def test_create_constellation(germaparl, discoursemes):
corpus_name = germaparl['corpus_name']
# parameters
parameters = discoursemes.pop('parameters')
flags = parameters['flags_query']
escape = parameters['escape_query']
p_query = parameters['p_query']
s_query = parameters['s_query']
s_context = parameters['s_context']
context = parameters['context']
# get topic and additional discoursemes
topic_items = discoursemes.pop('topic')
topic_discourseme = {'topic': topic_items}
discoursemes = discoursemes
# filter
const = create_constellation(
corpus_name,
# discoursemes
topic_discourseme,
discoursemes,
{},
# context settings
s_context,
context,
# query settings
p_query,
s_query,
flags,
escape,
# CWB setttings
registry_path=germaparl['registry_path'],
data_path=DATA_PATH)
assert len(const.df) == 10
# highlight
const = create_constellation(
corpus_name,
# discoursemes
topic_discourseme,
{},
discoursemes,
# context settings
s_context,
context,
# query settings
p_query,
s_query,
flags,
escape,
# CWB setttings
registry_path=germaparl['registry_path'],
data_path=DATA_PATH)
assert len(const.df) == 2990
def test_textual_constellation_association(germaparl, discoursemes):
corpus_name = germaparl['corpus_name']
# parameters
parameters = discoursemes.pop('parameters')
flags = parameters['flags_query']
escape = parameters['escape_query']
p_query = parameters['p_query']
s_query = parameters['s_query']
s_context = parameters['s_context']
context = parameters['context']
const = create_constellation(
corpus_name,
# discoursemes
{},
discoursemes,
{},
# context settings
s_context,
context,
# query settings
p_query,
s_query,
flags,
escape,
# CWB setttings
registry_path=germaparl['registry_path'],
data_path=DATA_PATH)
assoc = const.associations()
assert len(assoc) == 6
assert 'candidate' in assoc.columns
def test_constellation_conc(germaparl, discoursemes):
# parameters
parameters = discoursemes.pop('parameters')
# get topic and additional discoursemes
topic_items = discoursemes.pop('topic')
topic_discourseme = {'topic': topic_items}
discoursemes = discoursemes
# filter
const = create_constellation(
germaparl['corpus_name'],
# discoursemes
topic_discourseme,
discoursemes,
{},
# context settings
parameters['s_context'],
parameters['context'],
# query settings
parameters['p_query'],
parameters['s_query'],
parameters['flags_query'],
parameters['escape_query'],
# CWB setttings
registry_path=germaparl['registry_path'],
data_path=DATA_PATH)
lines = const.concordance(s_show=['text_id'])
assert len(lines) == 3
assert isinstance(lines[0], dict)
assert 'word' in lines[0]
assert isinstance(lines[0]['word'], list)
def test_create_textconstellation(germaparl, discoursemes):
corpus_name = germaparl['corpus_name']
# parameters
parameters = discoursemes.pop('parameters')
flags = parameters['flags_query']
escape = parameters['escape_query']
p_query = parameters['p_query']
s_query = parameters['s_query']
s_context = parameters['s_context']
context = parameters['context']
# create constellation
const = create_constellation(
corpus_name,
# discoursemes
{},
{},
discoursemes,
# context settings
s_context,
context,
# query settings
p_query,
s_query,
flags,
escape,
# CWB setttings
registry_path=germaparl['registry_path'],
data_path=DATA_PATH)
assert len(const.df) == 2198
def test_constellation_collocates(germaparl):
topic_name = 'topic'
topic_items = ['CDU', 'CSU']
p_query = 'lemma'
s_query = None
flags_query = '%cd'
flags_show = ''
min_freq = 2
s_context = 's'
context = 20
additional_discoursemes = {}
windows = [3, 5, 7]
cqp_bin = 'cqp'
lib_path = None
p_show = ['lemma']
ams = None
cut_off = 200
min_freq = 2
order = 'log_likelihood'
escape = True
frequencies = True
match_strategy = 'longest'
# preprocess parameters
s_query = s_context if s_query is None else s_query
topic_name = 'topic'
# create constellation
const = create_constellation(germaparl['corpus_name'],
{topic_name: topic_items}, {},
additional_discoursemes, s_context, context,
p_query, s_query, flags_query, escape,
match_strategy, lib_path, cqp_bin,
germaparl['registry_path'])
collocates = const.collocates(windows=windows,
p_show=p_show,
flags=flags_show,
ams=ams,
frequencies=frequencies,
min_freq=min_freq,
order=order,
cut_off=cut_off)
assert len(collocates) == 3
def test_textual_constellation_concordance(germaparl, discoursemes):
corpus_name = germaparl['corpus_name']
# parameters
parameters = discoursemes.pop('parameters')
flags = parameters['flags_query']
escape = parameters['escape_query']
p_query = parameters['p_query']
s_query = parameters['s_query']
s_context = parameters['s_context']
context = parameters['context']
# create constellation
const = create_constellation(
corpus_name,
# discoursemes
{},
discoursemes,
{},
# context settings
s_context,
context,
# query settings
p_query,
s_query,
flags,
escape,
# CWB setttings
registry_path=germaparl['registry_path'],
data_path=DATA_PATH)
# retrieve lines
lines = const.concordance(cut_off=None)
assert len(lines) == 2198