| 1 | const std = @import("std"); |
| 2 | const Token = @import("./tokenize.zig").Token; |
| 3 | const extras = @import("extras"); |
| 4 | |
| 5 | const string = []const u8; |
| 6 | |
| 7 | // |
| 8 | // |
| 9 | |
| 10 | pub const Value = union(enum) { |
| 11 | element: Element, |
| 12 | attr: Attr, |
| 13 | string: string, |
| 14 | replacement: Replacement, |
| 15 | block: Block, |
| 16 | body: Body, |
| 17 | function: Fn, |
| 18 | }; |
| 19 | |
| 20 | pub const Element = struct { |
| 21 | name: string, |
| 22 | attrs: []const Attr, |
| 23 | children: Body, |
| 24 | }; |
| 25 | |
| 26 | pub const Attr = struct { |
| 27 | key: string, |
| 28 | value: union(enum) { string: string, body: Body }, |
| 29 | }; |
| 30 | |
| 31 | pub const Replacement = struct { |
| 32 | arms: []const string, |
| 33 | raw: bool = false, |
| 34 | }; |
| 35 | |
| 36 | pub const Block = struct { |
| 37 | name: Type, |
| 38 | func: ?string, |
| 39 | args: []const Arg, |
| 40 | body: Body, |
| 41 | bttm: Body, |
| 42 | |
| 43 | pub const Type = enum { |
| 44 | each, |
| 45 | @"if", |
| 46 | ifnot, |
| 47 | ifequal, |
| 48 | ifnotequal, |
| 49 | }; |
| 50 | }; |
| 51 | |
| 52 | pub const Body = []const Value; |
| 53 | |
| 54 | pub const Fn = struct { |
| 55 | name: string, |
| 56 | raw: bool, |
| 57 | args: []const Arg, |
| 58 | }; |
| 59 | |
| 60 | pub const Arg = union(enum) { |
| 61 | lookup: []const string, |
| 62 | plain: string, |
| 63 | int: u64, |
| 64 | value: []const Value, |
| 65 | }; |
| 66 | |
| 67 | // |
| 68 | // |
| 69 | |
| 70 | pub fn do(comptime tokens: []const Token) Element { |
| 71 | var parser = Parser{ .tokens = tokens, .index = 0 }; |
| 72 | return parser.doElement(); |
| 73 | } |
| 74 | |
| 75 | // |
| 76 | // |
| 77 | |
| 78 | const Parser = struct { |
| 79 | tokens: []const Token, |
| 80 | index: usize, |
| 81 | |
| 82 | pub fn doElement(comptime self: *Parser) Element { |
| 83 | return Element{ |
| 84 | .name = self.eat(.word), |
| 85 | .attrs = self.doAttrs(), |
| 86 | .children = self.doChildren(), |
| 87 | }; |
| 88 | } |
| 89 | |
| 90 | pub fn doAttrs(comptime self: *Parser) []const Attr { |
| 91 | var ret: []const Attr = &[_]Attr{}; |
| 92 | |
| 93 | if (self.tryEatSymbol("[")) { |
| 94 | inline while (true) { |
| 95 | if (self.tryEatSymbol("]")) break; |
| 96 | ret = ret ++ &[_]Attr{self.doAttr()}; |
| 97 | } |
| 98 | } |
| 99 | |
| 100 | return ret; |
| 101 | } |
| 102 | |
| 103 | fn raise(comptime line: usize, comptime pos: usize, comptime expected: []const u8, comptime found: []const u8) void { |
| 104 | @compileError(std.fmt.comptimePrint("pek: file:{d}:{d}: expected {s}, found {s}", .{ line, pos, expected, found })); |
| 105 | } |
| 106 | |
| 107 | fn expect(comptime expected: []const u8, comptime found: []const u8, comptime line: usize, comptime pos: usize) void { |
| 108 | if (std.mem.eql(u8, found, expected)) return; |
