def post_process(x):
"Converts list of lists set into nicely formatted `Tree` objects."
if not isinstance(x, tuple):
return x
[body, label] = x
if isinstance(body, tuple):
return ImmutableTree(label, list(map(post_process, body)))
else:
return ImmutableTree(label, [body])
def derivation(x):
"""Post-process `Viterbi` and `Point` derivations into nicely formatted `Tree`
objects.
"""
if not isinstance(x.d, (list, tuple)):
return x.d
assert len(x.d) == 2
label = x.d[1].d
body = x.d[0].d
if isinstance(body, (list, tuple)):
return ImmutableTree(label, list(map(derivation, body)))
else:
return ImmutableTree(label, [body])
def main():
intree = ImmutableTree("(foo bar)")
outtree = ImmutableTree("(bar foo)")
mutable = Tree("(foo bar)")
imm = immutable(mutable)
print(imm)
rules = loadrules("testrules.yaml")
setofrules = set()
for rule in rules:
setofrules.add(rule)
print(setofrules)
setofrules = frozenset(setofrules)
## XXX(alexr): do we need to make rules be sets?
## ah geez. let's just use frozenrules.
print(produce(intree, outtree, setofrules, "q0", (0), (1)))
def test_grammar():
grammar = RuleSet()
grammar.add(BinarizedRule("S", "NP", "VP", 0, 0, (0, 0)))
grammar.add_unary(UnaryRule("NP", "N"))
grammar.add_unary(UnaryRule("VP", "V"))
grammar.finish()
tree = Tree.fromstring("(S^2 (NP^1 (N^1 dog^1)) (VP^2 (V^2 flies^2)))")
encoder = LexicalizedCFGEncoder(["dog", "flies"], ["N", "V"], grammar)
parts = encoder.transform_structure(tree)
print parts
tree2 = encoder.from_parts(parts)
print tree
print tree2
assert (tree == tree2)
def main(): """ a basic REPL for testing """ corpus = """(S (NP John) (VP (V likes) (NP Mary))) (S (NP Peter) (VP (V hates) (NP Susan))) (S (NP Harry) (VP (V eats) (NP pizza))) (S (NP Hermione) (VP (V eats)))""".splitlines() corpus = """(S (NP (DT The) (NN cat)) (VP (VBP saw) (NP (DT the) (JJ hungry) (NN dog)))) (S (NP (DT The) (JJ little) (NN mouse)) (VP (VBP ate) (NP (DT the) (NN cat))))""".splitlines() #corpus = """(S (NP mary) (VP walks) (AP quickly))""".splitlines() #(S (NP Harry) (VP (V likes) (NP Susan) (ADVP (RB very) (RB much)))) corpus = [Tree(a) for a in corpus] #d = GoodmanDOP(corpus, rootsymbol='S') from bitpar import BitParChartParser d = GoodmanDOP(corpus, rootsymbol='TOP', wrap='TOP', parser=BitParChartParser) #d = GoodmanDOP(corpus, rootsymbol='TOP', wrap='TOP') #print d.grammar print "corpus" for a in corpus: print a w = "foo!" while w: print "sentence:", w = raw_input().split() try: p = FreqDist() for n, a in enumerate(d.parser.nbest_parse(w)): if n > 1000: break print a p.inc(ImmutableTree.convert(removeids(a)), a.prob()) #for b, a in sorted((b,a) for (a,b) in p.items()): # print a, b print print 'best', p.max(), p[p.max()] #print d.parse(w) except Exception: # as e: print "error", #e
def immutable(tr):
"""Given a Tree, make it an ImmutableTree."""
str = tr.pprint(margin=10000)
return ImmutableTree(str)
def test_PreterminalJapaneseTypedVarImmutableTree(self):
tree = ImmutableTree.fromstring(u'(学生 hello)')
subtree = tree_or_string(u'?x0|学生')
self.assertTrue(TreeContains(tree, subtree))