def test_inp_cfg():
"""
Simple example:
arg1: CFG that only has two important nonterminals: WORD and SPACE
arg2: "WORD", meaning "return all parts of input that are WORD tokens"
"""
# -> space is included if it's part of regex ?
# here, there is NO SPACE after the colon
# however, each symbol must be separated by a space
# a space in your regex means there must be two spaces ?
in_cfg = ("\G :\WORD \SPACE \G | \\0\n" "\WORD :\w+\n" "\SPACE : ")
lines = [line.strip() for line in in_cfg.split('\n') if line.strip()]
mappings = Symbol.create_mappings(lines)
def test_cfg_manual():
"""
test:
- create mapping from cfg expr
- create aggregate Symbol with arithmetic and parse
"""
in_cfg = ("\G :\WORD \SPACE \G | \\0\n" "\WORD :\w+\n" "\SPACE : ")
lines = [line for line in in_cfg.split('\n') if line.strip()]
mappings = Symbol.create_mappings(lines)
sw = Symbol(mappings['WORD'])
ss = Symbol(mappings['SPACE'])
g = sw + ss
g.rrec()
#g = g | Symbol('')
st = "first second third fourth fifth"
g.parse(st)