Lines Matching +full:0 +full:xff01
34 RANGE_MARKER = 0x00
40 # Process BMP characters (0x0000 - 0xFFFF) to keep table size manageable
41 for cp in range(0x0080, 0x10000): # Skip ASCII range (0x00-0x7F)
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)
86 for cp in range(0x2596, 0x259F+1):
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 -> .
136 …for cp in (0x2204, 0x2209, 0x220C, 0x2224, 0x2226, 0x226E, 0x226F, 0x2280, 0x2281, 0x2284, 0x2285):
140 for cp in (0x2241, 0x2244, 0x2249, 0x2262, 0x2268, 0x2269, 0x226D, 0x228A, 0x228B):
144 for cp in (0x219A, 0x219B, 0x21AE, 0x21CD, 0x21CE, 0x21CF):
148 for cp in (0x2010, 0x2011, 0x2012, 0x2013, 0x2014, 0x2015, 0x2043, 0x2052):
152 for cp in (0x203D, 0x2047, 0x2048):
156 for cp in (0x203C, 0x2049):
160 for cp in (0x2042, 0x2051, 0x2055):
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
179 for cp in (0x2715, 0x2717):
181 for cp in (0x2716, 0x2718):
185 for cp in (0x2605, 0x2606, 0x262A, 0x269D, 0x2698):
187 for cp in range(0x2721, 0x2746+1):
189 for cp in range(0x2749, 0x274B+1):
191 for cp in (0x229B, 0x22C6, 0x235F, 0x2363):
194 # Special exclusions with fallback value of 0
198 overrides[0x2028] = 0 # LINE SEPARATOR (unidecode: '\n')
201 # 0xFF01 (！) to 0xFF5E (～) -> ASCII 33 (!) to 126 (~)
205 for cp in range(0xFF01, 0xFF5E + 1):
206 overrides[cp] = 0 # Double-width ASCII characters
215 # Skip characters with fallback value of 0 (excluded characters)
216 if fallback == 0:
220 offset = code & 0xFF # Get the low byte (offset within page)
237 i = 0
244 entries[j][0] == entries[j-1][0] + 1 and # consecutive offsets
249 end_offset = entries[j-1][0]
313 start_index = 0
318 f.write(f"\t{{ 0x{page:02X}, {count}, {start_index} }},\n")
331 page_hex = f"0x{page:02X}"
336 offset_hex = f"0x{offset:02X}"
337 fallback_hex = f"0x{fallback:02X}"
346 f.write(f"\t{{ 0x{offset:02X}, 0x{fallback:02X} }}, /* {comment} */\n")