lark/tests/test_parser.py

1637 lines
56 KiB
Python

# -*- coding: utf-8 -*-
from __future__ import absolute_import
import unittest
import logging
import os
import sys
try:
from cStringIO import StringIO as cStringIO
except ImportError:
# Available only in Python 2.x, 3.x only has io.StringIO from below
cStringIO = None
from io import (
StringIO as uStringIO,
open,
)
logging.basicConfig(level=logging.INFO)
from lark.lark import Lark
from lark.exceptions import GrammarError, ParseError, UnexpectedToken, UnexpectedInput, UnexpectedCharacters
from lark.tree import Tree
from lark.visitors import Transformer, Transformer_InPlace, v_args
from lark.grammar import Rule
from lark.lexer import TerminalDef, Lexer, TraditionalLexer
__path__ = os.path.dirname(__file__)
def _read(n, *args):
with open(os.path.join(__path__, n), *args) as f:
return f.read()
class TestParsers(unittest.TestCase):
def test_same_ast(self):
"Tests that Earley and LALR parsers produce equal trees"
g = Lark(r"""start: "(" name_list ("," "*" NAME)? ")"
name_list: NAME | name_list "," NAME
NAME: /\w+/ """, parser='lalr')
l = g.parse('(a,b,c,*x)')
g = Lark(r"""start: "(" name_list ("," "*" NAME)? ")"
name_list: NAME | name_list "," NAME
NAME: /\w/+ """)
l2 = g.parse('(a,b,c,*x)')
assert l == l2, '%s != %s' % (l.pretty(), l2.pretty())
def test_infinite_recurse(self):
g = """start: a
a: a | "a"
"""
self.assertRaises(GrammarError, Lark, g, parser='lalr')
# TODO: should it? shouldn't it?
# l = Lark(g, parser='earley', lexer='dynamic')
# self.assertRaises(ParseError, l.parse, 'a')
def test_propagate_positions(self):
g = Lark("""start: a
a: "a"
""", propagate_positions=True)
r = g.parse('a')
self.assertEqual( r.children[0].meta.line, 1 )
def test_expand1(self):
g = Lark("""start: a
?a: b
b: "x"
""")
r = g.parse('x')
self.assertEqual( r.children[0].data, "b" )
g = Lark("""start: a
?a: b -> c
b: "x"
""")
r = g.parse('x')
self.assertEqual( r.children[0].data, "c" )
g = Lark("""start: a
?a: B -> c
B: "x"
""")
self.assertEqual( r.children[0].data, "c" )
g = Lark("""start: a
?a: b b -> c
b: "x"
""")
r = g.parse('xx')
self.assertEqual( r.children[0].data, "c" )
def test_visit_tokens(self):
class T(Transformer):
def a(self, children):
return children[0] + "!"
def A(self, tok):
return tok.upper()
# Test regular
g = Lark("""start: a
a : A
A: "x"
""", parser='lalr')
r = T().transform(g.parse("x"))
self.assertEqual( r.children, ["x!"] )
r = T(True).transform(g.parse("x"))
self.assertEqual( r.children, ["X!"] )
def test_embedded_transformer(self):
class T(Transformer):
def a(self, children):
return "<a>"
def b(self, children):
return "<b>"
def c(self, children):
return "<c>"
# Test regular
g = Lark("""start: a
a : "x"
""", parser='lalr')
r = T().transform(g.parse("x"))
self.assertEqual( r.children, ["<a>"] )
g = Lark("""start: a
a : "x"
""", parser='lalr', transformer=T())
r = g.parse("x")
self.assertEqual( r.children, ["<a>"] )
# Test Expand1
g = Lark("""start: a
?a : b
b : "x"
""", parser='lalr')
r = T().transform(g.parse("x"))
self.assertEqual( r.children, ["<b>"] )
g = Lark("""start: a
?a : b
b : "x"
""", parser='lalr', transformer=T())
r = g.parse("x")
self.assertEqual( r.children, ["<b>"] )
# Test Expand1 -> Alias
g = Lark("""start: a
?a : b b -> c
b : "x"
""", parser='lalr')
r = T().transform(g.parse("xx"))
self.assertEqual( r.children, ["<c>"] )
g = Lark("""start: a
?a : b b -> c
b : "x"
""", parser='lalr', transformer=T())
r = g.parse("xx")
self.assertEqual( r.children, ["<c>"] )
def test_embedded_transformer_inplace(self):
@v_args(tree=True)
class T1(Transformer_InPlace):
def a(self, tree):
assert isinstance(tree, Tree), tree
tree.children.append("tested")
return tree
def b(self, tree):
return Tree(tree.data, tree.children + ['tested2'])
@v_args(tree=True)
class T2(Transformer):
def a(self, tree):
assert isinstance(tree, Tree)
tree.children.append("tested")
return tree
def b(self, tree):
return Tree(tree.data, tree.children + ['tested2'])
class T3(Transformer):
@v_args(tree=True)
def a(self, tree):
assert isinstance(tree, Tree)
tree.children.append("tested")
return tree
@v_args(tree=True)
def b(self, tree):
return Tree(tree.data, tree.children + ['tested2'])
for t in [T1(), T2(), T3()]:
for internal in [False, True]:
g = Lark("""start: a b
a : "x"
b : "y"
""", parser='lalr', transformer=t if internal else None)
r = g.parse("xy")
if not internal:
r = t.transform(r)
a, b = r.children
self.assertEqual(a.children, ["tested"])
self.assertEqual(b.children, ["tested2"])
def test_alias(self):
Lark("""start: ["a"] "b" ["c"] "e" ["f"] ["g"] ["h"] "x" -> d """)
def _make_full_earley_test(LEXER):
def _Lark(grammar, **kwargs):
return Lark(grammar, lexer=LEXER, parser='earley', propagate_positions=True, **kwargs)
class _TestFullEarley(unittest.TestCase):
def test_anon(self):