| 109 | raise(line, pos, expected, found); |
| 110 | } |
| 111 | |
| 112 | fn tryEatSymbol(comptime self: *Parser, comptime needle: string) bool { |
| 113 | if (self.index >= self.tokens.len) return false; |
| 114 | switch (self.tokens[self.index].data) { |
| 115 | .symbol => |sym| { |
| 116 | if (std.mem.eql(u8, sym, needle)) { |
| 117 | self.index += 1; |
| 118 | return true; |
| 119 | } |
| 120 | return false; |
| 121 | }, |
| 122 | else => { |
| 123 | return false; |
| 124 | }, |
| 125 | } |
| 126 | } |
| 127 | |
| 128 | pub fn doAttr(comptime self: *Parser) Attr { |
| 129 | const k = self.eat(.word); |
| 130 | self.eatSymbol("="); |
| 131 | const body = self.doChildren(); |
| 132 | if (body.len > 0) { |
| 133 | return Attr{ |
| 134 | .key = k, |
| 135 | .value = .{ .body = body }, |
| 136 | }; |
| 137 | } |
| 138 | const v = self.eat(.string); |
| 139 | return Attr{ |
| 140 | .key = k, |
| 141 | .value = .{ .string = v }, |
| 142 | }; |
| 143 | } |
| 144 | |
| 145 | pub fn eatSymbol(comptime self: *Parser, comptime needle: string) void { |
| 146 | const tok = self.tokens[self.index]; |
| 147 | expect(needle, self.eat(.symbol), tok.line, tok.pos); |
| 148 | } |
| 149 | |
| 150 | pub fn eatWord(comptime self: *Parser, comptime needle: string) void { |
| 151 | const tok = self.tokens[self.index]; |
| 152 | expect(needle, self.eat(.word), tok.line, tok.pos); |
| 153 | } |
| 154 | |
| 155 | pub fn doChildren(comptime self: *Parser) []const Value { |
| 156 | var ret: []const Value = &[_]Value{}; |
| 157 | |
| 158 | if (self.tryEatSymbol("(")) { |
| 159 | inline while (true) { |
| 160 | if (self.tryEatSymbol(")")) break; |
| 161 | ret = ret ++ &[_]Value{self.doValue()}; |
| 162 | } |
| 163 | } |
| 164 | |
| 165 | return ret; |
| 166 | } |
| 167 | |
| 168 | pub fn doValue(comptime self: *Parser) Value { |
| 169 | if (self.tokens[self.index].data == .string) { |
| 170 | return Value{ .string = self.eat(.string) }; |
| 171 | } |
| 172 | if (self.tryEatSymbol("{")) { |
| 173 | if (self.tryEatSymbol("#")) { |
| 174 | const fraw = self.tryEatSymbol("#"); |
| 175 | const w = self.eat(.word); |
| 176 | std.debug.assert(w.len > 0); |
| 177 | std.debug.assert(w[0] != '_'); |
| 178 | if (std.meta.stringToEnum(Block.Type, w)) |name| { |
| 179 | const func = if (self.tryEatSymbol("#")) self.eat(.word) else null; |
| 180 | const args = self.doArgs(); |
| 181 | var children: []const Value = &.{}; |
| 182 | var bottom: []const Value = &.{}; |
| 183 | var top = true; |
| 184 | while (!self.tryEatSymbol("/")) { |
| 185 | if (self.tryEatSymbol("<")) { |
| 186 | self.eatWord("else"); |
| 187 | self.eatSymbol(">"); |
| 188 | top = false; |
| 189 | } |
| 190 | if (top) { |
| 191 | children = children ++ &[_]Value{self.doValue()}; |
| 192 | } else { |
| 193 | bottom = bottom ++ &[_]Value{self.doValue()}; |
| 194 | } |
| 195 | } |
| 196 | self.eatWord(@tagName(name)); |
