678 lines
17 KiB
Zig
678 lines
17 KiB
Zig
const coral = @import("coral");
|
|
|
|
const tokens = @import("./tokens.zig");
|
|
|
|
pub const BinaryOperator = enum {
|
|
addition,
|
|
subtraction,
|
|
multiplication,
|
|
divsion,
|
|
equals_comparison,
|
|
greater_than_comparison,
|
|
greater_equals_comparison,
|
|
less_than_comparison,
|
|
less_equals_comparison,
|
|
|
|
fn builder(comptime build_next: ExpressionBuilder, comptime operators: []const BinaryOperator) ExpressionBuilder {
|
|
const Builder = struct {
|
|
fn build(self: *Tree) ParseError!Expression {
|
|
const allocator = self.arena.as_allocator();
|
|
var expression = try build_next(self);
|
|
|
|
inline for (operators) |operator| {
|
|
const token = @as(tokens.Token, switch (operator) {
|
|
.addition => .symbol_plus,
|
|
.subtraction => .symbol_minus,
|
|
.multiplication => .symbol_asterisk,
|
|
.divsion => .symbol_forward_slash,
|
|
.equals_comparison => .symbol_double_equals,
|
|
.greater_than_comparison => .symbol_greater_than,
|
|
.greater_equals_comparison => .symbol_greater_equals,
|
|
.less_than_comparison => .symbol_less_than,
|
|
.less_equals_comparison => .symbol_less_equals,
|
|
});
|
|
|
|
if (self.tokenizer.token == coral.io.tag_of(token)) {
|
|
self.tokenizer.step();
|
|
|
|
if (self.tokenizer.token == .end) {
|
|
return self.report(
|
|
"expected other half of expression after `" ++
|
|
comptime token.text() ++
|
|
"`");
|
|
}
|
|
|
|
// TODO: Remove once Zig has fixed struct self-reassignment.
|
|
const unnecessary_temp = try coral.io.allocate_one(allocator, expression);
|
|
|
|
expression = .{
|
|
.binary_operation = .{
|
|
.operator = operator,
|
|
.lhs_expression = unnecessary_temp,
|
|
.rhs_expression = try coral.io.allocate_one(allocator, try build_next(self)),
|
|
},
|
|
};
|
|
}
|
|
}
|
|
|
|
return expression;
|
|
}
|
|
};
|
|
|
|
return Builder.build;
|
|
}
|
|
};
|
|
|
|
pub const Expression = union (enum) {
|
|
nil_literal,
|
|
true_literal,
|
|
false_literal,
|
|
builtin: []const coral.io.Byte,
|
|
number_literal: []const coral.io.Byte,
|
|
string_literal: []const coral.io.Byte,
|
|
symbol_literal: []const coral.io.Byte,
|
|
table_literal: TableLiteral,
|
|
grouped_expression: *Expression,
|
|
|
|
local_get: struct {
|
|
identifier: []const coral.io.Byte,
|
|
},
|
|
|
|
local_set: struct {
|
|
identifier: []const coral.io.Byte,
|
|
value_expression: *Expression,
|
|
},
|
|
|
|
field_get: struct {
|
|
object_expression: *Expression,
|
|
identifier: []const coral.io.Byte,
|
|
},
|
|
|
|
field_set: struct {
|
|
object_expression: *Expression,
|
|
identifier: []const coral.io.Byte,
|
|
value_expression: *Expression,
|
|
},
|
|
|
|
subscript_get: struct {
|
|
object_expression: *Expression,
|
|
subscript_expression: *Expression,
|
|
},
|
|
|
|
subscript_set: struct {
|
|
object_expression: *Expression,
|
|
subscript_expression: *Expression,
|
|
value_expression: *Expression,
|
|
},
|
|
|
|
binary_operation: struct {
|
|
operator: BinaryOperator,
|
|
lhs_expression: *Expression,
|
|
rhs_expression: *Expression,
|
|
},
|
|
|
|
unary_operation: struct {
|
|
operator: UnaryOperator,
|
|
expression: *Expression,
|
|
},
|
|
|
|
invoke: struct {
|
|
object_expression: *Expression,
|
|
argument_expressions: List,
|
|
},
|
|
|
|
const List = coral.list.Stack(Expression);
|
|
|
|
const TableLiteral = coral.list.Stack(struct {
|
|
key_expression: Expression,
|
|
value_expression: Expression,
|
|
});
|
|
};
|
|
|
|
const ExpressionBuilder = fn (self: *Tree) ParseError!Expression;
|
|
|
|
pub const ParseError = error {
|
|
