T2: NameNormalizer testy + transliteratePolish, token-strip titles [checkpoint]

2026-07-07 23:49:04 +02:00
parent 3914669c14
commit afe107cea0
19 changed files with 1649 additions and 27 deletions
@@ -0,0 +1,132 @@
+package com.developx.szpitale.ingest.normalize;
+
+import org.junit.jupiter.api.Test;
+import org.junit.jupiter.params.ParameterizedTest;
+import org.junit.jupiter.params.provider.CsvSource;
+
+import java.util.List;
+
+import static org.junit.jupiter.api.Assertions.*;
+
+class NameNormalizerTest {
+
+    private final NameNormalizer normalizer = new NameNormalizer();
+
+    @Test
+    void normalize_polishDiacritics_strippedToAsciiSlug() {
+        String slug = normalizer.makeSlug("Łukasz Żółć");
+        assertEquals("lukasz-zolc", slug);
+    }
+
+    @Test
+    void normalize_slug_lowercaseSeparator() {
+        String slug = normalizer.makeSlug("Jan Kowalski");
+        assertEquals("jan-kowalski", slug);
+    }
+
+    @Test
+    void normalize_slug_multipleSpacesToOne() {
+        String slug = normalizer.makeSlug("Jan    Kowalski");
+        assertEquals("jan-kowalski", slug);
+    }
+
+    @Test
+    void normalize_slug_removesSpecialChars() {
+        String slug = normalizer.makeSlug("Dr. Jan Kowalski Jr.");
+        assertEquals("dr-jan-kowalski-jr", slug);
+    }
+
+    @ParameterizedTest
+    @CsvSource({
+        "'prof. dr hab. Jan Kochanowicz', 'Jan Kochanowicz'",
+        "'dr Jan Nowak', 'Jan Nowak'",
+        "'mgr inż. Anna Kowalska', 'Anna Kowalska'",
+        "'Jan Wiśniewski', 'Jan Wiśniewski'",
+        "'prof. Wojciech Łuczaj', 'Wojciech Łuczaj'"
+    })
+    void normalize_displayName_keepsOriginal(String rawName, String expectedFullName) {
+        String fullName = normalizer.extractFullName(rawName);
+        assertEquals(expectedFullName, fullName);
+    }
+
+    @ParameterizedTest
+    @CsvSource({
+        "'prof. dr hab. Jan Kochanowicz', 'prof.'",
+        "'prof. dr hab. Jan Kochanowicz', 'dr hab.'",
+        "'dr hab. n. med. Jan Wiśniewski', 'dr hab.'",
+        "'dr hab. n. med. Jan Wiśniewski', 'n. med.'",
+        "'mgr inż. Anna Kowalska', 'mgr inż.'"
+    })
+    void normalize_titlesContainExpected(String rawName, String expectedTitle) {
+        List<String> titles = normalizer.extractTitles(rawName);
+        assertTrue(titles.contains(expectedTitle), "Expected title '" + expectedTitle + "' not found in " + titles);
+    }
+
+    @Test
+    void normalize_titlesEmptyForPlainName() {
+        List<String> titles = normalizer.extractTitles("Jan Nowak");
+        assertTrue(titles.isEmpty());
+    }
+
+    @Test
+    void normalize_emptyTitlesWhenNoTitles() {
+        List<String> titles = normalizer.extractTitles("Jan Kowalski");
+        assertTrue(titles.isEmpty());
+    }
+
+    @Test
+    void makePersonId_createsPersonSlug() {
+        String id = normalizer.makePersonId("Jan Kowalski");
+        assertEquals("person:jan-kowalski", id);
+    }
+
+    @Test
+    void makePerson_idUsesSameSlugAsMakeSlug() {
+        String id = normalizer.makePersonId("Łukasz Żółć");
+        assertEquals("person:lukasz-zolc", id);
+    }
+
+    @Test
+    void makeHospitalId_createsHospitalSlug() {
+        String id = normalizer.makeHospitalId("podlaskie", "Uniwersytecki Szpital Kliniczny");
+        assertEquals("hosp:podlaskie:uniwersytecki-szpital-kliniczny", id);
+    }
+
+    @Test
+    void fuzzySimilarity_identicalStringsReturn1() {
+        double similarity = normalizer.fuzzySimilarity("Jan Kowalski", "Jan Kowalski");
+        assertEquals(1.0, similarity, 0.0);
+    }
+
+    @Test
+    void fuzzySimilarity_typoReturnsHigh() {
+        double similarity = normalizer.fuzzySimilarity("Jan Kowalski", "Jan Kowaski");
+        assertTrue(similarity > 0.9, "Typo 'Kowalski' vs 'Kowaski' should have very high similarity: " + similarity);
+    }
+
+    @Test
+    void fuzzySimilarity_differentNamesReturnLower() {
+        double similarity = normalizer.fuzzySimilarity("Jan Kowalski", "Anna Nowak");
+        assertTrue(similarity < 0.7, "Two completely different names should have moderate similarity: " + similarity);
+    }
+
+    @Test
+    void fuzzySimilarity_handlesPolishCharacters() {
+        double similarity = normalizer.fuzzySimilarity("Jan Kowalski", "Jan Kowalscy");
+        assertTrue(similarity > 0.8, "Names differing only by suffix should have high similarity: " + similarity);
+    }
+
+    @Test
+    void findDuplicateCandidates_returnsCandidatesAboveThreshold() {
+        List<String> names = List.of("Jan Kowalski", "Jan Kowaski", "Anna Nowak", "Jan Nowak");
+        List<String> candidates = normalizer.findDuplicateCandidates(names, "Jan Kowalski", 0.7);
+        assertTrue(candidates.contains("Jan Kowaski"), "Should contain similar name");
+    }
+
+    @Test
+    void findDuplicateCandidates_filtersTargetName() {
+        List<String> names = List.of("Jan Kowalski");
+        List<String> candidates = normalizer.findDuplicateCandidates(names, "Jan Kowalski", 0.7);
+        assertTrue(candidates.isEmpty(), "Should not return the target name itself");
+    }
+}