| 197 | self.eatSymbol("/"); |
| 198 | return Value{ .block = Block{ |
| 199 | .name = name, |
| 200 | .func = func, |
| 201 | .args = args, |
| 202 | .body = children, |
| 203 | .bttm = bottom, |
| 204 | } }; |
| 205 | } |
| 206 | return Value{ .function = .{ |
| 207 | .name = w, |
| 208 | .raw = fraw, |
| 209 | .args = self.doArgs(), |
| 210 | } }; |
| 211 | } |
| 212 | if (self.tryEatSymbol("{")) { |
| 213 | defer self.eatSymbol("}"); |
| 214 | return Value{ .replacement = .{ .arms = self.doReplacement(), .raw = true } }; |
| 215 | } |
| 216 | return Value{ .replacement = .{ .arms = self.doReplacement() } }; |
| 217 | } |
| 218 | return Value{ .element = self.doElement() }; |
| 219 | } |
| 220 | |
| 221 | pub fn doArgs(comptime self: *Parser) []const Arg { |
| 222 | var ret: []const Arg = &.{}; |
| 223 | var temp: []const string = &.{}; |
| 224 | while (!self.tryEatSymbol("}")) { |
| 225 | if (self.nextIs(.string)) { |
| 226 | if (temp.len > 0) { |
| 227 | ret = ret ++ &[_]Arg{.{ .lookup = temp }}; |
| 228 | temp = &.{}; |
| 229 | } |
| 230 | ret = ret ++ &[_]Arg{.{ .plain = self.eat(.string) }}; |
| 231 | continue; |
| 232 | } |
| 233 | if (self.tryEatSymbol("(")) { |
| 234 | self.index -= 1; |
| 235 | ret = ret ++ &[_]Arg{.{ .value = self.doChildren() }}; |
| 236 | continue; |
| 237 | } |
| 238 | if (temp.len == 0 and self.nextIs(.word)) { |
| 239 | const next_word = self.eat(.word); |
| 240 | if (extras.matchesAll(u8, next_word, std.ascii.isDigit)) { |
| 241 | ret = ret ++ &[_]Arg{.{ .int = extras.parseDigits(u64, next_word, 10) catch unreachable }}; |
| 242 | continue; |
| 243 | } |
| 244 | temp = temp ++ &[_]string{next_word}; |
| 245 | } |
| 246 | if (self.tryEatSymbol(".")) { |
| 247 | temp = temp ++ &[_]string{self.eat(.word)}; |
| 248 | } else { |
| 249 | ret = ret ++ &[_]Arg{.{ .lookup = temp }}; |
| 250 | temp = &.{}; |
| 251 | } |
| 252 | } |
| 253 | if (temp.len > 0) ret = ret ++ &[_]Arg{.{ .lookup = temp }}; |
| 254 | return ret; |
| 255 | } |
| 256 | |
| 257 | pub fn doReplacement(comptime self: *Parser) []const string { |
| 258 | var ret: []const string = &.{}; |
| 259 | ret = ret ++ &[_]string{self.eat(.word)}; |
| 260 | while (!self.tryEatSymbol("}")) { |
| 261 | self.eatSymbol("."); |
| 262 | ret = ret ++ &[_]string{self.eat(.word)}; |
| 263 | } |
| 264 | return ret; |
| 265 | } |
| 266 | |
| 267 | fn eat(comptime self: *Parser, comptime typ: std.meta.Tag(Token.Data)) string { |
| 268 | defer self.index += 1; |
| 269 | const tok = self.tokens[self.index]; |
| 270 | const tag = std.meta.activeTag(tok.data); |
| 271 | if (tag != typ) raise(tok.line, tok.pos, @tagName(typ), @tagName(tag)); |
| 272 | return @field(tok.data, @tagName(typ)); |
| 273 | } |
| 274 | |
| 275 | fn nextIs(comptime self: *Parser, comptime typ: std.meta.Tag(Token.Data)) bool { |
| 276 | return self.tokens[self.index].data == typ; |
| 277 | } |
| 278 | }; |