"""Tests for EPO OPS XML biblio and search response parsers.""" from __future__ import annotations from scholar_mcp._epo_xml import ( parse_biblio_xml, parse_citations_from_biblio, parse_claims_xml, parse_description_xml, parse_family_xml, parse_legal_xml, parse_search_xml, ) # --------------------------------------------------------------------------- # Fixtures — inline XML bytes # --------------------------------------------------------------------------- BIBLIO_XML_FULL = b""" EP 1234567 A1 20200115 EP 18123456 A 20181201 US 201762123456 P 20171105 ACME Corporation Beta Inc Smith, John Doe, Jane A Method for Testing XML Parsers Ein Verfahren zum Testen von XML-Parsern

This is the English abstract describing the invention in detail.

Dies ist die deutsche Zusammenfassung der Erfindung.
H
04 L 29 06 I
G
06 F 21 60 I """ BIBLIO_XML_NON_ENGLISH_ONLY = b""" DE 102020123456 A1 20210301 DE 102020123456 A 20200601 DE 102019123456 A 20190915 Firma GmbH Mueller, Hans Nur Deutsches Patent

Nur eine deutsche Zusammenfassung.

 """ BIBLIO_XML_MINIMAL = b""" WO 2023123456 A1 20230601 WO 2023US12345 A 20230101 """ SEARCH_XML_FULL = b""" EP 1234567 A1 WO 2020123456 A1 US 10234567 B2 """ SEARCH_XML_EMPTY = b""" """ SEARCH_XML_SINGLE = b""" EP 9876543 B1 """ # --------------------------------------------------------------------------- # Tests for parse_biblio_xml # --------------------------------------------------------------------------- class TestParseBiblioXmlBasicFields: def test_publication_number(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert result["publication_number"] == "EP.1234567.A1" def test_publication_date(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert result["publication_date"] == "2020-01-15" def test_filing_date(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert result["filing_date"] == "2018-12-01" def test_priority_date(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert result["priority_date"] == "2017-11-05" def test_family_id(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert result["family_id"] == "54321" def test_url_contains_publication_number(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert result["url"] == ( "https://worldwide.espacenet.com/patent/search/family/54321/publication/EP1234567A1" ) class TestParseBiblioXmlEnglishPreference: def test_title_prefers_english(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert result["title"] == "A Method for Testing XML Parsers" def test_abstract_prefers_english(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert "English abstract" in result["abstract"] def test_title_fallback_to_non_english(self) -> None: result = parse_biblio_xml(BIBLIO_XML_NON_ENGLISH_ONLY) assert result["title"] == "Nur Deutsches Patent" def test_abstract_fallback_to_non_english(self) -> None: result = parse_biblio_xml(BIBLIO_XML_NON_ENGLISH_ONLY) assert "deutsche Zusammenfassung" in result["abstract"] class TestParseBiblioXmlApplicantsInventors: def test_applicants_list(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert result["applicants"] == ["ACME Corporation", "Beta Inc"] def test_inventors_list(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert result["inventors"] == ["Smith, John", "Doe, Jane"] def test_single_applicant(self) -> None: result = parse_biblio_xml(BIBLIO_XML_NON_ENGLISH_ONLY) assert result["applicants"] == ["Firma GmbH"] def test_empty_applicants(self) -> None: result = parse_biblio_xml(BIBLIO_XML_MINIMAL) assert result["applicants"] == [] def test_empty_inventors(self) -> None: result = parse_biblio_xml(BIBLIO_XML_MINIMAL) assert result["inventors"] == [] class TestParseBiblioXmlClassifications: def test_classifications_list(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert "H04L29/06" in result["classifications"] assert "G06F21/60" in result["classifications"] def test_classifications_count(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert len(result["classifications"]) == 2 def test_empty_classifications(self) -> None: result = parse_biblio_xml(BIBLIO_XML_MINIMAL) assert result["classifications"] == [] class TestParseBiblioXmlMinimal: def test_minimal_no_priority_claims(self) -> None: result = parse_biblio_xml(BIBLIO_XML_MINIMAL) assert