def test_read_from_json_handles_parentheses_correctly(self):
json = {
'question': [],
'columns': ['Urban settlements'],
'cells': [['Dzhebariki-Khaya\\n(Джебарики-Хая)'],
['South Korea (KOR)'],
['Area (km²)']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.urban_settlements'])
assert neighbors == {'fb:cell.dzhebariki_khaya',
'fb:cell.south_korea_kor',
'fb:cell.area_km'}
json = {
'question': [],
'columns': ['Margin\\nof victory'],
'cells': [['−9 (67-67-68-69=271)']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.margin_of_victory'])
assert neighbors == {'fb:cell._9_67_67_68_69_271'}
json = {
'question': [],
'columns': ['Record'],
'cells': [['4.08 m (13 ft 41⁄2 in)']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.record'])
assert neighbors == {'fb:cell.4_08_m_13_ft_41_2_in'}
def test_read_from_json_handles_parentheses_correctly(self):
json = {
'question': [],
'columns': ['Urban settlements'],
'cells': [['Dzhebariki-Khaya\\n(Джебарики-Хая)'],
['South Korea (KOR)'], ['Area (km²)']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.urban_settlements'])
assert neighbors == {
'fb:cell.dzhebariki_khaya', 'fb:cell.south_korea_kor',
'fb:cell.area_km'
}
json = {
'question': [],
'columns': ['Margin\\nof victory'],
'cells': [['−9 (67-67-68-69=271)']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.margin_of_victory'])
assert neighbors == {'fb:cell._9_67_67_68_69_271'}
json = {
'question': [],
'columns': ['Record'],
'cells': [['4.08 m (13 ft 41⁄2 in)']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.record'])
assert neighbors == {'fb:cell.4_08_m_13_ft_41_2_in'}
def test_read_from_json_handles_newlines_in_columns(self):
# The TSV file we use has newlines converted to "\n", not actual escape characters. We
# need to be sure we catch this.
json = {
'question': [],
'columns': ['Peak\\nAUS', 'Peak\\nNZ'],
'cells': [['1', '2'],
['3', '4']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.peak_aus'])
assert neighbors == {'fb:cell.1', 'fb:cell.3'}
neighbors = set(graph.neighbors['fb:row.row.peak_nz'])
assert neighbors == {'fb:cell.2', 'fb:cell.4'}
neighbors = set(graph.neighbors['fb:cell.1'])
assert neighbors == {'fb:row.row.peak_aus'}
json = {
'question': [],
'columns': ['Title'],
'cells': [['Dance of the\\nSeven Veils']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.title'])
assert neighbors == {'fb:cell.dance_of_the_seven_veils'}
def test_read_from_json_handles_simple_cases(self):
json = {
'question': [Token(x) for x in ['where', 'is', 'mersin', '?']],
'columns': ['Name in English', 'Location'],
'cells': [['Paradeniz', 'Mersin'],
['Lake Gala', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:cell.mersin'])
assert graph.entities == ['-1', '0', '1', 'fb:cell.edirne', 'fb:cell.lake_gala',
'fb:cell.mersin', 'fb:cell.paradeniz', 'fb:row.row.location',
'fb:row.row.name_in_english']
assert neighbors == {'fb:row.row.location'}
neighbors = set(graph.neighbors['fb:row.row.name_in_english'])
assert neighbors == {'fb:cell.paradeniz', 'fb:cell.lake_gala'}
assert graph.entity_text['fb:cell.edirne'] == 'Edirne'
assert graph.entity_text['fb:cell.lake_gala'] == 'Lake Gala'
assert graph.entity_text['fb:cell.mersin'] == 'Mersin'
assert graph.entity_text['fb:cell.paradeniz'] == 'Paradeniz'
assert graph.entity_text['fb:row.row.location'] == 'Location'
assert graph.entity_text['fb:row.row.name_in_english'] == 'Name in English'
