Appearance
| 1 | namespace Syntax.Process is | |
| 2 | use Logging; | |
| 3 | use Source.LOCATION; | |
| 4 | ||
| 5 | use Semantic.Lookups.InnateSymbolLookup; | |
| 6 | use Semantic.Types.Type; | |
| 7 | ||
| 8 | use IR.Values; | |
| 9 | ||
| 10 | // Classify a parsed integer literal: pick the target type from | |
| 11 | // any trailing type-suffix (`b`/`s`/`i`/`l`/`w`/`c`, optionally | |
| 12 | // prefixed by `u` for unsigned or `s` for sign-without-width), | |
| 13 | // strip the suffix from the value string, and range-check the | |
| 14 | // result against the chosen type. The lexer accepts arbitrarily | |
| 15 | // long digit strings and any suffix combination; this is where | |
| 16 | // suffix → type and value → fits-in-type get decided. | |
| 17 | // | |
| 18 | // Bounds errors and "character literals cannot be unsigned" | |
| 19 | // structural errors are logged via the constructor-supplied | |
| 20 | // logger. Callers receive a non-null `Literal.NUMBER` on success | |
| 21 | // and null on either kind of failure. | |
| 22 | // | |
| 23 | // The `fits` / `parse_hex` / `max_unsigned_for` helpers are | |
| 24 | // exposed for unit testing in isolation from the suffix path. | |
| 25 | class NUMERIC_LITERAL_CLASSIFIER is | |
| 26 | _logger: Logger; | |
| 27 | _innate_symbol_lookup: InnateSymbolLookup; | |
| 28 | ||
| 29 | init(logger: Logger, innate_symbol_lookup: InnateSymbolLookup) is | |
| 30 | super.init(); | |
| 31 | ||
| 32 | _logger = logger; | |
| 33 | _innate_symbol_lookup = innate_symbol_lookup; | |
| 34 | si | |
| 35 | ||
| 36 | classify_integer(location: LOCATION, value_string: string) -> Literal.NUMBER? is | |
| 37 | let type: Type mut; | |
| 38 | let suffix mut = "i4"; | |
| 39 | let conv: string? mut = null; | |
| 40 | let working mut = value_string; | |
| 41 | ||
| 42 | let last_char = working.get_chars(working.length - 1); | |
| 43 | ||
| 44 | if "bBcCsSiIlLwWuU".contains(last_char) then | |
| 45 | let is_unsigned mut = false; | |
| 46 | let cutoff mut = working.length; | |
| 47 | ||
| 48 | if last_char == 'u' \/ last_char == 'U' then | |
| 49 | is_unsigned = true; | |
| 50 | cutoff = cutoff - 1; | |
| 51 | elif working.length > 2 then | |
| 52 | let sign_char = working.get_chars(working.length - 2); | |
| 53 | ||
| 54 | if sign_char == 'u' \/ sign_char == 'U' then | |
| 55 | is_unsigned = true; | |
| 56 | cutoff = cutoff - 1; | |
| 57 | elif sign_char == 's' \/ sign_char == 'S' then | |
| 58 | cutoff = cutoff - 1; | |
| 59 | fi | |
| 60 | fi | |
| 61 | ||
| 62 | case last_char | |
| 63 | when 'b', 'B' then | |
| 64 | if is_unsigned then | |
| 65 | type = _innate_symbol_lookup.get_ubyte_type(); | |
| 66 | else | |
| 67 | type = _innate_symbol_lookup.get_byte_type(); | |
| 68 | fi | |
| 69 | ||
| 70 | cutoff = cutoff - 1; | |
| 71 | ||
| 72 | when 'c', 'C' then | |
| 73 | if is_unsigned then | |
| 74 | _logger.error(location, "numeric character literal cannot be unsigned"); | |
| 75 | else | |
| 76 | type = _innate_symbol_lookup.get_char_type(); | |
| 77 | fi | |
| 78 | ||
| 79 | cutoff = cutoff - 1; | |
| 80 | ||
| 81 | when 's', 'S' then | |
| 82 | if is_unsigned then | |
| 83 | type = _innate_symbol_lookup.get_ushort_type(); | |
| 84 | else | |
| 85 | type = _innate_symbol_lookup.get_short_type(); | |
| 86 | fi | |
| 87 | ||
| 88 | cutoff = cutoff - 1; | |
| 89 | ||
| 90 | when 'i', 'I' then | |
| 91 | if is_unsigned then | |
| 92 | type = _innate_symbol_lookup.get_uint_type(); | |
| 93 | else | |
| 94 | type = _innate_symbol_lookup.get_int_type(); | |
| 95 | fi | |
| 96 | ||
| 97 | cutoff = cutoff - 1; | |
| 98 | ||
| 99 | when 'l', 'L' then | |
| 100 | suffix = "i8"; | |
| 101 | ||
| 102 | if is_unsigned then | |
| 103 | type = _innate_symbol_lookup.get_ulong_type(); | |
| 104 | else | |
| 105 | type = _innate_symbol_lookup.get_long_type(); | |
| 106 | fi | |
| 107 | ||
| 108 | cutoff = cutoff - 1; | |
| 109 | ||
| 110 | when 'w', 'W' then | |
| 111 | suffix = "i8"; | |
| 112 | ||
| 113 | if is_unsigned then | |
| 114 | type = _innate_symbol_lookup.get_uword_type(); | |
| 115 | conv = "conv.u"; | |
| 116 | else | |
| 117 | type = _innate_symbol_lookup.get_word_type(); | |
| 118 | conv = "conv.i"; | |
| 119 | fi | |
| 120 | ||
| 121 | cutoff = cutoff - 1; | |
| 122 | ||
| 123 | else | |
| 124 | if is_unsigned then | |
| 125 | type = _innate_symbol_lookup.get_uint_type(); | |
