diff options
Diffstat (limited to 'config-model/src/main/java/com/yahoo/schema')
18 files changed, 74 insertions, 40 deletions
diff --git a/config-model/src/main/java/com/yahoo/schema/RankProfile.java b/config-model/src/main/java/com/yahoo/schema/RankProfile.java index 82ed45028b3..60674b5487c 100644 --- a/config-model/src/main/java/com/yahoo/schema/RankProfile.java +++ b/config-model/src/main/java/com/yahoo/schema/RankProfile.java @@ -141,6 +141,8 @@ public class RankProfile implements Cloneable { private Boolean strict; + private Boolean useSignificanceModel; + private final ApplicationPackage applicationPackage; private final DeployLogger deployLogger; @@ -216,6 +218,16 @@ public class RankProfile implements Cloneable { this.strict = strict; } + public void setUseSignificanceModel(Boolean useSignificanceModel) { + this.useSignificanceModel = useSignificanceModel; + } + + public boolean useSignificanceModel() { + if (useSignificanceModel != null) return useSignificanceModel; + return uniquelyInherited(p -> p.useSignificanceModel(), "use-model") + .orElse(false); // Disabled by default + } + /** * Adds a profile to those inherited by this. * The profile must belong to this schema (directly or by inheritance). diff --git a/config-model/src/main/java/com/yahoo/schema/Schema.java b/config-model/src/main/java/com/yahoo/schema/Schema.java index 3402ba31be9..127d12594b4 100644 --- a/config-model/src/main/java/com/yahoo/schema/Schema.java +++ b/config-model/src/main/java/com/yahoo/schema/Schema.java @@ -721,7 +721,7 @@ public class Schema implements ImmutableSchema { "', but this schema does not exist"); // Require schema and document type inheritance to be consistent to keep things simple - // And require it to be explicit so we have the option to support other possibilities later + // and require it to be explicit, so we have the option to support other possibilities later var parentDocument = owner.schemas().get(inherited.get()).getDocument(); if ( ! getDocument().inheritedTypes().containsKey(new DataTypeName(parentDocument.getName()))) throw new IllegalArgumentException(this + " inherits '" + inherited.get() + diff --git a/config-model/src/main/java/com/yahoo/schema/derived/SchemaInfo.java b/config-model/src/main/java/com/yahoo/schema/derived/SchemaInfo.java index f996b2624db..b91404be2dd 100644 --- a/config-model/src/main/java/com/yahoo/schema/derived/SchemaInfo.java +++ b/config-model/src/main/java/com/yahoo/schema/derived/SchemaInfo.java @@ -183,10 +183,12 @@ public final class SchemaInfo extends Derived { private void addRankProfilesConfig(SchemaInfoConfig.Schema.Builder schemaBuilder) { for (RankProfileInfo rankProfile : rankProfiles().values()) { - var rankProfileConfig = new SchemaInfoConfig.Schema.Rankprofile.Builder(); - rankProfileConfig.name(rankProfile.name()); - rankProfileConfig.hasSummaryFeatures(rankProfile.hasSummaryFeatures()); - rankProfileConfig.hasRankFeatures(rankProfile.hasRankFeatures()); + var rankProfileConfig = new SchemaInfoConfig.Schema.Rankprofile.Builder() + .name(rankProfile.name()) + .hasSummaryFeatures(rankProfile.hasSummaryFeatures()) + .hasRankFeatures(rankProfile.hasRankFeatures()) + .significance(new SchemaInfoConfig.Schema.Rankprofile.Significance.Builder() + .useModel(rankProfile.useSignificanceModel())); for (var input : rankProfile.inputs().entrySet()) { var inputConfig = new SchemaInfoConfig.Schema.Rankprofile.Input.Builder(); inputConfig.name(input.getKey().toString()); @@ -226,6 +228,7 @@ public final class SchemaInfo extends Derived { private final String name; private final boolean hasSummaryFeatures; private final boolean hasRankFeatures; + private final boolean useSignificanceModel; private final Map<Reference, RankProfile.Input> inputs; public