# These are default numbers that should always be in the graph.
assert graph.neighbors['-1'] == []
assert graph.neighbors['0'] == []
assert graph.neighbors['1'] == []
assert graph.entity_text['-1'] == '-1'
assert graph.entity_text['0'] == '0'
assert graph.entity_text['1'] == '1'
def test_read_from_json_handles_simple_cases(self):
json = {
'question': [Token(x) for x in ['where', 'is', 'mersin', '?']],
'columns': ['Name in English', 'Location'],
'cells': [['Paradeniz', 'Mersin'], ['Lake Gala', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:cell.mersin'])
assert graph.entities == [
'-1', '0', '1', 'fb:cell.edirne', 'fb:cell.lake_gala',
'fb:cell.mersin', 'fb:cell.paradeniz', 'fb:row.row.location',
'fb:row.row.name_in_english'
]
assert neighbors == {'fb:row.row.location'}
neighbors = set(graph.neighbors['fb:row.row.name_in_english'])
assert neighbors == {'fb:cell.paradeniz', 'fb:cell.lake_gala'}
assert graph.entity_text['fb:cell.edirne'] == 'Edirne'
assert graph.entity_text['fb:cell.lake_gala'] == 'Lake Gala'
assert graph.entity_text['fb:cell.mersin'] == 'Mersin'
assert graph.entity_text['fb:cell.paradeniz'] == 'Paradeniz'
assert graph.entity_text['fb:row.row.location'] == 'Location'
assert graph.entity_text[
'fb:row.row.name_in_english'] == 'Name in English'
# These are default numbers that should always be in the graph.
assert graph.neighbors['-1'] == []
assert graph.neighbors['0'] == []
assert graph.neighbors['1'] == []
assert graph.entity_text['-1'] == '-1'
assert graph.entity_text['0'] == '0'
assert graph.entity_text['1'] == '1'
def setUp(self):
self.tokenizer = WordTokenizer(SpacyWordSplitter(pos_tags=True))
self.utterance = self.tokenizer.tokenize("where is mersin?")
self.token_indexers = {"tokens": SingleIdTokenIndexer("tokens")}
json = {
'question': self.utterance,
'columns': ['Name in English', 'Location in English'],
'cells': [['Paradeniz', 'Mersin'],
['Lake Gala', 'Edirne']]
}
self.graph = TableQuestionKnowledgeGraph.read_from_json(json)
self.vocab = Vocabulary()
self.name_index = self.vocab.add_token_to_namespace("name", namespace='tokens')
self.in_index = self.vocab.add_token_to_namespace("in", namespace='tokens')
self.english_index = self.vocab.add_token_to_namespace("english", namespace='tokens')
self.location_index = self.vocab.add_token_to_namespace("location", namespace='tokens')
self.paradeniz_index = self.vocab.add_token_to_namespace("paradeniz", namespace='tokens')
self.mersin_index = self.vocab.add_token_to_namespace("mersin", namespace='tokens')
self.lake_index = self.vocab.add_token_to_namespace("lake", namespace='tokens')
self.gala_index = self.vocab.add_token_to_namespace("gala", namespace='tokens')
self.negative_one_index = self.vocab.add_token_to_namespace("-1", namespace='tokens')
self.zero_index = self.vocab.add_token_to_namespace("0", namespace='tokens')
self.one_index = self.vocab.add_token_to_namespace("1", namespace='tokens')
self.oov_index = self.vocab.get_token_index('random OOV string', namespace='tokens')
self.edirne_index = self.oov_index
self.field = KnowledgeGraphField(self.graph, self.utterance, self.token_indexers, self.tokenizer)
super(KnowledgeGraphFieldTest, self).setUp()
def setUp(self):
self.tokenizer = WordTokenizer(SpacyWordSplitter(pos_tags=True))
self.utterance = self.tokenizer.tokenize("where is mersin?")
