| |
| from __future__ import annotations |
| from lexer.lexer import tokenize, Token |
| from mantra_ast.nodes import * |
| from typing import List, Iterator |
|
|
| class ParseError(Exception): |
| pass |
|
|
| class Parser: |
| def __init__(self, tokens): |
| self.tokens = list(tokens) |
| self.i = 0 |
|
|
| def peek(self) -> Token: |
| return self.tokens[self.i] if self.i < len(self.tokens) else Token('EOF', '', 0, 0) |
|
|
| def peek2(self) -> Token: |
| return self.tokens[self.i+1] if self.i+1 < len(self.tokens) else Token('EOF', '', 0, 0) |
|
|
| def advance(self) -> Token: |
| t = self.peek() |
| self.i += 1 |
| return t |
|
|
| def expect(self, typ: str) -> Token: |
| t = self.advance() |
| if t.type != typ: |
| raise ParseError(f"Expected {typ} at {t.line}:{t.col}, got {t.type!r} ({t.value!r})") |
| return t |
|
|
| def match(self, *types) -> bool: |
| return self.peek().type in types |
|
|
| |
|
|
| def parse(self) -> Program: |
| modules, items = [], [] |
| while not self.match('EOF'): |
| if self.match('MODULE'): |
| modules.append(self._parse_module()) |
| else: |
| items.append(self._parse_item()) |
| if not modules: |
| modules = [Module(name='main', items=items)] |
| return Program(modules=modules) |
|
|
| def _parse_module(self) -> Module: |
| self.expect('MODULE') |
| name = self.expect('IDENT').value |
| self.expect('LBRACE') |
| items = [] |
| while not self.match('RBRACE', 'EOF'): |
| items.append(self._parse_item()) |
| self.expect('RBRACE') |
| return Module(name=name, items=items) |
|
|
| |
|
|
| def _parse_item(self) -> Node: |
| ann = self._parse_annotations() |
| t = self.peek() |
| if t.type == 'CONST': |
| return self._parse_const(ann) |
| if t.type == 'LET': |
| return self._parse_let(ann) |
| if t.type == 'DEFINE': |
| return self._parse_define(ann) |
| if t.type == 'ASSERT': |
| self.advance() |
| return Assert(expr=self._parse_expr()) |
| |
| return self._parse_expr() |
|
|
| def _parse_annotations(self) -> List[Annotation]: |
| ann = [] |
| while self.match('ANNOT'): |
| raw = self.advance().value |
| if '(' in raw: |
| name, rest = raw.split('(', 1) |
| ann.append(Annotation(name=name[1:], args=rest.rstrip(')'))) |
| else: |
| ann.append(Annotation(name=raw[1:], args=None)) |
| return ann |
|
|
| def _parse_const(self, ann) -> ConstDecl: |
| self.expect('CONST') |
| name = self.expect('IDENT').value |
| self.expect('ASSIGN') |
| val = self._parse_expr() |
| return ConstDecl(name=name, value=val, annotations=ann) |
|
|
| def _parse_let(self, ann) -> LetDecl: |
| self.expect('LET') |
| name = self.expect('IDENT').value |
| self.expect('ASSIGN') |
| val = self._parse_expr() |
| return LetDecl(name=name, value=val, annotations=ann) |
|
|
| def _parse_define(self, ann) -> FunctionDecl: |
| self.expect('DEFINE') |
| name = self.expect('IDENT').value |
| self.expect('LPAREN') |
| params = [] |
| while not self.match('RPAREN', 'EOF'): |
| params.append(self.expect('IDENT').value) |
| if self.match('COMMA'): |
| self.advance() |
| self.expect('RPAREN') |
| self.expect('BEGIN') |
| body = [] |
| while not self.match('END', 'EOF'): |
| if self.match('RETURN'): |
| self.advance() |
| body.append(Return(expr=self._parse_expr())) |
| elif self.match('ASSERT'): |
| self.advance() |
| body.append(Assert(expr=self._parse_expr())) |
| elif self.match('LET'): |
| body.append(self._parse_let([])) |
| elif self.match('CONST'): |
| body.append(self._parse_const([])) |
| else: |
| body.append(self._parse_expr()) |
| self.expect('END') |
| return FunctionDecl(name=name, params=params, body=body, annotations=ann) |
|
|
| |
|
|
| def _parse_expr(self) -> Node: |
| return self._parse_or() |
|
|
| def _parse_or(self) -> Node: |
| left = self._parse_and() |
| while self.match('OR', 'PIPE'): |
| op = self.advance().value |
| left = BinaryExpr(op='or', left=left, right=self._parse_and()) |
| return left |
|
|
| def _parse_and(self) -> Node: |
| left = self._parse_equality() |
| while self.match('AND', 'AMP'): |
| op = self.advance().value |
| left = BinaryExpr(op='and', left=left, right=self._parse_equality()) |
| return left |
|
|
| def _parse_equality(self) -> Node: |
| left = self._parse_relational() |
| while self.match('EQ', 'NEQ'): |
| op = self.advance().value |
| left = BinaryExpr(op=op, left=left, right=self._parse_relational()) |
| return left |
|
|
| def _parse_relational(self) -> Node: |
| left = self._parse_shift() |
| while self.match('LT', 'GT', 'LE', 'GE'): |
| op = self.advance().value |
