justinzeus/scholarr
1
0
Fork 0
Code Issues 1 Pull requests 18 Projects Releases Packages Wiki Activity Actions

m

Browse source
This commit is contained in:
Justin Visser 2026-02-20 23:29:44 +01:00
parent 01454162bb
commit 4240ad38e2
2 changed files with 39 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

12
app/services/domains/scholar/profile_rows.py
View file

@ -106,15 +106,13 @@ def parse_cluster_id_from_href(href: str | None) -> str | None:
if citation_for_view: if citation_for_view:
token = citation_for_view[0].strip() token = citation_for_view[0].strip()
if token: if token:
if ":" in token: return f"cfv:{token}"
return token.rsplit(":", 1)[-1] or None
return token
cluster = query.get("cluster") cluster = query.get("cluster")
if cluster: if cluster:
token = cluster[0].strip() token = cluster[0].strip()
if token: if token:
return token return f"cluster:{token}"
return None return None
@ -133,10 +131,10 @@ def parse_citation_count(parts: list[str]) -> int | None:
text = normalize_space(" ".join(parts)) text = normalize_space(" ".join(parts))
if not text: if not text:
return 0 return 0
match = re.search(r"\d+", text) digits = re.sub(r"\D+", "", text)
if not match: if not digits:
return None return None
return int(match.group(0)) return int(digits)
def _parse_publication_row(row_html: str) -> tuple[PublicationCandidate | None, list[str]]: def _parse_publication_row(row_html: str) -> tuple[PublicationCandidate | None, list[str]]:

34
tests/unit/test_scholar_parser.py
View file

@ -97,10 +97,44 @@ def test_parse_profile_page_handles_missing_optional_metadata() -> None:
assert parsed.state_reason == "publications_extracted" assert parsed.state_reason == "publications_extracted"
assert len(parsed.publications) == 1 assert len(parsed.publications) == 1
publication = parsed.publications[0] publication = parsed.publications[0]
assert publication.cluster_id == "cfv:abc:def123"
assert publication.year is None assert publication.year is None
assert publication.venue_text is None assert publication.venue_text is None
def test_parse_profile_page_parses_comma_formatted_citation_counts() -> None:
html = """
<html>
<div id="gsc_prf_in">Citation Formatting Test</div>
<span id="gsc_a_nn">Articles 1-1</span>
<table>
<tbody id="gsc_a_b">
<tr class="gsc_a_tr">
<td class="gsc_a_t">
<a class="gsc_a_at" href="/citations?view_op=view_citation&citation_for_view=abc:def123">Paper</a>
</td>
<td class="gsc_a_c"><a class="gsc_a_ac">Cited by 1,234</a></td>
<td class="gsc_a_y"><span class="gsc_a_h">2024</span></td>
</tr>
</tbody>
</table>
</html>
"""
fetch_result = FetchResult(
requested_url="https://scholar.google.com/citations?hl=en&user=abcDEF123456",
status_code=200,
final_url="https://scholar.google.com/citations?hl=en&user=abcDEF123456",
body=html,
error=None,
)
parsed = parse_profile_page(fetch_result)
assert parsed.state == ParseState.OK
assert len(parsed.publications) == 1
assert parsed.publications[0].citation_count == 1234
def test_parse_profile_page_ignores_direct_pdf_link_markup() -> None: def test_parse_profile_page_ignores_direct_pdf_link_markup() -> None:
html = """ html = """
<html> <html>