self.token_indexers = {"tokens": SingleIdTokenIndexer("tokens")}
json = {
'question': self.utterance,
'columns': ['Name in English', 'Location in English'],
'cells': [['Paradeniz', 'Mersin'],
['Lake Gala', 'Edirne']]
}
self.graph = TableQuestionKnowledgeGraph.read_from_json(json)
self.vocab = Vocabulary()
self.name_index = self.vocab.add_token_to_namespace("name", namespace='tokens')
self.in_index = self.vocab.add_token_to_namespace("in", namespace='tokens')
self.english_index = self.vocab.add_token_to_namespace("english", namespace='tokens')
self.location_index = self.vocab.add_token_to_namespace("location", namespace='tokens')
self.paradeniz_index = self.vocab.add_token_to_namespace("paradeniz", namespace='tokens')
self.mersin_index = self.vocab.add_token_to_namespace("mersin", namespace='tokens')
self.lake_index = self.vocab.add_token_to_namespace("lake", namespace='tokens')
self.gala_index = self.vocab.add_token_to_namespace("gala", namespace='tokens')
self.negative_one_index = self.vocab.add_token_to_namespace("-1", namespace='tokens')
self.zero_index = self.vocab.add_token_to_namespace("0", namespace='tokens')
self.one_index = self.vocab.add_token_to_namespace("1", namespace='tokens')
self.oov_index = self.vocab.get_token_index('random OOV string', namespace='tokens')
self.edirne_index = self.oov_index
self.field = KnowledgeGraphField(self.graph, self.utterance, self.token_indexers, self.tokenizer)
super(KnowledgeGraphFieldTest, self).setUp()
def test_read_from_json_handles_crazy_unicode(self):
json = {
'question': [],
'columns': ['Town'],
'cells': [['Viðareiði'], ['Funningsfjørður'], ['Froðba']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.town'])
assert neighbors == {
'fb:cell.funningsfj_r_ur',
'fb:cell.vi_arei_i',
'fb:cell.fro_ba',
}
json = {
'question': [],
'columns': ['Fate'],
'cells': [['Sunk at 45°00′N 11°21′W / 45.000°N 11.350°W'],
['66°22′32″N 29°20′19″E / 66.37556°N 29.33861°E']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.fate'])
assert neighbors == {
'fb:cell.sunk_at_45_00_n_11_21_w_45_000_n_11_350_w',
'fb:cell.66_22_32_n_29_20_19_e_66_37556_n_29_33861_e'
}
json = {
'question': [],
'columns': ['€0.01', 'Σ Points'],
'cells': [['6,000', '9.5']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row._0_01'])
assert neighbors == {'fb:cell.6_000'}
neighbors = set(graph.neighbors['fb:row.row._points'])
assert neighbors == {'fb:cell.9_5'}
json = {
'question': [],
'columns': ['Division'],
'cells': [['1ª Aut. Pref.']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.division'])
assert neighbors == {'fb:cell.1_aut_pref'}
def test_read_from_json_handles_crazy_unicode(self):
json = {
'question': [],
'columns': ['Town'],
'cells': [['Viðareiði'],
['Funningsfjørður'],
['Froðba']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.town'])
assert neighbors == {
'fb:cell.funningsfj_r_ur',
'fb:cell.vi_arei_i',
'fb:cell.fro_ba',
}
json = {
'question': [],
'columns': ['Fate'],
'cells': [['Sunk at 45°00′N 11°21′W / 45.000°N 11.350°W'],
['66°22′32″N 29°20′19″E / 66.37556°N 29.33861°E']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.fate'])
assert neighbors == {'fb:cell.sunk_at_45_00_n_11_21_w_45_000_n_11_350_w',
'fb:cell.66_22_32_n_29_20_19_e_66_37556_n_29_33861_e'}
json = {
'question': [],
'columns': ['€0.01', 'Σ Points'],
'cells': [['6,000', '9.5']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row._0_01'])
assert neighbors == {'fb:cell.6_000'}
neighbors = set(graph.neighbors['fb:row.row._points'])
assert neighbors == {'fb:cell.9_5'}
json = {
'question': [],
'columns': ['Division'],
'cells': [['1ª Aut. Pref.']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.division'])
assert neighbors == {'fb:cell.1_aut_pref'}
def test_get_linked_agenda_items(self):
json = {
'question': [Token(x) for x in ['where', 'is', 'mersin', '?']],
'columns': ['Name in English', 'Location'],
'cells': [['Paradeniz', 'Mersin'],
