"
base = Nokogiri::HTML5::DocumentFragment.parse(frag)
html5_fragment = Nokogiri::HTML5.fragment(frag)
assert_equal(base.to_html, html5_fragment.to_html)
end
def test_url
html = "
"
shift_jis = utf8.encode(Encoding::SHIFT_JIS)
raw = shift_jis.dup
raw.force_encoding(Encoding::ASCII_8BIT)
assert_match(/おはようございます/, Nokogiri::HTML5.fragment(utf8).to_s)
assert_match(/おはようございます/, Nokogiri::HTML5.fragment(shift_jis).to_s)
refute_match(/おはようございます/, Nokogiri::HTML5.fragment(raw).to_s)
assert_match(/おはようございます/, Nokogiri::HTML5.fragment(raw, Encoding::SHIFT_JIS).to_s)
assert_match(/おはようございます/, Nokogiri::HTML5::DocumentFragment.parse(raw, Encoding::SHIFT_JIS).to_s)
# with kwargs
assert_match(/おはようございます/, Nokogiri::HTML5.fragment(raw, encoding: Encoding::SHIFT_JIS).to_s)
assert_match(/おはようございます/, Nokogiri::HTML5::DocumentFragment.parse(raw, encoding: Encoding::SHIFT_JIS).to_s)
end
def test_fragment_serialization_encoding
frag = Nokogiri::HTML5.fragment("아는 길도 물어가라")
html = frag.serialize(encoding: "US-ASCII")
assert_equal("아는 길도 물어가라", html)
frag = Nokogiri::HTML5.fragment(html)
assert_equal("아는 길도 물어가라", frag.serialize)
end
def test_serialization_encoding
html = "ฉันไม่พูดภาษาไทย"
doc = Nokogiri::HTML5(html)
span = doc.at("/html/body/span")
serialized = span.inner_html(encoding: "US-ASCII")
assert_match(/^(?:&#(?:\d+|x\h+);)*$/, serialized)
assert_equal(
"ฉันไม่พูดภาษาไทย".each_char.map(&:ord),
serialized.scan(/&#(\d+|x\h+);/).map do |s|
s = s.first
if s.start_with?("x")
s[1..-1].to_i(16)
else
s.to_i
end
end,
)
doc2 = Nokogiri::HTML5(doc.serialize(encoding: "Big5"))
html2 = doc2.serialize(encoding: "UTF-8")
assert_match("ฉันไม่พูดภาษาไทย", html2)
end
def test_parse_noscript_as_elements_in_head
# isn't allowed in noscript so the noscript element is popped off
# the stack of open elements and the token is reprocessed in `head`
# which causes the `head` element to be popped off the stack of open
# elements and a `body` element to be inserted. Then the `img` element is
# inserted in the body.
html = ""
doc = Nokogiri::HTML5(html, parse_noscript_content_as_text: false, max_errors: 100)
noscript = doc.at("/html/head/noscript")
assert_equal(3, doc.errors.length, doc.errors.join("\n"))
# Start tag 'img' isn't allowed here
# End tag 'noscript' isn't allowed here
# End tag head isn't allowed here
assert_empty(noscript.children)
img = doc.at("/html/body/img")
refute_nil(img)
end
def test_parse_noscript_as_text_in_head
# In contrast to the previous test, when the scripting flag is enabled, the content
# of the noscript element is parsed as raw text.
html = ""
doc = Nokogiri::HTML5(html, parse_noscript_content_as_text: true, max_errors: 100)
noscript = doc.at("/html/head/noscript")
assert_empty(doc.errors)
assert_equal(1, noscript.children.length)
assert_kind_of(Nokogiri::XML::Text, noscript.children.first)
end
def test_parse_noscript_as_elements_in_body
html = ""
doc = Nokogiri::HTML5(html, parse_noscript_content_as_text: false, max_errors: 100)
assert_empty(doc.errors)
img = doc.at("/html/body/noscript/img")
refute_nil(img)
end
def test_parse_noscript_as_text_in_body
html = ""
doc = Nokogiri::HTML5(html, parse_noscript_content_as_text: true, max_errors: 100)
noscript = doc.at("/html/body/noscript")
assert_empty(doc.errors, doc.errors.join("\n"))
assert_equal(1, noscript.children.length)
assert_kind_of(Nokogiri::XML::Text, noscript.children.first)
end
def test_parse_noscript_fragment_as_elements
html = ""
frag = Nokogiri::HTML5::DocumentFragment.new(Nokogiri::HTML5::Document.new, html, "noscript", parse_noscript_content_as_text: false, max_errors: 100)
assert_empty(frag.errors)
assert_equal(2, frag.children.length)
end
def test_parse_noscript_fragment_as_text
html = ""
frag = Nokogiri::HTML5::DocumentFragment.new(Nokogiri::HTML5::Document.new, html, "noscript", parse_noscript_content_as_text: true, max_errors: 100)
assert_empty(frag.errors)
assert_equal(1, frag.children.length)
assert_kind_of(Nokogiri::XML::Text, frag.children.first)
end
def test_parse_noscript_content_default
html = ""
doc = Nokogiri::HTML5(html, max_errors: 100)
assert_empty(doc.errors)
img = doc.at("/html/body/noscript/img")