OutOfMemory,
|
|
BadSyntax,
|
|
};
|
|
|
|
pub const Statement = union (enum) {
|
|
@"return": ?Expression,
|
|
|
|
@"if": struct {
|
|
condition_expression: Expression,
|
|
block_statements: List,
|
|
else_statement: ?*Statement,
|
|
},
|
|
|
|
@"while": struct {
|
|
condition_expression: Expression,
|
|
block_statements: List,
|
|
},
|
|
|
|
block: List,
|
|
expression: Expression,
|
|
|
|
const List = coral.list.Stack(Statement);
|
|
};
|
|
|
|
pub const Tree = struct {
|
|
name: []const coral.io.Byte,
|
|
allocator: coral.io.Allocator,
|
|
arena: coral.arena.Stacking,
|
|
error_buffer: coral.list.ByteStack,
|
|
tokenizer: tokens.Tokenizer,
|
|
parsed_statements: Statement.List,
|
|
has_returned: bool,
|
|
|
|
pub fn error_message(self: Tree) []const coral.io.Byte {
|
|
return self.error_buffer.values;
|
|
}
|
|
|
|
pub fn free(self: *Tree) void {
|
|
self.parsed_statements.free();
|
|
self.error_buffer.free();
|
|
self.arena.free();
|
|
}
|
|
|
|
pub fn make(allocator: coral.io.Allocator, ast_name: []const coral.io.Byte) Tree {
|
|
return .{
|
|
.arena = coral.arena.Stacking.make(allocator, 4096),
|
|
.error_buffer = coral.list.ByteStack.make(allocator),
|
|
.parsed_statements = Statement.List.make(allocator),
|
|
.tokenizer = .{.source = ""},
|
|
.allocator = allocator,
|
|
.name = ast_name,
|
|
.has_returned = false,
|
|
};
|
|
}
|
|
|
|
pub fn parse(self: *Tree, data: []const coral.io.Byte) ParseError![]const Statement {
|
|
self.free();
|
|
|
|
self.tokenizer = .{.source = data};
|
|
self.has_returned = false;
|
|
|
|
self.tokenizer.skip_newlines();
|
|
|
|
while (self.tokenizer.token != .end) {
|
|
try self.parsed_statements.push_one(try self.parse_statement());
|
|
}
|
|
|
|
return self.parsed_statements.values;
|
|
}
|
|
|
|
const parse_additive = BinaryOperator.builder(parse_equality, &.{
|
|
.addition,
|
|
.subtraction,
|
|
});
|
|
|
|
fn parse_branch(self: *Tree) ParseError!Statement {
|
|
const allocator = self.arena.as_allocator();
|
|
|
|
defer self.tokenizer.skip_newlines();
|
|
|
|
self.tokenizer.step();
|
|
|
|
const condition_expression = try self.parse_expression();
|
|
|
|
if (self.tokenizer.token != .symbol_colon) {
|
|
return self.report("expected `:` after `if` statement condition");
|
|
}
|
|
|
|
var statements = Statement.List.make(allocator);
|
|
|
|
self.tokenizer.skip_newlines();
|
|
|
|
while (true) {
|
|
switch (self.tokenizer.token) {
|
|
.keyword_end => {
|
|
return .{
|
|
.@"if" = .{
|
|
.condition_expression = condition_expression,
|
|
.block_statements = statements,
|
|
.else_statement = null,
|
|
},
|
|
};
|
|
},
|
|
|
|
.keyword_else => {
|
|
self.tokenizer.step();
|
|
|
|
if (self.tokenizer.token != .symbol_colon) {
|
|
return self.report("expected `:` after `if` statement condition");
|
|
}
|
|
|
|
var else_statements = Statement.List.make(allocator);
|
|
|
|
self.tokenizer.skip_newlines();
|
|
|
|
while (self.tokenizer.token != .keyword_end) {
|
|
try else_statements.push_one(try self.parse_statement());
|
|
}
|
|
|
|
return .{
|
|
.@"if" = .{
|
|
.else_statement = try coral.io.allocate_one(allocator, Statement{.block = else_statements}),
|
|
.condition_expression = condition_expression,
|
|
.block_statements = statements,
|
|
},
|
|
};
|
|
},
|
|
|
|
.keyword_elif => {
|
|
return .{
|
|
.@"if" = .{
|
|
.else_statement = try coral.io.allocate_one(allocator, try self.parse_branch()),
|
|
.condition_expression = condition_expression,
|
|
.block_statements = statements,
|
|
},
|
|
};
|
|
},
|
|
|
|
else => try statements.push_one(try self.parse_statement()),