['Lake Gala', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
assert graph.get_linked_agenda_items() == ['fb:cell.mersin', 'fb:row.row.location']
def test_read_from_json_handles_diacritics_and_newlines(self):
json = {
'question': [],
'columns': ['Notes'],
'cells': [['8 districts\nFormed from Orūzgān Province in 2004']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.notes'])
assert neighbors == {'fb:cell.8_districts_formed_from_oruzgan_province_in_2004'}
def test_get_longest_span_matching_entities(self):
json = {
'question': [Token(x) for x in ['where', 'is', 'lake', 'big', 'gala', '?']],
'columns': ['Name in English', 'Location'],
'cells': [['Paradeniz', 'Lake Big'],
['Lake Big Gala', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
assert graph._get_longest_span_matching_entities() == ['fb:cell.lake_big_gala']
def test_read_from_json_handles_diacritics_and_newlines(self):
json = {
u'question': [],
u'columns': [u'Notes'],
u'cells': [[u'8 districts\nFormed from Orūzgān Province in 2004']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors[u'fb:row.row.notes'])
assert neighbors == set(
[u'fb:cell.8_districts_formed_from_oruzgan_province_in_2004'])
def test_get_linked_agenda_items(self):
json = {
'question': [Token(x) for x in ['where', 'is', 'mersin', '?']],
'columns': ['Name in English', 'Location'],
'cells': [['Paradeniz', 'Mersin'], ['Lake Gala', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
assert graph.get_linked_agenda_items() == [
'fb:cell.mersin', 'fb:row.row.location'
]
def test_get_longest_span_matching_entities(self):
json = {
'question':
[Token(x) for x in ['where', 'is', 'lake', 'big', 'gala', '?']],
'columns': ['Name in English', 'Location'],
'cells': [['Paradeniz', 'Lake Big'], ['Lake Big Gala', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
assert graph._get_longest_span_matching_entities() == [
'fb:cell.lake_big_gala'
]
def test_read_from_json_handles_cells_with_duplicate_normalizations(self):
json = {
'question': [],
'columns': ['answer'],
'cells': [['yes'], ['yes*'], ['yes'], ['yes '], ['yes*']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
# There are three unique text strings that all normalize to "yes", so there are three
# fb:cell.yes entities. Hopefully we produce them in the same order as SEMPRE does...
assert graph.entities == ['-1', '0', '1', 'fb:cell.yes', 'fb:cell.yes_2', 'fb:cell.yes_3',
'fb:row.row.answer']
def test_read_from_json_handles_numbers_in_question(self):
# The TSV file we use has newlines converted to "\n", not actual escape characters. We
# need to be sure we catch this.
json = {
'question': [Token(x) for x in ['one', '4']],
'columns': [],
'cells': []
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
assert graph.neighbors['1'] == []
assert graph.neighbors['4'] == []
assert graph.entity_text['1'] == 'one'
assert graph.entity_text['4'] == '4'
def test_read_from_json_handles_diacritics(self):
json = {
'question': [],
'columns': ['Name in English', 'Name in Turkish', 'Location'],
'cells': [['Lake Van', 'Van Gölü', 'Mersin'],
['Lake Gala', 'Gala Gölü', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.name_in_turkish'])
assert neighbors == {'fb:cell.van_golu', 'fb:cell.gala_golu'}
json = {
'question': [],
'columns': ['Notes'],
'cells': [['Ordained as a priest at\nReșița on March, 29th 1936']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.notes'])
assert neighbors == {
'fb:cell.ordained_as_a_priest_at_resita_on_march_29th_1936'
}
json = {
'question': [],
'columns': ['Player'],
'cells': [['Mateja Kežman']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.player'])
assert neighbors == {'fb:cell.mateja_kezman'}
json = {
'question': [],
'columns': ['Venue'],
'cells': [['Arena Națională, Bucharest, Romania']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.venue'])
assert neighbors == {'fb:cell.arena_nationala_bucharest_romania'}
def test_read_from_json_handles_numbers_in_question(self):
