def test_xpath(xpath):
document = Document("<root><a/><b/></root>")
transformation = Transformation(Rule(xpath, lib.set_text("x")))
result = transformation(document).root
assert not result._data_node._exists
assert first(result.css_select("a")).full_text == "x"
assert first(result.css_select("b")).full_text == ""
def get_publication(node: TagNode, publicationStmt: TagNode):
date = first(node.css_select("mods|dateIssued")).full_text
name = first(node.css_select("mods|publisher")).full_text
place = first(
node.xpath('./mods:place/mods:placeTerm[@type="text"]')).full_text
publicationStmt.append_child(
tag("publisher", tag("name", name)),
tag("pubPlace", place),
tag("date", {"type": "creation"}, date),
)
def get_title(node: TagNode, titleStmt: TagNode, biblFull_titleStmt: TagNode):
non_sort = first(node.css_select("mods|nonSort"))
main_title = (non_sort.full_text + " ") if non_sort is not None else ""
main_title += first(node.css_select("mods|title")).full_text
sub_title = first(node.css_select("mods|subTitle")).full_text
for target in (titleStmt, biblFull_titleStmt):
target.append_child(
tag("title", {"type": "main"}, main_title),
tag("title", {"type": "sub"}, sub_title),
)
def test_first():
assert first([]) is None
assert first([1]) == 1
assert first((1, 2)) == 1
root = Document("<root><a/><a/><b/></root>").root
assert first(root.child_nodes(is_tag_node, lambda x: x.local_name == "c")) is None
assert (
first(root.child_nodes(is_tag_node, lambda x: x.local_name == "b")) is root[2]
)
assert (
first(root.child_nodes(is_tag_node, lambda x: x.local_name == "a")) is root[0]
)
with pytest.raises(TypeError):
first({})
def list_persons(previous_result, html: TagNode):
first(html.css_select("html|body html|ul")).append_child(
*(html.new_tag_node("li", children=[f"{x[1]}, {x[0]}"])
for x in previous_result))
def extract_person(node: TagNode, persons):
persons.append((
first(node.css_select("name")).full_text,
first(node.css_select("family-name")).full_text,
))
def extract_person(node: TagNode):
return node.attributes["username"], first(
node.css_select("name")).full_text