def test_extract_article_full_page_content_digest_node_indexes():
check_extract_article(
"addictinginfo.com-1_full_page.html",
"addictinginfo.com-1_simple_article_from_full_page_content_digest_node_indexes.json",
content_digests=True,
node_indexes=True)
def test_extract_article_list_items_node_indexes():
check_extract_article(
"list_items_full_page.html",
"list_items_simple_article_from_full_page_node_indexes.json",
node_indexes=True)
def test_extract_article_full_page():
check_extract_article(
"addictinginfo.com-1_full_page.html",
"addictinginfo.com-1_simple_article_from_full_page.json")
def test_extract_article_headers_and_non_paragraph_blockquote_text():
check_extract_article(
"davidwolfe.com-1_full_page.html",
"davidwolfe.com-1_simple_article_from_full_page.json")
def test_extract_article_list_items_content_digests():
check_extract_article(
"list_items_full_page.html",
"list_items_simple_article_from_full_page_content_digests.json",
content_digests=True)
def test_extract_article_list_items():
check_extract_article("list_items_full_page.html",
"list_items_simple_article_from_full_page.json")
def test_extract_article_unicode_normalisation():
check_extract_article(
"conservativehq.com-1_full_page.html",
"conservativehq.com-1_simple_article_from_full_page.json")
def test_extract_article_non_article():
check_extract_article("non_article_full_page.html",
"non_article_full_page.json")
def test_extract_simple_article_with_readability_js():
check_extract_article(
"plain-content-test_full_article.html",
"plain-content-test_full_article_javascript.json",
use_readability_js=True
)
def test_extract_article_from_page_with_readability_js():
check_extract_article(
"addictinginfo.com-1_full_page.html",
"addictinginfo.com-1_full_page_javascript.json",
use_readability_js=True
)
