def test_roots_share_suffix(self):
graphs = Pair(AlpinoGraph(), AlpinoGraph())
graphs.source.add_node(1, "x")
graphs.target.add_node(2, "y")
nodes = Pair(1, 2)
# no roots
self.assertEqual(ff_roots_share_suffix(nodes, graphs), "-")
graphs.source.node[1]["root"] = "woon_wagen"
graphs.target.node[2]["root"] = "eet_tafel"
self.assertEqual(ff_roots_share_suffix(nodes, graphs), "F")
graphs.source.node[1]["root"] = "woon_wagen"
graphs.target.node[2]["root"] = "woon_wagen"
self.assertEqual(ff_roots_share_suffix(nodes, graphs), "F")
graphs.source.node[1]["root"] = "woon_wagen"
graphs.target.node[2]["root"] = "wagen"
self.assertEqual(ff_roots_share_suffix(nodes, graphs), "F")
graphs.source.node[1]["root"] = "woon_wagen_bewoner"
graphs.target.node[2]["root"] = "wagen_bewoner"
self.assertEqual(ff_roots_share_suffix(nodes, graphs), "F")
graphs.source.node[1]["root"] = "woon_wagen"
graphs.target.node[2]["root"] = "mest_wagen"
self.assertEqual(ff_roots_share_suffix(nodes, graphs), "T")
graphs.source.node[1]["root"] = "woon_wagen_trekker"
graphs.target.node[2]["root"] = "mest_wagen_trekker"
self.assertEqual(ff_roots_share_suffix(nodes, graphs), "T")
def test_roots_subsumption(self):
graphs = Pair(AlpinoGraph(), AlpinoGraph())
graphs.source.add_node(1, "x")
graphs.target.add_node(2, "y")
nodes = Pair(1, 2)
# no roots
self.assertEqual(ff_roots_subsumption(nodes, graphs), "-")
graphs.source.node[1]["root"] = "wagen"
graphs.target.node[2]["root"] = "wagen"
self.assertEqual(ff_roots_subsumption(nodes, graphs), "equals")
graphs.source.node[1]["root"] = "brandweer_wagen"
graphs.target.node[2]["root"] = "brandweer"
self.assertEqual(ff_roots_subsumption(nodes, graphs), "has_prefix")
graphs.source.node[1]["root"] = "brandweer"
graphs.target.node[2]["root"] = "brandweer_wagen"
self.assertEqual(ff_roots_subsumption(nodes, graphs), "is_prefix")
graphs.source.node[1]["root"] = "brandweer_wagen"
graphs.target.node[2]["root"] = "wagen"
self.assertEqual(ff_roots_subsumption(nodes, graphs), "has_suffix")
graphs.source.node[1]["root"] = "wagen"
graphs.target.node[2]["root"] = "brandweer_wagen"
self.assertEqual(ff_roots_subsumption(nodes, graphs), "is_suffix")
graphs.source.node[1]["root"] = "woon_wagen_bewoners_kamp_ingang"
graphs.target.node[2]["root"] = "wagen_bewoners"
self.assertEqual(ff_roots_subsumption(nodes, graphs), "has_infix")
graphs.source.node[1]["root"] = "wagen_bewoners"
graphs.target.node[2]["root"] = "woon_wagen_bewoners_kamp_ingang"
self.assertEqual(ff_roots_subsumption(nodes, graphs), "is_infix")
# no subsumption
graphs.source.node[1]["root"] = "brandweer_wagen"
graphs.target.node[2]["root"] = "kamp_ingang"
self.assertEqual(ff_roots_subsumption(nodes, graphs), "none")
def setUp(self):
self.ag = AlpinoGraph(root="0")
self.ag.add_node("0", "top", cat="top",
tokens="Ik wilde weten of hij echt begrip had .".split())
self.ag.add_node("1", "smain", cat="smain",
tokens="Ik wilde weten of hij echt begrip had".split())
self.ag.add_node("2", "pron", pos="pron", root="ik", index="1",
tokens="Ik".split())
self.ag.add_node("3", "verb", pos="verb", root="willen",
tokens="wilde".split())
self.ag.add_node("4", "inf", cat="inf",
tokens="weten of hij echt begrip had".split())
