2019-03-10 04:55:12 +00:00
|
|
|
from typing import Optional, Any, List
|
|
|
|
|
|
|
|
from jortsc.parser.lexer import Token, TokenType
|
2019-03-10 05:38:52 +00:00
|
|
|
from jortsc.parser.ast_nodes import (
|
2019-03-10 06:04:37 +00:00
|
|
|
Function, TypedVar, Identifier, Import, ReturnType, String, Number,
|
|
|
|
FunctionCall
|
2019-03-10 05:38:52 +00:00
|
|
|
)
|
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
|
|
|
|
|
2019-03-10 05:38:52 +00:00
|
|
|
def __repr__(self):
|
2019-03-10 06:04:37 +00:00
|
|
|
return (f'<Reader cur={self.cur} tot={len(self.tokens)} '
|
|
|
|
f'cur_tok={self.peek()}>')
|
2019-03-10 05:38:52 +00:00
|
|
|
|
|
|
|
def peek(self) -> Optional[Token]:
|
|
|
|
"""Peek at the current token."""
|
2019-03-10 04:55:12 +00:00
|
|
|
try:
|
|
|
|
token = self.tokens[self.cur]
|
2019-03-10 05:38:52 +00:00
|
|
|
return token
|
2019-03-10 04:55:12 +00:00
|
|
|
except IndexError:
|
|
|
|
return None
|
|
|
|
|
2019-03-10 05:38:52 +00:00
|
|
|
def next(self) -> Optional[Token]:
|
|
|
|
"""Fetch the current token then skip to the next one."""
|
|
|
|
token = self.peek()
|
2019-03-10 04:55:12 +00:00
|
|
|
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:
|
2019-03-10 05:38:52 +00:00
|
|
|
raise ParseError(f'Expected {token_type}, got '
|
|
|
|
f'{token.type_} {token.value!r}')
|
2019-03-10 05:13:22 +00:00
|
|
|
|
|
|
|
return token
|
|
|
|
|
2019-03-10 05:38:52 +00:00
|
|
|
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
|
|
|
|
|
|
|
|
|
2019-03-10 05:13:22 +00:00
|
|
|
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 05:38:52 +00:00
|
|
|
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
|
|
|
|
|
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)
|
|
|
|
|
|
|
|
|
2019-03-10 05:38:52 +00:00
|
|
|
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)
|
|
|
|
|
|
|
|
|
2019-03-10 04:55:12 +00:00
|
|
|
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
|
2019-03-10 05:38:52 +00:00
|
|
|
|
|
|
|
reader.expect(TokenType.whitespace)
|
|
|
|
reader.expect_val('(')
|
|
|
|
|
2019-03-10 04:55:12 +00:00
|
|
|
fn_args = _fn_read_args(reader)
|
2019-03-10 05:38:52 +00:00
|
|
|
|
|
|
|
reader.expect(TokenType.whitespace)
|
|
|
|
fn_ret_type = _fn_ret_type(reader)
|
|
|
|
|
|
|
|
# only skip whitespace if we see it
|
|
|
|
reader.ignore(TokenType.whitespace)
|
2019-03-10 04:55:12 +00:00
|
|
|
block = read_start(reader)
|
|
|
|
elif token.value == '(':
|
|
|
|
fn_args = _fn_read_args(reader)
|
2019-03-10 05:38:52 +00:00
|
|
|
fn_ret_type = _fn_ret_type(reader)
|
2019-03-10 04:55:12 +00:00
|
|
|
block = read_start(reader)
|
|
|
|
|
2019-03-10 05:38:52 +00:00
|
|
|
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,
|
|
|
|
}
|
2019-03-10 04:55:12 +00:00
|
|
|
|
|
|
|
|
|
|
|
def read_reserved(token: Token, reader: Reader):
|
|
|
|
"""Read reserved statements."""