result["priority_date"] == "" def test_minimal_no_title(self) -> None: result = parse_biblio_xml(BIBLIO_XML_MINIMAL) assert result["title"] == "" def test_minimal_no_abstract(self) -> None: result = parse_biblio_xml(BIBLIO_XML_MINIMAL) assert result["abstract"] == "" def test_minimal_publication_number(self) -> None: result = parse_biblio_xml(BIBLIO_XML_MINIMAL) assert result["publication_number"] == "WO.2023123456.A1" def test_minimal_family_id(self) -> None: result = parse_biblio_xml(BIBLIO_XML_MINIMAL) assert result["family_id"] == "77777" class TestParseBiblioXmlOutputKeys: def test_all_expected_keys_present(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) expected_keys = { "title", "abstract", "applicants", "inventors", "publication_number", "publication_date", "filing_date", "priority_date", "family_id", "classifications", "url", } assert set(result.keys()) == expected_keys def test_applicants_is_list(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert isinstance(result["applicants"], list) def test_inventors_is_list(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert isinstance(result["inventors"], list) def test_classifications_is_list(self) -> None: result = parse_biblio_xml(BIBLIO_XML_FULL) assert isinstance(result["classifications"], list) # --------------------------------------------------------------------------- # Tests for parse_search_xml # --------------------------------------------------------------------------- class TestParseSearchXmlCount: def test_total_count(self) -> None: result = parse_search_xml(SEARCH_XML_FULL) assert result["total_count"] == 42 def test_total_count_empty(self) -> None: result = parse_search_xml(SEARCH_XML_EMPTY) assert result["total_count"] == 0 def test_total_count_single(self) -> None: result = parse_search_xml(SEARCH_XML_SINGLE) assert result["total_count"] == 1 class TestParseSearchXmlReferences: def test_references_count(self) -> None: result = parse_search_xml(SEARCH_XML_FULL) assert len(result["references"]) == 3 def test_first_reference_ep(self) -> None: result = parse_search_xml(SEARCH_XML_FULL) ref = result["references"][0] assert ref["country"] == "EP" assert ref["number"] == "1234567" assert ref["kind"] == "A1" def test_second_reference_wo(self) -> None: result = parse_search_xml(SEARCH_XML_FULL) ref = result["references"][1] assert ref["country"] == "WO" assert ref["number"] == "2020123456" assert ref["kind"] == "A1" def test_third_reference_us(self) -> None: result = parse_search_xml(SEARCH_XML_FULL) ref = result["references"][2] assert ref["country"] == "US" assert ref["number"] == "10234567" assert ref["kind"] == "B2" def test_empty_results(self) -> None: result = parse_search_xml(SEARCH_XML_EMPTY) assert result["references"] == [] def test_single_result(self) -> None: result = parse_search_xml(SEARCH_XML_SINGLE) assert len(result["references"]) == 1 ref = result["references"][0] assert ref["country"] == "EP" assert ref["number"] == "9876543" assert ref["kind"] == "B1" class TestParseSearchXmlOutputKeys: def test_all_expected_keys_present(self) -> None: result = parse_search_xml(SEARCH_XML_FULL) assert set(result.keys()) == {"total_count", "references"} def test_references_is_list(self) -> None: result = parse_search_xml(SEARCH_XML_FULL) assert isinstance(result["references"], list) def test_reference_keys(self) -> None: result = parse_search_xml(SEARCH_XML_FULL) ref = result["references"][0] assert set(ref.keys()) == {"country", "number", "kind"} # --------------------------------------------------------------------------- # Fixtures for claims and description parsers # --------------------------------------------------------------------------- CLAIMS_XML = b"""\ 1. A method for processing widgets comprising: a) receiving input data; b) transforming the input data. 2. The method of claim 1, wherein the input data is digital. 1. Ein Verfahren zur Verarbeitung... """ DESCRIPTION_XML = b"""\

FIELD OF THE INVENTION

The present invention relates to widget processing.

BACKGROUND

Prior art widgets have limitations.

 """ DESCRIPTION_MULTILANG_XML = b"""\

GEBIET DER ERFINDUNG

Die vorliegende Erfindung betrifft die Widget-Verarbeitung.

FIELD OF THE INVENTION

The present invention relates to widget processing.