# Fails an Earley implementation without special handling for empty rules,
# or re-processing of already completed rules.
g = Lark(r"""start: B
B: ("ab"|/[^b]/)+
""", lexer=LEXER)
self.assertEqual( g.parse('abc').children[0], 'abc')
def test_earley(self):
g = Lark("""start: A "b" c
A: "a"+
c: "abc"
""", parser="earley", lexer=LEXER)
x = g.parse('aaaababc')
def test_earley2(self):
grammar = """
start: statement+
statement: "r"
| "c" /[a-z]/+
%ignore " "
"""
program = """c b r"""
l = Lark(grammar, parser='earley', lexer=LEXER)
l.parse(program)
@unittest.skipIf(LEXER=='dynamic', "Only relevant for the dynamic_complete parser")
def test_earley3(self):
"""Tests prioritization and disambiguation for pseudo-terminals (there should be only one result)
By default, `+` should immitate regexp greedy-matching
"""
grammar = """
start: A A
A: "a"+
"""
l = Lark(grammar, parser='earley', lexer=LEXER)
res = l.parse("aaa")
self.assertEqual(set(res.children), {'aa', 'a'})
# XXX TODO fix Earley to maintain correct order
# i.e. terminals it imitate greedy search for terminals, but lazy search for rules
# self.assertEqual(res.children, ['aa', 'a'])
def test_earley4(self):
grammar = """
start: A A?
A: "a"+
"""
l = Lark(grammar, parser='earley', lexer=LEXER)
res = l.parse("aaa")
assert set(res.children) == {'aa', 'a'} or res.children == ['aaa']
# XXX TODO fix Earley to maintain correct order
# i.e. terminals it imitate greedy search for terminals, but lazy search for rules
# self.assertEqual(res.children, ['aaa'])
def test_earley_repeating_empty(self):
# This was a sneaky bug!
grammar = """
!start: "a" empty empty "b"
empty: empty2
empty2:
"""
parser = Lark(grammar, parser='earley', lexer=LEXER)
res = parser.parse('ab')
empty_tree = Tree('empty', [Tree('empty2', [])])
self.assertSequenceEqual(res.children, ['a', empty_tree, empty_tree, 'b'])
@unittest.skipIf(LEXER=='standard', "Requires dynamic lexer")
def test_earley_explicit_ambiguity(self):
# This was a sneaky bug!
grammar = """
start: a b | ab
a: "a"
b: "b"
ab: "ab"
"""
parser = Lark(grammar, parser='earley', lexer=LEXER, ambiguity='explicit')
ambig_tree = parser.parse('ab')
self.assertEqual( ambig_tree.data, '_ambig')
self.assertEqual( len(ambig_tree.children), 2)
@unittest.skipIf(LEXER=='standard', "Requires dynamic lexer")
def test_ambiguity1(self):
grammar = """
start: cd+ "e"
!cd: "c"
| "d"
| "cd"
"""
l = Lark(grammar, parser='earley', ambiguity='explicit', lexer=LEXER)
ambig_tree = l.parse('cde')
assert ambig_tree.data == '_ambig', ambig_tree
assert len(ambig_tree.children) == 2
@unittest.skipIf(LEXER=='standard', "Requires dynamic lexer")
def test_ambiguity2(self):
grammar = """
ANY: /[a-zA-Z0-9 ]+/
a.2: "A" b+
b.2: "B"
c: ANY
start: (a|c)*
"""
l = Lark(grammar, parser='earley', lexer=LEXER)
res = l.parse('ABX')
expected = Tree('start', [
Tree('a', [
Tree('b', [])
]),
Tree('c', [
'X'
])
])
self.assertEqual(res, expected)
def test_fruitflies_ambig(self):
grammar = """
start: noun verb noun -> simple
| noun verb "like" noun -> comparative
noun: adj? NOUN
verb: VERB
adj: ADJ
NOUN: "flies" | "bananas" | "fruit"
VERB: "like" | "flies"
ADJ: "fruit"
%import common.WS
%ignore WS
"""
parser = Lark(grammar, ambiguity='explicit', lexer=LEXER)
tree = parser.parse('fruit flies like bananas')
expected = Tree('_ambig', [
Tree('comparative', [
Tree('noun', ['fruit']),
Tree('verb', ['flies']),
Tree('noun', ['bananas'])
]),
Tree('simple', [
Tree('noun', [Tree('adj', ['fruit']), 'flies']),
Tree('verb', ['like']),
Tree('noun', ['bananas'])
])
])
# self.assertEqual(tree, expected)
self.assertEqual(tree.data, expected.data)
self.assertEqual(set(tree.children), set(expected.children))
@unittest.skipIf(LEXER!='dynamic_complete', "Only relevant for the dynamic_complete parser")
def test_explicit_ambiguity2(self):
grammar = r"""
start: NAME+
NAME: /\w+/
%ignore " "
"""
text = """cat"""
parser = _Lark(grammar, start='start', ambiguity='explicit')
tree = parser.parse(text)
self.assertEqual(tree.data, '_ambig')
combinations = {tuple(str(s) for s in t.children) for t in tree.children}
self.assertEqual(combinations, {
('cat',),
('ca', 't'),
('c', 'at'),
('c', 'a' ,'t')
})
def test_term_ambig_resolve(self):
grammar = r"""
!start: NAME+
NAME: /\w+/
%ignore " "
"""
text = """foo bar"""
parser = Lark(grammar)
tree = parser.parse(text)
self.assertEqual(tree.children, ['foo', 'bar'])
# @unittest.skipIf(LEXER=='dynamic', "Not implemented in Dynamic Earley yet") # TODO
# def test_not_all_derivations(self):
# grammar = """
# start: cd+ "e"
# !cd: "c"
# | "d"
# | "cd"
# """
# l = Lark(grammar, parser='earley', ambiguity='explicit', lexer=LEXER, earley__all_derivations=False)
# x = l.parse('cde')
# assert x.data != '_ambig', x
# assert len(x.children) == 1
_NAME = "TestFullEarley" + LEXER.capitalize()
_TestFullEarley.__name__ = _NAME
globals()[_NAME] = _TestFullEarley
class CustomLexer(Lexer):
"""
Purpose of this custom lexer is to test the integration,
so it uses the traditionalparser as implementation without custom lexing behaviour.
