diff options
author | Harald Musum <musum@vespa.ai> | 2023-11-13 21:34:45 +0100 |
---|---|---|
committer | GitHub <noreply@github.com> | 2023-11-13 21:34:45 +0100 |
commit | ef5be496bc4857c5923f566251dd527873b248bf (patch) | |
tree | 657d51a4166d3f7cf40e04f0a5972f11d0261afd /indexinglanguage | |
parent | 944d635d00e165166508ef23399e9ed65a87a9c8 (diff) |
Revert "Bratseth/casing take 2"
Diffstat (limited to 'indexinglanguage')
5 files changed, 68 insertions, 83 deletions
diff --git a/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/expressions/NGramExpression.java b/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/expressions/NGramExpression.java index fdfadf65400..26058eeb8f3 100644 --- a/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/expressions/NGramExpression.java +++ b/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/expressions/NGramExpression.java @@ -15,8 +15,6 @@ import com.yahoo.vespa.indexinglanguage.linguistics.LinguisticsAnnotator; import java.util.Iterator; -import static com.yahoo.language.LinguisticsCase.toLowerCase; - /** * A filter which splits incoming text into n-grams. * @@ -70,9 +68,8 @@ public final class NGramExpression extends Expression { // annotate gram as a word term String gramString = gram.extractFrom(output.getString()); - typedSpan(gram.getStart(), - gram.getCodePointCount(), - TokenType.ALPHABETIC, spanList).annotate(LinguisticsAnnotator.termAnnotation(toLowerCase(gramString), gramString)); + typedSpan(gram.getStart(), gram.getCodePointCount(), TokenType.ALPHABETIC, spanList). + annotate(LinguisticsAnnotator.lowerCaseTermAnnotation(gramString, gramString)); lastPosition = gram.getStart() + gram.getCodePointCount(); } diff --git a/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/linguistics/AnnotatorConfig.java b/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/linguistics/AnnotatorConfig.java index 5c1bf0813c4..684bae3bf97 100644 --- a/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/linguistics/AnnotatorConfig.java +++ b/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/linguistics/AnnotatorConfig.java @@ -13,7 +13,7 @@ public class AnnotatorConfig implements Cloneable { private Language language; private StemMode stemMode; private boolean removeAccents; - private int maxTermOccurrences; + private int maxTermOccurences; private int maxTokenizeLength; public static final int DEFAULT_MAX_TERM_OCCURRENCES; @@ -29,7 +29,7 @@ public class AnnotatorConfig implements Cloneable { language = Language.ENGLISH; stemMode = StemMode.NONE; removeAccents = false; - maxTermOccurrences = DEFAULT_MAX_TERM_OCCURRENCES; + maxTermOccurences = DEFAULT_MAX_TERM_OCCURRENCES; maxTokenizeLength = DEFAULT_MAX_TOKENIZE_LENGTH; } @@ -37,7 +37,7 @@ public class AnnotatorConfig implements Cloneable { language = rhs.language; stemMode = rhs.stemMode; removeAccents = rhs.removeAccents; - maxTermOccurrences = rhs.maxTermOccurrences; + maxTermOccurences = rhs.maxTermOccurences; maxTokenizeLength = rhs.maxTokenizeLength; } @@ -74,11 +74,11 @@ public class AnnotatorConfig implements Cloneable { } public int getMaxTermOccurrences() { - return maxTermOccurrences; + return maxTermOccurences; } public AnnotatorConfig setMaxTermOccurrences(int maxTermCount) { - this.maxTermOccurrences = maxTermCount; + this.maxTermOccurences = maxTermCount; return this; } @@ -109,7 +109,7 @@ public class AnnotatorConfig implements Cloneable { if (removeAccents != rhs.removeAccents) { return false; } - if (maxTermOccurrences != rhs.maxTermOccurrences) { + if (maxTermOccurences != rhs.maxTermOccurences) { return false; } if (maxTokenizeLength != rhs.maxTokenizeLength) { @@ -121,7 +121,6 @@ public class AnnotatorConfig implements Cloneable { @Override public int hashCode() { return getClass().hashCode() + language.hashCode() + stemMode.hashCode() + - Boolean.valueOf(removeAccents).hashCode() + maxTermOccurrences + maxTokenizeLength; + Boolean.valueOf(removeAccents).hashCode() + maxTermOccurences + maxTokenizeLength; } - } diff --git a/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/linguistics/LinguisticsAnnotator.java b/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/linguistics/LinguisticsAnnotator.java index 52cd8a8ff54..191d067effe 100644 --- a/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/linguistics/LinguisticsAnnotator.java +++ b/indexinglanguage/src/main/java/com/yahoo/vespa/indexinglanguage/linguistics/LinguisticsAnnotator.java @@ -34,8 +34,8 @@ public class LinguisticsAnnotator { final Map<String, Integer> termOccurrences = new HashMap<>(); final int maxOccurrences; - public TermOccurrences(int maxOccurrences) { - this.maxOccurrences = maxOccurrences; + public TermOccurrences(int maxOccurences) { + this.maxOccurrences = maxOccurences; } boolean termCountBelowLimit(String term) { @@ -86,23 +86,24 @@ public class LinguisticsAnnotator { } /** - * Creates a TERM annotation which has the term as annotation (only) if it is different from the + * Creates a TERM annotation which has the lowercase value as annotation (only) if it is different from the * original. * - * @param term the term - * @param origTerm the original term + * @param termToLowerCase the term to lower case + * @param origTerm the original term * @return the created TERM annotation */ - public static Annotation termAnnotation(String term, String origTerm) { - if (term.equals(origTerm)) + public static Annotation lowerCaseTermAnnotation(String termToLowerCase, String origTerm) { + String annotationValue = toLowerCase(termToLowerCase); + if (annotationValue.equals(origTerm)) { return new Annotation(AnnotationTypes.TERM); - else - return new Annotation(AnnotationTypes.TERM, new StringFieldValue(term)); + } + return new Annotation(AnnotationTypes.TERM, new StringFieldValue(annotationValue)); } private static void addAnnotation(Span here, String term, String orig, TermOccurrences termOccurrences) { if (termOccurrences.termCountBelowLimit(term)) { - here.annotate(termAnnotation(term, orig)); + here.annotate(lowerCaseTermAnnotation(term, orig)); } } @@ -126,24 +127,29 @@ public class LinguisticsAnnotator { } if (mode == StemMode.ALL) { Span where = parent.span((int)token.getOffset(), token.getOrig().length()); - String lowercasedOrig = toLowerCase(token.getOrig()); + addAnnotation(where, token.getOrig(), token.getOrig(), termOccurrences); + + String lowercasedTerm = lowercasedOrig; String term = token.getTokenString(); if (term != null) { + lowercasedTerm = toLowerCase(term); + } + if (! lowercasedOrig.equals(lowercasedTerm)) { addAnnotation(where, term, token.getOrig(), termOccurrences); - if ( ! term.equals(lowercasedOrig)) - addAnnotation(where, token.getOrig(), token.getOrig(), termOccurrences); } for (int i = 0; i < token.getNumStems(); i++) { String stem = token.getStem(i); - if (! (stem.equals(lowercasedOrig) || stem.equals(term))) + String lowercasedStem = toLowerCase(stem); + if (! (lowercasedOrig.equals(lowercasedStem) || lowercasedTerm.equals(lowercasedStem))) { addAnnotation(where, stem, token.getOrig(), termOccurrences); + } } } else { String term = token.getTokenString(); if (term == null || term.trim().isEmpty()) return; if (termOccurrences.termCountBelowLimit(term)) { - parent.span((int)token.getOffset(), token.getOrig().length()).annotate(termAnnotation(term, token.getOrig())); + parent.span((int)token.getOffset(), token.getOrig().length()).annotate(lowerCaseTermAnnotation(term, token.getOrig())); } } } diff --git a/indexinglanguage/src/test/java/com/yahoo/vespa/indexinglanguage/expressions/NGramTestCase.java b/indexinglanguage/src/test/java/com/yahoo/vespa/indexinglanguage/expressions/NGramTestCase.java index b4e266ab3eb..bcde8751de8 100644 --- a/indexinglanguage/src/test/java/com/yahoo/vespa/indexinglanguage/expressions/NGramTestCase.java +++ b/indexinglanguage/src/test/java/com/yahoo/vespa/indexinglanguage/expressions/NGramTestCase.java @@ -57,8 +57,8 @@ public class NGramTestCase { new NGramExpression(new SimpleLinguistics(), 3).execute(context); StringFieldValue value = (StringFieldValue)context.getValue(); - assertEquals("Grams are pure annotations - field value is unchanged", - "en gul Bille sang... ", value.getString()); + assertEquals("Grams are pure annotations - field value is unchanged", "en gul Bille sang... ", + value.getString()); SpanTree gramTree = value.getSpanTree(SpanTrees.LINGUISTICS); assertNotNull(gramTree); SpanList grams = (SpanList)gramTree.getRoot(); diff --git a/indexinglanguage/src/test/java/com/yahoo/vespa/indexinglanguage/linguistics/LinguisticsAnnotatorTestCase.java b/indexinglanguage/src/test/java/com/yahoo/vespa/indexinglanguage/linguistics/LinguisticsAnnotatorTestCase.java index a4dbe1fe826..67bff3843ee 100644 --- a/indexinglanguage/src/test/java/com/yahoo/vespa/indexinglanguage/linguistics/LinguisticsAnnotatorTestCase.java +++ b/indexinglanguage/src/test/java/com/yahoo/vespa/indexinglanguage/linguistics/LinguisticsAnnotatorTestCase.java @@ -19,6 +19,7 @@ import org.junit.Test; import org.mockito.Mockito; import java.util.*; +import java.util.stream.Collectors; import static org.junit.Assert.assertEquals; import static org.junit.Assert.assertTrue; @@ -28,6 +29,8 @@ import static org.junit.Assert.assertTrue; */ public class LinguisticsAnnotatorTestCase { + private static final AnnotatorConfig CONFIG = new AnnotatorConfig(); + @Test public void requireThatAnnotateFailsWithZeroTokens() { assertAnnotations(null, "foo"); @@ -39,7 +42,7 @@ public class LinguisticsAnnotatorTestCase { if (type.isIndexable()) { continue; } - assertAnnotations(null, "foo", token("foo", "bar", type)); + assertAnnotations(null, "foo", newToken("foo", "bar", type)); } } @@ -51,27 +54,7 @@ public class LinguisticsAnnotatorTestCase { if (!type.isIndexable()) { continue; } - assertAnnotations(expected, "foo", token("foo", "bar", type)); - } - } - - @Test - public void requireThatIndexableTokenStringsAreAnnotatedWithModeALL() { - SpanTree expected = new SpanTree(SpanTrees.LINGUISTICS); - expected.spanList().span(0, 5).annotate(new Annotation(AnnotationTypes.TERM, new StringFieldValue("tesla"))); - var span2 = expected.spanList().span(0, 4); - span2.annotate(new Annotation(AnnotationTypes.TERM)); - span2.annotate(new Annotation(AnnotationTypes.TERM, new StringFieldValue("car"))); - var span3 = expected.spanList().span(0, 8); - span3.annotate(new Annotation(AnnotationTypes.TERM, new StringFieldValue("modelxes"))); - span3.annotate(new Annotation(AnnotationTypes.TERM, new StringFieldValue("modelx"))); - span3.annotate(new Annotation(AnnotationTypes.TERM, new StringFieldValue("mex"))); - for (TokenType type : TokenType.values()) { - if (!type.isIndexable()) continue; - assertAnnotations(expected, "Tesla cars", new AnnotatorConfig().setStemMode("ALL"), - token("Tesla", "tesla", type), - token("cars", "car", type), - SimpleToken.fromStems("ModelXes", List.of("modelxes", "modelx", "mex"))); + assertAnnotations(expected, "foo", newToken("foo", "bar", type)); } } @@ -80,7 +63,7 @@ public class LinguisticsAnnotatorTestCase { SpanTree expected = new SpanTree(SpanTrees.LINGUISTICS); expected.spanList().span(0, 3).annotate(new Annotation(AnnotationTypes.TERM, new StringFieldValue("bar"))); for (TokenType type : TokenType.values()) { - assertAnnotations(expected, "foo", token("foo", "bar", type, true)); + assertAnnotations(expected, "foo", newToken("foo", "bar", type, true)); } } @@ -93,21 +76,21 @@ public class LinguisticsAnnotatorTestCase { if (!specialToken && !type.isIndexable()) { continue; } - assertAnnotations(expected, "foo", token("foo", "foo", type, specialToken)); + assertAnnotations(expected, "foo", newToken("foo", "foo", type, specialToken)); } } } @Test - public void requireThatTermAnnotationsPreserveCasing() { + public void requireThatTermAnnotationsAreLowerCased() { SpanTree expected = new SpanTree(SpanTrees.LINGUISTICS); - expected.spanList().span(0, 3).annotate(new Annotation(AnnotationTypes.TERM, new StringFieldValue("BaR"))); + expected.spanList().span(0, 3).annotate(new Annotation(AnnotationTypes.TERM, new StringFieldValue("bar"))); for (boolean specialToken : Arrays.asList(true, false)) { for (TokenType type : TokenType.values()) { if (!specialToken && !type.isIndexable()) { continue; } - assertAnnotations(expected, "foo", token("foo", "BaR", type, specialToken)); + assertAnnotations(expected, "foo", newToken("foo", "BAR", type, specialToken)); } } } @@ -119,11 +102,11 @@ public class LinguisticsAnnotatorTestCase { expected.spanList().span(3, 3).annotate(new Annotation(AnnotationTypes.TERM, new StringFieldValue("bar"))); expected.spanList().span(6, 3).annotate(new Annotation(AnnotationTypes.TERM, new StringFieldValue("baz"))); - SimpleToken token = token("FOOBARBAZ", "foobarbaz", TokenType.ALPHABETIC) - .addComponent(token("FOO", "foo", TokenType.ALPHABETIC).setOffset(0)) - .addComponent(token("BARBAZ", "barbaz", TokenType.ALPHABETIC).setOffset(3) - .addComponent(token("BAR", "bar", TokenType.ALPHABETIC).setOffset(3)) - .addComponent(token("BAZ", "baz", TokenType.ALPHABETIC).setOffset(6))); + SimpleToken token = newToken("FOOBARBAZ", "foobarbaz", TokenType.ALPHABETIC) + .addComponent(newToken("FOO", "foo", TokenType.ALPHABETIC).setOffset(0)) + .addComponent(newToken("BARBAZ", "barbaz", TokenType.ALPHABETIC).setOffset(3) + .addComponent(newToken("BAR", "bar", TokenType.ALPHABETIC).setOffset(3)) + .addComponent(newToken("BAZ", "baz", TokenType.ALPHABETIC).setOffset(6))); assertAnnotations(expected, "foobarbaz", token); } @@ -133,11 +116,11 @@ public class LinguisticsAnnotatorTestCase { expected.spanList().span(0, 9).annotate(new Annotation(AnnotationTypes.TERM, new StringFieldValue("foobarbaz"))); - SimpleToken token = token("FOOBARBAZ", "foobarbaz", TokenType.ALPHABETIC).setSpecialToken(true) - .addComponent(token("FOO", "foo", TokenType.ALPHABETIC).setOffset(0)) - .addComponent(token("BARBAZ", "barbaz", TokenType.ALPHABETIC).setOffset(3) - .addComponent(token("BAR", "bar", TokenType.ALPHABETIC).setOffset(3)) - .addComponent(token("BAZ", "baz", TokenType.ALPHABETIC).setOffset(6))); + SimpleToken token = newToken("FOOBARBAZ", "foobarbaz", TokenType.ALPHABETIC).setSpecialToken(true) + .addComponent(newToken("FOO", "foo", TokenType.ALPHABETIC).setOffset(0)) + .addComponent(newToken("BARBAZ", "barbaz", TokenType.ALPHABETIC).setOffset(3) + .addComponent(newToken("BAR", "bar", TokenType.ALPHABETIC).setOffset(3)) + .addComponent(newToken("BAZ", "baz", TokenType.ALPHABETIC).setOffset(6))); assertAnnotations(expected, "foobarbaz", token); } @@ -157,8 +140,7 @@ public class LinguisticsAnnotatorTestCase { continue; } assertAnnotations(expected, "foo", - new AnnotatorConfig(), - newLinguistics(List.of(token("foo", "foo", type, specialToken)), + newLinguistics(List.of(newToken("foo", "foo", type, specialToken)), Collections.singletonMap("foo", "bar"))); } } @@ -172,9 +154,11 @@ public class LinguisticsAnnotatorTestCase { StringFieldValue val = new StringFieldValue("foo"); val.setSpanTree(spanTree); - Linguistics linguistics = newLinguistics(List.of(token("foo", "bar", TokenType.ALPHABETIC, false)), + Linguistics linguistics = newLinguistics(List.of(newToken("foo", "bar", TokenType.ALPHABETIC, false)), Collections.<String, String>emptyMap()); - assertTrue(new LinguisticsAnnotator(linguistics, new AnnotatorConfig()).annotate(val)); + new LinguisticsAnnotator(linguistics, CONFIG).annotate(val); + + assertTrue(new LinguisticsAnnotator(linguistics, CONFIG).annotate(val)); assertEquals(spanTree, val.getSpanTree(SpanTrees.LINGUISTICS)); } @@ -202,7 +186,7 @@ public class LinguisticsAnnotatorTestCase { } @Test - public void requireThatMaxTermOccurrencesIsHonored() { + public void requireThatMaxTermOccurencesIsHonored() { final String inputTerm = "foo"; final String stemmedInputTerm = "bar"; // completely different from // inputTerm for safer test @@ -220,7 +204,7 @@ public class LinguisticsAnnotatorTestCase { StringBuilder input = new StringBuilder(); Token[] tokens = new Token[inputTermOccurence]; for (int i = 0; i < inputTermOccurence; ++i) { - SimpleToken t = token(inputTerm, stemmedInputTerm, type); + SimpleToken t = newToken(inputTerm, stemmedInputTerm, type); t.setOffset(i * paddedInputTerm.length()); tokens[i] = t; input.append(paddedInputTerm); @@ -230,29 +214,28 @@ public class LinguisticsAnnotatorTestCase { } // -------------------------------------------------------------------------------- + // // Utilities + // + // -------------------------------------------------------------------------------- - private static SimpleToken token(String orig, String stem, TokenType type) { - return token(orig, stem, type, false); + private static SimpleToken newToken(String orig, String stem, TokenType type) { + return newToken(orig, stem, type, false); } - private static SimpleToken token(String orig, String stem, TokenType type, boolean specialToken) { + private static SimpleToken newToken(String orig, String stem, TokenType type, boolean specialToken) { return new SimpleToken(orig).setTokenString(stem) .setType(type) .setSpecialToken(specialToken); } private static void assertAnnotations(SpanTree expected, String value, Token... tokens) { - assertAnnotations(expected, value, new AnnotatorConfig(), newLinguistics(Arrays.asList(tokens), Collections.emptyMap())); - } - - private static void assertAnnotations(SpanTree expected, String value, AnnotatorConfig config, Token... tokens) { - assertAnnotations(expected, value, config, newLinguistics(Arrays.asList(tokens), Collections.emptyMap())); + assertAnnotations(expected, value, newLinguistics(Arrays.asList(tokens), Collections.<String, String>emptyMap())); } - private static void assertAnnotations(SpanTree expected, String str, AnnotatorConfig config, Linguistics linguistics) { + private static void assertAnnotations(SpanTree expected, String str, Linguistics linguistics) { StringFieldValue val = new StringFieldValue(str); - assertEquals(expected != null, new LinguisticsAnnotator(linguistics, config).annotate(val)); + assertEquals(expected != null, new LinguisticsAnnotator(linguistics, CONFIG).annotate(val)); assertEquals(expected, val.getSpanTree(SpanTrees.LINGUISTICS)); } |