Lines Matching full:overrides
58 # Apply manual overrides for special cases
64 """Get special case overrides that need different handling than unidecode
67 overrides = {}
72 overrides[0x00C6] = ord('E') # Æ LATIN CAPITAL LETTER AE -> E (unidecode: "AE")
73 overrides[0x00E6] = ord('e') # æ LATIN SMALL LETTER AE -> e (unidecode: "ae")
74 overrides[0x0152] = ord('E') # ΠLATIN CAPITAL LIGATURE OE -> E (unidecode: "OE")
75 overrides[0x0153] = ord('e') # œ LATIN SMALL LETTER LIGATURE OE -> e (unidecode: "oe")
76 overrides[0x00DF] = ord('s') # ß LATIN SMALL LETTER SHARP S -> s (unidecode: "ss")
79 overrides[0x2264] = ord('<') # ≤ LESS-THAN OR EQUAL TO -> < (unidecode: "<=")
80 overrides[0x2265] = ord('>') # ≥ GREATER-THAN OR EQUAL TO -> > (unidecode: ">=")
83 overrides[0x2260] = ord('#') # ≠ NOT EQUAL TO -> # (unidecode: empty string)
87 overrides[cp] = ord('#') # ▖ ▗ ▘ ▙ etc. - map to # (unidecode: empty string)
91 overrides[0x2192] = ord('>') # → RIGHTWARDS ARROW -> > (unidecode: "-")
92 overrides[0x2190] = ord('<') # ← LEFTWARDS ARROW -> < (unidecode: "-")
93 overrides[0x2191] = ord('^') # ↑ UPWARDS ARROW -> ^ (unidecode: "|")
94 overrides[0x2193] = ord('v') # ↓ DOWNWARDS ARROW -> v (unidecode: "|")
97 overrides[0x21D0] = ord('<') # ⇐ LEFTWARDS DOUBLE ARROW -> <
98 overrides[0x21D1] = ord('^') # ⇑ UPWARDS DOUBLE ARROW -> ^
99 overrides[0x21D2] = ord('>') # ⇒ RIGHTWARDS DOUBLE ARROW -> >
100 overrides[0x21D3] = ord('v') # ⇓ DOWNWARDS DOUBLE ARROW -> v
104 overrides[0xFFE9] = ord('<') # ￩ HALFWIDTH LEFTWARDS ARROW -> < (unidecode: "-")
105 overrides[0xFFEA] = ord('^') # ￪ HALFWIDTH UPWARDS ARROW -> ^ (unidecode: "|")
106 overrides[0xFFEB] = ord('>') # ￫ HALFWIDTH RIGHTWARDS ARROW -> > (unidecode: "-")
107 overrides[0xFFEC] = ord('v') # ￬ HALFWIDTH DOWNWARDS ARROW -> v (unidecode: "|")
110 overrides[0x00A2] = ord('c') # ¢ CENT SIGN -> c
111 overrides[0x00A3] = ord('L') # £ POUND SIGN -> L
112 overrides[0x00A5] = ord('Y') # ¥ YEN SIGN -> Y
113 overrides[0x20AC] = ord('E') # € EURO SIGN -> E
116 overrides[0x00A7] = ord('S') # § SECTION SIGN -> S
117 overrides[0x00A9] = ord('C') # © COPYRIGHT SIGN -> C
118 overrides[0x00AE] = ord('R') # ® REGISTERED SIGN -> R
119 overrides[0x2122] = ord('T') # ™ TRADE MARK SIGN -> T
122 overrides[0x00B0] = ord('o') # ° DEGREE SIGN -> o
123 overrides[0x2103] = ord('C') # ℃ DEGREE CELSIUS -> C
124 overrides[0x2109] = ord('F') # ℉ DEGREE FAHRENHEIT -> F
127 overrides[0x00AB] = ord('<') # « LEFT-POINTING DOUBLE ANGLE QUOTATION MARK -> <
128 overrides[0x00BB] = ord('>') # » RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK -> >
131 overrides[0x2218] = ord('o') # ∘ RING OPERATOR -> o
132 overrides[0x2219] = ord('.') # ∙ BULLET OPERATOR -> .
137 overrides[cp] = ord('!') # Negated math symbols -> ! (not)
141 overrides[cp] = ord('#') # Negated equality symbols -> # (not equal)
145 overrides[cp] = ord('!') # Negated arrows -> ! (not)
149 overrides[cp] = ord('-') # Dashes and hyphens -> -
153 overrides[cp] = ord('?') # Question marks -> ?
157 overrides[cp] = ord('!') # Exclamation marks -> !
161 overrides[cp] = ord('*')
164 overrides[0x201E] = ord('"') # „ DOUBLE LOW-9 QUOTATION MARK
165 overrides[0x2023] = ord('>') # ‣ TRIANGULAR BULLET
166 overrides[0x2026] = ord('.') # … HORIZONTAL ELLIPSIS
167 overrides[0x2033] = ord('"') # ″ DOUBLE PRIME
168 overrides[0x204B] = ord('P') # ⁋ REVERSED PILCROW SIGN
169 overrides[0x204C] = ord('<') # ⁌ BLACK LEFTWARDS BULLET
170 overrides[0x204D] = ord('>') # ⁍ BLACK RIGHTWARDS BULLET
171 overrides[0x204F] = ord(';') # ⁏ REVERSED SEMICOLON
172 overrides[0x205B] = ord(':') # ⁛ FOUR DOT MARK
175 overrides[0x2713] = ord('v') # ✓ CHECK MARK
176 overrides[0x2714] = ord('V') # ✔ HEAVY CHECK MARK
180 overrides[cp] = ord('x') # Regular X marks -> x
182 overrides[cp] = ord('X') # Heavy X marks -> X
186 overrides[cp] = ord('*') # All star and asterisk symbols -> *
188 overrides[cp] = ord('*') # All star and asterisk symbols -> *
190 overrides[cp] = ord('*') # Last set of asterisk symbols -> *
192 overrides[cp] = ord('*') # Star operators -> *
198 overrides[0x2028] = 0 # LINE SEPARATOR (unidecode: '\n')
206 overrides[cp] = 0 # Double-width ASCII characters
208 return overrides