self.ag.add_node("5", "index", index="1")
self.ag.add_node("6", "verb", pos="verb", root="weten",
tokens="weten".split())
self.ag.add_node("7", "cp", cat="cp",
tokens="of hij echt begrip had".split())
self.ag.add_node("8", "comp", pos="comp", root="of",
tokens="of".split())
self.ag.add_node("9", "ssub", cat="ssub",
tokens="hij echt begrip had".split())
self.ag.add_node("10", "pron", pos="pron", root="hij",
tokens="hij".split())
self.ag.add_node("11", "np", cat="np",
tokens="echt begrip".split())
self.ag.add_node("12", "adj", pos="adj", root="echt",
tokens="echt".split())
self.ag.add_node("13", "noun", pos="noun", root="begrip",
tokens="begrip".split())
self.ag.add_node("14", "verb", pos="verb", root="hebben",
tokens="had".split())
self.ag.add_node("15", "punt", pos="punct", root=".",
tokens=".".split())
self.ag.add_edge("0", "1", "--")
self.ag.add_edge("1", "2", "su")
self.ag.add_edge("1", "3", "hd")
self.ag.add_edge("1", "4", "vc")
self.ag.add_edge("4", "5", "su")
self.ag.add_edge("4", "6", "hd")
self.ag.add_edge("4", "7", "vc")
self.ag.add_edge("7", "8", "cmp")
self.ag.add_edge("7", "9", "body")
self.ag.add_edge("9", "10", "su")
self.ag.add_edge("9", "11", "obj1")
self.ag.add_edge("11", "12", "mod")
self.ag.add_edge("11", "13", "hd")
self.ag.add_edge("9", "14", "hd")
self.ag.add_edge("0", "15", "punct")
def _start_graph(self, attrs):
# to enable parsing graphs without id attribute,
# id defaults to graph number
id = attrs.get("id", self._n)
self._n += 1
if not self._sparse or id in self._id2graph:
self._graph = self._id2graph[id] = AlpinoGraph(id=id)
# stack that records ancestor nodes,
# because we need a node's parent to add a new edge
self._ancestors = []
# list of tokens in the order that they are encountered in the tree
# (which is often different from the order in the sentence)
self._tokens = []
# tracks the current position in self._tokens
self._index = 0
# stack which records self._index upon entering a non-terminal node,
# which is required to calculate the tokens belonging to non-terminal
# when exiting a non-terminal
self._index_stack = []
class Test_AlpinoGraph(unittest.TestCase):
def setUp(self):
self.ag = AlpinoGraph(root="0")
self.ag.add_node(
"0",
"top",
cat="top",
tokens="Ik wilde weten of hij echt begrip had .".split())
self.ag.add_node(
"1",
"smain",
cat="smain",
tokens="Ik wilde weten of hij echt begrip had".split())
self.ag.add_node("2",
"pron",
pos="pron",
root="ik",
index="1",
tokens="Ik".split())
self.ag.add_node("3",
"verb",
pos="verb",
root="willen",
tokens="wilde".split())
self.ag.add_node("4",
"inf",
cat="inf",
tokens="weten of hij echt begrip had".split())
self.ag.add_node("5", "index", index="1")
self.ag.add_node("6",
"verb",
pos="verb",
root="weten",
tokens="weten".split())
self.ag.add_node("7",
"cp",
cat="cp",
tokens="of hij echt begrip had".split())
self.ag.add_node("8",
"comp",
pos="comp",
root="of",
tokens="of".split())
self.ag.add_node("9",
"ssub",
cat="ssub",
tokens="hij echt begrip had".split())
self.ag.add_node("10",
"pron",
pos="pron",
root="hij",
tokens="hij".split())
self.ag.add_node("11", "np", cat="np", tokens="echt begrip".split())
self.ag.add_node("12",
"adj",
pos="adj",
