def get_jats_abstract(abstract):
# Convert the abstract to jats abstract tags
abstract = etoolsutils.remove_tag_and_text("object-id", abstract)
abstract = etoolsutils.remove_tag("abstract", abstract)
abstract = utils_html.remove_comment_tags(abstract)
abstract = etoolsutils.escape_ampersand(abstract)
abstract = etoolsutils.escape_unmatched_angle_brackets(
abstract, utils.allowed_tags())
abstract = replace_jats_tag("sec", "jats:sec", abstract)
abstract = replace_jats_tag("related-object", "jats:related-object",
abstract)
abstract = replace_jats_tag("title", "jats:title", abstract)
abstract = eautils.replace_tags(abstract, "p", "jats:p")
abstract = eautils.replace_tags(abstract, "italic", "jats:italic")
abstract = eautils.replace_tags(abstract, "bold", "jats:bold")
abstract = eautils.replace_tags(abstract, "underline", "jats:underline")
abstract = eautils.replace_tags(abstract, "sub", "jats:sub")
abstract = eautils.replace_tags(abstract, "sup", "jats:sup")
abstract = eautils.replace_tags(abstract, "sc", "jats:sc")
abstract = replace_jats_tag("inline-formula", "jats:inline-formula",
abstract)
abstract = replace_jats_tag("ext-link", "jats:ext-link", abstract)
abstract = replace_jats_tag("xref", "jats:xref", abstract)
# remove rid attributes
abstract = remove_tag_attr("rid", abstract)
return abstract
def convert_inline_tags(original_string):
tag_converted_string = etoolsutils.escape_ampersand(original_string)
tag_converted_string = etoolsutils.escape_unmatched_angle_brackets(
tag_converted_string, utils.allowed_tags()
)
tag_converted_string = eautils.replace_tags(tag_converted_string, "italic", "i")
tag_converted_string = eautils.replace_tags(tag_converted_string, "bold", "b")
tag_converted_string = eautils.replace_tags(tag_converted_string, "underline", "u")
return tag_converted_string
def test_escape_unmatched_angle_brackets(self, value, expected):
"""
Test some additional examples of unmatched angle brackets specifically
"""
self.assertEqual(
utils.escape_unmatched_angle_brackets(value,
allowed_xml_tag_fragments()),
expected,
)
def get_basic_abstract(abstract):
# Strip inline tags, keep the p tags
abstract = etoolsutils.remove_tag_and_text("object-id", abstract)
abstract = etoolsutils.remove_tag("related-object", abstract)
abstract = etoolsutils.remove_tag("abstract", abstract)
abstract = utils_html.remove_comment_tags(abstract)
abstract = etoolsutils.escape_ampersand(abstract)
abstract = etoolsutils.escape_unmatched_angle_brackets(
abstract, utils.allowed_tags())
abstract = convert_sec_tags(abstract)
abstract = tags.clean_tags(abstract,
do_not_clean=["<p>", "</p>", "<mml:", "</mml:"])
abstract = eautils.replace_tags(abstract, "p", "jats:p")
return abstract
def convert_to_xml_string(string):
"""
For input strings with escaped tags and special characters
issue a set of conversion functions to prepare it prior
to adding it to an article object
"""
string = entity_to_unicode(string)
string = decode_brackets(string)
string = eautils.replace_tags(string, "i", "italic")
string = eautils.replace_tags(string, "u", "underline")
string = eautils.replace_tags(string, "b", "bold")
string = eautils.replace_tags(string, "em", "italic")
string = etoolsutils.escape_unmatched_angle_brackets(
string, allowed_tags())
return string
def add_clean_tag(
parent,
tag_name,
original_string,
namespaces=REPARSING_NAMESPACES,
attributes=None,
attributes_text="",
):
"""remove allowed tags and then add a tag the parent"""
tag_converted_string = clean_tags(original_string)
tag_converted_string = etoolsutils.escape_ampersand(tag_converted_string)
tag_converted_string = etoolsutils.escape_unmatched_angle_brackets(
tag_converted_string
)
minidom_tag = xmlio.reparsed_tag(
tag_name, tag_converted_string, namespaces, attributes_text
)
append_tag(parent, minidom_tag, attributes=attributes)
def escape_xml(xml_string):
"escape ampersands and unmatched angle brackets in HTML string allowing some whitelisted tags"
xml_string = escape_ampersand(xml_string)
return escape_unmatched_angle_brackets(xml_string,
allowed_xml_tag_fragments())