def query_html_doc(html_body, hquery, preserve_space=False, wrap_body=True):
soup = soup_with_body(html_body) if wrap_body else make_soup(html_body)
raw_result = HqueryProcessor(hquery,
preserve_space=preserve_space).query(soup)
return eliminate_blank_lines(
convert_results_to_output_text(raw_result,
preserve_space=preserve_space).strip())
def test_selects_the_para_element_descendants_of_the_context_node():
html = """
<para>
<context>
<para>selected</para>
<not-para>not selected</not-para>
<para>
<para>also selected</para>
</para>
</context>
</para>"""
soup = make_soup(html)
assert query_context_node(soup.para.context,
'.//para') == expected_result("""
<para>
selected
</para>
<para>
<para>
also selected
</para>
</para>
<para>
also selected
</para>""")
def test_selects_the_lang_attribute_of_the_parent_of_the_context_node():
html = """
<root lang="English">
<context></context>
</root>"""
soup = make_soup(html)
assert query_context_node(soup.root.context, '../@lang') == expected_result('lang="English"')
def test_selects_all_the_item_elements_in_the_same_document_as_the_context_node_that_have_an_olist_parent(
):
html = """
<root>
<olist>no items</olist>
<item>not selected</item>
<context></context>
<olist>
<item>first</item>
</olist>
<item>
<olist>
<item>second</item>
<olist>
</item>
</root>"""
soup = make_soup(html)
assert query_context_node(soup.root.context,
'//olist/item') == expected_result("""
<item>
first
</item>
<item>
second
</item>""")
def test_selects_the_div_ancestors_of_the_context_node_and_if_the_context_node_is_a_div_element_the_context_node_as_well(
):
html = """
<div>
<div/>
<notdiv/>
</div>"""
soup = make_soup(html)
assert query_context_node(soup.div.div,
'ancestor-or-self::div') == expected_result("""
<div>
<div>
</div>
<notdiv>
</notdiv>
</div>
<div>
</div>""")
assert query_context_node(soup.div.notdiv,
'ancestor-or-self::div') == expected_result("""
<div>
<div>
</div>
<notdiv>
</notdiv>
</div>""")
def test_selects_the_lang_attribute_of_the_parent_of_the_context_node():
html = """
<root lang="English">
<context></context>
</root>"""
soup = make_soup(html)
assert query_context_node(soup.root.context,
'../@lang') == expected_result('lang="English"')
def test_selects_all_div_ancestors_of_the_context_node():
html = """
<div>
<notdiv/>
</div>"""
assert query_context_node(make_soup(html).div.notdiv, 'ancestor::div') == expected_result("""
<div>
<notdiv>
</notdiv>
</div>""")
def test_name_test_at_root_ignores_all_but_root_element():
html = """
<!DOCTYPE html>
<!-- html -->
<html id="root">
</html>"""
raw_result = HqueryProcessor('/html').query(make_soup(html))
actual = convert_results_to_output_text(raw_result)
assert actual == expected_result("""
<html id="root">
</html>""")
def test_selects_all_the_para_elements_in_the_same_document_as_the_context_node():
html = """
<root>
<notpara/>
<para>selected</para>
</root>"""
soup = make_soup(html)
assert query_context_node(soup.root.notpara, '/descendant::para') == expected_result("""
<para>
selected
</para>""")
def test_selects_the_parent_of_the_context_node():
html = """
<root>
<context></context>
</root>"""
soup = make_soup(html)
assert query_context_node(html, '..') == expected_result("""
<root>
<context>
</context>
</root>""")
def test_name_test_at_root_ignores_all_but_root_element():
html = """
<!DOCTYPE html>
<!-- html -->
<html id="root">
</html>"""
raw_result = HqueryProcessor('/html').query(make_soup(html))
actual = convert_results_to_output_text(raw_result)
assert actual == expected_result("""
<html id="root">
</html>""")
def test_selects_the_parent_of_the_context_node():
html = """
<root>
<context></context>
</root>"""
soup = make_soup(html)
assert query_context_node(html, '..') == expected_result("""
<root>
<context>
</context>
</root>""")
def test_selects_all_div_ancestors_of_the_context_node():
html = """
<div>
<notdiv/>
</div>"""
assert query_context_node(make_soup(html).div.notdiv,
'ancestor::div') == expected_result("""
<div>
<notdiv>
</notdiv>
</div>""")
def test_selects_the_previous_chapter_sibling_of_the_context_node():
html = """
<root>
<chapter>not selected</chapter>
<chapter>selected</chapter>
<context/>
</root>"""
soup = make_soup(html)
assert query_context_node(soup.root.context, 'preceding-sibling::chapter[position()=1]') == expected_result("""
<chapter>
selected
</chapter>""")
def test_path_to_root_tag_succeeds_despite_other_root_level_objects():
html = """
<!DOCTYPE html>
<!-- outside -->
<html>
<!-- inside -->
</html>"""
raw_result = HqueryProcessor('/*').query(make_soup(html))
actual = convert_results_to_output_text(raw_result)
assert actual == expected_result("""
<html>
<!-- inside -->
</html>""")