root="echt",
tokens="echt".split())
self.ag.add_node("13",
"noun",
pos="noun",
root="begrip",
tokens="begrip".split())
self.ag.add_node("14",
"verb",
pos="verb",
root="hebben",
tokens="had".split())
self.ag.add_node("15",
"punt",
pos="punct",
root=".",
tokens=".".split())
self.ag.add_edge("0", "1", "--")
self.ag.add_edge("1", "2", "su")
self.ag.add_edge("1", "3", "hd")
self.ag.add_edge("1", "4", "vc")
self.ag.add_edge("4", "5", "su")
self.ag.add_edge("4", "6", "hd")
self.ag.add_edge("4", "7", "vc")
self.ag.add_edge("7", "8", "cmp")
self.ag.add_edge("7", "9", "body")
self.ag.add_edge("9", "10", "su")
self.ag.add_edge("9", "11", "obj1")
self.ag.add_edge("11", "12", "mod")
self.ag.add_edge("11", "13", "hd")
self.ag.add_edge("9", "14", "hd")
self.ag.add_edge("0", "15", "punct")
def test_print_subtree(self):
print "\n", self.ag
def test_node_is_nominal(self):
self.assertTrue(self.ag.node_is_nominal("13"))
self.assertFalse(self.ag.node_is_nominal("3"))
def test_node_is_punct(self):
self.assertTrue(self.ag.node_is_punct("15"))
self.assertFalse(self.ag.node_is_punct("14"))
def test_node_is_index(self):
self.assertTrue(self.ag.node_is_index("5"))
self.assertFalse(self.ag.node_is_index("1"))
self.assertFalse(self.ag.node_is_index("2"))
def test_get_root_node(self):
self.assertEqual(self.ag.root, "0")
def test_get_parent_node(self):
self.assertEqual(self.ag.get_parent_node("0"), None)
self.assertEqual(self.ag.get_parent_node("1"), "0")
def test_get_node_deprel(self):
self.assertEqual(self.ag.get_node_deprel("0"), None)
self.assertEqual(self.ag.get_node_deprel("15"), "punct")
def test_node_is_complement(self):
self.assertTrue(self.ag.node_is_complement("11"))
self.assertFalse(self.ag.node_is_complement("12"))
def setUp(self):
self.ag = AlpinoGraph(root="0")
self.ag.add_node(
"0",
"top",
cat="top",
tokens="Ik wilde weten of hij echt begrip had .".split())
self.ag.add_node(
"1",
"smain",
cat="smain",
tokens="Ik wilde weten of hij echt begrip had".split())
self.ag.add_node("2",
"pron",
pos="pron",
root="ik",
index="1",
tokens="Ik".split())
self.ag.add_node("3",
"verb",
pos="verb",
root="willen",
tokens="wilde".split())
self.ag.add_node("4",
"inf",
cat="inf",
tokens="weten of hij echt begrip had".split())
self.ag.add_node("5", "index", index="1")
self.ag.add_node("6",
"verb",
pos="verb",
root="weten",
tokens="weten".split())
self.ag.add_node("7",
"cp",
cat="cp",
tokens="of hij echt begrip had".split())
self.ag.add_node("8",
"comp",
pos="comp",
root="of",
tokens="of".split())
self.ag.add_node("9",
"ssub",
cat="ssub",
tokens="hij echt begrip had".split())
self.ag.add_node("10",
"pron",
pos="pron",
root="hij",
tokens="hij".split())
self.ag.add_node("11", "np", cat="np", tokens="echt begrip".split())
self.ag.add_node("12",
"adj",
pos="adj",
root="echt",
tokens="echt".split())
self.ag.add_node("13",
"noun",
pos="noun",
root="begrip",
tokens="begrip".split())
self.ag.add_node("14",
"verb",
pos="verb",
root="hebben",
tokens="had".split())
self.ag.add_node("15",
"punt",
pos="punct",
root=".",
tokens=".".split())
self.ag.add_edge("0", "1", "--")
self.ag.add_edge("1", "2", "su")
self.ag.add_edge("1", "3", "hd")
self.ag.add_edge("1", "4", "vc")
