def test_trace(self):
buffer = StringIO()
with TraceVariables(out=buffer):
word = ~Lookahead('OR') & Word()
phrase = String()
with DroppedSpace():
text = (phrase | word)[1:] > list
query = text[:, Drop('OR')]
expected = ''' phrase failed stream = 'spicy meatballs OR...
word = ['spicy'] stream = ' meatballs OR "el ...
phrase failed stream = 'meatballs OR "el b...
word = ['meatballs'] stream = ' OR "el bulli rest...
phrase failed stream = 'OR "el bulli resta...
word failed stream = 'OR "el bulli resta...
phrase failed stream = ' OR "el bulli rest...
word failed stream = ' OR "el bulli rest...
text = [['spicy', 'meatballs']] stream = ' OR "el bulli rest...
phrase = ['el bulli restaurant'] stream = ''
phrase failed stream = ''
word failed stream = ''
text = [['el bulli restaurant']] stream = ''
query = [['spicy', 'meatballs'], ['el... stream = ''
'''
query.config.auto_memoize(full=True)
query.parse('spicy meatballs OR "el bulli restaurant"')
trace = buffer.getvalue()
assert trace == expected, '"""' + trace + '"""'
# check caching works
query.parse('spicy meatballs OR "el bulli restaurant"')
trace = buffer.getvalue()
assert trace == expected, '"""' + trace + '"""'
def __init__(self, clean_html=True):
self.clean_html = clean_html
self._punctuation = '!"#&\'()*+,.;<=>?@[\\]^_`{|}~'
self._lctx_1_exceptions = set('/ :'.split())
self._lctx_2_exceptions = set('discount redeem voucher'.split())
self._rctx_1_exceptions = set('/ : th am pm hour hours %'.split())
self._rctx_2_exceptions = set('discount redeem voucher'.split())
# LEPL Real Number Matchers (w/thousands)
_comma_three_digits = Join(Drop(','), Add(Digit()[3]))[:]
_thousand_group = Or(
Join(_comma_three_digits, Any('.'), UnsignedInteger()),
Join(_comma_three_digits, Optional(Any('.'))))
_real = Or(Join(UnsignedInteger(), _thousand_group),
UnsignedReal()) >> float
_any = Join(Star(AnyBut(_real)))
self._real_partition_matcher = Star(And(_any, _real, _any))
self._real_simple_matcher = _real[:,
Drop(
Star(Or(Whitespace(), Any(',-')))
)]
def test_node(self):
class Term(Node): pass
number = Any('1') > 'number'
term = number > Term
factor = term | Drop(Optional(term))
factor.config.clear().compose_transforms()
p = factor.get_parse_string()
ast = p('1')[0]
assert type(ast) == Term, type(ast)
assert ast[0] == '1', ast[0]
assert str26(ast) == """Term
`- number '1'""", ast
def mkhex(char, n):
from lepl.matchers.derived import Drop
return Drop(Any(char)) + Any('0123456789abcdefABCDEF')[n,...] >> \
(lambda x: chr(int(x, 16)))
def __init__(self, space=None):
from lepl.matchers.derived import Space, Drop
if space is None:
space = Space()[:]
space = Drop(space)
super(DroppedSpace, self).__init__(space)
def test_not(self):
matcher = PostMatch(Drop(Any()[:]) & Any(), r'[0-9]', not_=True)
matcher.config.no_full_first_match()
results = list(matcher.parse_all('12a'))
assert results == [['a']], results
def test_normal(self):
matcher = PostMatch(Drop(Any()[:]) & Any(), r'[0-9]')
matcher.config.no_full_first_match()
results = list(matcher.parse_all('12a'))
assert results == [['2'], ['1']], results
def build(self):
'''
Construct the parser.
'''
# Avoid dependency loops
from lepl.matchers.derived import Drop, Eos, AnyBut, Upper
from lepl.matchers.core import Any, Lookahead, Literal, Delayed
from lepl.matchers.error import make_error
from lepl.matchers.variables import TraceVariables
from lepl.support.node import node_throw
with TraceVariables(False):
# these two definitions enforce the conditions above, providing only
# special characters appear as literals in the grammar
escaped = Drop(self.alphabet.escape) & self.alphabet.escaped
raw = ~Lookahead(self.alphabet.escape) & \
AnyBut(self.alphabet.illegal)
close = Drop(')')
extend = (Drop('(*') & Upper()[1:, ...] & close) >> self.extend
single = escaped | raw | extend
any_ = Literal('.') >> self.dot
letter = single >> self.dup
pair = single & Drop('-') & single > self.tup
interval = pair | letter
brackets = Drop('[') & interval[1:] & Drop(']')
inverted = Drop('[^') & interval[1:] & Drop(']') >= self.invert
char = inverted | brackets | letter | any_ | extend > self.char
item = Delayed()
open = Drop('(?:')
range = Drop(self.alphabet.escape) & self.alphabet.range
seq = (char | item | range)[0:] > self.sequence
group = open & seq & close
alts = open & seq[2:, Drop('|')] & close > self.choice
star = (alts | group | char) & Drop('*') > self.star
plus = (alts | group | char) & Drop('+') > self.plus
opt = (alts | group | char) & Drop('?') > self.option
bad_grp = (Drop('(') & ~Lookahead('?:') & seq & close) \
** make_error(
"Lepl's own regular expressions do not currently "
"support matched groups.\n"
"Use '(?:...)' to group expressions without "
"matching.")
item += alts | group | star | plus | opt | bad_grp
expr = ((char | item)[:] & Drop(Eos())) >> node_throw
# Empty config here avoids loops if the default config includes
# references to alphabets
expr.config.clear()
return expr.parse_string
def build(self):
'''
Construct the parser.
'''
# Avoid dependency loops
from lepl.matchers.derived import Drop, Eos, AnyBut, Upper
from lepl.matchers.core import Any, Lookahead, Literal, Delayed
# these two definitions enforce the conditions above, providing only
# special characters appear as literals in the grammar
escaped = Drop(self.alphabet.escape) + Any(self.alphabet.escaped)
raw = ~Lookahead(self.alphabet.escape) + \
AnyBut(self.alphabet.escaped)
single = escaped | raw
any_ = Literal('.') >> self.dot
letter = single >> self.dup
pair = single & Drop('-') & single > self.tup
extend = (Drop('(*') & Upper()[1:, ...] & Drop(')')) >> self.extend
interval = pair | letter | extend
brackets = Drop('[') & interval[1:] & Drop(']')
inverted = Drop('[^') & interval[1:] & Drop(']') >= self.invert
char = inverted | brackets | letter | any_ | extend > self.char
item = Delayed()
seq = (char | item)[0:] > self.sequence
group = Drop('(') & seq & Drop(')')
alts = Drop('(') & seq[2:, Drop('|')] & Drop(')') > self.choice
star = (alts | group | char) & Drop('*') > self.star
plus = (alts | group | char) & Drop('+') > self.plus
opt = (alts | group | char) & Drop('?') > self.option
item += alts | group | star | plus | opt
expr = (char | item)[:] & Drop(Eos())
# Empty config here avoids loops if the default config includes
# references to alphabets
expr.config.clear()
return expr.parse_string