def _test_to_dataframe(self, mdoc, test):
msdf = to_mapping_set_dataframe(mdoc)
df = to_dataframe(msdf)
self.assertEqual(
len(df),
test.ct_data_frame_rows,
f"The pandas data frame has less elements than the orginal one for {test.filename}",
)
df.to_csv(test.get_out_file("roundtrip.tsv"), sep="\t")
# data = pd.read_csv(test.get_out_file("roundtrip.tsv"), sep="\t")
data = read_pandas(test.get_out_file("roundtrip.tsv"))
self.assertEqual(
len(data),
test.ct_data_frame_rows,
f"The re-serialised pandas data frame has less elements than the orginal one for {test.filename}",
)
path = test.get_out_file("tsv")
with open(path, "w") as file:
write_table(msdf, file)
# self._test_files_equal(test.get_out_file("tsv"), test.get_validate_file("tsv"))
df = read_pandas(path)
self.assertEqual(
len(df),
test.ct_data_frame_rows,
f"The exported pandas data frame has less elements than the orginal one for {test.filename}",
)
def test_parse_sssom_dataframe_url(self):
"""Test parsing a TSV from a URL."""
msdf = read_sssom_table(self.df_url)
output_path = os.path.join(test_out_dir, "test_parse_sssom_dataframe_url.tsv")
with open(output_path, "w") as file:
write_table(msdf, file)
self.assertEqual(
len(msdf.df),
141,
f"{self.df_url} has the wrong number of mappings.",
)
def test_write_sssom_dataframe(self):
"""Test writing as a dataframe."""
tmp_path = os.path.join(test_out_dir, "test_write_sssom_dataframe.tsv")
with open(tmp_path, "w") as tmp_file:
write_table(self.msdf, tmp_file)
msdf = read_sssom_table(tmp_path)
self.assertEqual(
len(msdf.df),
self.mapping_count,
f"{tmp_file} has the wrong number of mappings.",
)
def test_parse_sssom_dataframe(self):
"""Test parsing a TSV."""
input_path = f"{test_data_dir}/basic.tsv"
msdf = read_sssom_table(input_path)
output_path = os.path.join(test_out_dir, "test_parse_sssom_dataframe.tsv")
with open(output_path, "w") as file:
write_table(msdf, file)
self.assertEqual(
len(msdf.df),
141,
f"{input_path} has the wrong number of mappings.",
)
def test_parse_sssom_rdf(self):
"""Test parsing RDF."""
msdf = from_sssom_rdf(
g=self.rdf_graph, prefix_map=self.df_prefix_map, meta=self.metadata
)
path = os.path.join(test_out_dir, "test_parse_sssom_rdf.tsv")
with open(path, "w") as file:
write_table(msdf, file)
self.assertEqual(
len(msdf.df),
136,
f"{self.rdf_graph_file} has the wrong number of mappings.",
)
def test_parse_tsv(self):
"""Test parsing TSV."""
msdf = from_sssom_dataframe(
df=self.df, prefix_map=self.df_prefix_map, meta=self.df_meta
)
path = os.path.join(test_out_dir, "test_parse_tsv.tsv")
with open(path, "w") as file:
write_table(msdf, file)
self.assertEqual(
len(msdf.df),
141,
f"{self.df_file} has the wrong number of mappings.",
)
def test_parse_obographs(self):
"""Test parsing OBO Graph JSON."""
msdf = from_obographs(
jsondoc=self.obographs,
prefix_map=self.prefix_map,
meta=self.metadata,
)
path = os.path.join(test_out_dir, "test_parse_obographs.tsv")
with open(path, "w") as file:
write_table(msdf, file)
self.assertEqual(
len(msdf.df),
9878,
f"{self.obographs_file} has the wrong number of mappings.",
)
def test_parse_sssom_json(self):
"""Test parsing JSON."""
msdf = from_sssom_json(
jsondoc=self.json,
prefix_map=self.df_prefix_map,
meta=self.metadata,
)
path = os.path.join(test_out_dir, "test_parse_sssom_json.tsv")
with open(path, "w") as file:
write_table(msdf, file)
self.assertEqual(
len(msdf.df),
141,
f"{self.json_file} has the wrong number of mappings.",
)
def test_parse_alignment_minidom(self):
"""Test parsing an alignment XML."""
msdf = from_alignment_minidom(
dom=self.alignmentxml,
prefix_map=self.prefix_map,
meta=self.metadata,
)
path = os.path.join(test_out_dir, "test_parse_alignment_minidom.tsv")
with open(path, "w") as file:
write_table(msdf, file)
self.assertEqual(
len(msdf.df),
646,
f"{self.alignmentxml_file} has the wrong number of mappings.",
)
def test_write_sssom_rdf(self):
"""Test writing as RDF."""
path_1 = os.path.join(test_out_dir, "test_write_sssom_rdf.rdf")
with open(path_1, "w") as file:
write_rdf(self.msdf, file)
msdf = read_sssom_rdf(path_1, self.msdf.prefix_map)
self.assertEqual(
len(msdf.df),
self.mapping_count,
f"{path_1} has the wrong number of mappings.",
)
# TODO this test doesn't make sense
path_2 = os.path.join(test_out_dir, "test_write_sssom_rdf.rdf.tsv")
with open(path_2, "w") as file:
write_table(self.msdf, file)