|
2019-03-10 05:38:52 +00:00
|
|
|
try:
|
|
|
|
handler = HANDLERS[token.value]
|
|
|
|
except KeyError:
|
|
|
|
raise ParseError(f'Unexpected reserved word {token.value!r}')
|
|
|
|
|
|
|
|
return handler(reader)
|
2019-03-10 04:55:12 +00:00
|
|
|
|
|
|
|
|
2019-03-10 06:04:37 +00:00
|
|
|
def read_value(token: Token, _reader: Reader):
|
|
|
|
"""Read a given value"""
|
|
|
|
if token.type_ == TokenType.string:
|
|
|
|
return String(token.value)
|
|
|
|
elif token.type_ == TokenType.number:
|
|
|
|
return Number(token.value)
|
|
|
|
|
|
|
|
|
|
|
|
def read_statement(token: Token, reader: Reader):
|
|
|
|
"""Read a statement"""
|
|
|
|
# token is an identifier, so first check for a function call
|
|
|
|
|
|
|
|
# TODO: handle more things than a function call
|
|
|
|
call_fn_name = token.value
|
|
|
|
token = reader.expect_val('(')
|
|
|
|
res = []
|
|
|
|
|
|
|
|
while True:
|
|
|
|
token = reader.next_safe()
|
|
|
|
|
|
|
|
if token.value == ')':
|
|
|
|
break
|
|
|
|
|
|
|
|
res.append(read_value(token, reader))
|
|
|
|
|
|
|
|
return FunctionCall(call_fn_name, res)
|
|
|
|
|
|
|
|
|
2019-03-10 04:55:12 +00:00
|
|
|
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()
|
2019-03-10 05:38:52 +00:00
|
|
|
|
|
|
|
if token is None:
|
2019-03-10 05:47:02 +00:00
|
|
|
print('eof!')
|
|
|
|
return None
|
2019-03-10 05:38:52 +00:00
|
|
|
|
2019-03-10 04:55:12 +00:00
|
|
|
ast = []
|
|
|
|
res = []
|
|
|
|
|
2019-03-10 05:13:22 +00:00
|
|
|
# handle blocks
|
2019-03-10 05:38:52 +00:00
|
|
|
if token.value == '{':
|
2019-03-10 05:47:02 +00:00
|
|
|
# next can be a whitespace, or a }
|
|
|
|
token = reader.next()
|
|
|
|
|
|
|
|
print('block start!, next:', token)
|
2019-03-10 05:13:22 +00:00
|
|
|
|
2019-03-10 05:47:02 +00:00
|
|
|
if token.type_ == TokenType.whitespace:
|
|
|
|
# keep going on reading
|
|
|
|
while True:
|
2019-03-10 06:04:37 +00:00
|
|
|
token = reader.peek()
|
|
|
|
print('block append', token)
|
2019-03-10 05:13:22 +00:00
|
|
|
|
2019-03-10 05:47:02 +00:00
|
|
|
if token.value == '}':
|
2019-03-10 06:04:37 +00:00
|
|
|
print('block end')
|
|
|
|
reader.next()
|
2019-03-10 05:47:02 +00:00
|
|
|
break
|
|
|
|
|
|
|
|
res.extend(read_start(reader))
|
|
|
|
elif token.value == '}':
|
|
|
|
res = []
|
2019-03-10 05:13:22 +00:00
|
|
|
|
2019-03-10 05:38:52 +00:00
|
|
|
# import, fn, etc
|
|
|
|
elif token.type_ == TokenType.reserved:
|
|
|
|
res = read_reserved(token, reader)
|
|
|
|
|
|
|
|
elif token.type_ == TokenType.comment:
|
|
|
|
return []
|
|
|
|
|
|
|
|
elif token.type_ == TokenType.identifier:
|
2019-03-10 06:04:37 +00:00
|
|
|
res = read_statement(token, reader)
|
|
|
|
else:
|
|
|
|
res = read_value(token, reader)
|
2019-03-10 05:38:52 +00:00
|
|
|
|
2019-03-10 05:13:22 +00:00
|
|
|
ast.append(res)
|
2019-03-10 04:55:12 +00:00
|
|
|
return ast
|
|
|
|
|
2019-03-10 05:47:02 +00:00
|
|
|
|
2019-03-10 05:38:52 +00:00
|
|
|
def read_loop(reader: Reader):
|
2019-03-10 05:47:02 +00:00
|
|
|
"""Read the AST."""
|
2019-03-10 05:38:52 +00:00
|
|
|
final_ast = []
|
|
|
|
|
|
|
|
while True:
|
|
|
|
ast = read_start(reader)
|
|
|
|
|
2019-03-10 05:47:02 +00:00
|
|
|
# break when eof
|
|
|
|
if ast is None:
|
2019-03-10 05:38:52 +00:00
|
|
|
break
|
|
|
|
|
2019-03-10 06:04:37 +00:00
|
|
|
# TODO: better ast cleanup
|
2019-03-10 05:38:52 +00:00
|
|
|
|
2019-03-10 06:04:37 +00:00
|
|
|
final_ast.append(ast)
|
2019-03-10 05:38:52 +00:00
|
|
|
|
|
|
|
return final_ast
|
|
|
|
|
2019-03-10 04:55:12 +00:00
|
|
|
|
|
|
|
def syntatic(tokens: List[Token]):
|
|
|
|
"""Create an AST out of the tokens."""
|
2019-03-10 05:38:52 +00:00
|
|
|
return read_loop(Reader(tokens))
|