def _html_table_processor() -> HTMLTableProcessor:
root_url = 'https://127.0.0.1'
headings_xpath = '{TABLE_XPATH}/thead//th[contains(@class, "rgHeader")]/descendant-or-self::*/text()'
data_rows_xpath = '{TABLE_XPATH}/tbody//tr'
data_cell_xpath = 'td/descendant-or-self::*/text()'
table_class = 'rgMasterTable'
return HTMLTableProcessor(root_url, headings_xpath, data_rows_xpath,
data_cell_xpath, table_class)
def test_get_table_html(self, html_table_processor: HTMLTableProcessor,
sample_html: str,
sample_processed_result: pd.DataFrame) -> None:
actual_result = html_table_processor.get_tabular_data_from_html(
sample_html,
column_name_mapper=GenericColumnMapper().map_list,
known_percentages=['Percentage']).reset_index(drop=True)
pd.testing.assert_frame_equal(sample_processed_result, actual_result)
def test_get_table_element(self, html_table_processor: HTMLTableProcessor,
sample_html: str,
sample_processed_result: pd.DataFrame) -> None:
html_dom = lxml.etree.HTML(sample_html)
actual_result = html_table_processor.get_tabular_data_from_element(
html_dom,
column_name_mapper=GenericColumnMapper().map_list,
known_percentages=['Percentage']).reset_index(drop=True)
pd.testing.assert_frame_equal(sample_processed_result,
actual_result,
check_dtype=False)
def test_get_table_url(self, html_table_processor: HTMLTableProcessor,
sample_html: str,
sample_processed_result: pd.DataFrame,
response_get_monkeypatch: Callable) -> None:
response_get_monkeypatch(sample_html)
actual_result = html_table_processor.get_tabular_data_from_url(
'/test',
column_name_mapper=GenericColumnMapper().map_list,
known_percentages=['Percentage']).reset_index(drop=True)
pd.testing.assert_frame_equal(sample_processed_result,
actual_result,
check_dtype=False)