""" A CPython inspired RPython parser. """ from rpython.rlib.objectmodel import not_rpython class Grammar(object): """ Base Grammar object. Pass this to ParserGenerator.build_grammar to fill it with useful values for the Parser. """ def __init__(self): self.symbol_ids = {} self.symbol_names = {} self.symbol_to_label = {} self.keyword_ids = {} self.token_to_error_string = {} self.dfas = [] self.labels = [0] self.token_ids = {} self.start = -1 def shared_copy(self): new = self.__class__() new.symbol_ids = self.symbol_ids new.symbols_names = self.symbol_names new.keyword_ids = self.keyword_ids new.token_to_error_string = self.token_to_error_string new.dfas = self.dfas new.labels = self.labels new.token_ids = self.token_ids return new def classify(self, token): """Find the label for a token.""" if token.token_type == self.KEYWORD_TOKEN: label_index = self.keyword_ids.get(token.value, -1) if label_index != -1: return label_index label_index = self.token_ids.get(token.token_type, -1) if label_index == -1: raise ParseError("invalid token", token) return label_index def _freeze_(self): # Remove some attributes not used in parsing. try: del self.symbol_to_label del self.symbol_names del self.symbol_ids except AttributeError: pass return True class DFA(object): def __init__(self, grammar, symbol_id, states, first): self.grammar = grammar self.symbol_id = symbol_id self.states = states self.first = self._first_to_string(first) self.grammar = grammar def could_match_token(self, label_index): pos = label_index >> 3 bit = 1 << (label_index & 0b111) return bool(ord(self.first[label_index >> 3]) & bit) @staticmethod @not_rpython def _first_to_string(first): l = sorted(first.keys()) b = bytearray(32) for label_index in l: pos = label_index >> 3 bit = 1 << (label_index & 0b111) b[pos] |= bit return str(b) class TokenASTBase(object): _attrs_ = [] class Token(TokenASTBase): def __init__(self, token_type, value, lineno, column, line, end_lineno=-1, end_column=-1): self.token_type = token_type self.value = value self.lineno = lineno # 0-based offset self.column = column self.line = line self.end_lineno = end_lineno self.end_column = end_column def __repr__(self): from pypy.interpreter.pyparser.pytoken import token_names return "Token(%s, %s)" % (token_names.get(self.token_type, self.token_type), self.value) def __eq__(self, other): # for tests return ( self.token_type == other.token_type and self.value == other.value and self.lineno == other.lineno and self.column == other.column and self.line == other.line and self.end_lineno == other.end_lineno and self.end_column == other.end_column ) def __ne__(self, other): return not self == other class Node(object): __slots__ = ("grammar", "type") def __init__(self, grammar, type): assert grammar is None or isinstance(grammar, Grammar) assert isinstance(type, int) self.grammar = grammar self.type = type def __eq__(self, other): raise NotImplementedError("abstract base class") def __ne__(self, other): return not self == other def get_value(self): return None def get_child(self, i): raise NotImplementedError("abstract base class") def num_children(self): return 0 def append_child(self, child): raise NotImplementedError("abstract base class") def get_lineno(self): raise NotImplementedError("abstract base class") def get_column(self): raise NotImplementedError("abstract base class") def get_line(self): raise NotImplementedError("abstract base class") def flatten(self, res=None): if res is None: res = [] for i in range(self.num_children()): child = self.get_child(i) if isinstance(child, Terminal): res.append(child) else: child.flatten(res) return res def view(self): from dotviewer import graphclient import pytest r = ["digraph G {"] self._dot(r) r.append("}") p = pytest.ensuretemp("pyparser").join("temp.dot") p.write("\n".join(r)) graphclient.display_dot_file(str(p)) def _dot(self, result): raise NotImplementedError("abstract base class") class Terminal(Node): __slots__ = ("value", "lineno", "column", "line", "end_lineno", "end_column") def __init__(self, grammar, type, value, lineno, column, line=None, end_lineno=-1, end_column=-1): Node.__init__(self, grammar, type) self.value = value self.lineno = lineno self.column = column self.line = line self.end_lineno = end_lineno self.end_column = end_column @staticmethod def fromtoken(grammar, token): return Terminal( grammar, token.token_type, token.value, token.lineno, token.column, token.line, token.end_lineno, token.end_column) def __repr__(self): return "Terminal(type=%s, value=%r)" % (self.type, self.value) def __eq__(self, other): # For tests. return (type(self) == type(other) and self.type == other.type and self.value == other.value) def get_value(self): return self.value def get_lineno(self): return self.lineno def get_column(self): return self.column def get_end_lineno(self): return self.end_lineno def get_end_column(self): return self.end_column def get_line(self): return self.line def _dot(self, result): result.append('%s [label="%r", shape=box];' % (id(self), self.value)) class AbstractNonterminal(Node): __slots__ = () def get_lineno(self): return self.get_child(0).get_lineno() def get_column(self): return self.get_child(0).get_column() def get_line(self): return self.get_child(0).get_line() def get_end_lineno(self): return self.get_child(self.num_children() - 1).get_end_lineno() def get_end_column(self): return self.get_child(self.num_children() - 1).get_end_column() def __eq__(self, other): # For tests. # grumble, annoying if not isinstance(other, AbstractNonterminal): return False if self.type != other.type: return False if self.num_children() != other.num_children(): return False for i in range(self.num_children()): if self.get_child(i) != other.get_child(i): return False return True def _dot(self, result): for i in range(self.num_children()): child = self.get_child(i) result.append('%s [label=%s, shape=box]' % (id(self), self.grammar.symbol_names[self.type])) result.append('%s -> %s [label="%s"]' % (id(self), id(child), i)) child._dot(result) class Nonterminal(AbstractNonterminal): __slots__ = ("_children", ) def __init__(self, grammar, type, children=None): Node.