| 1 | const std = @import("std"); |
| 2 | const string = []const u8; |
| 3 | |
| 4 | // |
| 5 | // |
| 6 | |
| 7 | pub const Token = struct { |
| 8 | data: Data, |
| 9 | line: usize, |
| 10 | pos: usize, |
| 11 | |
| 12 | pub const skippedChars = &[_]u8{ ' ', '\n', '\t', '\r' }; |
| 13 | |
| 14 | pub const Data = union(enum) { |
| 15 | word: string, |
| 16 | symbol: string, |
| 17 | string: string, |
| 18 | }; |
| 19 | }; |
| 20 | |
| 21 | // |
| 22 | // |
| 23 | |
| 24 | pub fn do(comptime input: string, comptime symbols: []const u8) []const Token { |
| 25 | var ret: []const Token = &[_]Token{}; |
| 26 | |
| 27 | var line = 1; |
| 28 | var pos = 1; |
| 29 | |
| 30 | var start = 0; |
| 31 | var end = 0; |
| 32 | var mode = 0; |
| 33 | |
| 34 | @setEvalBranchQuota(1000000); |
| 35 | if (!@inComptime()) @compileError("must be at comptime!"); |
| 36 | |
| 37 | inline for (input, 0..) |c, i| { |
| 38 | const s = &[_]u8{c}; |
| 39 | |
| 40 | var shouldFlush: bool = undefined; |
| 41 | |
| 42 | blk: { |
| 43 | if (mode == 0) { |
| 44 | if (c == '/' and input[i + 1] == '/') { |
| 45 | mode = 1; |
| 46 | shouldFlush = false; |
| 47 | break :blk; |
| 48 | } |
| 49 | if (c == '"') { |
| 50 | mode = 2; |
| 51 | shouldFlush = false; |
| 52 | break :blk; |
| 53 | } |
| 54 | if (c == '\'') { |
| 55 | mode = 2; |
| 56 | shouldFlush = false; |
| 57 | break :blk; |
| 58 | } |
| 59 | } |
| 60 | if (mode == 1) { |
| 61 | if (c == '\n') { |
| 62 | // skip comments |
| 63 | // f(v.handle(TTCom, in[s:i])) |
| 64 | start = i; |
| 65 | end = i; |
| 66 | mode = 0; |
| 67 | } |
| 68 | shouldFlush = c == '\n'; |
| 69 | break :blk; |
| 70 | } |
| 71 | if (mode == 2) { |
| 72 | if (c == input[start]) { |
| 73 | const data = input[start .. i + 1][0..].*; |
| 74 | ret = ret ++ &[_]Token{.{ |
| 75 | .data = .{ .string = &data }, |
| 76 | .line = line, |
| 77 | .pos = pos, |
| 78 | }}; |
| 79 | start = i + 1; |
| 80 | end = i; |
| 81 | mode = 0; |
| 82 | } |
| 83 | shouldFlush = false; |
| 84 | break :blk; |
| 85 | } |
| 86 | if (std.mem.indexOfScalar(u8, Token.skippedChars, c)) |_| { |
| 87 | shouldFlush = true; |
| 88 | break :blk; |
| 89 | } |
| 90 | if (std.mem.indexOfScalar(u8, symbols, c)) |_| { |
| 91 | shouldFlush = true; |
| 92 | break :blk; |
| 93 | } |
| 94 | shouldFlush = false; |
| 95 | break :blk; |
| 96 | } |
| 97 | |
| 98 | if (!shouldFlush) { |
| 99 | end += 1; |
| 100 | } |
| 101 | if (shouldFlush) { |
| 102 | if (mode == 0) { |
| 103 | if (end - start > 0) { |
| 104 | const data = input[start..end][0..].*; |
| 105 | ret = ret ++ &[_]Token{.{ |
| 106 | .data = .{ .word = &data }, |
| 107 | .line = line, |
| 108 | .pos = pos, |
| 109 | }}; |
| 110 | start = i; |
| 111 | end = i; |
| 112 | } |
| 113 | if (std.mem.indexOfScalar(u8, Token.skippedChars, c)) |_| { |
| 114 | start += 1; |
| 115 | end += 1; |
| 116 | } |
| 117 | if (std.mem.indexOfScalar(u8, symbols, c)) |_| { |
| 118 | ret = ret ++ &[_]Token{.{ |
| 119 | .data = .{ .symbol = s }, |
| 120 | .line = line, |
| 121 | .pos = pos, |
| 122 | }}; |
| 123 | start += 1; |
| 124 | end += 1; |
| 125 | } |
| 126 | } |
| 127 | } |
| 128 | |
| 129 | pos += 1; |
| 130 | if (c != '\n') continue; |
| 131 | line += 1; |
| 132 | pos = 1; |
| 133 | } |
| 134 | |
| 135 | return ret; |
| 136 | } |