"""
def __init__(self, lexer_conf):
self.lexer = TraditionalLexer(lexer_conf.tokens, ignore=lexer_conf.ignore, user_callbacks=lexer_conf.callbacks)
def lex(self, *args, **kwargs):
return self.lexer.lex(*args, **kwargs)
def _make_parser_test(LEXER, PARSER):
lexer_class_or_name = CustomLexer if LEXER == 'custom' else LEXER
def _Lark(grammar, **kwargs):
return Lark(grammar, lexer=lexer_class_or_name, parser=PARSER, propagate_positions=True, **kwargs)
def _Lark_open(gfilename, **kwargs):
return Lark.open(gfilename, lexer=lexer_class_or_name, parser=PARSER, propagate_positions=True, **kwargs)
class _TestParser(unittest.TestCase):
def test_basic1(self):
g = _Lark("""start: a+ b a* "b" a*
b: "b"
a: "a"
""")
r = g.parse('aaabaab')
self.assertEqual( ''.join(x.data for x in r.children), 'aaabaa' )
r = g.parse('aaabaaba')
self.assertEqual( ''.join(x.data for x in r.children), 'aaabaaa' )
self.assertRaises(ParseError, g.parse, 'aaabaa')
def test_basic2(self):
# Multiple parsers and colliding tokens
g = _Lark("""start: B A
B: "12"
A: "1" """)
g2 = _Lark("""start: B A
B: "12"
A: "2" """)
x = g.parse('121')
assert x.data == 'start' and x.children == ['12', '1'], x
x = g2.parse('122')
assert x.data == 'start' and x.children == ['12', '2'], x
@unittest.skipIf(cStringIO is None, "cStringIO not available")
def test_stringio_bytes(self):
"""Verify that a Lark can be created from file-like objects other than Python's standard 'file' object"""
_Lark(cStringIO(b'start: a+ b a* "b" a*\n b: "b"\n a: "a" '))
def test_stringio_unicode(self):
"""Verify that a Lark can be created from file-like objects other than Python's standard 'file' object"""
_Lark(uStringIO(u'start: a+ b a* "b" a*\n b: "b"\n a: "a" '))
def test_unicode(self):
g = _Lark(u"""start: UNIA UNIB UNIA
UNIA: /\xa3/
UNIB: /\u0101/
""")
g.parse(u'\xa3\u0101\u00a3')
def test_unicode2(self):
g = _Lark(r"""start: UNIA UNIB UNIA UNIC
UNIA: /\xa3/
UNIB: "a\u0101b\ "
UNIC: /a?\u0101c\n/
""")
g.parse(u'\xa3a\u0101b\\ \u00a3\u0101c\n')
def test_unicode3(self):
g = _Lark(r"""start: UNIA UNIB UNIA UNIC
UNIA: /\xa3/
UNIB: "\u0101"
UNIC: /\u0203/ /\n/
""")
g.parse(u'\xa3\u0101\u00a3\u0203\n')
def test_hex_escape(self):
g = _Lark(r"""start: A B C
A: "\x01"
B: /\x02/
C: "\xABCD"
""")
g.parse('\x01\x02\xABCD')
def test_unicode_literal_range_escape(self):
g = _Lark(r"""start: A+
A: "\u0061".."\u0063"
""")
g.parse('abc')
def test_hex_literal_range_escape(self):
g = _Lark(r"""start: A+
A: "\x01".."\x03"
""")
g.parse('\x01\x02\x03')
@unittest.skipIf(PARSER == 'cyk', "Takes forever")
def test_stack_for_ebnf(self):
"""Verify that stack depth isn't an issue for EBNF grammars"""
g = _Lark(r"""start: a+
a : "a" """)
g.parse("a" * (sys.getrecursionlimit()*2 ))
def test_expand1_lists_with_one_item(self):
g = _Lark(r"""start: list
?list: item+
item : A
A: "a"
""")
r = g.parse("a")
# because 'list' is an expand-if-contains-one rule and we only provided one element it should have expanded to 'item'
self.assertSequenceEqual([subtree.data for subtree in r.children], ('item',))
# regardless of the amount of items: there should be only *one* child in 'start' because 'list' isn't an expand-all rule
self.assertEqual(len(r.children), 1)
def test_expand1_lists_with_one_item_2(self):
g = _Lark(r"""start: list
?list: item+ "!"
item : A
A: "a"
""")
r = g.parse("a!")
# because 'list' is an expand-if-contains-one rule and we only provided one element it should have expanded to 'item'
self.assertSequenceEqual([subtree.data for subtree in r.children], ('item',))
# regardless of the amount of items: there should be only *one* child in 'start' because 'list' isn't an expand-all rule
self.assertEqual(len(r.children), 1)
def test_dont_expand1_lists_with_multiple_items(self):
g = _Lark(r"""start: list
?list: item+
item : A
A: "a"
""")
r = g.parse("aa")
# because 'list' is an expand-if-contains-one rule and we've provided more than one element it should *not* have expanded
self.assertSequenceEqual([subtree.data for subtree in r.children], ('list',))
# regardless of the amount of items: there should be only *one* child in 'start' because 'list' isn't an expand-all rule
self.assertEqual(len(r.children), 1)
# Sanity check: verify that 'list' contains the two 'item's we've given it
[list] = r.children
self.assertSequenceEqual([item.data for item in list.children], ('item', 'item'))
def test_dont_expand1_lists_with_multiple_items_2(self):
g = _Lark(r"""start: list
?list: item+ "!"
item : A
A: "a"
""")
r = g.parse("aa!")
# because 'list' is an expand-if-contains-one rule and we've provided more than one element it should *not* have expanded
self.assertSequenceEqual([subtree.data for subtree in r.children], ('list',))
# regardless of the amount of items: there should be only *one* child in 'start' because 'list' isn't an expand-all rule
self.assertEqual(len(r.children), 1)
# Sanity check: verify that 'list' contains the two 'item's we've given it
[list] = r.children
self.assertSequenceEqual([item.data for item in list.children], ('item', 'item'))
@unittest.skipIf(PARSER == 'cyk', "No empty rules")
def test_empty_expand1_list(self):
g = _Lark(r"""start: list
?list: item*
item : A
A: "a"
""")
r = g.parse("")
# because 'list' is an expand-if-contains-one rule and we've provided less than one element (i.e. none) it should *not* have expanded
self.assertSequenceEqual([subtree.data for subtree in r.children], ('list',))
# regardless of the amount of items: there should be only *one* child in 'start' because 'list' isn't an expand-all rule
self.assertEqual(len(r.children), 1)
# Sanity check: verify that 'list' contains no 'item's as we've given it none
[list] = r.children
self.assertSequenceEqual([item.data for item in list.children], ())
@unittest.skipIf(PARSER == 'cyk', "No empty rules")
def test_empty_expand1_list_2(self):
g = _Lark(r"""start: list
?list: item* "!"?
