def test_count_items(germaparl):
corpus = Corpus(germaparl['corpus_name'],
registry_path=germaparl['registry_path'])
cqp = corpus.start_cqp()
items = ["Helmut", "Kohl", "CDU"]
queries = [formulate_cqp_query([item]) for item in items]
# whole corpus
counts1 = corpus.marginals(items)
counts2 = corpus.counts.mwus(cqp, queries)
assert (list(counts1["freq"]) == list(counts2["freq"]))
# subcorpus
cqp.nqr_from_query(query='[lemma="und"] expand to s', name='und')
cqp.nqr_activate(corpus.corpus_name, 'und')
counts1 = corpus.marginals(items)
counts2 = corpus.counts.mwus(cqp, queries)
assert (counts1.loc[items[0], 'freq'] > counts2.loc[queries[0], 'freq'])
# whole corpus
cqp.nqr_activate(corpus.corpus_name)
counts1 = corpus.marginals(items)
counts2 = corpus.counts.mwus(cqp, queries)
assert (list(counts1["freq"]) == list(counts2["freq"]))
cqp.__kill__()
def __init__(self,
corpus,
items,
p_query,
s_query,
s_context,
context=20,
flags="%cd",
escape=False):
"""
.items
.parameters
.idx
.dump
"""
self.items = items
self.parameters = {
'p_query': p_query,
's_query': s_query,
's_context': s_context,
'context': context,
'flags': flags,
'escape': escape
}
# run query
query = formulate_cqp_query(items, p_query, s_query, flags, escape)
dump = corpus.query(query, context, context_break=s_context)
self.dump = dump
self.idx = dump.name_cache
self._context = None
self._matches = None
def test_count_mwus_3(germaparl):
# whole corpus
corpus = Corpus(germaparl['corpus_name'],
registry_path=germaparl['registry_path'])
items = ["Horst Seehofer", r"( CSU )", "CSU", "WES324", "CSU"]
queries = [formulate_cqp_query([item]) for item in items]
cqp = corpus.start_cqp()
counts3 = corpus.counts.mwus(cqp, queries, strategy=3, fill_missing=False)
cqp.__kill__()
assert (counts3['freq']['CSU'] == 635)
def test_count_mwus_strategies(germaparl):
# whole corpus
corpus = Corpus(germaparl['corpus_name'],
registry_path=germaparl['registry_path'])
items = ["Horst Seehofer", r"( CSU )", "CSU", "WES324", "CSU"]
queries = [formulate_cqp_query([item]) for item in items]
cqp = corpus.start_cqp()
counts1 = corpus.counts.mwus(cqp, queries, strategy=1, fill_missing=False)
assert ('([word="CSU"])' in counts1.index)
counts2 = corpus.counts.mwus(cqp, queries, strategy=2, fill_missing=False)
counts3 = corpus.counts.mwus(cqp, queries, strategy=3, fill_missing=False)
cqp.__kill__()
assert (counts2.equals(counts3))
assert (sum(counts1['freq']) == sum(counts2['freq']))
def test_count_items_subcorpora(germaparl):
# subcorpus
corpus = Corpus(germaparl['corpus_name'],
registry_path=germaparl['registry_path'])
cqp = corpus.start_cqp()
dump = corpus.dump_from_s_att("text_role", ["presidency"])
cqp.nqr_from_dump(dump.df, 'presidency')
cqp.nqr_activate(corpus.corpus_name, 'presidency')
items = ["Horst Seehofer", r"( CSU )", "CSU", "WES324", "CSU"]
queries = [formulate_cqp_query([item]) for item in items]
counts1 = corpus.counts.mwus(cqp, queries, strategy=1, fill_missing=False)
assert (sum(counts1['freq']) > 0)
counts2 = corpus.counts.mwus(cqp, queries, strategy=2, fill_missing=False)
counts3 = corpus.counts.mwus(cqp, queries, strategy=3, fill_missing=False)
assert (counts2.equals(counts3))
cqp.__kill__()