| 126 | else | |
| 127 | type = _innate_symbol_lookup.get_int_type(); | |
| 128 | fi | |
| 129 | esac | |
| 130 | ||
| 131 | working = working.substring(0, cutoff); | |
| 132 | else | |
| 133 | type = _innate_symbol_lookup.get_int_type(); | |
| 134 | fi | |
| 135 | ||
| 136 | // the resolved type may be null | |
| 137 | @suppress("presence-test-non-optional") | |
| 138 | if !type? then | |
| 139 | return null; | |
| 140 | fi | |
| 141 | ||
| 142 | if !fits(location, working, type) then | |
| 143 | return null; | |
| 144 | fi | |
| 145 | ||
| 146 | return Literal.NUMBER(working, type, suffix, conv); | |
| 147 | si | |
| 148 | ||
| 149 | // Range-check a parsed integer literal against its target | |
| 150 | // type. Comparing as a ulong using each type's unsigned max | |
| 151 | // preserves the standard "literal fits in N unsigned bits, | |
| 152 | // unary minus negates separately" convention. Without this, | |
| 153 | // oversize literals reach IL emission and ilasm rejects the | |
| 154 | // `ldc.i4 <bignum>` with an opaque syntax error (issue #580). | |
| 155 | fits(location: LOCATION, value_string: string, type: Type) -> bool is | |
| 156 | let (ok, parsed) = | |
| 157 | if value_string.starts_with("0x") \/ value_string.starts_with("0X") then | |
| 158 | parse_hex(value_string.substring(2, value_string.length - 2)); | |
| 159 | else | |
| 160 | let v: ulong mut = _; | |
| 161 | let parsed_ok = ulong.try_parse(value_string, v ref); | |
| 162 | (parsed_ok, v); | |
| 163 | fi; | |
| 164 | ||
| 165 | if !ok then | |
| 166 | _logger.error(location, "numeric literal {value_string} does not fit in {type}"); | |
| 167 | return false; | |
| 168 | fi | |
| 169 | ||
| 170 | let max_value = max_unsigned_for(type); | |
| 171 | ||
| 172 | if max_value > 0UL /\ parsed > max_value then | |
| 173 | _logger.error(location, "numeric literal {value_string} does not fit in {type}"); | |
| 174 | return false; | |
| 175 | fi | |
| 176 | ||
| 177 | return true; | |
| 178 | si | |
| 179 | ||
| 180 | // Parse a hex digit string (no `0x` prefix) into a ulong. | |
| 181 | // Returns (false, 0) for empty, > 16 digits, or any non-hex | |
| 182 | // character. | |
| 183 | parse_hex(s: string) -> (ok: bool, value: ulong) is | |
| 184 | if s.length == 0 \/ s.length > 16 then | |
| 185 | return (false, 0UL); | |
| 186 | fi | |
| 187 | ||
| 188 | let value: ulong mut = 0UL; | |
| 189 | ||
| 190 | for c in s do | |
| 191 | let digit: int mut; | |
| 192 | ||
| 193 | if c >= '0' /\ c <= '9' then | |
| 194 | digit = cast int(c) - cast int('0'); | |
| 195 | elif c >= 'a' /\ c <= 'f' then | |
| 196 | digit = cast int(c) - cast int('a') + 10; | |
| 197 | elif c >= 'A' /\ c <= 'F' then | |
| 198 | digit = cast int(c) - cast int('A') + 10; | |
| 199 | else | |
| 200 | return (false, 0UL); | |
| 201 | fi | |
| 202 | ||
| 203 | value = (value << 4UL) + cast ulong(digit); | |
| 204 | od | |
| 205 | ||
| 206 | return (true, value); | |
| 207 | si | |
| 208 | ||
| 209 | // Largest non-negative literal value the type accepts, or 0 | |
| 210 | // to signal "no narrower bound than the ulong-parse range" | |
| 211 | // (used for ulong / uword: ulong.try_parse already enforces | |
| 212 | // the 64-bit ceiling). 0 makes `fits` skip the narrower- | |
| 213 | // bound check. | |
| 214 | max_unsigned_for(type: Type) -> ulong is | |
| 215 | if type.matches(_innate_symbol_lookup.get_byte_type()) then | |
| 216 | return 127UL; | |
| 217 | elif type.matches(_innate_symbol_lookup.get_ubyte_type()) then | |
| 218 | return 255UL; | |
| 219 | elif type.matches(_innate_symbol_lookup.get_short_type()) then | |
| 220 | return 32767UL; | |
| 221 | elif type.matches(_innate_symbol_lookup.get_ushort_type()) then | |
| 222 | return 65535UL; | |
| 223 | elif type.matches(_innate_symbol_lookup.get_char_type()) then | |
| 224 | return 65535UL; | |
| 225 | elif type.matches(_innate_symbol_lookup.get_int_type()) then | |
| 226 | return 2147483647UL; | |
| 227 | elif type.matches(_innate_symbol_lookup.get_uint_type()) then | |
| 228 | return 4294967295UL; | |
| 229 | elif type.matches(_innate_symbol_lookup.get_long_type()) then | |
| 230 | return 9223372036854775807UL; | |
| 231 | elif type.matches(_innate_symbol_lookup.get_word_type()) then | |
| 232 | return 9223372036854775807UL; | |
| 233 | fi | |
| 234 | ||
| 235 | return 0UL; | |
| 236 | si | |
| 237 | si | |
| 238 | si |