def test_constellation_add(germaparl, discoursemes):
corpus = get_corpus(germaparl)
# init constellation
topic_query = format_cqp_query(
discoursemes['topic'],
p_query=discoursemes['parameters']['p_query'],
s_query=discoursemes['parameters']['s_query'],
flags=discoursemes['parameters']['flags_query'],
escape=discoursemes['parameters']['escape_query'])
topic_dump = corpus.query(
topic_query,
context=None,
context_break=discoursemes['parameters']['s_context'])
const = Constellation(topic_dump)
# add discourseme
disc1_query = format_cqp_query(
discoursemes['disc1'],
p_query=discoursemes['parameters']['p_query'],
s_query=discoursemes['parameters']['s_query'],
flags=discoursemes['parameters']['flags_query'],
escape=discoursemes['parameters']['escape_query'])
disc1_dump = corpus.query(
disc1_query,
context=None,
context_break=discoursemes['parameters']['s_context'])
const.add_discourseme(disc1_dump)
assert len(const.df) == 1599
assert 'topic' in const.discoursemes
assert len(const.discoursemes) == 2
def test_constellation_add():
topic_query = format_cqp_query(TOPIC_ITEMS,
p_query=P_QUERY,
s_query=S_QUERY,
flags="%cd",
escape=False)
# init constellation
topic_dump = CORPUS.query(topic_query,
context=None,
context_break=S_CONTEXT)
const = Constellation(topic_dump)
# add discourseme
disc1_query = format_cqp_query(DISC1_ITEMS,
p_query=P_QUERY,
s_query=S_QUERY,
flags="%cd",
escape=False)
disc1_dump = CORPUS.query(disc1_query,
context=None,
context_break=S_CONTEXT)
const.add_discourseme(disc1_dump)
print(const.df)
print(const.discoursemes.keys())
def test_constellation_init():
# init constellation
topic_query = format_cqp_query(TOPIC_ITEMS,
p_query=P_QUERY,
s_query=S_QUERY,
flags="%cd",
escape=False)
topic_dump = CORPUS.query(topic_query,
context=None,
context_break=S_CONTEXT)
const = Constellation(topic_dump)
print(const.df)
def test_constellation_coll(germaparl, discoursemes):
corpus = get_corpus(germaparl)
# init constellation
topic_query = format_cqp_query(
discoursemes['topic'],
p_query=discoursemes['parameters']['p_query'],
s_query=discoursemes['parameters']['s_query'],
flags=discoursemes['parameters']['flags_query'],
escape=discoursemes['parameters']['escape_query'])
topic_dump = corpus.query(
topic_query,
context=None,
context_break=discoursemes['parameters']['s_context'])
const = Constellation(topic_dump)
# add discourseme 1
disc1_query = format_cqp_query(
discoursemes['disc1'],
p_query=discoursemes['parameters']['p_query'],
s_query=discoursemes['parameters']['s_query'],
flags=discoursemes['parameters']['flags_query'],
escape=discoursemes['parameters']['escape_query'])
disc1_dump = corpus.query(
disc1_query,
context=None,
context_break=discoursemes['parameters']['s_context'])
const.add_discourseme(disc1_dump, name='disc1')
# add discourseme 2
disc2_query = format_cqp_query(
discoursemes['disc2'],
p_query=discoursemes['parameters']['p_query'],
s_query=discoursemes['parameters']['s_query'],
flags=discoursemes['parameters']['flags_query'],
escape=discoursemes['parameters']['escape_query'])
disc2_dump = corpus.query(
disc2_query,
context=None,
context_break=discoursemes['parameters']['s_context'])
const.add_discourseme(disc2_dump, name='disc2')
dfs = const.collocates(windows=list(range(1, 21)))
assert len(dfs) == 20
assert len(dfs[1]) == 2
assert len(dfs[20]) == 5
def test_constellation_init(germaparl, discoursemes):
corpus = get_corpus(germaparl)
# init constellation
topic_query = format_cqp_query(
discoursemes['topic'],
p_query=discoursemes['parameters']['p_query'],
s_query=discoursemes['parameters']['s_query'],
flags=discoursemes['parameters']['flags_query'],
escape=discoursemes['parameters']['escape_query'])
topic_dump = corpus.query(
topic_query,
context=None,
context_break=discoursemes['parameters']['s_context'])
const = Constellation(topic_dump)
assert isinstance(const.df, DataFrame)
assert len(const.df) == 2777
def test_constellation_conc():
topic_query = format_cqp_query(TOPIC_ITEMS,
p_query=P_QUERY,
s_query=S_QUERY,
flags="%cd",
escape=False)
# init constellation
topic_dump = CORPUS.query(topic_query,
context=None,
context_break=S_CONTEXT)
const = Constellation(topic_dump)
# add discourseme 1
disc1_query = format_cqp_query(DISC1_ITEMS,
p_query=P_QUERY,
s_query=S_QUERY,
flags="%cd",
escape=False)
disc1_dump = CORPUS.query(disc1_query,
context=None,
context_break=S_CONTEXT)
const.add_discourseme(disc1_dump, name='disc1')
# add discourseme 2
disc2_query = format_cqp_query(DISC2_ITEMS,
p_query=P_QUERY,
s_query=S_QUERY,
flags="%cd",
escape=False)
disc2_dump = CORPUS.query(disc2_query,
context=None,
context_break=S_CONTEXT)
const.add_discourseme(disc2_dump, name='disc2')
lines = const.concordance(s_show=['text_id'])
print(lines)
def test_constellation_coll():
topic_query = format_cqp_query(TOPIC_ITEMS,
p_query=P_QUERY,
s_query=S_QUERY,
flags="%cd",
escape=False)
# init constellation
topic_dump = CORPUS.query(topic_query,
context=None,
context_break=S_CONTEXT)
const = Constellation(topic_dump)
# add discourseme 1
disc1_query = format_cqp_query(DISC1_ITEMS,
p_query=P_QUERY,
s_query=S_QUERY,
flags="%cd",
escape=False)
disc1_dump = CORPUS.query(disc1_query,
context=None,
context_break=S_CONTEXT)
const.add_discourseme(disc1_dump, name='disc1')
# add discourseme 2
disc2_query = format_cqp_query(DISC2_ITEMS,
p_query=P_QUERY,
s_query=S_QUERY,
flags="%cd",
escape=False)
disc2_dump = CORPUS.query(disc2_query,
context=None,
context_break=S_CONTEXT)
const.add_discourseme(disc2_dump, name='disc2')
lines = const.collocates(windows=list(range(1, 20)))
print(lines)