2019-03-10 04:55:12 +00:00
|
|
|
from typing import Optional, Any, List
|
|
|
|
|
|
|
|
from jortsc.parser.lexer import Token, TokenType
|
2019-03-10 05:13:22 +00:00
|
|
|
from jortsc.parser.ast_nodes import Function, TypedVar
|
2019-03-10 04:55:12 +00:00
|
|
|
|
|
|
|
|
|
|
|
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 next(self) -> Optional[Token]:
|
|
|
|
"""Fetch the current token then skip to the next one."""
|
|
|
|
try:
|
|
|
|
token = self.tokens[self.cur]
|
|
|
|
except IndexError:
|
|
|
|
return None
|
|
|
|
|
|
|
|
self.cur += 1
|
|
|
|
return token
|
|
|
|
|
2019-03-10 05:13:22 +00:00
|
|
|
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 {token.type_}')
|
|
|
|
|
|
|
|
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
|
|
|
|
|
2019-03-10 04:55:12 +00:00
|
|
|
|
|
|
|
def _fn_read_args(reader: Reader, cur: List = None) -> List:
|
|
|
|
"""Recursively read the arguments of the function."""
|
|
|
|
if cur is None:
|
|
|
|
cur = []
|
|
|
|
|
2019-03-10 05:13:22 +00:00
|
|
|
# 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()
|
2019-03-10 04:55:12 +00:00
|
|
|
|
|
|
|
if token.value == ')':
|
|
|
|
return cur
|
|
|
|
|
2019-03-10 05:13:22 +00:00
|
|
|
argtype = token
|
|
|
|
reader.expect(TokenType.whitespace)
|
|
|
|
argname = reader.next_safe()
|
|
|
|
|
|
|
|
cur.append(TypedVar(argtype.value, argname.value))
|
2019-03-10 04:55:12 +00:00
|
|
|
return _fn_read_args(reader, cur)
|
|
|
|
|
|
|
|
|
|
|
|
def read_function(reader: Reader):
|
|
|
|
"""Read a function block."""
|
2019-03-10 05:13:22 +00:00
|
|
|
reader.expect(TokenType.whitespace)
|
2019-03-10 04:55:12 +00:00
|
|
|
|
|
|
|
token = reader.next()
|
|
|
|
|
|
|
|
fn_name = '_anonymous'
|
|
|
|
fn_args = []
|
|
|
|
|
2019-03-10 05:13:22 +00:00
|
|
|
print('function token', token)
|
|
|
|
|
2019-03-10 04:55:12 +00:00
|
|
|
if token.type_ == TokenType.identifier:
|
|
|
|
fn_name = token.value
|
|
|
|
fn_args = _fn_read_args(reader)
|
|
|
|
block = read_start(reader)
|
|
|
|
elif token.value == '(':
|
|
|
|
fn_args = _fn_read_args(reader)
|
|
|
|
block = read_start(reader)
|
|
|
|
|
2019-03-10 05:13:22 +00:00
|
|
|
return Function(fn_name, fn_args, block)
|
2019-03-10 04:55:12 +00:00
|
|
|
|
|
|
|
|
|
|
|
def read_reserved(token: Token, reader: Reader):
|
|
|
|
"""Read reserved statements."""
|
|
|
|
if token.value == 'fn':
|
|
|
|
return read_function(reader)
|
|
|
|
|
|
|
|
|
|
|
|
def read_start(reader: Reader):
|
|
|
|
"""Read the start of a program."""
|
2019-03-10 05:13:22 +00:00
|
|
|
print('reader', reader)
|
|
|
|
|
2019-03-10 04:55:12 +00:00
|
|
|
token = reader.next()
|
|
|
|
ast = []
|
|
|
|
res = []
|
|
|
|
|
|
|
|
print('cur', token)
|
|
|
|
|
2019-03-10 05:13:22 +00:00
|
|
|
# TODO: statement handling
|
2019-03-10 04:55:12 +00:00
|
|
|
if token.type_ == TokenType.reserved:
|
|
|
|
res = read_reserved(token, reader)
|
|
|
|
|
2019-03-10 05:13:22 +00:00
|
|
|
# handle blocks
|
|
|
|
elif token.value == '{':
|
|
|
|
while True:
|
|
|
|
token = reader.peek()
|
|
|
|
|
|
|
|
if token.value == '}':
|
|
|
|
break
|
|
|
|
|
|
|
|
res.extend(read_start(reader))
|
|
|
|
|
|
|
|
ast.append(res)
|
2019-03-10 04:55:12 +00:00
|
|
|
return ast
|
|
|
|
|
|
|
|
|
|
|
|
def syntatic(tokens: List[Token]):
|
|
|
|
"""Create an AST out of the tokens."""
|
|
|
|
return read_start(Reader(tokens))
|