def test_selects_all_the_para_elements_in_the_same_document_as_the_context_node(
):
html = """
<root>
<notpara/>
<para>selected</para>
</root>"""
soup = make_soup(html)
assert query_context_node(soup.root.notpara,
'/descendant::para') == expected_result("""
<para>
selected
</para>""")
def test_selects_the_previous_chapter_sibling_of_the_context_node():
html = """
<root>
<chapter>not selected</chapter>
<chapter>selected</chapter>
<context/>
</root>"""
soup = make_soup(html)
assert query_context_node(
soup.root.context,
'preceding-sibling::chapter[position()=1]') == expected_result("""
<chapter>
selected
</chapter>""")
def test_selects_all_the_item_elements_that_have_an_olist_parent_and_that_are_in_the_same_document_as_the_context_node():
html = """
<root>
<notolist/>
<olist>
<notitem>not selected</notitem>
<item>selected</item>
<olist>
</root>"""
soup = make_soup(html)
assert query_context_node(soup.root.notolist, '/descendant::olist/child::item') == expected_result("""
<item>
selected
</item>""")
def test_selects_all_the_item_elements_that_have_an_olist_parent_and_that_are_in_the_same_document_as_the_context_node(
):
html = """
<root>
<notolist/>
<olist>
<notitem>not selected</notitem>
<item>selected</item>
<olist>
</root>"""
soup = make_soup(html)
assert query_context_node(
soup.root.notolist,
'/descendant::olist/child::item') == expected_result("""
<item>
selected
</item>""")
def test_selects_all_the_para_descendants_of_the_document_root_and_thus_selects_all_para_elements_in_the_same_document_as_the_context_node():
html = """
<root>
<para>
<para>selected</para>
</para>
<context></context>
<para>also selected</para>
</root>"""
soup = make_soup(html)
assert query_context_node(soup.root.context, '//para') == expected_result("""
<para>
<para>
selected
</para>
</para>
<para>
selected
</para>
<para>
also selected
</para>""")
def test_selects_all_the_item_elements_in_the_same_document_as_the_context_node_that_have_an_olist_parent():
html = """
<root>
<olist>no items</olist>
<item>not selected</item>
<context></context>
<olist>
<item>first</item>
</olist>
<item>
<olist>
<item>second</item>
<olist>
</item>
</root>"""
soup = make_soup(html)
assert query_context_node(soup.root.context, '//olist/item') == expected_result("""
<item>
first
</item>
<item>
second
</item>""")
def test_selects_the_div_ancestors_of_the_context_node_and_if_the_context_node_is_a_div_element_the_context_node_as_well():
html = """
<div>
<div/>
<notdiv/>
</div>"""
soup = make_soup(html)
assert query_context_node(soup.div.div, 'ancestor-or-self::div') == expected_result("""
<div>
<div>
</div>
<notdiv>
</notdiv>
</div>
<div>
</div>""")
assert query_context_node(soup.div.notdiv, 'ancestor-or-self::div') == expected_result("""
<div>
<div>
</div>
<notdiv>
</notdiv>
</div>""")
def test_selects_all_the_para_descendants_of_the_document_root_and_thus_selects_all_para_elements_in_the_same_document_as_the_context_node(
):
html = """
<root>
<para>
<para>selected</para>
</para>
<context></context>
<para>also selected</para>
</root>"""
soup = make_soup(html)
assert query_context_node(soup.root.context,
'//para') == expected_result("""
<para>
<para>
selected
</para>
</para>
<para>
selected
</para>
<para>
also selected
</para>""")
def test_selects_the_para_element_descendants_of_the_context_node():
html = """
<para>
<context>
<para>selected</para>
<not-para>not selected</not-para>
<para>
<para>also selected</para>
</para>
</context>
</para>"""
soup = make_soup(html)
assert query_context_node(soup.para.context, './/para') == expected_result("""
<para>
selected
</para>
<para>
<para>
also selected
</para>
</para>
<para>
also selected
</para>""")
def soup_with_body(contents):
return make_soup(wrap_html_body(contents))
def query_html_doc(html_body, hquery, preserve_space=False, wrap_body=True):
soup = soup_with_body(html_body) if wrap_body else make_soup(html_body)
raw_result = HqueryProcessor(hquery, preserve_space=preserve_space).query(soup)
return eliminate_blank_lines(convert_results_to_output_text(raw_result, preserve_space=preserve_space).strip())
def soup_with_body(contents):
return make_soup(wrap_html_body(contents))
def query_context_node(node_or_source, hquery):
if not is_any_node(node_or_source):
node_or_source = root_tag_from_soup(make_soup(node_or_source))
raw_result = HqueryProcessor(hquery).query(node_or_source)
return eliminate_blank_lines(convert_results_to_output_text(raw_result).strip())
def query_context_node(node_or_source, hquery):
if not is_any_node(node_or_source):
node_or_source = root_tag_from_soup(make_soup(node_or_source))
raw_result = HqueryProcessor(hquery).query(node_or_source)
return eliminate_blank_lines(
convert_results_to_output_text(raw_result).strip())