|
|
}
|
|
}
|
|
}
|
|
|
|
const parse_comparison = BinaryOperator.builder(parse_term, &.{
|
|
.greater_than_comparison,
|
|
.greater_equals_comparison,
|
|
.less_than_comparison,
|
|
.less_equals_comparison
|
|
});
|
|
|
|
const parse_equality = BinaryOperator.builder(parse_comparison, &.{
|
|
.equals_comparison,
|
|
});
|
|
|
|
pub fn parse_expression(self: *Tree) ParseError!Expression {
|
|
const allocator = self.arena.as_allocator();
|
|
const expression = try self.parse_additive();
|
|
|
|
if (self.tokenizer.token == .symbol_equals) {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
if (self.tokenizer.token == .end) {
|
|
return self.report("expected assignment after `=`");
|
|
}
|
|
|
|
return switch (expression) {
|
|
.local_get => |local_get| .{
|
|
.local_set = .{
|
|
.identifier = local_get.identifier,
|
|
.value_expression = try coral.io.allocate_one(allocator, try self.parse_expression()),
|
|
},
|
|
},
|
|
|
|
.field_get => |field_get| .{
|
|
.field_set = .{
|
|
.object_expression = field_get.object_expression,
|
|
.identifier = field_get.identifier,
|
|
.value_expression = try coral.io.allocate_one(allocator, try self.parse_expression()),
|
|
},
|
|
},
|
|
|
|
.subscript_get => |subscript_get| .{
|
|
.subscript_set = .{
|
|
.object_expression = subscript_get.object_expression,
|
|
.subscript_expression = subscript_get.subscript_expression,
|
|
.value_expression = try coral.io.allocate_one(allocator, try self.parse_expression()),
|
|
},
|
|
},
|
|
|
|
else => self.report("expected local or field on left-hand side of expression"),
|
|
};
|
|
}
|
|
|
|
return expression;
|
|
}
|
|
|
|
fn parse_factor(self: *Tree) ParseError!Expression {
|
|
const allocator = self.arena.as_allocator();
|
|
|
|
var expression = @as(Expression, parse: {
|
|
switch (self.tokenizer.token) {
|
|
.symbol_paren_left => {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
if (self.tokenizer.token == .end) {
|
|
return self.report("expected an expression after `(`");
|
|
}
|
|
|
|
const expression = try self.parse_expression();
|
|
|
|
if (self.tokenizer.token != .symbol_paren_right) {
|
|
return self.report("expected a closing `)` after expression");
|
|
}
|
|
|
|
self.tokenizer.skip_newlines();
|
|
|
|
break: parse .{.grouped_expression = try coral.io.allocate_one(allocator, expression)};
|
|
},
|
|
|
|
.keyword_nil => {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
break: parse .nil_literal;
|
|
},
|
|
|
|
.keyword_true => {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
break: parse .true_literal;
|
|
},
|
|
|
|
.keyword_false => {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
break: parse .false_literal;
|
|
},
|
|
|
|
.number => |value| {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
break: parse .{.number_literal = value};
|
|
},
|
|
|
|
.string => |value| {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
break: parse .{.string_literal = value};
|
|
},
|
|
|
|
.identifier => |identifier| {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
break: parse .{.local_get = .{.identifier = identifier}};
|
|
},
|
|
|
|
.builtin => |builtin| {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
break: parse .{.builtin = builtin};
|
|
},
|
|
|
|
.symbol_brace_left => {
|
|
var table_literal = Expression.TableLiteral.make(allocator);
|
|
|
|
self.tokenizer.skip_newlines();
|
|
|
|
while (true) {
|
|
switch (self.tokenizer.token) {
|
|
.symbol_brace_right => {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
break: parse .{.table_literal = table_literal};
|
|
},
|
|