RankProfileInfo(RankProfile profile) { @@ -233,11 +236,13 @@ public final class SchemaInfo extends Derived { this.hasSummaryFeatures = ! profile.getSummaryFeatures().isEmpty(); this.hasRankFeatures = ! profile.getRankFeatures().isEmpty(); this.inputs = profile.inputs(); + useSignificanceModel = profile.useSignificanceModel(); } public String name() { return name; } public boolean hasSummaryFeatures() { return hasSummaryFeatures; } public boolean hasRankFeatures() { return hasRankFeatures; } + public boolean useSignificanceModel() { return useSignificanceModel; } public Map<Reference, RankProfile.Input> inputs() { return inputs; } } diff --git a/config-model/src/main/java/com/yahoo/schema/derived/SummaryClass.java b/config-model/src/main/java/com/yahoo/schema/derived/SummaryClass.java index 398897c6b78..73495c066b9 100644 --- a/config-model/src/main/java/com/yahoo/schema/derived/SummaryClass.java +++ b/config-model/src/main/java/com/yahoo/schema/derived/SummaryClass.java @@ -11,7 +11,6 @@ import com.yahoo.vespa.documentmodel.DocumentSummary; import com.yahoo.vespa.documentmodel.SummaryField; import com.yahoo.vespa.documentmodel.SummaryTransform; -import java.io.IOException; import java.util.Collections; import java.util.Map; import java.util.logging.Level; diff --git a/config-model/src/main/java/com/yahoo/schema/document/Matching.java b/config-model/src/main/java/com/yahoo/schema/document/Matching.java index 9d68553fa80..33256fa8586 100644 --- a/config-model/src/main/java/com/yahoo/schema/document/Matching.java +++ b/config-model/src/main/java/com/yahoo/schema/document/Matching.java @@ -33,6 +33,8 @@ public class Matching implements Cloneable, Serializable { private Integer maxLength; /** Maximum number of occurrences for each term */ private Integer maxTermOccurrences; + /** Maximum number of characters in a token. */ + private Integer maxTokenLength; private String exactMatchTerminator = null; @@ -61,6 +63,8 @@ public class Matching implements Cloneable, Serializable { public Matching maxLength(int maxLength) { this.maxLength = maxLength; return this; } public Integer maxTermOccurrences() { return maxTermOccurrences; } public Matching maxTermOccurrences(int maxTermOccurrences) { this.maxTermOccurrences = maxTermOccurrences; return this; } + public Integer maxTokenLength() { return maxTokenLength; } + public Matching maxTokenLength(int maxTokenLength) { this.maxTokenLength = maxTokenLength; return this; } public boolean isTypeUserSet() { return typeUserSet; } public MatchAlgorithm getAlgorithm() { return algorithm; } diff --git a/config-model/src/main/java/com/yahoo/schema/document/SDField.java b/config-model/src/main/java/com/yahoo/schema/document/SDField.java index f165141b16e..2483fa47667 100644 --- a/config-model/src/main/java/com/yahoo/schema/document/SDField.java +++ b/config-model/src/main/java/com/yahoo/schema/document/SDField.java @@ -46,7 +46,7 @@ import java.util.TreeMap; * * @author bratseth */ -public class SDField extends Field implements TypedKey, ImmutableSDField { +public class SDField extends Field implements ImmutableSDField { /** Use this field for modifying index-structure, even if it doesn't have any indexing code */ private boolean indexStructureField = false; @@ -315,7 +315,7 @@ public class SDField extends Field implements TypedKey, ImmutableSDField { supplyStructField.accept(field.getName(), field.getDataType()); } } - if ((subType == null) && (structFields.size() > 0)) { + if ((subType == null) && (!structFields.isEmpty())) { throw new IllegalArgumentException("Cannot find matching (repo=" + sdoc + ") for subfields in " + this + " [" + getDataType() + getDataType().getClass() + "] with " + structFields.size() + " struct fields"); @@ -627,7 +627,7 @@ public class SDField extends Field implements TypedKey, ImmutableSDField { public Attribute addAttribute(Attribute attribute) { String name = attribute.getName(); - if (name == null || "".equals(name)) { + if (name == null || name.isEmpty()) { name = getName(); attribute.setName(name); } diff --git a/config-model/src/main/java/com/yahoo/schema/document/TypedKey.java b/config-model/src/main/java/com/yahoo/schema/document/TypedKey.java deleted file mode 100644 index 652d21d7f7d..00000000000 --- a/config-model/src/main/java/com/yahoo/schema/document/TypedKey.java +++ /dev/null @@ -1,20 +0,0 @@ -// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. -package com.yahoo.schema.document; - -import com.yahoo.document.DataType; - -/** - * Common interface for various typed key (or field definitions). - * Used by code which wants to use common algorithms for dealing with typed keys, like the logical mapping - * - * @author bratseth - */ -public interface TypedKey { - - String getName(); - - void setDataType(DataType type); - - DataType getDataType(); - -} diff --git a/config-model/src/main/java/com/yahoo/schema/parser/ConvertParsedFields.java b/config-model/src/main/java/com/yahoo/schema/parser/ConvertParsedFields.java index 7659a1e6562..173eebe2a94 100644 --- a/config-model/src/main/java/com/yahoo/schema/parser/ConvertParsedFields.java +++ b/config-model/src/main/java/com/yahoo/schema/parser/ConvertParsedFields.java @@ -44,6 +44,7 @@ public class ConvertParsedFields { parsed.getGramSize().ifPresent(gramSize -> field.getMatching().setGramSize(gramSize)); parsed.getMaxLength().ifPresent(maxLength -> field.getMatching().maxLength(maxLength)); parsed.getMaxTermOccurrences().ifPresent(maxTermOccurrences -> field.getMatching().maxTermOccurrences(maxTermOccurrences)); + parsed.getMaxTokenLength().ifPresent(maxTokenLength -> field.getMatching().maxTokenLength(maxTokenLength)); parsed.getMatchAlgorithm().ifPresent (matchingAlgorithm -> field.setMatchingAlgorithm(matchingAlgorithm)); parsed.getExactTerminator().ifPresent diff --git a/config-model/src/main/java/com/yahoo/schema/parser/ConvertParsedRanking.java b/config-model/src/main/java/com/yahoo/schema/parser/ConvertParsedRanking.java index 5ccbb7b19a4..77a10862f9c 100644 --- a/config-model/src/main/java/com/yahoo/schema/parser/ConvertParsedRanking.java +++ b/config-model/src/main/java/com/yahoo/schema/parser/ConvertParsedRanking.java @@ -39,6 +39,7 @@ public class ConvertParsedRanking { profile.inherit(name); parsed.isStrict().ifPresent(value -> profile.setStrict(value)); + parsed.isUseSignificanceModel().ifPresent(value -> profile.setUseSignificanceModel(value)); for (var constant : parsed.getConstants().values()) profile.add(constant); diff --git a/config-model/src/main/java/com/yahoo/schema/parser/IntermediateCollection.java b/config-model/src/main/java/com/yahoo/schema/parser/IntermediateCollection.java index 789f7023aed..063962bf0c4 100644 --- a/config-model/src/main/java/com/yahoo/schema/parser/IntermediateCollection.java +++ b/config-model/src/main/java/com/yahoo/schema/parser/IntermediateCollection.java @@ -133,9 +133,9 @@ public class IntermediateCollection { var parser = new SchemaParser(stream, deployLogger, modelProperties); try { parser.rankProfile(schema); - } catch (ParseException pe) { + } catch (ParseException | TokenMgrException e) { throw new ParseException("Failed parsing rank-profile from '" + reader.getName() + "': " + - stream.formatException(Exceptions.toMessageString(pe))); + stream.formatException(Exceptions.toMessageString(e))); } } catch (java.io.IOException ex) { throw new IllegalArgumentException("Failed reading from '" + reader.getName() + "': " + ex.getMessage()); diff --git