__init__(self, grammar, type) if children is None: children = [] self._children = children def __repr__(self): return "Nonterminal(type=%s, children=%r)" % ( self.grammar.symbol_names[self.type] if self.grammar is not None else self.type, self._children) def get_child(self, i): assert self._children is not None return self._children[i] def num_children(self): return len(self._children) def append_child(self, child): self._children.append(child) class Nonterminal1(AbstractNonterminal): __slots__ = ("_child", ) def __init__(self, grammar, type, child): Node.__init__(self, grammar, type) self._child = child def __repr__(self): return "Nonterminal(type=%s, children=[%r])" % ( self.grammar.symbol_names[self.type] if self.grammar is not None else self.type, self._child) def get_child(self, i): assert i == 0 or i == -1 return self._child def num_children(self): return 1 def append_child(self, child): assert 0, "should be unreachable" class ParseError(Exception): def __init__(self, msg, token, expected=-1, expected_str=None): self.msg = msg self.token = token self.expected = expected self.expected_str = expected_str def __str__(self): return "ParserError(%s)" % (self.token, ) class StackEntry(object): def __init__(self, next, dfa, state): self.next = next self.dfa = dfa self.state = state self.node = None def push(self, dfa, state): return StackEntry(self, dfa, state) def pop(self): return self.next def node_append_child(self, child): node = self.node if node is None: self.node = Nonterminal1(self.dfa.grammar, self.dfa.symbol_id, child) elif isinstance(node, Nonterminal1): newnode = self.node = Nonterminal( self.dfa.grammar, self.dfa.symbol_id, [node._child, child]) else: self.node.append_child(child) def view(self): from dotviewer import graphclient import pytest r = ["digraph G {"] self._dot(r) r.append("}") p = pytest.ensuretemp("pyparser").join("temp.dot") p.write("\n".join(r)) graphclient.display_dot_file(str(p)) def _dot(self, result): result.append('%s [label=%s, shape=box, color=white]' % (id(self), self.dfa.grammar.symbol_names[self.dfa.symbol_id])) if self.next: result.append('%s -> %s [label="next"]' % (id(self), id(self.next))) self.next._dot(result) if self.node: result.append('%s -> %s [label="node"]' % (id(self), id(self.node))) self.node._dot(result) class Parser(object): def __init__(self, grammar): self.grammar = grammar self.root = None def prepare(self, start=-1): """Setup the parser for parsing. Takes the starting symbol as an argument. """ if start == -1: start = self.grammar.start self.root = None self.stack = StackEntry(None, self.grammar.dfas[start - 256], 0) def add_token(self, token): label_index = self.grammar.classify(token) sym_id = 0 # for the annotator while True: dfa = self.stack.dfa state_index = self.stack.state states = dfa.states arcs, is_accepting = states[state_index] for i, next_state in arcs: sym_id = self.grammar.labels[i] if label_index == i: # We matched a non-terminal. self.shift(dfa.grammar, next_state, token) state = states[next_state] # While the only possible action is to accept, pop nodes off # the stack. while state[1] and not state[0]: self.pop() if self.stack is None: # Parsing is done. return True dfa = self.stack.dfa state_index = self.stack.state state = dfa.states[state_index] return False elif sym_id >= 256: sub_node_dfa = self.grammar.dfas[sym_id - 256] # Check if this token can start a child node. if sub_node_dfa.could_match_token(label_index): self.push(sub_node_dfa, next_state, sym_id) break else: # We failed to find any arcs to another state, so unless this # state is accepting, it's invalid input. if is_accepting: self.pop() if self.stack is None: raise ParseError("too much input", token) else: # If only one possible input would satisfy, attach it to the # error. possible_arcs = self._get_possible_arcs(arcs) if len(possible_arcs) == 1: possible_arc = possible_arcs[0] expected = self.grammar.labels[possible_arc[0]] expected_str = self.grammar.token_to_error_string.get( possible_arc[0], None) else: expected = -1 expected_str = None raise ParseError("bad input", token, expected, expected_str) def _get_possible_arcs(self, arcs): """Filter out pseudo tokens from the possible paths to be taken in the grammar, in order to determine most precise token type for syntax errors.""" return arcs def shift(self, grammar, next_state, token): """Shift a non-terminal and prepare for the next state.""" new_node = Terminal.fromtoken(self.grammar, token) self.stack.node_append_child(new_node) self.stack.state = next_state def push(self, next_dfa, next_state, node_type): """Push a terminal and adjust the current state.""" self.stack.state = next_state self.stack = self.stack.push(next_dfa, 0) def pop(self): """Pop an entry off the stack and make its node a child of the last.""" top = self.stack self.stack = top.pop() node = top.node assert node is not None if self.stack: self.stack.node_append_child(node) else: self.root = node def reset(self): """Reset the state-bound data"""