self.ag.add_edge("4", "5", "su")
self.ag.add_edge("4", "6", "hd")
self.ag.add_edge("4", "7", "vc")
self.ag.add_edge("7", "8", "cmp")
self.ag.add_edge("7", "9", "body")
self.ag.add_edge("9", "10", "su")
self.ag.add_edge("9", "11", "obj1")
self.ag.add_edge("11", "12", "mod")
self.ag.add_edge("11", "13", "hd")
self.ag.add_edge("9", "14", "hd")
self.ag.add_edge("0", "15", "punct")
class Test_AlpinoGraph(unittest.TestCase):
def setUp(self):
self.ag = AlpinoGraph(root="0")
self.ag.add_node("0", "top", cat="top",
tokens="Ik wilde weten of hij echt begrip had .".split())
self.ag.add_node("1", "smain", cat="smain",
tokens="Ik wilde weten of hij echt begrip had".split())
self.ag.add_node("2", "pron", pos="pron", root="ik", index="1",
tokens="Ik".split())
self.ag.add_node("3", "verb", pos="verb", root="willen",
tokens="wilde".split())
self.ag.add_node("4", "inf", cat="inf",
tokens="weten of hij echt begrip had".split())
self.ag.add_node("5", "index", index="1")
self.ag.add_node("6", "verb", pos="verb", root="weten",
tokens="weten".split())
self.ag.add_node("7", "cp", cat="cp",
tokens="of hij echt begrip had".split())
self.ag.add_node("8", "comp", pos="comp", root="of",
tokens="of".split())
self.ag.add_node("9", "ssub", cat="ssub",
tokens="hij echt begrip had".split())
self.ag.add_node("10", "pron", pos="pron", root="hij",
tokens="hij".split())
self.ag.add_node("11", "np", cat="np",
tokens="echt begrip".split())
self.ag.add_node("12", "adj", pos="adj", root="echt",
tokens="echt".split())
self.ag.add_node("13", "noun", pos="noun", root="begrip",
tokens="begrip".split())
self.ag.add_node("14", "verb", pos="verb", root="hebben",
tokens="had".split())
self.ag.add_node("15", "punt", pos="punct", root=".",
tokens=".".split())
self.ag.add_edge("0", "1", "--")
self.ag.add_edge("1", "2", "su")
self.ag.add_edge("1", "3", "hd")
self.ag.add_edge("1", "4", "vc")
self.ag.add_edge("4", "5", "su")
self.ag.add_edge("4", "6", "hd")
self.ag.add_edge("4", "7", "vc")
self.ag.add_edge("7", "8", "cmp")
self.ag.add_edge("7", "9", "body")
self.ag.add_edge("9", "10", "su")
self.ag.add_edge("9", "11", "obj1")
self.ag.add_edge("11", "12", "mod")
self.ag.add_edge("11", "13", "hd")
self.ag.add_edge("9", "14", "hd")
self.ag.add_edge("0", "15", "punct")
def test_print_subtree(self):
print "\n", self.ag
def test_node_is_nominal(self):
self.assertTrue(self.ag.node_is_nominal("13"))
self.assertFalse(self.ag.node_is_nominal("3"))
def test_node_is_punct(self):
self.assertTrue(self.ag.node_is_punct("15"))
self.assertFalse(self.ag.node_is_punct("14"))
def test_node_is_index(self):
self.assertTrue(self.ag.node_is_index("5"))
self.assertFalse(self.ag.node_is_index("1"))
self.assertFalse(self.ag.node_is_index("2"))
def test_get_root_node(self):
self.assertEqual(self.ag.root, "0")
def test_get_parent_node(self):
self.assertEqual(self.ag.get_parent_node("0"), None)
self.assertEqual(self.ag.get_parent_node("1"), "0")
def test_get_node_deprel(self):
self.assertEqual(self.ag.get_node_deprel("0"), None)
self.assertEqual(self.ag.get_node_deprel("15"), "punct")
def test_node_is_complement(self):
self.assertTrue(self.ag.node_is_complement("11"))
self.assertFalse(self.ag.node_is_complement("12"))