|
|
.symbol_period => {
|
|
self.tokenizer.step();
|
|
|
|
const identifier = switch (self.tokenizer.token) {
|
|
.identifier => |identifier| identifier,
|
|
else => return self.report("expected identifier after `.`"),
|
|
};
|
|
|
|
self.tokenizer.skip_newlines();
|
|
|
|
if (self.tokenizer.token != .symbol_equals) {
|
|
return self.report("expected `=` after symbol");
|
|
}
|
|
|
|
self.tokenizer.skip_newlines();
|
|
|
|
try table_literal.push_one(.{
|
|
.value_expression = try self.parse_expression(),
|
|
.key_expression = .{.symbol_literal = identifier},
|
|
});
|
|
|
|
switch (self.tokenizer.token) {
|
|
.symbol_comma => self.tokenizer.skip_newlines(),
|
|
|
|
.symbol_brace_right => {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
break: parse .{.table_literal = table_literal};
|
|
},
|
|
|
|
else => return self.report("expected `,` or `}` after expression"),
|
|
}
|
|
},
|
|
|
|
.symbol_bracket_left => {
|
|
self.tokenizer.step();
|
|
|
|
const subscript_expression = try self.parse_expression();
|
|
|
|
if (self.tokenizer.token != .symbol_bracket_right) {
|
|
return self.report("expected `]` after subscript expression");
|
|
}
|
|
|
|
self.tokenizer.skip_newlines();
|
|
|
|
if (self.tokenizer.token != .symbol_equals) {
|
|
return self.report("expected `=` after `]`");
|
|
}
|
|
|
|
self.tokenizer.skip_newlines();
|
|
|
|
try table_literal.push_one(.{
|
|
.value_expression = try self.parse_expression(),
|
|
.key_expression = subscript_expression,
|
|
});
|
|
|
|
switch (self.tokenizer.token) {
|
|
.symbol_comma => self.tokenizer.skip_newlines(),
|
|
|
|
.symbol_brace_right => {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
break: parse .{.table_literal = table_literal};
|
|
},
|
|
|
|
else => return self.report("expected `,` or `}` after expression"),
|
|
}
|
|
},
|
|
|
|
else => return self.report("expected `}` or fields in table literal"),
|
|
}
|
|
}
|
|
},
|
|
|
|
.symbol_minus => {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
if (self.tokenizer.token == .end) {
|
|
return self.report("expected expression after numeric negation (`-`)");
|
|
}
|
|
|
|
break: parse .{
|
|
.unary_operation = .{
|
|
.expression = try coral.io.allocate_one(allocator, try self.parse_factor()),
|
|
.operator = .numeric_negation,
|
|
},
|
|
};
|
|
},
|
|
|
|
.symbol_bang => {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
if (self.tokenizer.token == .end) {
|
|
return self.report("expected expression after boolean negation (`!`)");
|
|
}
|
|
|
|
break: parse .{
|
|
.unary_operation = .{
|
|
.expression = try coral.io.allocate_one(allocator, try self.parse_factor()),
|
|
.operator = .boolean_negation,
|
|
},
|
|
};
|
|
},
|
|
|
|
else => return self.report("unexpected token in expression"),
|
|
}
|
|
});
|
|
|
|
while (true) {
|
|
switch (self.tokenizer.token) {
|
|
.symbol_period => {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
// TODO: Remove when Zig fixes miscompilation with in-place struct re-assignment.
|
|
const unnecessary_temp = try coral.io.allocate_one(allocator, expression);
|
|
|
|
expression = .{
|
|
.field_get = .{
|
|
.identifier = switch (self.tokenizer.token) {
|
|
.identifier => |field_identifier| field_identifier,
|
|
else => return self.report("expected identifier after `.`"),
|
|
},
|
|
|
|
.object_expression = unnecessary_temp,
|
|
},
|
|
};
|
|
|
|
self.tokenizer.skip_newlines();
|
|
},
|
|
|
|
.symbol_bracket_left => {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
// TODO: Remove when Zig fixes miscompilation with in-place struct re-assignment.
