|
| 1 | +from collections import deque |
| 2 | +from typing import Iterable |
| 3 | + |
| 4 | +from syntactes import Token |
| 5 | +from syntactes._action import Action, ActionType |
| 6 | +from syntactes._state import LR0State |
| 7 | +from syntactes.parser import ( |
| 8 | + ExecutablesRegistry, |
| 9 | + NotAcceptedError, |
| 10 | + ParserError, |
| 11 | + UnexpectedTokenError, |
| 12 | +) |
| 13 | +from syntactes.table import LR0ParsingTable, SLRParsingTable |
| 14 | + |
| 15 | + |
| 16 | +class LR0Parser: |
| 17 | + """ |
| 18 | + Parses streams of tokens based on the configured parsing table. |
| 19 | + """ |
| 20 | + |
| 21 | + def __init__(self, table: LR0ParsingTable) -> None: |
| 22 | + self._table = table |
| 23 | + self._token_stack: deque[Token] = deque() |
| 24 | + self._state_stack: deque[LR0State] = deque() |
| 25 | + self._token_stream: deque[Token] = deque() |
| 26 | + |
| 27 | + def parse(self, stream: Iterable[Token]) -> None: |
| 28 | + """ |
| 29 | + Parses the given stream of tokens. Expects the EOF token as the last one. |
| 30 | +
|
| 31 | + Raises `syntactes.parser.UnexpectedTokenError` if an unexpected token is |
| 32 | + received. |
| 33 | +
|
| 34 | + Raises `syntactes.parser.NotAcceptedError` if the stream of token has been |
| 35 | + parsed and the parser did not receive an accept action. |
| 36 | + """ |
| 37 | + self._set_state(self._table.initial_state) |
| 38 | + self._token_stream.extend(stream) |
| 39 | + |
| 40 | + while len(self._token_stream) > 0: |
| 41 | + token = self._token_stream.popleft() |
| 42 | + self._apply_action(token, self._get_action(token)) |
| 43 | + |
| 44 | + if token != Token.eof(): |
| 45 | + self._raise(NotAcceptedError("Expected EOF token. ")) |
| 46 | + |
| 47 | + if not self._get_state().is_final: |
| 48 | + actions = self._table.get(self._get_state()) |
| 49 | + expected_tokens = [] if actions is None else list(actions.keys()) |
| 50 | + self._raise(UnexpectedTokenError(Token.eof(), expected_tokens)) |
| 51 | + |
| 52 | + def _apply_action(self, token: Token, action: Action) -> None: |
| 53 | + if action.action_type == ActionType.SHIFT: |
| 54 | + self._token_stack.append(token) |
| 55 | + self._set_state(action.actionable) |
| 56 | + elif action.action_type == ActionType.REDUCE: |
| 57 | + rule = action.actionable |
| 58 | + args = [self._token_stack.pop() for _ in reversed(rule.rhs)] |
| 59 | + self._token_stack.append(rule.lhs) |
| 60 | + |
| 61 | + {self._state_stack.pop() for _ in rule.rhs} |
| 62 | + |
| 63 | + executable = ExecutablesRegistry.get(rule) |
| 64 | + executable(*args) |
| 65 | + |
| 66 | + self._token_stream.appendleft(token) # reduce actions do not consume tokenA |
| 67 | + |
| 68 | + shift = self._get_action(rule.lhs) |
| 69 | + self._set_state(shift.actionable) |
| 70 | + |
| 71 | + def _get_action(self, token: Token) -> Action: |
| 72 | + actions = self._table.get_actions(self._get_state(), token) |
| 73 | + if actions is None: |
| 74 | + actions = self._table.get(self._get_state()) |
| 75 | + expected_tokens = [] if actions is None else list(actions.keys()) |
| 76 | + self._raise(UnexpectedTokenError(token, expected_tokens)) |
| 77 | + |
| 78 | + action = self._resolve_conflict(actions) |
| 79 | + return action |
| 80 | + |
| 81 | + def _resolve_conflict(self, actions: list[Action]) -> Action: |
| 82 | + return actions[0] |
| 83 | + |
| 84 | + def _set_state(self, state: LR0State) -> None: |
| 85 | + self._state_stack.append(state) |
| 86 | + |
| 87 | + def _get_state(self) -> LR0State: |
| 88 | + return self._state_stack[-1] |
| 89 | + |
| 90 | + def _cleanup(self) -> None: |
| 91 | + self._token_stack.clear() if len(self._token_stack) > 0 else ... |
| 92 | + self._state_stack.clear() if len(self._state_stack) > 0 else ... |
| 93 | + self._token_stream.clear() if len(self._token_stream) > 0 else ... |
| 94 | + |
| 95 | + def _raise(self, error: ParserError) -> None: |
| 96 | + self._cleanup() |
| 97 | + raise error from None |
| 98 | + |
| 99 | + |
| 100 | +class SLRParser(LR0Parser): |
| 101 | + """ |
| 102 | + Parses streams of tokens based on the configured parsing table. |
| 103 | + """ |
0 commit comments