# The TSV file we use has newlines converted to "\n", not actual escape characters. We
# need to be sure we catch this.
json = {
'question': [Token(x) for x in ['one', '4']],
'columns': [],
'cells': []
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
assert graph.neighbors['1'] == []
assert graph.neighbors['4'] == []
assert graph.entity_text['1'] == 'one'
assert graph.entity_text['4'] == '4'
def test_read_from_json_handles_cells_with_duplicate_normalizations(self):
json = {
'question': [],
'columns': ['answer'],
'cells': [['yes'], ['yes*'], ['yes'], ['yes '], ['yes*']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
# There are three unique text strings that all normalize to "yes", so there are three
# fb:cell.yes entities. Hopefully we produce them in the same order as SEMPRE does...
assert graph.entities == [
'-1', '0', '1', 'fb:cell.yes', 'fb:cell.yes_2', 'fb:cell.yes_3',
'fb:row.row.answer'
]
def test_read_from_json_handles_columns_with_duplicate_normalizations(
self):
json = {
'question': [],
'columns': ['# of votes', '% of votes'],
'cells': [['1', '2'], ['3', '4']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row._of_votes'])
assert neighbors == {'fb:cell.1', 'fb:cell.3'}
neighbors = set(graph.neighbors['fb:row.row._of_votes_2'])
assert neighbors == {'fb:cell.2', 'fb:cell.4'}
neighbors = set(graph.neighbors['fb:cell.1'])
assert neighbors == {'fb:row.row._of_votes'}
def test_read_from_json_handles_columns_with_duplicate_normalizations(self):
json = {
'question': [],
'columns': ['# of votes', '% of votes'],
'cells': [['1', '2'],
['3', '4']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row._of_votes'])
assert neighbors == {'fb:cell.1', 'fb:cell.3'}
neighbors = set(graph.neighbors['fb:row.row._of_votes_2'])
assert neighbors == {'fb:cell.2', 'fb:cell.4'}
neighbors = set(graph.neighbors['fb:cell.1'])
assert neighbors == {'fb:row.row._of_votes'}
def test_read_from_json_handles_diacritics(self):
json = {
'question': [],
'columns': ['Name in English', 'Name in Turkish', 'Location'],
'cells': [['Lake Van', 'Van Gölü', 'Mersin'],
['Lake Gala', 'Gala Gölü', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.name_in_turkish'])
assert neighbors == {'fb:cell.van_golu', 'fb:cell.gala_golu'}
json = {
'question': [],
'columns': ['Notes'],
'cells': [['Ordained as a priest at\nReșița on March, 29th 1936']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.notes'])
assert neighbors == {'fb:cell.ordained_as_a_priest_at_resita_on_march_29th_1936'}
json = {
'question': [],
'columns': ['Player'],
'cells': [['Mateja Kežman']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.player'])
assert neighbors == {'fb:cell.mateja_kezman'}
json = {
'question': [],
'columns': ['Venue'],
'cells': [['Arena Națională, Bucharest, Romania']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.venue'])
assert neighbors == {'fb:cell.arena_nationala_bucharest_romania'}
def test_read_from_json_handles_newlines_in_columns(self):
# The TSV file we use has newlines converted to "\n", not actual escape characters. We
# need to be sure we catch this.
json = {
'question': [],
'columns': ['Peak\\nAUS', 'Peak\\nNZ'],
'cells': [['1', '2'], ['3', '4']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.peak_aus'])
assert neighbors == {'fb:cell.1', 'fb:cell.3'}
neighbors = set(graph.neighbors['fb:row.row.peak_nz'])
assert neighbors == {'fb:cell.2', 'fb:cell.4'}
neighbors = set(graph.neighbors['fb:cell.1'])
assert neighbors == {'fb:row.row.peak_aus'}
json = {
'question': [],
'columns': ['Title'],
'cells': [['Dance of the\\nSeven Veils']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
neighbors = set(graph.neighbors['fb:row.row.title'])
assert neighbors == {'fb:cell.dance_of_the_seven_veils'}
def test_read_from_json_splits_columns_when_necessary(self):
json = {
'question': [Token(x) for x in ['where', 'is', 'mersin', '?']],
'columns': ['Name in English', 'Location'],
'cells': [['Paradeniz', 'Mersin with spaces'],
['Lake, Gala', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
assert graph.entities == ['-1', '0', '1', 'fb:cell.edirne', 'fb:cell.lake_gala',
'fb:cell.mersin_with_spaces', 'fb:cell.paradeniz', 'fb:part.gala',
'fb:part.lake', 'fb:part.paradeniz', 'fb:row.row.location',
'fb:row.row.name_in_english']
assert graph.neighbors['fb:part.lake'] == []
assert graph.neighbors['fb:part.gala'] == []
assert graph.neighbors['fb:part.paradeniz'] == []
def test_read_from_json_replaces_newlines(self):