item : A
A: "a"
""")
r = g.parse("")
# because 'list' is an expand-if-contains-one rule and we've provided less than one element (i.e. none) it should *not* have expanded
self.assertSequenceEqual([subtree.data for subtree in r.children], ('list',))
# regardless of the amount of items: there should be only *one* child in 'start' because 'list' isn't an expand-all rule
self.assertEqual(len(r.children), 1)
# Sanity check: verify that 'list' contains no 'item's as we've given it none
[list] = r.children
self.assertSequenceEqual([item.data for item in list.children], ())
@unittest.skipIf(PARSER == 'cyk', "No empty rules")
def test_empty_flatten_list(self):
g = _Lark(r"""start: list
list: | item "," list
item : A
A: "a"
""")
r = g.parse("")
# Because 'list' is a flatten rule it's top-level element should *never* be expanded
self.assertSequenceEqual([subtree.data for subtree in r.children], ('list',))
# Sanity check: verify that 'list' contains no 'item's as we've given it none
[list] = r.children
self.assertSequenceEqual([item.data for item in list.children], ())
@unittest.skipIf(True, "Flattening list isn't implemented (and may never be)")
def test_single_item_flatten_list(self):
g = _Lark(r"""start: list
list: | item "," list
item : A
A: "a"
""")
r = g.parse("a,")
# Because 'list' is a flatten rule it's top-level element should *never* be expanded
self.assertSequenceEqual([subtree.data for subtree in r.children], ('list',))
# Sanity check: verify that 'list' contains exactly the one 'item' we've given it
[list] = r.children
self.assertSequenceEqual([item.data for item in list.children], ('item',))
@unittest.skipIf(True, "Flattening list isn't implemented (and may never be)")
def test_multiple_item_flatten_list(self):
g = _Lark(r"""start: list
#list: | item "," list
item : A
A: "a"
""")
r = g.parse("a,a,")
# Because 'list' is a flatten rule it's top-level element should *never* be expanded
self.assertSequenceEqual([subtree.data for subtree in r.children], ('list',))
# Sanity check: verify that 'list' contains exactly the two 'item's we've given it
[list] = r.children
self.assertSequenceEqual([item.data for item in list.children], ('item', 'item'))
@unittest.skipIf(True, "Flattening list isn't implemented (and may never be)")
def test_recurse_flatten(self):
"""Verify that stack depth doesn't get exceeded on recursive rules marked for flattening."""
g = _Lark(r"""start: a | start a
a : A
A : "a" """)
# Force PLY to write to the debug log, but prevent writing it to the terminal (uses repr() on the half-built
# STree data structures, which uses recursion).
g.parse("a" * (sys.getrecursionlimit() // 4))
def test_token_collision(self):
g = _Lark(r"""start: "Hello" NAME
NAME: /\w/+
%ignore " "
""")
x = g.parse('Hello World')
self.assertSequenceEqual(x.children, ['World'])
x = g.parse('Hello HelloWorld')
self.assertSequenceEqual(x.children, ['HelloWorld'])
def test_token_collision_WS(self):
g = _Lark(r"""start: "Hello" NAME
NAME: /\w/+
%import common.WS
%ignore WS
""")
x = g.parse('Hello World')
self.assertSequenceEqual(x.children, ['World'])
x = g.parse('Hello HelloWorld')
self.assertSequenceEqual(x.children, ['HelloWorld'])
def test_token_collision2(self):
g = _Lark("""
!start: "starts"
%import common.LCASE_LETTER
""")
x = g.parse("starts")
self.assertSequenceEqual(x.children, ['starts'])
# def test_string_priority(self):
# g = _Lark("""start: (A | /a?bb/)+
# A: "a" """)
# x = g.parse('abb')
# self.assertEqual(len(x.children), 2)
# # This parse raises an exception because the lexer will always try to consume
# # "a" first and will never match the regular expression
# # This behavior is subject to change!!
# # Thie won't happen with ambiguity handling.
# g = _Lark("""start: (A | /a?ab/)+
# A: "a" """)
# self.assertRaises(LexError, g.parse, 'aab')
def test_undefined_rule(self):
self.assertRaises(GrammarError, _Lark, """start: a""")
def test_undefined_token(self):
self.assertRaises(GrammarError, _Lark, """start: A""")
def test_rule_collision(self):
g = _Lark("""start: "a"+ "b"
| "a"+ """)
x = g.parse('aaaa')
x = g.parse('aaaab')
def test_rule_collision2(self):
g = _Lark("""start: "a"* "b"
| "a"+ """)
x = g.parse('aaaa')
x = g.parse('aaaab')
x = g.parse('b')
def test_token_not_anon(self):
"""Tests that "a" is matched as an anonymous token, and not A.
"""
g = _Lark("""start: "a"
A: "a" """)
x = g.parse('a')
self.assertEqual(len(x.children), 0, '"a" should be considered anonymous')
g = _Lark("""start: "a" A
A: "a" """)
x = g.parse('aa')
self.assertEqual(len(x.children), 1, 'only "a" should be considered anonymous')
self.assertEqual(x.children[0].type, "A")
g = _Lark("""start: /a/
A: /a/ """)
x = g.parse('a')
self.assertEqual(len(x.children), 1)
self.assertEqual(x.children[0].type, "A", "A isn't associated with /a/")
@unittest.skipIf(PARSER == 'cyk', "No empty rules")
def test_maybe(self):
g = _Lark("""start: ["a"] """)
x = g.parse('a')
x = g.parse('')
def test_start(self):
g = _Lark("""a: "a" a? """, start='a')
x = g.parse('a')
x = g.parse('aa')
x = g.parse('aaa')
def test_alias(self):
g = _Lark("""start: "a" -> b """)
x = g.parse('a')
self.assertEqual(x.data, "b")
def test_token_ebnf(self):
g = _Lark("""start: A
A: "a"* ("b"? "c".."e")+
""")
x = g.parse('abcde')
x = g.parse('dd')
def test_backslash(self):
g = _Lark(r"""start: "\\" "a"
""")
x = g.parse(r'\a')
g = _Lark(r"""start: /\\/ /a/
""")
x = g.parse(r'\a')
def test_backslash2(self):
g = _Lark(r"""start: "\"" "-"
""")
x = g.parse('"-')
g = _Lark(r"""start: /\// /-/
""")
x = g.parse('/-')
def test_special_chars(self):
g = _Lark(r"""start: "\n"
""")
x = g.parse('\n')
g = _Lark(r"""start: /\n/
""")
x = g.parse('\n')
# def test_token_recurse(self):
# g = _Lark("""start: A
# A: B
# B: A
# """)
@unittest.skipIf(PARSER == 'cyk', "No empty rules")
def test_empty(self):
