def test_returnsBeamTreeOfSize1For1ValidInterpretation(
self, esperantifier):
sentence_tokens = [
BeamToken("assign", {ReservedWord.PUT: 1}),
BeamToken("7", {UnalphabeticTerminal.NUMBER: 1}),
BeamToken("to", {ReservedWord.TO: 1}),
BeamToken("small", {PartOfSpeech.ADJECTIVE: 1}),
BeamToken("cat", {PartOfSpeech.NOUN: 1}),
]
interpretations = esperantifier.try_interpreting(sentence_tokens)
assert isinstance(interpretations, BeamTree)
assert interpretations.number_of_leaves() == 1
def test_whenMidwayNodeHasInvalidInterpretationsItIsPruned(
self, esperantifier):
sentence_tokens = [
BeamToken("assign", {ReservedWord.PUT: 1}),
BeamToken("7", {UnalphabeticTerminal.NUMBER: 1}),
BeamToken("to", {
ReservedWord.TO: 1,
PartOfSpeech.V_IMP: 0.5
}),
BeamToken("small", {PartOfSpeech.ADJECTIVE: 1}),
BeamToken("cat", {PartOfSpeech.NOUN: 0.5}),
]
interpretations = esperantifier.try_interpreting(sentence_tokens)
remaining_interpretation = interpretations.get_next_interpretation()
assert isinstance(interpretations, BeamTree)
assert interpretations.number_of_leaves() == 1
assert len(remaining_interpretation) == len(sentence_tokens)
assert remaining_interpretation[2].tag == ReservedWord.TO
def test_pruningOneBranchDoesNotAffectOtherSubtrees(self, esperantifier):
"""The bug that this checks: pruning [PUT, ADJECTIVE, TO] has pruned TO
from the [PUT,NOUN] branch as well, resulting in:
└- BeamTree- root (None: 1)
└- BeamTreeNode- assign (ReservedWord.PUT: 1)
├- BeamTreeNode- love (PartOfSpeech.ADJECTIVE: 1) -- real mistake
│ ├- BeamTreeNode- to (ReservedWord.TO: 1)
│ │ └- BeamTreeNode- small (PartOfSpeech.ADJECTIVE: 1)
│ │ └- BeamTreeNode- cat (PartOfSpeech.NOUN: 0.5)
│ └- BeamTreeNode- to (PartOfSpeech.V_IMP: 0.5)
│ └- BeamTreeNode- small (PartOfSpeech.ADJECTIVE: 1)
│ └- BeamTreeNode- cat (PartOfSpeech.NOUN: 0.5)
└- BeamTreeNode- love (PartOfSpeech.NOUN: 1)
├- BeamTreeNode- to (ReservedWord.TO: 1) -- pruned unnecessarily
│ └- BeamTreeNode- small (PartOfSpeech.ADJECTIVE: 1)
│ └- BeamTreeNode- cat (PartOfSpeech.NOUN: 0.5)
└- BeamTreeNode- to (PartOfSpeech.V_IMP: 0.5)
└- BeamTreeNode- small (PartOfSpeech.ADJECTIVE: 1)
└- BeamTreeNode- cat (PartOfSpeech.NOUN: 0.5)"""
sentence_tokens = [
BeamToken("assign", {ReservedWord.PUT: 1}),
BeamToken("love", {
PartOfSpeech.ADJECTIVE: 1,
PartOfSpeech.NOUN: 1
}),
BeamToken("to", {
ReservedWord.TO: 1,
PartOfSpeech.V_IMP: 0.5
}),
BeamToken("small", {PartOfSpeech.ADJECTIVE: 1}),
BeamToken("cat", {PartOfSpeech.NOUN: 0.5}),
]
interpretations = esperantifier.try_interpreting(sentence_tokens)
remaining_interpretation = interpretations.get_next_interpretation()
assert isinstance(interpretations, BeamTree)
assert interpretations.number_of_leaves() == 1
assert len(remaining_interpretation) == len(sentence_tokens)
assert remaining_interpretation[2].tag == ReservedWord.TO
def test_whenWrongInterpretationIsAfter1RuleHasBeenFoldedPruningActionDoesNotMiss(
self, esperantifier):
sentence_tokens = [
BeamToken("assign", {ReservedWord.PUT: 1}),
BeamToken("40", {UnalphabeticTerminal.NUMBER: 1}),
BeamToken("+", {UnalphabeticTerminal.PLUS: 1}),
BeamToken("1", {UnalphabeticTerminal.NUMBER: 1}),
BeamToken("+", {UnalphabeticTerminal.PLUS: 1}),
BeamToken("1", {UnalphabeticTerminal.NUMBER: 1}),
BeamToken("to", {
ReservedWord.TO: 1,
PartOfSpeech.V_IMP: 0.5
}),
BeamToken("small", {PartOfSpeech.ADJECTIVE: 1}),
BeamToken("cat", {PartOfSpeech.NOUN: 0.5}),
]
interpretations = esperantifier.try_interpreting(sentence_tokens)
remaining_interpretation = interpretations.get_next_interpretation()
assert isinstance(interpretations, BeamTree)
assert interpretations.number_of_leaves() == 1
assert len(remaining_interpretation) == len(sentence_tokens)
assert remaining_interpretation[6].tag == ReservedWord.TO
def test_listOfMultipleBeamTokensIsPrintedWithAsMuchLinesAsTheMostPossibleTagsPlusOne(
self, kite_noun_token, love_noun_or_verb_token):
token_list = [kite_noun_token, love_noun_or_verb_token]
list_format = BeamToken.list_pretty_format(token_list)
assert len(
list_format.splitlines()) == len(love_noun_or_verb_token.tags) + 1
def token(self):
return BeamToken.from_tuple(self._stack.pop())
def love_noun_or_verb_token(self):
return BeamToken("love", {
PartOfSpeech.NOUN: 0.5,
PartOfSpeech.V_IMP: 0.125
})
def kite_noun_token(self):
return BeamToken("kite", {PartOfSpeech.NOUN: 1})