 """ DESCRIPTION_FALLBACK_XML = b"""\

GEBIET DER ERFINDUNG

 """ # --------------------------------------------------------------------------- # Tests for parse_claims_xml # --------------------------------------------------------------------------- class TestParseClaimsXml: def test_english_preferred(self) -> None: result = parse_claims_xml(CLAIMS_XML) assert "processing widgets" in result assert "Verfahren" not in result def test_multiple_claims(self) -> None: result = parse_claims_xml(CLAIMS_XML) assert "1." in result assert "2." in result def test_empty_xml(self) -> None: empty = b'' result = parse_claims_xml(empty) assert result == "" # --------------------------------------------------------------------------- # Tests for parse_description_xml # --------------------------------------------------------------------------- class TestParseDescriptionXml: def test_paragraphs_joined(self) -> None: result = parse_description_xml(DESCRIPTION_XML) assert "FIELD OF THE INVENTION" in result assert "widget processing" in result def test_not_empty(self) -> None: result = parse_description_xml(DESCRIPTION_XML) assert len(result) > 50 def test_english_preferred(self) -> None: result = parse_description_xml(DESCRIPTION_MULTILANG_XML) assert "widget processing" in result assert "Erfindung" not in result def test_fallback_to_non_english(self) -> None: result = parse_description_xml(DESCRIPTION_FALLBACK_XML) assert "GEBIET DER ERFINDUNG" in result def test_empty_xml(self) -> None: empty = b'' result = parse_description_xml(empty) assert result == "" # --------------------------------------------------------------------------- # Fixtures for family and legal parsers # --------------------------------------------------------------------------- FAMILY_XML = b"""\ EP 1234567 A1 20200115 US 11234567 B2 20210301 """ LEGAL_XML = b"""\ 20190501 APPLICATION Application filed 20200115 PUBLICATION Publication of application """ # --------------------------------------------------------------------------- # Tests for parse_family_xml # --------------------------------------------------------------------------- class TestParseFamilyXml: def test_family_members(self) -> None: result = parse_family_xml(FAMILY_XML) assert len(result) == 2 assert result[0]["country"] == "EP" assert result[1]["country"] == "US" def test_family_member_fields(self) -> None: result = parse_family_xml(FAMILY_XML) ep = result[0] assert ep["number"] == "1234567" assert ep["kind"] == "A1" assert ep["date"] == "2020-01-15" def test_empty_xml(self) -> None: empty = b'' result = parse_family_xml(empty) assert result == [] # --------------------------------------------------------------------------- # Tests for parse_legal_xml # --------------------------------------------------------------------------- class TestParseLegalXml: def test_legal_events(self) -> None: result = parse_legal_xml(LEGAL_XML) assert len(result) == 2 def test_event_fields(self) -> None: result = parse_legal_xml(LEGAL_XML) assert result[0]["date"] == "2019-05-01" assert result[0]["code"] == "APPLICATION" assert "filed" in result[0]["description"] def test_empty_xml(self) -> None: empty = b'' result = parse_legal_xml(empty) assert result == [] # --------------------------------------------------------------------------- # Fixtures for citations parser # --------------------------------------------------------------------------- BIBLIO_WITH_CITATIONS_XML = b"""\ US 9876543 B2 Smith et al., "Widget Processing", Journal of Widgets, 2018, doi:10.1234/widgets.2018 Doe, J., "Advanced Widgets", Conference on Widgets, 2019 """ # --------------------------------------------------------------------------- # Tests for parse_citations_from_biblio # --------------------------------------------------------------------------- class TestParseCitationsFromBiblio: def test_patent_citations(self) -> None: result = parse_citations_from_biblio(BIBLIO_WITH_CITATIONS_XML) assert len(result["patent_refs"]) == 1 assert result["patent_refs"][0]["country"] == "US" assert result["patent_refs"][0]["number"] == "9876543" assert result["patent_refs"][0]["kind"] == "B2" def test_npl_citations(self) -> None: result = parse_citations_from_biblio(BIBLIO_WITH_CITATIONS_XML) assert len(result["npl_refs"]) == 2 assert "Widget Processing" in result["npl_refs"][0]["raw"] def test_doi_extraction(self) -> None: result = parse_citations_from_biblio(BIBLIO_WITH_CITATIONS_XML) assert result["npl_refs"][0]["doi"] == "10.1234/widgets.2018" assert result["npl_refs"][1]["doi"] is None def test_no_citations(self) -> None: xml = b"""\ """ result = parse_citations_from_biblio(xml) assert result["patent_refs"] == [] assert result["npl_refs"] == [] def test_empty_xml(self) -> None: empty = b'' result = parse_citations_from_biblio(empty) assert result == {"patent_refs": [], "npl_refs": []}