# Fails an Earley implementation without special handling for empty rules,
# or re-processing of already completed rules.
g = _Lark(r"""start: _empty a "B"
a: _empty "A"
_empty:
""")
x = g.parse('AB')
def test_regex_quote(self):
g = r"""
start: SINGLE_QUOTED_STRING | DOUBLE_QUOTED_STRING
SINGLE_QUOTED_STRING : /'[^']*'/
DOUBLE_QUOTED_STRING : /"[^"]*"/
"""
g = _Lark(g)
self.assertEqual( g.parse('"hello"').children, ['"hello"'])
self.assertEqual( g.parse("'hello'").children, ["'hello'"])
def test_lexer_token_limit(self):
"Python has a stupid limit of 100 groups in a regular expression. Test that we handle this limitation"
tokens = {'A%d'%i:'"%d"'%i for i in range(300)}
g = _Lark("""start: %s
%s""" % (' '.join(tokens), '\n'.join("%s: %s"%x for x in tokens.items())))
def test_float_without_lexer(self):
expected_error = UnexpectedCharacters if LEXER.startswith('dynamic') else UnexpectedToken
if PARSER == 'cyk':
expected_error = ParseError
g = _Lark("""start: ["+"|"-"] float
float: digit* "." digit+ exp?
| digit+ exp
exp: ("e"|"E") ["+"|"-"] digit+
digit: "0"|"1"|"2"|"3"|"4"|"5"|"6"|"7"|"8"|"9"
""")
g.parse("1.2")
g.parse("-.2e9")
g.parse("+2e-9")
self.assertRaises( expected_error, g.parse, "+2e-9e")
def test_keep_all_tokens(self):
l = _Lark("""start: "a"+ """, keep_all_tokens=True)
tree = l.parse('aaa')
self.assertEqual(tree.children, ['a', 'a', 'a'])
def test_token_flags(self):
l = _Lark("""!start: "a"i+
"""
)
tree = l.parse('aA')
self.assertEqual(tree.children, ['a', 'A'])
l = _Lark("""!start: /a/i+
"""
)
tree = l.parse('aA')
self.assertEqual(tree.children, ['a', 'A'])
# g = """!start: "a"i "a"
# """
# self.assertRaises(GrammarError, _Lark, g)
# g = """!start: /a/i /a/
# """
# self.assertRaises(GrammarError, _Lark, g)
g = """start: NAME "," "a"
NAME: /[a-z_]/i /[a-z0-9_]/i*
"""
l = _Lark(g)
tree = l.parse('ab,a')
self.assertEqual(tree.children, ['ab'])
tree = l.parse('AB,a')
self.assertEqual(tree.children, ['AB'])
def test_token_flags3(self):
l = _Lark("""!start: ABC+
ABC: "abc"i
"""
)
tree = l.parse('aBcAbC')
self.assertEqual(tree.children, ['aBc', 'AbC'])
def test_token_flags2(self):
g = """!start: ("a"i | /a/ /b/?)+
"""
l = _Lark(g)
tree = l.parse('aA')
self.assertEqual(tree.children, ['a', 'A'])
@unittest.skipIf(PARSER == 'cyk', "No empty rules")
def test_twice_empty(self):
g = """!start: [["A"]]
"""
l = _Lark(g)
tree = l.parse('A')
self.assertEqual(tree.children, ['A'])
tree = l.parse('')
self.assertEqual(tree.children, [])
def test_undefined_ignore(self):
g = """!start: "A"
%ignore B
"""
self.assertRaises( GrammarError, _Lark, g)
def test_alias_in_terminal(self):
g = """start: TERM
TERM: "a" -> alias
"""
self.assertRaises( GrammarError, _Lark, g)
def test_line_and_column(self):
g = r"""!start: "A" bc "D"
!bc: "B\nC"
"""
l = _Lark(g)
a, bc, d = l.parse("AB\nCD").children
self.assertEqual(a.line, 1)
self.assertEqual(a.column, 1)
bc ,= bc.children
self.assertEqual(bc.line, 1)
self.assertEqual(bc.column, 2)
self.assertEqual(d.line, 2)
self.assertEqual(d.column, 2)
if LEXER != 'dynamic':
self.assertEqual(a.end_line, 1)
self.assertEqual(a.end_column, 2)
self.assertEqual(bc.end_line, 2)
self.assertEqual(bc.end_column, 2)
self.assertEqual(d.end_line, 2)
self.assertEqual(d.end_column, 3)
def test_reduce_cycle(self):
"""Tests an edge-condition in the LALR parser, in which a transition state looks exactly like the end state.
It seems that the correct solution is to explicitely distinguish finalization in the reduce() function.
