # 'k'
+ 108: 56, # 'l'
+ 109: 65, # 'm'
+ 110: 54, # 'n'
+ 111: 49, # 'o'
+ 112: 66, # 'p'
+ 113: 110, # 'q'
+ 114: 51, # 'r'
+ 115: 43, # 's'
+ 116: 44, # 't'
+ 117: 63, # 'u'
+ 118: 81, # 'v'
+ 119: 77, # 'w'
+ 120: 98, # 'x'
+ 121: 75, # 'y'
+ 122: 108, # 'z'
+ 123: 253, # '{'
+ 124: 253, # '|'
+ 125: 253, # '}'
+ 126: 253, # '~'
+ 127: 253, # '\x7f'
+ 128: 124, # '€'
+ 129: 202, # None
+ 130: 203, # '‚'
+ 131: 204, # 'ƒ'
+ 132: 205, # '„'
+ 133: 40, # '…'
+ 134: 58, # '†'
+ 135: 206, # '‡'
+ 136: 207, # 'ˆ'
+ 137: 208, # '‰'
+ 138: 209, # None
+ 139: 210, # '‹'
+ 140: 211, # None
+ 141: 212, # None
+ 142: 213, # None
+ 143: 214, # None
+ 144: 215, # None
+ 145: 83, # '‘'
+ 146: 52, # '’'
+ 147: 47, # '“'
+ 148: 46, # '”'
+ 149: 72, # '•'
+ 150: 32, # '–'
+ 151: 94, # '—'
+ 152: 216, # '˜'
+ 153: 113, # '™'
+ 154: 217, # None
+ 155: 109, # '›'
+ 156: 218, # None
+ 157: 219, # None
+ 158: 220, # None
+ 159: 221, # None
+ 160: 34, # '\xa0'
+ 161: 116, # '¡'
+ 162: 222, # '¢'
+ 163: 118, # '£'
+ 164: 100, # '₪'
+ 165: 223, # '¥'
+ 166: 224, # '¦'
+ 167: 117, # '§'
+ 168: 119, # '¨'
+ 169: 104, # '©'
+ 170: 125, # '×'
+ 171: 225, # '«'
+ 172: 226, # '¬'
+ 173: 87, # '\xad'
+ 174: 99, # '®'
+ 175: 227, # '¯'
+ 176: 106, # '°'
+ 177: 122, # '±'
+ 178: 123, # '²'
+ 179: 228, # '³'
+ 180: 55, # '´'
+ 181: 229, # 'µ'
+ 182: 230, # '¶'
+ 183: 101, # '·'
+ 184: 231, # '¸'
+ 185: 232, # '¹'
+ 186: 120, # '÷'
+ 187: 233, # '»'
+ 188: 48, # '¼'
+ 189: 39, # '½'
+ 190: 57, # '¾'
+ 191: 234, # '¿'
+ 192: 30, # 'ְ'
+ 193: 59, # 'ֱ'
+ 194: 41, # 'ֲ'
+ 195: 88, # 'ֳ'
+ 196: 33, # 'ִ'
+ 197: 37, # 'ֵ'
+ 198: 36, # 'ֶ'
+ 199: 31, # 'ַ'
+ 200: 29, # 'ָ'
+ 201: 35, # 'ֹ'
+ 202: 235, # None
+ 203: 62, # 'ֻ'
+ 204: 28, # 'ּ'
+ 205: 236, # 'ֽ'
+ 206: 126, # '־'
+ 207: 237, # 'ֿ'
+ 208: 238, # '׀'
+ 209: 38, # 'ׁ'
+ 210: 45, # 'ׂ'
+ 211: 239, # '׃'
+ 212: 240, # 'װ'
+ 213: 241, # 'ױ'
+ 214: 242, # 'ײ'
+ 215: 243, # '׳'
+ 216: 127, # '״'
+ 217: 244, # None
+ 218: 245, # None
+ 219: 246, # None
+ 220: 247, # None
+ 221: 248, # None
+ 222: 249, # None
+ 223: 250, # None
+ 224: 9, # 'א'
+ 225: 8, # 'ב'
+ 226: 20, # 'ג'
+ 227: 16, # 'ד'
+ 228: 3, # 'ה'
+ 229: 2, # 'ו'
+ 230: 24, # 'ז'
+ 231: 14, # 'ח'
+ 232: 22, # 'ט'
+ 233: 1, # 'י'
+ 234: 25, # 'ך'
+ 235: 15, # 'כ'
+ 236: 4, # 'ל'
+ 237: 11, # 'ם'
+ 238: 6, # 'מ'
+ 239: 23, # 'ן'
+ 240: 12, # 'נ'
+ 241: 19, # 'ס'
+ 242: 13, # 'ע'
+ 243: 26, # 'ף'
+ 244: 18, # 'פ'
+ 245: 27, # 'ץ'
+ 246: 21, # 'צ'
+ 247: 17, # 'ק'
+ 248: 7, # 'ר'
+ 249: 10, # 'ש'
+ 250: 5, # 'ת'
+ 251: 251, # None
+ 252: 252, # None
+ 253: 128, # '\u200e'
+ 254: 96, # '\u200f'
+ 255: 253, # None
}
-WINDOWS_1255_HEBREW_MODEL = SingleByteCharSetModel(charset_name='windows-1255',
- language='Hebrew',
- char_to_order_map=WINDOWS_1255_HEBREW_CHAR_TO_ORDER,
- language_model=HEBREW_LANG_MODEL,
- typical_positive_ratio=0.984004,
- keep_ascii_letters=False,
- alphabet='אבגדהוזחטיךכלםמןנסעףפץצקרשתװױײ')
-
+WINDOWS_1255_HEBREW_MODEL = SingleByteCharSetModel(
+ charset_name="windows-1255",
+ language="Hebrew",
+ char_to_order_map=WINDOWS_1255_HEBREW_CHAR_TO_ORDER,
+ language_model=HEBREW_LANG_MODEL,
+ typical_positive_ratio=0.984004,
+ keep_ascii_letters=False,
+ alphabet="אבגדהוזחטיךכלםמןנסעףפץצקרשתװױײ",
+)