def test_duplicate_headers_to_ab_stream_ignores_duplicates(self):
sheet_name = "sheet1"
header_values = ["h1", "h1", "h3"]
# h1 is ignored because it is duplicate
expected_stream_header_values = ["h3"]
expected_stream = AirbyteStream(
name=sheet_name,
json_schema={
"$schema": "http://json-schema.org/draft-07/schema#",
"type": "object",
# For simplicity, the type of every cell is a string
"properties": {
header: {
"type": "string"
}
for header in expected_stream_header_values
},
},
supported_sync_modes=["full_refresh"],
)
actual_stream = Helpers.headers_to_airbyte_stream(
logger, sheet_name, header_values)
self.assertEqual(expected_stream, actual_stream)
def test_headers_to_airbyte_stream_blank_values_terminate_row(self):
sheet_name = "sheet1"
header_values = ["h1", "", "h3"]
expected_stream = AirbyteStream(
name=sheet_name,
json_schema={
"$schema": "http://json-schema.org/draft-07/schema#",
"type": "object",
# For simplicity, the type of every cell is a string
"properties": {"h1": {"type": "string"}},
},
)
actual_stream = Helpers.headers_to_airbyte_stream(logger, sheet_name, header_values)
self.assertEqual(expected_stream, actual_stream)
def test_duplicate_headers_to_ab_stream_fails(self):
sheet_name = "sheet1"
header_values = ["h1", "h1", "h3"]
with self.assertRaises(BaseException):
Helpers.headers_to_airbyte_stream(sheet_name, header_values)