# The csv -> tsv conversion renders '\n' as r'\n' (with a literal slash character), that
# gets read in a two characters instead of one. We need to make sure we convert it back to
# one newline character, so our splitting and other processing works correctly.
json = {
'question': [Token(x) for x in ['where', 'is', 'mersin', '?']],
'columns': ['Name\\nin English', 'Location'],
'cells': [['Paradeniz', 'Mersin'],
['Lake\\nGala', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
assert graph.entities == ['-1', '0', '1', 'fb:cell.edirne', 'fb:cell.lake_gala',
'fb:cell.mersin', 'fb:cell.paradeniz', 'fb:part.gala',
'fb:part.lake', 'fb:part.paradeniz', 'fb:row.row.location',
'fb:row.row.name_in_english']
assert graph.entity_text['fb:row.row.name_in_english'] == 'Name\nin English'
def test_read_from_json_splits_columns_when_necessary(self):
json = {
'question': [Token(x) for x in ['where', 'is', 'mersin', '?']],
'columns': ['Name in English', 'Location'],
'cells': [['Paradeniz', 'Mersin with spaces'],
['Lake, Gala', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
assert graph.entities == [
'-1', '0', '1', 'fb:cell.edirne', 'fb:cell.lake_gala',
'fb:cell.mersin_with_spaces', 'fb:cell.paradeniz', 'fb:part.gala',
'fb:part.lake', 'fb:part.paradeniz', 'fb:row.row.location',
'fb:row.row.name_in_english'
]
assert graph.neighbors['fb:part.lake'] == []
assert graph.neighbors['fb:part.gala'] == []
assert graph.neighbors['fb:part.paradeniz'] == []
def test_read_from_json_replaces_newlines(self):
# The csv -> tsv conversion renders '\n' as r'\n' (with a literal slash character), that
# gets read in a two characters instead of one. We need to make sure we convert it back to
# one newline character, so our splitting and other processing works correctly.
json = {
'question': [Token(x) for x in ['where', 'is', 'mersin', '?']],
'columns': ['Name\\nin English', 'Location'],
'cells': [['Paradeniz', 'Mersin'], ['Lake\\nGala', 'Edirne']]
}
graph = TableQuestionKnowledgeGraph.read_from_json(json)
assert graph.entities == [
'-1', '0', '1', 'fb:cell.edirne', 'fb:cell.lake_gala',
'fb:cell.mersin', 'fb:cell.paradeniz', 'fb:part.gala',
'fb:part.lake', 'fb:part.paradeniz', 'fb:row.row.location',
'fb:row.row.name_in_english'
]
assert graph.entity_text[
'fb:row.row.name_in_english'] == 'Name\nin English'
def text_to_instance(
self, # type: ignore
question: str,
table_info: Union[str, JsonDict],
example_lisp_string: str = None,
dpd_output: List[str] = None,
tokenized_question: List[Token] = None) -> Instance:
"""
Reads text inputs and makes an instance. WikitableQuestions dataset provides tables as TSV
files, which we use for training. For running a demo, we may want to provide tables in a
JSON format. To make this method compatible with both, we take ``table_info``, which can
either be a filename, or a dict. We check the argument's type and call the appropriate
method in ``TableQuestionKnowledgeGraph``.
Parameters
----------
question : ``str``
Input question
table_info : ``str`` or ``JsonDict``
Table filename or the table content itself, as a dict. See
``TableQuestionKnowledgeGraph.read_from_json`` for the expected format.
example_lisp_string : ``str``, optional
The original (lisp-formatted) example string in the WikiTableQuestions dataset. This
comes directly from the ``.examples`` file provided with the dataset. We pass this to
SEMPRE for evaluating logical forms during training. It isn't otherwise used for
anything.
dpd_output : List[str], optional
List of logical forms, produced by dynamic programming on denotations. Not required
during test.
tokenized_question : ``List[Token]``, optional
If you have already tokenized the question, you can pass that in here, so we don't
duplicate that work. You might, for example, do batch processing on the questions in
the whole dataset, then pass the result in here.