|
|
const unnecessary_temp = try coral.io.allocate_one(allocator, expression);
|
|
|
|
expression = .{
|
|
.subscript_get = .{
|
|
.subscript_expression = try coral.io.allocate_one(allocator, try self.parse_expression()),
|
|
.object_expression = unnecessary_temp,
|
|
},
|
|
};
|
|
|
|
if (self.tokenizer.token != .symbol_bracket_right) {
|
|
return self.report("expected `]` subscript expression");
|
|
}
|
|
|
|
self.tokenizer.skip_newlines();
|
|
},
|
|
|
|
.symbol_paren_left => {
|
|
var argument_expressions = Expression.List.make(allocator);
|
|
|
|
while (true) {
|
|
self.tokenizer.skip_newlines();
|
|
|
|
switch (self.tokenizer.token) {
|
|
.symbol_paren_right => break,
|
|
|
|
else => {
|
|
try argument_expressions.push_one(try self.parse_expression());
|
|
|
|
switch (self.tokenizer.token) {
|
|
.symbol_comma => continue,
|
|
.symbol_paren_right => break,
|
|
else => return self.report("expected `,` or `)` after function argument expression"),
|
|
}
|
|
},
|
|
}
|
|
}
|
|
|
|
self.tokenizer.skip_newlines();
|
|
|
|
// TODO: Remove when Zig fixes miscompilation with in-place struct re-assignment.
|
|
const unnecessary_temp = try coral.io.allocate_one(allocator, expression);
|
|
|
|
expression = .{
|
|
.invoke = .{
|
|
.argument_expressions = argument_expressions,
|
|
.object_expression = unnecessary_temp,
|
|
},
|
|
};
|
|
},
|
|
|
|
else => break,
|
|
}
|
|
}
|
|
|
|
return expression;
|
|
}
|
|
|
|
fn parse_statement(self: *Tree) ParseError!Statement {
|
|
const allocator = self.arena.as_allocator();
|
|
|
|
switch (self.tokenizer.token) {
|
|
.keyword_return => {
|
|
defer self.tokenizer.skip_newlines();
|
|
|
|
if (self.has_returned) {
|
|
return self.report("multiple returns in lambda scope but expected only one");
|
|
}
|
|
|
|
self.tokenizer.step();
|
|
|
|
if (self.tokenizer.token != .end and self.tokenizer.token != .newline) {
|
|
return .{.@"return" = try self.parse_expression()};
|
|
}
|
|
|
|
if (self.tokenizer.token != .end and self.tokenizer.token != .newline) {
|
|
return self.report("expected end or newline after return statement");
|
|
}
|
|
|
|
self.has_returned = true;
|
|
|
|
return .{.@"return" = null};
|
|
},
|
|
|
|
.keyword_while => {
|
|
defer self.tokenizer.skip_newlines();
|
|
|
|
self.tokenizer.step();
|
|
|
|
const condition_expression = try self.parse_expression();
|
|
|
|
if (self.tokenizer.token != .symbol_colon) {
|
|
return self.report("expected `:` after `while` statement");
|
|
}
|
|
|
|
var statements = Statement.List.make(allocator);
|
|
|
|
self.tokenizer.skip_newlines();
|
|
|
|
while (self.tokenizer.token != .keyword_end) {
|
|
try statements.push_one(try self.parse_statement());
|
|
}
|
|
|
|
return .{
|
|
.@"while" = .{
|
|
.block_statements = statements,
|
|
.condition_expression = condition_expression,
|
|
},
|
|
};
|
|
},
|
|
|
|
.keyword_if => return self.parse_branch(),
|
|
else => return .{.expression = try self.parse_expression()},
|
|
}
|
|
}
|
|
|
|
const parse_term = BinaryOperator.builder(parse_factor, &.{
|
|
.multiplication,
|
|
.divsion,
|
|
});
|
|
|
|
fn report(self: *Tree, message: []const coral.io.Byte) ParseError {
|
|
coral.utf8.print_formatted(coral.list.stack_as_writer(&self.error_buffer), "{name}@{line}: {message}", .{
|
|
.name = self.name,
|
|
.line = self.tokenizer.lines_stepped,
|
|
.message = message,
|
|
}) catch return error.OutOfMemory;
|
|
|
|
return error.BadSyntax;
|
|
}
|
|
};
|
|
|
|
pub const UnaryOperator = enum {
|
|
boolean_negation,
|
|
numeric_negation,
|
|
};
|