| left = BinaryExpr(op=op, left=left, right=self._parse_shift()) |
| return left |
|
|
| def _parse_shift(self) -> Node: |
| left = self._parse_bitor() |
| while self.match('LSH', 'RSH'): |
| op = self.advance().value |
| left = BinaryExpr(op=op, left=left, right=self._parse_bitor()) |
| return left |
|
|
| def _parse_bitor(self) -> Node: |
| left = self._parse_bitxor() |
| while self.match('PIPE'): |
| self.advance() |
| left = BinaryExpr(op='|', left=left, right=self._parse_bitxor()) |
| return left |
|
|
| def _parse_bitxor(self) -> Node: |
| left = self._parse_bitand() |
| while self.match('CARET'): |
| self.advance() |
| left = BinaryExpr(op='^', left=left, right=self._parse_bitand()) |
| return left |
|
|
| def _parse_bitand(self) -> Node: |
| left = self._parse_add() |
| while self.match('AMP'): |
| self.advance() |
| left = BinaryExpr(op='&', left=left, right=self._parse_add()) |
| return left |
|
|
| def _parse_add(self) -> Node: |
| left = self._parse_mul() |
| while self.match('PLUS', 'MINUS'): |
| op = self.advance().value |
| left = BinaryExpr(op=op, left=left, right=self._parse_mul()) |
| return left |
|
|
| def _parse_mul(self) -> Node: |
| left = self._parse_unary() |
| while self.match('STAR', 'SLASH', 'PERCENT', 'POW'): |
| op = self.advance().value |
| left = BinaryExpr(op=op, left=left, right=self._parse_unary()) |
| return left |
|
|
| def _parse_unary(self) -> Node: |
| if self.match('MINUS', 'TILDE', 'NOT'): |
| op = self.advance().value |
| return UnaryExpr(op=op, operand=self._parse_unary()) |
| return self._parse_postfix() |
|
|
| def _parse_postfix(self) -> Node: |
| node = self._parse_primary() |
| while self.match('DOT'): |
| self.advance() |
| field = self.expect('IDENT').value |
| node = FieldAccess(obj=node, field=field) |
| return node |
|
|
| def _parse_primary(self) -> Node: |
| t = self.peek() |
|
|
| if t.type == 'NUMBER': |
| self.advance() |
| return IntegerLiteral(int(t.value)) |
|
|
| if t.type == 'HEX': |
| self.advance() |
| return HexLiteral(raw=t.value) |
|
|
| if t.type in ('TRUE', 'FALSE'): |
| self.advance() |
| return BoolLiteral(t.type == 'TRUE') |
|
|
| if t.type == 'STRING': |
| self.advance() |
| s = t.value[1:-1].encode('utf-8').decode('unicode_escape') |
| return StringLiteral(s) |
|
|
| if t.type == 'BYTES': |
| self.advance() |
| raw = t.value[2:-1] |
| b = raw.encode('utf-8').decode('unicode_escape').encode('latin1') |
| return BytesLiteral(b) |
|
|
| if t.type == 'SOME': |
| self.advance() |
| self.expect('LPAREN') |
| v = self._parse_expr() |
| self.expect('RPAREN') |
| return OptionLiteral(tag='some', value=v) |
|
|
| if t.type == 'NONE': |
| self.advance() |
| return OptionLiteral(tag='none', value=None) |
|
|
| |
| _CALLABLE_KEYWORDS = {'CONCAT', 'LEN', 'SLICE', 'SOME', 'OK', 'ERR', |
| 'BYTES', 'BYTE', 'OPTION', 'RESULT'} |
| if t.type == 'IDENT' or t.type in _CALLABLE_KEYWORDS: |
| name = self.advance().value |
| if self.match('LPAREN'): |
| self.advance() |
| args = [] |
| while not self.match('RPAREN', 'EOF'): |
| args.append(self._parse_expr()) |
| if self.match('COMMA'): |
| self.advance() |
| self.expect('RPAREN') |
| return Call(callee=name, args=args) |
| return Identifier(name=name) |
|
|
| if t.type == 'LPAREN': |
| self.advance() |
| expr = self._parse_expr() |
| self.expect('RPAREN') |
| return expr |
|
|
| if t.type == 'LBRACE': |
| return self._parse_dict_or_list() |
|
|
| if t.type == 'LBRACKET': |
| self.advance() |
| items = [] |
| while not self.match('RBRACKET', 'EOF'): |
| items.append(self._parse_expr()) |
| if self.match('COMMA'): |
| self.advance() |
| self.expect('RBRACKET') |
| return ListLiteral(items=items) |
|
|
| raise ParseError(f"Unexpected token {t.type!r} ({t.value!r}) at {t.line}:{t.col}") |
|
|
| def _parse_dict_or_list(self) -> Node: |
| self.expect('LBRACE') |
| |
| if self.peek().type == 'IDENT' and self.peek2().type == 'COLON': |
| pairs = [] |
| while not self.match('RBRACE', 'EOF'): |
| key = self.expect('IDENT').value |
| self.expect('COLON') |
| val = self._parse_expr() |
| pairs.append((key, val)) |
| if self.match('COMMA'): |
| self.advance() |
| self.expect('RBRACE') |
| return DictLiteral(pairs=pairs) |
| else: |
| items = [] |
| while not self.match('RBRACE', 'EOF'): |
| items.append(self._parse_expr()) |
| if self.match('COMMA'): |
| self.advance() |
| self.expect('RBRACE') |
| return ListLiteral(items=items) |
|
|
|
|
| def parse(code: str) -> Program: |
| tokens = tokenize(code) |
| return Parser(tokens).parse() |
|
|