"""
l = _Lark("""
term: A
| term term
A: "a"
""", start='term')
tree = l.parse("aa")
self.assertEqual(len(tree.children), 2)
@unittest.skipIf(LEXER != 'standard', "Only standard lexers care about token priority")
def test_lexer_prioritization(self):
"Tests effect of priority on result"
grammar = """
start: A B | AB
A.2: "a"
B: "b"
AB: "ab"
"""
l = _Lark(grammar)
res = l.parse("ab")
self.assertEqual(res.children, ['a', 'b'])
self.assertNotEqual(res.children, ['ab'])
grammar = """
start: A B | AB
A: "a"
B: "b"
AB.3: "ab"
"""
l = _Lark(grammar)
res = l.parse("ab")
self.assertNotEqual(res.children, ['a', 'b'])
self.assertEqual(res.children, ['ab'])
grammar = """
start: A B | AB
A: "a"
B.-20: "b"
AB.-10: "ab"
"""
l = _Lark(grammar)
res = l.parse("ab")
self.assertEqual(res.children, ['a', 'b'])
grammar = """
start: A B | AB
A.-99999999999999999999999: "a"
B: "b"
AB: "ab"
"""
l = _Lark(grammar)
res = l.parse("ab")
self.assertEqual(res.children, ['ab'])
def test_import(self):
grammar = """
start: NUMBER WORD
%import common.NUMBER
%import common.WORD
%import common.WS
%ignore WS
"""
l = _Lark(grammar)
x = l.parse('12 elephants')
self.assertEqual(x.children, ['12', 'elephants'])
def test_import_rename(self):
grammar = """
start: N W
%import common.NUMBER -> N
%import common.WORD -> W
%import common.WS
%ignore WS
"""
l = _Lark(grammar)
x = l.parse('12 elephants')
self.assertEqual(x.children, ['12', 'elephants'])
def test_relative_import(self):
l = _Lark_open('test_relative_import.lark', rel_to=__file__)
x = l.parse('12 lions')
self.assertEqual(x.children, ['12', 'lions'])
def test_relative_import_rename(self):
l = _Lark_open('test_relative_import_rename.lark', rel_to=__file__)
x = l.parse('12 lions')
self.assertEqual(x.children, ['12', 'lions'])
def test_relative_rule_import(self):
l = _Lark_open('test_relative_rule_import.lark', rel_to=__file__)
x = l.parse('xaabby')
self.assertEqual(x.children, [
'x',
Tree('expr', ['a', Tree('expr', ['a', 'b']), 'b']),
'y'])
def test_relative_rule_import_drop_ignore(self):
# %ignore rules are dropped on import
l = _Lark_open('test_relative_rule_import_drop_ignore.lark',
rel_to=__file__)
self.assertRaises((ParseError, UnexpectedInput),
l.parse, 'xa abby')
def test_relative_rule_import_subrule(self):
l = _Lark_open('test_relative_rule_import_subrule.lark',
rel_to=__file__)
x = l.parse('xaabby')
self.assertEqual(x.children, [
'x',
Tree('startab', [
Tree('grammars__ab__expr', [
'a', Tree('grammars__ab__expr', ['a', 'b']), 'b',
]),
]),
'y'])
def test_relative_rule_import_subrule_no_conflict(self):
l = _Lark_open(
'test_relative_rule_import_subrule_no_conflict.lark',
rel_to=__file__)
x = l.parse('xaby')
self.assertEqual(x.children, [Tree('expr', [
'x',
Tree('startab', [
Tree('grammars__ab__expr', ['a', 'b']),
]),
'y'])])
self.assertRaises((ParseError, UnexpectedInput),
l.parse, 'xaxabyby')
def test_relative_rule_import_rename(self):
l = _Lark_open('test_relative_rule_import_rename.lark',
rel_to=__file__)
x = l.parse('xaabby')
self.assertEqual(x.children, [
'x',
Tree('ab', ['a', Tree('ab', ['a', 'b']), 'b']),
'y'])
def test_multi_import(self):
grammar = """
start: NUMBER WORD
%import common (NUMBER, WORD, WS)
%ignore WS
"""
l = _Lark(grammar)
x = l.parse('12 toucans')
self.assertEqual(x.children, ['12', 'toucans'])
def test_relative_multi_import(self):
l = _Lark_open("test_relative_multi_import.lark", rel_to=__file__)
x = l.parse('12 capybaras')
self.assertEqual(x.children, ['12', 'capybaras'])
def test_relative_import_preserves_leading_underscore(self):
l = _Lark_open("test_relative_import_preserves_leading_underscore.lark", rel_to=__file__)
x = l.parse('Ax')
self.assertEqual(next(x.find_data('c')).children, ['A'])
def test_relative_import_of_nested_grammar(self):
l = _Lark_open("grammars/test_relative_import_of_nested_grammar.lark", rel_to=__file__)
x = l.parse('N')
self.assertEqual(next(x.find_data('rule_to_import')).children, ['N'])
def test_relative_import_rules_dependencies_imported_only_once(self):
l = _Lark_open("test_relative_import_rules_dependencies_imported_only_once.lark", rel_to=__file__)
x = l.parse('AAA')
self.assertEqual(next(x.find_data('a')).children, ['A'])
self.assertEqual(next(x.find_data('b')).children, ['A'])
self.assertEqual(next(x.find_data('d')).children, ['A'])
def test_import_errors(self):
grammar = """
start: NUMBER WORD
%import .grammars.bad_test.NUMBER
"""
self.assertRaises(IOError, _Lark, grammar)
grammar = """
start: NUMBER WORD
%import bad_test.NUMBER
"""
self.assertRaises(IOError, _Lark, grammar)
@unittest.skipIf(PARSER != 'earley', "Currently only Earley supports priority in rules")
def test_earley_prioritization(self):
"Tests effect of priority on result"
grammar = """
start: a | b
a.1: "a"
b.2: "a"
"""
# l = Lark(grammar, parser='earley', lexer='standard')
l = _Lark(grammar)
res = l.parse("a")
self.assertEqual(res.children[0].data, 'b')
grammar = """
start: a | b
a.2: "a"
b.1: "a"
"""
l = _Lark(grammar)
# l = Lark(grammar, parser='earley', lexer='standard')
res = l.parse("a")
self.assertEqual(res.children[0].data, 'a')
@unittest.skipIf(PARSER != 'earley', "Currently only Earley supports priority in rules")
def test_earley_prioritization_sum(self):
"Tests effect of priority on result"
grammar = """
start: ab_ b_ a_ | indirection
indirection: a_ bb_ a_
a_: "a"
b_: "b"
ab_: "ab"
bb_.1: "bb"
"""
l = Lark(grammar, priority="invert")
res = l.parse('abba')
self.assertEqual(''.join(child.data for child in res.children), 'ab_b_a_')
grammar = """
start: ab_ b_ a_ | indirection
indirection: a_ bb_ a_
a_: "a"
b_: "b"
ab_.1: "ab"
bb_: "bb"
"""
l = Lark(grammar, priority="invert")
res = l.parse('abba')
self.assertEqual(''.join(child.data for child in res.children), 'indirection')
grammar = """
start: ab_ b_ a_ | indirection
indirection: a_ bb_ a_
a_.2: "a"
b_.1: "b"
ab_.3: "ab"
bb_.3: "bb"
"""
l = Lark(grammar, priority="invert")
res = l.parse('abba')
self.assertEqual(''.join(child.data for child in res.children), 'ab_b_a_')
grammar = """
start: ab_ b_ a_ | indirection
indirection: a_ bb_ a_
a_.1: "a"
b_.1: "b"
ab_.4: "ab"
bb_.3: "bb"
"""
l = Lark(grammar, priority="invert")
res = l.parse('abba')
self.assertEqual(''.join(child.data for child in res.children), 'indirection')
def test_utf8(self):
g = u"""start: a
a: "±a"
"""
l = _Lark(g)
self.assertEqual(l.parse(u'±a'), Tree('start', [Tree('a', [])]))
g = u"""start: A
A: "±a"
"""
l = _Lark(g)
self.assertEqual(l.parse(u'±a'), Tree('start', [u'\xb1a']))
@unittest.skipIf(PARSER == 'cyk', "No empty rules")
def test_ignore(self):
grammar = r"""
COMMENT: /(!|(\/\/))[^\n]*/
%ignore COMMENT
%import common.WS -> _WS
%import common.INT
start: "INT"i _WS+ INT _WS*
"""
parser = _Lark(grammar)
tree = parser.parse("int 1 ! This is a comment\n")
self.assertEqual(tree.children, ['1'])
tree = parser.parse("int 1 ! This is a comment") # A trailing ignore token can be tricky!