a/config-model/src/main/java/com/yahoo/schema/parser/ParsedMatchSettings.java b/config-model/src/main/java/com/yahoo/schema/parser/ParsedMatchSettings.java index c7d1a215ce3..bac2c894283 100644 --- a/config-model/src/main/java/com/yahoo/schema/parser/ParsedMatchSettings.java +++ b/config-model/src/main/java/com/yahoo/schema/parser/ParsedMatchSettings.java @@ -23,6 +23,7 @@ public class ParsedMatchSettings { private Integer gramSize = null; private Integer maxLength = null; private Integer maxTermOccurrences = null; + private Integer maxTokenLength = null; Optional<MatchType> getMatchType() { return Optional.ofNullable(matchType); } Optional<Case> getMatchCase() { return Optional.ofNullable(matchCase); } @@ -31,6 +32,7 @@ public class ParsedMatchSettings { Optional<Integer> getGramSize() { return Optional.ofNullable(gramSize); } Optional<Integer> getMaxLength() { return Optional.ofNullable(maxLength); } Optional<Integer> getMaxTermOccurrences() { return Optional.ofNullable(maxTermOccurrences); } + Optional<Integer> getMaxTokenLength() { return Optional.ofNullable(maxTokenLength); } // TODO - consider allowing each set only once: void setType(MatchType value) { this.matchType = value; } @@ -40,5 +42,6 @@ public class ParsedMatchSettings { void setGramSize(int value) { this.gramSize = value; } void setMaxLength(int value) { this.maxLength = value; } void setMaxTermOccurrences(int value) { this.maxTermOccurrences = value; } + void setMaxTokenLength(int value) { this.maxTokenLength = value; } } diff --git a/config-model/src/main/java/com/yahoo/schema/parser/ParsedRankProfile.java b/config-model/src/main/java/com/yahoo/schema/parser/ParsedRankProfile.java index fbbb0c7fe83..93319e82076 100644 --- a/config-model/src/main/java/com/yahoo/schema/parser/ParsedRankProfile.java +++ b/config-model/src/main/java/com/yahoo/schema/parser/ParsedRankProfile.java @@ -44,6 +44,7 @@ class ParsedRankProfile extends ParsedBlock { private String inheritedMatchFeatures = null; private String secondPhaseExpression = null; private Boolean strict = null; + private Boolean useSignificanceModel = null; private final List<MutateOperation> mutateOperations = new ArrayList<>(); private final List<String> inherited = new ArrayList<>(); private final Map<String, Boolean> fieldsRankFilter = new LinkedHashMap<>(); @@ -96,6 +97,8 @@ class ParsedRankProfile extends ParsedBlock { Optional<String> getSecondPhaseExpression() { return Optional.ofNullable(this.secondPhaseExpression); } Optional<Boolean> isStrict() { return Optional.ofNullable(this.strict); } + Optional<Boolean> isUseSignificanceModel() { return Optional.ofNullable(this.useSignificanceModel); } + void addSummaryFeatures(FeatureList features) { this.summaryFeatures.add(features); } void addMatchFeatures(FeatureList features) { this.matchFeatures.add(features); } void addRankFeatures(FeatureList features) { this.rankFeatures.add(features); } @@ -218,6 +221,10 @@ class ParsedRankProfile extends ParsedBlock { this.strict = strict; } + void setUseSignificanceModel(boolean useSignificanceModel) { + verifyThat(this.useSignificanceModel == null, "already has use-model"); + this.useSignificanceModel = useSignificanceModel; + } void setTermwiseLimit(double limit) { verifyThat(termwiseLimit == null, "already has termwise-limit"); this.termwiseLimit = limit; diff --git a/config-model/src/main/java/com/yahoo/schema/processing/AddDataTypeAndTransformToSummaryOfImportedFields.java b/config-model/src/main/java/com/yahoo/schema/processing/AddDataTypeAndTransformToSummaryOfImportedFields.java index 762279e3871..e66cd62caa8 100644 --- a/config-model/src/main/java/com/yahoo/schema/processing/AddDataTypeAndTransformToSummaryOfImportedFields.java +++ b/config-model/src/main/java/com/yahoo/schema/processing/AddDataTypeAndTransformToSummaryOfImportedFields.java @@ -32,7 +32,7 @@ public class AddDataTypeAndTransformToSummaryOfImportedFields extends Processor @Override public void process(boolean validate, boolean documentsOnly) { schema.allImportedFields() - .forEach(field -> setTransform(field)); + .forEach(this::setTransform); } private Stream<SummaryField> getSummaryFieldsForImportedField(ImmutableSDField importedField) { diff --git a/config-model/src/main/java/com/yahoo/schema/processing/AttributesImplicitWord.java b/config-model/src/main/java/com/yahoo/schema/processing/AttributesImplicitWord.java index 767593b82d0..769f0c9de92 100644 --- a/config-model/src/main/java/com/yahoo/schema/processing/AttributesImplicitWord.java +++ b/config-model/src/main/java/com/yahoo/schema/processing/AttributesImplicitWord.java @@ -2,6 +2,7 @@ package com.yahoo.schema.processing; import com.yahoo.config.application.api.DeployLogger; +import com.yahoo.document.TensorDataType; import com.yahoo.schema.RankProfileRegistry; import com.yahoo.document.DataType; import com.yahoo.schema.Schema; @@ -45,6 +46,8 @@ public class AttributesImplicitWord extends Processor { private boolean fieldImplicitlyWordMatch(ImmutableSDField field) { // numeric types should not trigger exact-match query parsing if (field.getDataType().getPrimitiveType() instanceof NumericDataType) return false; + // Tensor type should not trigger exact-match query parsing + if (field.getDataType() instanceof TensorDataType) return false; return (! field.hasIndex() && !field.getAttributes().isEmpty() diff --git a/config-model/src/main/java/com/yahoo/schema/processing/ExactMatch.java b/config-model/src/main/java/com/yahoo/schema/processing/ExactMatch.java index 056c37a9830..4313ceb4be1 100644 --- a/config-model/src/main/java/com/yahoo/schema/processing/ExactMatch.java +++ b/config-model/src/main/java/com/yahoo/schema/processing/ExactMatch.java @@ -16,6 +16,7 @@ import com.yahoo.vespa.indexinglanguage.expressions.ForEachExpression; import com.yahoo.vespa.indexinglanguage.expressions.IndexExpression; import com.yahoo.vespa.indexinglanguage.expressions.OutputExpression; import com.yahoo.vespa.indexinglanguage.expressions.ScriptExpression; +import com.yahoo.vespa.indexinglanguage.linguistics.AnnotatorConfig; import com.yahoo.vespa.model.container.search.QueryProfiles; /** @@ -75,7 +76,11 @@ public class ExactMatch extends Processor { } ScriptExpression script = field.getIndexingScript(); if (new ExpressionSearcher<>(IndexExpression.class).containedIn(script)) { - field.setIndexingScript(schema.getName(), (ScriptExpression)new MyProvider(schema).convert(field.getIndexingScript())); + var maxTokenLength = field.getMatching().maxTokenLength(); + if (maxTokenLength == null) { + maxTokenLength = AnnotatorConfig.getDefaultMaxTokenLength(); + } + field.setIndexingScript(schema.getName(), (ScriptExpression)new MyProvider(schema, maxTokenLength).convert(field.getIndexingScript())); } } @@ -85,8 +90,12 @@ public class ExactMatch extends Processor { private static class MyProvider extends TypedTransformProvider { - MyProvider(Schema schema) { + private int maxTokenLength; + + MyProvider(Schema schema, int maxTokenLength) + { super(ExactExpression.class, schema); + this.maxTokenLength = maxTokenLength; } @Override @@ -96,7 +105,7 @@ public class ExactMatch extends Processor { @Override protected Expression newTransform(DataType fieldType) { - Expression exp = new ExactExpression(); + Expression exp = new ExactExpression(maxTokenLength); if (fieldType instanceof CollectionDataType) { exp = new ForEachExpression(exp); } diff --git a/config-model/src/main/java/com/yahoo/schema/processing/Processor.java