def unpack_element(
tree: etree._ElementTree,
element: Union[etree._Element, etree._ElementUnicodeResult, Any]
) -> Tuple[str, str]:
"""Returns path in the tree and string representation for the given XPath query element.
"""
if isinstance(element, etree._Element):
path = tree.getpath(element)
text = etree.tostring(element,
encoding='unicode',
pretty_print=True)
else:
text = str(element)
try:
parent = element.getparent()
except AttributeError:
path = ''
else:
path = tree.getpath(parent)
return (path, text)
def _check_hidden_attributes(
tree: _ElementTree, element: HtmlElement,
defined_language: str) -> Optional[LanguageInfraction]:
"""Check for an infraction in the hidden attributes of an element.
Parameters
----------
tree : _ElementTree
The root element tree of the web page, used to calculate the xpath
element : HtmlElement
The current element to check for an infraction
defined_language : str
The defined language of the current element
Returns
-------
Optional[LanguageInfraction]
Either an infraction against WCAG 3.1.2 or `None`
"""
for attribute_name in HIDDEN_ATTRIBUTES:
if attribute_name in element.attrib:
attribute_value = element.attrib[attribute_name].strip()
if count_words(attribute_value) < MIN_WORDS_HIDDEN:
# If the hidden attribute's value is too short, we don't check its language
continue
detected_language = predict_language(attribute_value)
if detected_language and defined_language != detected_language:
# The hidden attribute is wrong, return an infraction
return LanguageInfraction(
wcag_criterion="WCAG_3_1_2",
xpath=tree.getpath(element),
html_language=defined_language,
predicted_language=detected_language,
text=attribute_value,
)
return None
def _dfs(
tree: _ElementTree,
element: HtmlElement,
parent_language: str,
infractions: List[LanguageInfraction],
) -> Tuple[str, Optional[str], Optional[str], List[LanguageInfraction]]:
"""Check for infractions against WCAG 3.1.2 using a recursive DFS.
Parameters
----------
tree : _ElementTree
The root element tree of the web page, used to calculate the xpath
element : HtmlElement
The current element to check for infractions
parent_language : str
The defined language of the current element's parent
infractions : List[LanguageInfraction]
The infractions found until now
Returns
-------
Tuple[str, Optional[str], Optional[str], List[LanguageInfraction]]
A tuple containing:
- the explicitly defined language of the current element
- the detected language of the current element
- the text of the current element
- the infractions found
"""
# If the current element does not have a `lang` attribute, take the parent's language
defined_language = element.get("lang", parent_language).lower()[:2]
# The text contained in this element (and in this element alone)
text = (element.text or "").replace("\n", " ").strip()
# Check for infractions against WCAG 3.1.2 in hidden attributes
hidden_infraction = _check_hidden_attributes(tree, element,
defined_language)
if hidden_infraction is not None:
infractions.append(hidden_infraction)
# Check for infractions against WCAG 3.1.2 in the current element's children
children = element.getchildren()
if children:
# Run this function on each of the children (= recursion)
children_results = [
_dfs(tree, child, defined_language, infractions)
for child in children
]
# Differ between children that are similar and children that are different
if len({(defined, detected)
for defined, detected, _, _ in children_results}) == 1:
# All children have the same language defined and detected
child_defined_language, child_detected_language, _, _ = children_results[
0]
children_text = " ".join(
str(child_result[2]) for child_result in children_results)
if child_detected_language and child_defined_language != child_detected_language:
# All children are wrong
# Give a warning for the current element instead of for each of its children
infractions.append(
LanguageInfraction(
wcag_criterion="WCAG_3_1_2",
xpath=tree.getpath(element),
html_language=child_defined_language,
predicted_language=child_detected_language,
text=children_text,
))
else:
# The children have different values for their defined and detected languages
for child, child_result in zip(children, children_results):
child_defined_language, child_detected_language, child_text, _ = child_result
if child_detected_language and child_detected_language != child_defined_language:
# This child is wrong, give a warning for only this child
infractions.append(
LanguageInfraction(
wcag_criterion="WCAG_3_1_2",
xpath=tree.getpath(child),
html_language=child_defined_language,
predicted_language=child_detected_language,
text=str(child_text),
))
# If any of the children of the current element contains a very short piece of text, add it
# to the current element's text
for _, _, child_text, _ in children_results:
if child_text is None:
continue
child_text = child_text.strip()
# We only add the child text if it is short
if count_words(child_text) >= MIN_WORDS_DEFAULT:
continue
current_text = (text or "").strip()
# We only add the child text if the current text doesn't end with the child text
if not current_text.endswith(child_text):
text = current_text + " " + child_text
# If the current element's text is long enough, predict its language
if count_words(text) >= MIN_WORDS_DEFAULT:
detected_language = predict_language(text)
else:
detected_language = None
return defined_language, detected_language, text, infractions