self.assertEqual(tree.children, ['1'])
parser = _Lark(r"""
start : "a"*
%ignore "b"
""")
tree = parser.parse("bb")
self.assertEqual(tree.children, [])
def test_regex_escaping(self):
g = _Lark("start: /[ab]/")
g.parse('a')
g.parse('b')
self.assertRaises( UnexpectedInput, g.parse, 'c')
_Lark(r'start: /\w/').parse('a')
g = _Lark(r'start: /\\w/')
self.assertRaises( UnexpectedInput, g.parse, 'a')
g.parse(r'\w')
_Lark(r'start: /\[/').parse('[')
_Lark(r'start: /\//').parse('/')
_Lark(r'start: /\\/').parse('\\')
_Lark(r'start: /\[ab]/').parse('[ab]')
_Lark(r'start: /\\[ab]/').parse('\\a')
_Lark(r'start: /\t/').parse('\t')
_Lark(r'start: /\\t/').parse('\\t')
_Lark(r'start: /\\\t/').parse('\\\t')
_Lark(r'start: "\t"').parse('\t')
_Lark(r'start: "\\t"').parse('\\t')
_Lark(r'start: "\\\t"').parse('\\\t')
def test_ranged_repeat_rules(self):
g = u"""!start: "A"~3
"""
l = _Lark(g)
self.assertEqual(l.parse(u'AAA'), Tree('start', ["A", "A", "A"]))
self.assertRaises(ParseError, l.parse, u'AA')
self.assertRaises((ParseError, UnexpectedInput), l.parse, u'AAAA')
g = u"""!start: "A"~0..2
"""
if PARSER != 'cyk': # XXX CYK currently doesn't support empty grammars
l = _Lark(g)
self.assertEqual(l.parse(u''), Tree('start', []))
self.assertEqual(l.parse(u'A'), Tree('start', ['A']))
self.assertEqual(l.parse(u'AA'), Tree('start', ['A', 'A']))
self.assertRaises((UnexpectedToken, UnexpectedInput), l.parse, u'AAA')
g = u"""!start: "A"~3..2
"""
self.assertRaises(GrammarError, _Lark, g)
g = u"""!start: "A"~2..3 "B"~2
"""
l = _Lark(g)
self.assertEqual(l.parse(u'AABB'), Tree('start', ['A', 'A', 'B', 'B']))
self.assertEqual(l.parse(u'AAABB'), Tree('start', ['A', 'A', 'A', 'B', 'B']))
self.assertRaises(ParseError, l.parse, u'AAAB')
self.assertRaises((ParseError, UnexpectedInput), l.parse, u'AAABBB')
self.assertRaises((ParseError, UnexpectedInput), l.parse, u'ABB')
self.assertRaises((ParseError, UnexpectedInput), l.parse, u'AAAABB')
def test_ranged_repeat_terms(self):
g = u"""!start: AAA
AAA: "A"~3
"""
l = _Lark(g)
self.assertEqual(l.parse(u'AAA'), Tree('start', ["AAA"]))
self.assertRaises((ParseError, UnexpectedInput), l.parse, u'AA')
self.assertRaises((ParseError, UnexpectedInput), l.parse, u'AAAA')
g = u"""!start: AABB CC
AABB: "A"~0..2 "B"~2
CC: "C"~1..2
"""
l = _Lark(g)
self.assertEqual(l.parse(u'AABBCC'), Tree('start', ['AABB', 'CC']))
self.assertEqual(l.parse(u'BBC'), Tree('start', ['BB', 'C']))
self.assertEqual(l.parse(u'ABBCC'), Tree('start', ['ABB', 'CC']))
self.assertRaises((ParseError, UnexpectedInput), l.parse, u'AAAB')
self.assertRaises((ParseError, UnexpectedInput), l.parse, u'AAABBB')
self.assertRaises((ParseError, UnexpectedInput), l.parse, u'ABB')
self.assertRaises((ParseError, UnexpectedInput), l.parse, u'AAAABB')
@unittest.skipIf(PARSER=='earley', "Priority not handled correctly right now") # TODO XXX
def test_priority_vs_embedded(self):
g = """
A.2: "a"
WORD: ("a".."z")+
start: (A | WORD)+
"""
l = _Lark(g)
t = l.parse('abc')
self.assertEqual(t.children, ['a', 'bc'])
self.assertEqual(t.children[0].type, 'A')
def test_line_counting(self):
p = _Lark("start: /[^x]+/")
text = 'hello\nworld'
t = p.parse(text)
tok = t.children[0]
self.assertEqual(tok, text)
self.assertEqual(tok.line, 1)
self.assertEqual(tok.column, 1)
if _LEXER != 'dynamic':
self.assertEqual(tok.end_line, 2)
self.assertEqual(tok.end_column, 6)
@unittest.skipIf(PARSER=='cyk', "Empty rules")
def test_empty_end(self):
p = _Lark("""
start: b c d
b: "B"
c: | "C"
d: | "D"
""")
res = p.parse('B')
self.assertEqual(len(res.children), 3)
@unittest.skipIf(PARSER=='cyk', "Empty rules")
def test_maybe_placeholders(self):
# Anonymous tokens shouldn't count
p = _Lark("""start: ["a"] ["b"] ["c"] """, maybe_placeholders=True)
self.assertEqual(p.parse("").children, [])
# All invisible constructs shouldn't count
p = _Lark("""start: [A] ["b"] [_c] ["e" "f" _c]
A: "a"
_c: "c" """, maybe_placeholders=True)
self.assertEqual(p.parse("").children, [None])
self.assertEqual(p.parse("c").children, [None])
self.assertEqual(p.parse("aefc").children, ['a'])
