from pygea import pangeaservice
ARTICLE_WITH_TWEET_SNIPPETS = """
Lead paragraph.
Middle paragraph.
Tail paragraph.
""".strip()
SNIPPET_PAGE_58108 = """
""".strip()
SNIPPET_PAGE_58109 = """
""".strip()
def test_snippet_urls_from_article_content_finds_supported_infographics_snippets() -> (
None
):
urls = pangeaservice.snippet_urls_from_article_content(
ARTICLE_WITH_TWEET_SNIPPETS, "www.martinoticias.com"
)
assert urls == [
"https://www.martinoticias.com/a/29036.html?parameterid=58108",
"https://www.martinoticias.com/a/29036.html?parameterid=58109",
]
def test_snippet_urls_from_article_content_ignores_malformed_or_unsupported_snippets() -> (
None
):
raw_html = """
No supported snippet target is present here.
""".strip()
urls = pangeaservice.snippet_urls_from_article_content(
raw_html, "www.martinoticias.com"
)
assert urls == []
def test_extract_embed_html_from_snippet_page_returns_decoded_blockquote() -> None:
embed_html = pangeaservice.extract_embed_html_from_snippet_page(SNIPPET_PAGE_58108)
assert embed_html is not None
assert " None:
embed_html = pangeaservice.extract_embed_html_from_snippet_page(
""
)
assert embed_html is None
def test_resolve_content_snippets_replaces_supported_tweet_placeholders() -> None:
def fetch_html(url: str) -> str | None:
pages = {
"https://www.martinoticias.com/a/29036.html?parameterid=58108": SNIPPET_PAGE_58108,
"https://www.martinoticias.com/a/29036.html?parameterid=58109": SNIPPET_PAGE_58109,
}
return pages.get(url)
resolved = pangeaservice.resolve_content_snippets(
ARTICLE_WITH_TWEET_SNIPPETS, "www.martinoticias.com", fetch_html
)
assert 'class="twitter-tweet"' in resolved
assert "@HenryAlviarez" in resolved
assert "@MariaCorinaYA" in resolved
assert "?parameterid=58108" not in resolved
assert "?parameterid=58109" not in resolved
assert "Lead paragraph." in resolved
assert "Tail paragraph." in resolved
def test_resolve_content_snippets_leaves_placeholder_when_resolution_fails() -> None:
resolved = pangeaservice.resolve_content_snippets(
ARTICLE_WITH_TWEET_SNIPPETS,
"www.martinoticias.com",
lambda _url: None,
)
assert 'mode="infographics|plain|29036|large||"' in resolved
assert "?parameterid=58108" in resolved
assert "?parameterid=58109" in resolved
assert 'class="twitter-tweet"' not in resolved
def test_rss_article_from_pangea_article_resolves_supported_snippets(
monkeypatch,
) -> None:
service = object.__new__(pangeaservice.PangeaService)
service._verbose_p = False
service._domain = "www.martinoticias.com"
service._rev_categories = {}
monkeypatch.setattr(
pangeaservice.utilities,
"get_media_metadata",
lambda _url: None,
)
monkeypatch.setattr(
service,
"_fetch_snippet_page_html",
lambda url: {
"https://www.martinoticias.com/a/29036.html?parameterid=58108": SNIPPET_PAGE_58108,
"https://www.martinoticias.com/a/29036.html?parameterid=58109": SNIPPET_PAGE_58109,
}.get(url),
)
rss_article = service.rss_article_from_pangea_article(
{
"url": "https://www.martinoticias.com/a/reabre-sede-del-partido-de-maria-corina-machado-en-caracas/453944.html",
"title": "Reabre sede del partido de María Corina Machado en Caracas (VIDEO)",
"content": ARTICLE_WITH_TWEET_SNIPPETS,
"pubDate": "2026-03-29T16:29:38",
}
)
assert 'class="twitter-tweet"' in rss_article["content"]
assert "@HenryAlviarez" in rss_article["content"]
assert "@MariaCorinaYA" in rss_article["content"]