"""
# pylint: disable=arguments-differ
tokenized_question = tokenized_question or self._tokenizer.tokenize(
question.lower())
question_field = TextField(tokenized_question,
self._question_token_indexers)
if isinstance(table_info, str):
table_knowledge_graph = TableQuestionKnowledgeGraph.read_from_file(
table_info, tokenized_question)
table_metadata = MetadataField(open(table_info).readlines())
else:
table_knowledge_graph = TableQuestionKnowledgeGraph.read_from_json(
table_info)
table_metadata = MetadataField(table_info)
table_field = KnowledgeGraphField(
table_knowledge_graph,
tokenized_question,
self._table_token_indexers,
tokenizer=self._tokenizer,
feature_extractors=self._linking_feature_extractors,
include_in_vocab=self._use_table_for_vocab,
max_table_tokens=self._max_table_tokens)
world = WikiTablesWorld(table_knowledge_graph)
world_field = MetadataField(world)
production_rule_fields: List[Field] = []
for production_rule in world.all_possible_actions():
_, rule_right_side = production_rule.split(' -> ')
is_global_rule = not world.is_table_entity(rule_right_side)
field = ProductionRuleField(production_rule, is_global_rule)
production_rule_fields.append(field)
action_field = ListField(production_rule_fields)
fields = {
'question': question_field,
'table': table_field,
'world': world_field,
'actions': action_field
}
if self._include_table_metadata:
fields['table_metadata'] = table_metadata
if example_lisp_string:
fields['example_lisp_string'] = MetadataField(example_lisp_string)
# We'll make each target action sequence a List[IndexField], where the index is into
# the action list we made above. We need to ignore the type here because mypy doesn't
# like `action.rule` - it's hard to tell mypy that the ListField is made up of
# ProductionRuleFields.
action_map = {
action.rule: i
for i, action in enumerate(action_field.field_list)
} # type: ignore
if dpd_output:
action_sequence_fields: List[Field] = []
for logical_form in dpd_output:
if not self._should_keep_logical_form(logical_form):
logger.debug(f'Question was: {question}')
logger.debug(f'Table info was: {table_info}')
continue
try:
expression = world.parse_logical_form(logical_form)
except ParsingError as error:
logger.debug(
f'Parsing error: {error.message}, skipping logical form'
)
logger.debug(f'Question was: {question}')
logger.debug(f'Logical form was: {logical_form}')
logger.debug(f'Table info was: {table_info}')
continue
except:
logger.error(logical_form)
raise
action_sequence = world.get_action_sequence(expression)
try:
index_fields: List[Field] = []
for production_rule in action_sequence:
index_fields.append(
IndexField(action_map[production_rule],
action_field))
action_sequence_fields.append(ListField(index_fields))
except KeyError as error:
logger.debug(
f'Missing production rule: {error.args}, skipping logical form'
)
logger.debug(f'Question was: {question}')
logger.debug(f'Table info was: {table_info}')
logger.debug(f'Logical form was: {logical_form}')
continue
if len(action_sequence_fields) >= self._max_dpd_logical_forms:
break
if not action_sequence_fields:
# This is not great, but we're only doing it when we're passed logical form
# supervision, so we're expecting labeled logical forms, but we can't actually
# produce the logical forms. We should skip this instance. Note that this affects
# _dev_ and _test_ instances, too, so your metrics could be over-estimates on the
# full test data.
return None
fields['target_action_sequences'] = ListField(
action_sequence_fields)
if self._output_agendas:
agenda_index_fields: List[Field] = []
for agenda_string in world.get_agenda():
agenda_index_fields.append(
IndexField(action_map[agenda_string], action_field))
if not agenda_index_fields:
agenda_index_fields = [IndexField(-1, action_field)]
fields['agenda'] = ListField(agenda_index_fields)
return Instance(fields)