# ? shouldn't apply
p = _Lark("""!start: ["a"] "b"? ["c"] """, maybe_placeholders=True)
self.assertEqual(p.parse("").children, [None, None])
self.assertEqual(p.parse("b").children, [None, 'b', None])
p = _Lark("""!start: ["a"] ["b"] ["c"] """, maybe_placeholders=True)
self.assertEqual(p.parse("").children, [None, None, None])
self.assertEqual(p.parse("a").children, ['a', None, None])
self.assertEqual(p.parse("b").children, [None, 'b', None])
self.assertEqual(p.parse("c").children, [None, None, 'c'])
self.assertEqual(p.parse("ab").children, ['a', 'b', None])
self.assertEqual(p.parse("ac").children, ['a', None, 'c'])
self.assertEqual(p.parse("bc").children, [None, 'b', 'c'])
self.assertEqual(p.parse("abc").children, ['a', 'b', 'c'])
p = _Lark("""!start: (["a"] "b" ["c"])+ """, maybe_placeholders=True)
self.assertEqual(p.parse("b").children, [None, 'b', None])
self.assertEqual(p.parse("bb").children, [None, 'b', None, None, 'b', None])
self.assertEqual(p.parse("abbc").children, ['a', 'b', None, None, 'b', 'c'])
self.assertEqual(p.parse("babbcabcb").children,
[None, 'b', None,
'a', 'b', None,
None, 'b', 'c',
'a', 'b', 'c',
None, 'b', None])
p = _Lark("""!start: ["a"] ["c"] "b"+ ["a"] ["d"] """, maybe_placeholders=True)
self.assertEqual(p.parse("bb").children, [None, None, 'b', 'b', None, None])
self.assertEqual(p.parse("bd").children, [None, None, 'b', None, 'd'])
self.assertEqual(p.parse("abba").children, ['a', None, 'b', 'b', 'a', None])
self.assertEqual(p.parse("cbbbb").children, [None, 'c', 'b', 'b', 'b', 'b', None, None])
def test_escaped_string(self):
"Tests common.ESCAPED_STRING"
grammar = r"""
start: ESCAPED_STRING+
%import common (WS_INLINE, ESCAPED_STRING)
%ignore WS_INLINE
"""
parser = _Lark(grammar)
parser.parse(r'"\\" "b" "c"')
parser.parse(r'"That" "And a \"b"')
def test_meddling_unused(self):
"Unless 'unused' is removed, LALR analysis will fail on reduce-reduce collision"
grammar = """
start: EKS* x
x: EKS
unused: x*
EKS: "x"
"""
parser = _Lark(grammar)
@unittest.skipIf(PARSER!='lalr' or LEXER=='custom', "Serialize currently only works for LALR parsers without custom lexers (though it should be easy to extend)")
def test_serialize(self):
grammar = """
start: _ANY b "C"
_ANY: /./
b: "B"
"""
parser = _Lark(grammar)
d = parser.serialize()
parser2 = Lark.deserialize(d, {}, {})
self.assertEqual(parser2.parse('ABC'), Tree('start', [Tree('b', [])]) )
namespace = {'Rule': Rule, 'TerminalDef': TerminalDef}
d, m = parser.memo_serialize(namespace.values())
parser3 = Lark.deserialize(d, namespace, m)
self.assertEqual(parser3.parse('ABC'), Tree('start', [Tree('b', [])]) )
def test_multi_start(self):
parser = _Lark('''
a: "x" "a"?
b: "x" "b"?
''', start=['a', 'b'])
self.assertEqual(parser.parse('xa', 'a'), Tree('a', []))
self.assertEqual(parser.parse('xb', 'b'), Tree('b', []))
def test_lexer_detect_newline_tokens(self):
# Detect newlines in regular tokens
g = _Lark(r"""start: "go" tail*
!tail : SA "@" | SB "@" | SC "@" | SD "@"
SA : "a" /\n/
SB : /b./s
SC : "c" /[^a-z]/
SD : "d" /\s/
""")
a,b,c,d = [x.children[1] for x in g.parse('goa\n@b\n@c\n@d\n@').children]
self.assertEqual(a.line, 2)
self.assertEqual(b.line, 3)
self.assertEqual(c.line, 4)
self.assertEqual(d.line, 5)
# Detect newlines in ignored tokens
for re in ['/\\n/', '/[^a-z]/', '/\\s/']:
g = _Lark('''!start: "a" "a"
%ignore {}'''.format(re))
a, b = g.parse('a\na').children
self.assertEqual(a.line, 1)
self.assertEqual(b.line, 2)
_NAME = "Test" + PARSER.capitalize() + LEXER.capitalize()
_TestParser.__name__ = _NAME
globals()[_NAME] = _TestParser
# Note: You still have to import them in __main__ for the tests to run
_TO_TEST = [
('standard', 'earley'),
('standard', 'cyk'),
('dynamic', 'earley'),
('dynamic_complete', 'earley'),
('standard', 'lalr'),
('contextual', 'lalr'),
('custom', 'lalr'),
# (None, 'earley'),
]
for _LEXER, _PARSER in _TO_TEST:
_make_parser_test(_LEXER, _PARSER)
for _LEXER in ('dynamic', 'dynamic_complete'):
_make_full_earley_test(_LEXER)
if __name__ == '__main__':
unittest.main()