jorts/jortsc/parser/syntatic.py

from typing import Optional, Any, List

from jortsc.parser.lexer import Token, TokenType
from jortsc.parser.ast_nodes import (
    Function, TypedVar, Identifier, Import, ReturnType
)


class ParseError(Exception):
    """Represents a parse error."""
    pass


class Reader:
    """Main reader class"""
    def __init__(self, tokens: List[Token]):
        self.tokens = tokens
        self.cur = 0

    def __repr__(self):
        return f'<Reader cur={self.cur} cur_tok={self.peek()}>'

    def peek(self) -> Optional[Token]:
        """Peek at the current token."""
        try:
            token = self.tokens[self.cur]
            return token
        except IndexError:
            return None

    def next(self) -> Optional[Token]:
        """Fetch the current token then skip to the next one."""
        token = self.peek()
        self.cur += 1
        return token

    def expect(self, token_type: TokenType) -> Token:
        """Check for a specific token type and error if it fails"""
        token = self.next()

        if token.type_ != token_type:
            raise ParseError(f'Expected {token_type}, got '
                             f'{token.type_} {token.value!r}')

        return token

    def expect_val(self, value: str) -> Token:
        """Check the next token to see if it matches against a given value,
        instead of a type."""
        token = self.next()

        if token.value != value:
            raise ParseError(f'Expected {value!r}, got '
                             f'{token.type_} {token.value!r}')

        return token


    def next_safe(self) -> Token:
        """'Safe' version of next().

        Raises an 'Unexpected EOF' error if next() returns None.
        """
        token = self.next()

        if token is None:
            raise ParseError('Unexpected EOF')

        return token

    def ignore(self, token_type: TokenType):
        """Only increase self.cur if token_type is the upcoming token."""
        try:
            assert self.tokens[self.cur].type_ == token_type
            self.cur += 1
        except AssertionError:
            pass


def _fn_read_args(reader: Reader, cur: List = None) -> List:
    """Recursively read the arguments of the function."""
    if cur is None:
        cur = []

    # it can be an identifier for the arguments' type, OR a RPAREN
    # if it is rparen, we stop
    # if it isnt, we keep going until that happens
    token = reader.next_safe()

    if token.value == ')':
        return cur

    argtype = token
    reader.expect(TokenType.whitespace)
    argname = reader.next_safe()

    cur.append(TypedVar(argtype.value, argname.value))
    return _fn_read_args(reader, cur)


def _fn_ret_type(reader: Reader) -> ReturnType:
    """Fetch the return type of a function. Defaults to void."""
    try:
        reader.expect_val('->')
    except ParseError:
        return ReturnType('void')

    reader.ignore(TokenType.whitespace)
    token = reader.expect(TokenType.identifier)
    return ReturnType(token.value)


def read_function(reader: Reader):
    """Read a function block."""
    reader.expect(TokenType.whitespace)

    token = reader.next()

    fn_name = '_anonymous'
    fn_args = []

    print('function token', token)

    if token.type_ == TokenType.identifier:
        fn_name = token.value

        reader.expect(TokenType.whitespace)
        reader.expect_val('(')

        fn_args = _fn_read_args(reader)

        reader.expect(TokenType.whitespace)
        fn_ret_type = _fn_ret_type(reader)

        # only skip whitespace if we see it
        reader.ignore(TokenType.whitespace)
        block = read_start(reader)
    elif token.value == '(':
        fn_args = _fn_read_args(reader)
        fn_ret_type = _fn_ret_type(reader)
        block = read_start(reader)

    print('final function', fn_name, fn_args, fn_ret_type, block)

    return Function(fn_name, fn_args, fn_ret_type, block)


def read_import(reader):
    """Read an import"""
    reader.expect(TokenType.whitespace)
    module = reader.next_safe()
    return Import(module.value)


HANDLERS = {
    'fn': read_function,
    'import': read_import,
}


def read_reserved(token: Token, reader: Reader):
    """Read reserved statements."""
    try:
        handler = HANDLERS[token.value]
    except KeyError:
        raise ParseError(f'Unexpected reserved word {token.value!r}')

    return handler(reader)


def read_start(reader: Reader):
    """Read the start of a program."""
    print('reader', reader)

    token = reader.next()

    if token is None:
        return []

    ast = []
    res = []

    print('cur', token)

    # handle blocks
    if token.value == '{':
        while True:
            token = reader.next()

            if token.value == '}':
                break

            res.extend(read_start(reader))

    # import, fn, etc
    elif token.type_ == TokenType.reserved:
        res = read_reserved(token, reader)

    elif token.type_ == TokenType.comment:
        return []

    elif token.type_ == TokenType.identifier:
        res = [Identifier(token.value)]

    ast.append(res)
    return ast

def read_loop(reader: Reader):
    final_ast = []

    while True:
        ast = read_start(reader)

        if not ast:
            break

        inner = ast[0]

        if not inner:
            break

        final_ast.append(ast)

    return final_ast


def syntatic(tokens: List[Token]):
    """Create an AST out of the tokens."""
    return read_loop(Reader(tokens))