def _write_read_file_into_url(base_url, filename, compression=None, pkg=None):
if not _is_installed(pkg, filename):
return
source_url = _build_source_url_from(base_url, filename, compression)
if source_url is None:
return
actual = None
if ".avro" in filename:
toavro(_table, source_url)
actual = fromavro(source_url)
elif ".xlsx" in filename:
toxlsx(_table, source_url, 'test1', mode='overwrite')
toxlsx(_table2, source_url, 'test2', mode='add')
actual = fromxlsx(source_url, 'test1')
elif ".xls" in filename:
toxls(_table, source_url, 'test')
actual = fromxls(source_url, 'test')
elif ".json" in filename:
tojson(_table, source_url)
actual = fromjson(source_url)
elif ".csv" in filename:
tocsv(_table, source_url, encoding="ascii", lineterminator="\n")
actual = fromcsv(source_url, encoding="ascii")
if actual is not None:
_show__rows_from("Expected:", _table)
_show__rows_from("Actual:", actual)
ieq(_table, actual)
ieq(_table, actual) # verify can iterate twice
else:
print("\n - %s SKIPPED " % filename, file=sys.stderr, end="")
def _write_read_file_into_url(base_url, filename, compression=None):
if ".avro" in filename and not _has_avro:
return
is_local = base_url.startswith("./")
if compression is not None:
if is_local:
return
filename = filename + "." + compression
codec = fsspec.utils.infer_compression(filename)
if codec is None:
print("\n - %s SKIPPED " % filename, file=sys.stderr, end="")
return
print("\n - %s " % filename, file=sys.stderr, end="")
if is_local:
source_url = base_url + filename
else:
source_url = os.path.join(base_url, filename)
_show__rows_from("Expected:", _table)
if ".avro" in filename:
toavro(_table, source_url)
actual = fromavro(source_url)
else:
tocsv(_table, source_url, encoding="ascii", lineterminator="\n")
actual = fromcsv(source_url, encoding="ascii")
_show__rows_from("Actual:", actual)
ieq(_table, actual)
ieq(_table, actual) # verify can iterate twice
def _append_to_avro_file(test_rows1,
test_rows2,
test_schema,
test_expect=None,
print_tables=True):
_show__expect_rows(test_rows1, print_tables)
_show__expect_rows(test_rows2, print_tables)
test_filename = _get_tempfile_path()
toavro(test_rows1, test_filename, schema=test_schema)
appendavro(test_rows2, test_filename, schema=test_schema)
test_actual = fromavro(test_filename)
if test_expect is not None:
test_expect2 = test_expect
else:
test_expect2 = cat(test_rows1, test_rows2)
_assert_rows_are_equals(test_expect2, test_actual, print_tables)
def _write_read_file_into_url(base_url, filename, compression=None, pkg=None):
if pkg is not None:
if not _is_installed(pkg):
print("\n - %s SKIPPED " % filename, file=sys.stderr, end="")
return
is_local = base_url.startswith("./")
if compression is not None:
if is_local:
return
filename = filename + "." + compression
codec = fsspec.utils.infer_compression(filename)
if codec is None:
print("\n - %s SKIPPED " % filename, file=sys.stderr, end="")
return
print("\n - %s " % filename, file=sys.stderr, end="")
if is_local:
source_url = base_url + filename
else:
source_url = os.path.join(base_url, filename)
actual = None
if ".avro" in filename:
toavro(_table, source_url)
actual = fromavro(source_url)
elif ".xlsx" in filename:
toxlsx(_table, source_url, 'test1', mode='overwrite')
toxlsx(_table2, source_url, 'test2', mode='add')
actual = fromxlsx(source_url, 'test1')
elif ".xls" in filename:
toxls(_table, source_url, 'test')
actual = fromxls(source_url, 'test')
elif ".json" in filename:
tojson(_table, source_url)
actual = fromjson(source_url)
elif ".csv" in filename:
tocsv(_table, source_url, encoding="ascii", lineterminator="\n")
actual = fromcsv(source_url, encoding="ascii")
if actual is not None:
_show__rows_from("Expected:", _table)
_show__rows_from("Actual:", actual)
ieq(_table, actual)
ieq(_table, actual) # verify can iterate twice
else:
print("\n - %s SKIPPED " % filename, file=sys.stderr, end="")
def _write_temp_avro_file(test_rows, test_schema):
test_filename = _get_tempfile_path()
print("Writing avro file:", test_filename)
toavro(test_rows, test_filename, schema=test_schema)
return test_filename