def test_item_merge_into_list_processor_wrong_field_exception():
test_item = {"category": "electronics", "subcategory": "phones"}
with pytest.raises(KeyError):
processors.ItemKeysMergeIntoListProcessor(
new_item_key="breadcrumbs",
item_keys=["category", "subcategory", "type"]).parse(test_item)
def test_item_merge_into_list_processor():
test_item = {"category": "electronics", "subcategory": "phones"}
item = processors.ItemKeysMergeIntoListProcessor(
new_item_key="breadcrumbs", item_keys=["category",
"subcategory"]).parse(test_item)
assert item["breadcrumbs"] == ["electronics", "phones"]
# test that old parsers have been removed
assert all(k not in item for k in ["category", "subcategory"])
def test_item_merge_into_list_processor_ignore_none():
test_item = {
"category": "electronics",
"subcategory": "phones",
"type": None
}
item = processors.ItemKeysMergeIntoListProcessor(
new_item_key="breadcrumbs",
item_keys=["category", "subcategory", "type"]).parse(test_item)
assert item["breadcrumbs"] == ["electronics", "phones"]
def test_item_merge_into_list_processor_preserve_original():
test_item = {"category": "electronics", "subcategory": "phones"}
item = processors.ItemKeysMergeIntoListProcessor(
new_item_key="breadcrumbs",
item_keys=["category", "subcategory"],
preserve_original=True,
).parse(test_item)
assert item["breadcrumbs"] == ["electronics", "phones"]
# test that old parsers remain
assert all(k in item for k in ["category", "subcategory"])