refute_nil(img)
end
["pre", "listing", "textarea"].each do |tag|
define_method("test_serialize_preserve_newline_#{tag}".to_sym) do
doc = Nokogiri::HTML5("<#{tag}>\n\nContent")
html = doc.at("/html/body/#{tag}").serialize(preserve_newline: true)
assert_equal "<#{tag}>\n\nContent", html
end
define_method("test_inner_html_preserve_newline_#{tag}".to_sym) do
doc = Nokogiri::HTML5("<#{tag}>\n\nContent")
html = doc.at("/html/body/#{tag}").inner_html(preserve_newline: true)
assert_equal "\n\nContent", html
end
end
def test_document_io
html = StringIO.new("test", "r")
doc = Nokogiri::HTML5::Document.read_io(html)
refute_nil(doc.at_xpath("/html/body/span"))
end
def test_document_memory
html = "test"
doc = Nokogiri::HTML5::Document.read_memory(html)
refute_nil(doc)
refute_nil(doc.at_xpath("/html/body/span"))
end
def test_document_io_failure
html = "test"
assert_raises(ArgumentError) { Nokogiri::HTML5::Document.read_io(html) }
end
def test_document_memory_failure
html = StringIO.new("test", "r")
assert_raises(ArgumentError) { Nokogiri::HTML5::Document.read_memory(html) }
end
def test_document_parse_failure
html = ["Neither a string, nor I/O"]
assert_raises(ArgumentError) { Nokogiri::HTML5::Document.parse(html) }
end
def test_ownership
# Test that we don't change the passed in string, even if we need to
# re-encode it.
html = ""
refute_nil(Nokogiri::HTML5.parse(html))
iso8859_1 = html.encode(Encoding::ISO_8859_1).freeze
refute_nil(Nokogiri::HTML5.parse(iso8859_1))
ascii_8bit = html.encode(Encoding::ASCII_8BIT).freeze
refute_nil(Nokogiri::HTML5.parse(ascii_8bit))
end
def test_fragment_from_node
doc = Nokogiri.HTML5("")
span = doc.at_xpath("/html/body/form/span")
refute_nil(span)
frag = span.fragment("")
assert_kind_of(Nokogiri::HTML5::DocumentFragment, frag)
assert_equal(1, frag.children.length)
nested_form = frag.at_xpath("form")
assert_nil(nested_form)
assert_predicate(frag.children[0], :text?)
end
def test_fragment_from_node_no_form
doc = Nokogiri.HTML5("")
span = doc.at_xpath("/html/body/span")
refute_nil(span)
frag = span.fragment("")
assert_kind_of(Nokogiri::HTML5::DocumentFragment, frag)
assert_equal(1, frag.children.length)
form = frag.at_xpath("form")
refute_nil(form)
end
def test_empty_fragment
doc = Nokogiri.HTML5("")
frag = doc.fragment
assert_kind_of(Nokogiri::HTML5::DocumentFragment, frag)
assert_empty(frag.children)
end
def test_fragment_with_annotation_xml_context
# An annotation-xml element with an encoding of text/html or application/xhtml+xml
# is an HTML integration point so its children are not in the MathML namespace.
# Other encodings are not HTML integration points so children are in the MathML namespace.
doc = Nokogiri.HTML5("")
a_xml = doc.xpath("//math:annotation-xml")[0]
frag = a_xml.fragment("x")
mi = frag.children[0]
assert_equal("mi", mi.name)
refute_nil(mi.namespace)
assert_equal("math", mi.namespace.prefix)
doc = Nokogiri.HTML5("")
a_xml = doc.xpath("//math:annotation-xml")[0]
frag = a_xml.fragment("x")
mi = frag.children[0]
assert_equal("mi", mi.name)
assert_nil(mi.namespace)
end
def test_html_eh
doc = Nokogiri.HTML5("")
assert_predicate(doc, :html?)
refute_predicate(doc, :xml?)
end
def test_node_wrap
doc = Nokogiri.HTML5("")
div = doc.at_css("div")
div.wrap("")
assert_equal("section", div.parent.name)
assert_equal("body", div.parent.parent.name)
end
def test_node_wrap_uses_parent_node_as_parsing_context_node
doc = Nokogiri.HTML5("")
el = doc.at_css("option")
# fails to parse because `div` is not valid in the context of a `select` element
exception = assert_raises(RuntimeError) { el.wrap("") }
assert_match(/Failed to parse .* in the context of a 'select' element/, exception.message)
# parses because `optgroup` is valid in the context of a `select` element
el.wrap("")
assert_equal("optgroup", el.parent.name)
assert_equal("select", el.parent.parent.name)
end
def test_parse_in_context_of_foreign_namespace
# https://github.com/sparklemotion/nokogiri/issues/3112
# https://gitlab.gnome.org/GNOME/libxml2/-/issues/672
# released upstream in v2.12.5
skip if Nokogiri.uses_libxml?(["~> 2.12.0", "< 2.12.5"])
doc = Nokogiri::HTML5::Document.parse("