b/config-model/src/main/java/com/yahoo/schema/processing/Processor.java index dd36bbb3b61..8b0446364ea 100644 --- a/config-model/src/main/java/com/yahoo/schema/processing/Processor.java +++ b/config-model/src/main/java/com/yahoo/schema/processing/Processor.java @@ -13,6 +13,7 @@ import com.yahoo.schema.document.RankType; import com.yahoo.schema.document.SDField; import com.yahoo.schema.document.Stemming; import com.yahoo.vespa.model.container.search.QueryProfiles; +import com.yahoo.vespa.objects.FieldBase; import java.util.Iterator; import java.util.List; @@ -110,8 +111,8 @@ public abstract class Processor { List<RankProfile.RankSetting> someRankSettings = new java.util.ArrayList<>(); for (RankProfile profile : rankProfileRegistry.rankProfilesOf(schema)) { - for (Iterator j = profile.declaredRankSettingIterator(); j.hasNext(); ) { - RankProfile.RankSetting setting = (RankProfile.RankSetting)j.next(); + for (Iterator<RankProfile.RankSetting> j = profile.declaredRankSettingIterator(); j.hasNext(); ) { + RankProfile.RankSetting setting = j.next(); if (setting.getType().equals(type)) { someRankSettings.add(setting); } @@ -128,7 +129,7 @@ public abstract class Processor { return new IllegalArgumentException(formatError(schemaName, fieldName, msg)); } - protected RuntimeException newProcessException(Schema schema, Field field, String msg) { + protected RuntimeException newProcessException(Schema schema, FieldBase field, String msg) { return newProcessException(schema.getName(), field.getName(), msg); } diff --git a/config-model/src/main/java/com/yahoo/schema/processing/TextMatch.java b/config-model/src/main/java/com/yahoo/schema/processing/TextMatch.java index e29f683761f..3f23cbc9b2d 100644 --- a/config-model/src/main/java/com/yahoo/schema/processing/TextMatch.java +++ b/config-model/src/main/java/com/yahoo/schema/processing/TextMatch.java @@ -64,12 +64,16 @@ public class TextMatch extends Processor { if (fieldMatching != null) { var maxLength = fieldMatching.maxLength(); if (maxLength != null) { - ret.setMaxTokenLength(maxLength); + ret.setMaxTokenizeLength(maxLength); } var maxTermOccurrences = fieldMatching.maxTermOccurrences(); if (maxTermOccurrences != null) { ret.setMaxTermOccurrences(maxTermOccurrences); } + var maxTokenLength = fieldMatching.maxTokenLength(); + if (maxTokenLength != null) { + ret.setMaxTokenLength(maxTokenLength); + } } return ret; } diff --git a/config-model/src/main/java/com/yahoo/schema/processing/TypedTransformProvider.java b/config-model/src/main/java/com/yahoo/schema/processing/TypedTransformProvider.java index 8ccc8870419..8be7be02135 100644 --- a/config-model/src/main/java/com/yahoo/schema/processing/TypedTransformProvider.java +++ b/config-model/src/main/java/com/yahoo/schema/processing/TypedTransformProvider.java @@ -5,6 +5,7 @@ import com.yahoo.document.DataType; import com.yahoo.document.Field; import com.yahoo.schema.Schema; import com.yahoo.schema.document.Attribute; +import com.yahoo.vespa.documentmodel.SummaryField; import com.yahoo.vespa.indexinglanguage.ValueTransformProvider; import com.yahoo.vespa.indexinglanguage.expressions.AttributeExpression; import com.yahoo.vespa.indexinglanguage.expressions.Expression; @@ -29,6 +30,10 @@ public abstract class TypedTransformProvider extends ValueTransformProvider { protected final boolean requiresTransform(Expression exp) { if (exp instanceof OutputExpression) { String fieldName = ((OutputExpression)exp).getFieldName(); + if (fieldName == null) { + // Incomplete output expressions never require a transform. + return false; + } if (exp instanceof AttributeExpression) { Attribute attribute = schema.getAttribute(fieldName); if (attribute == null) @@ -42,7 +47,7 @@ public abstract class TypedTransformProvider extends ValueTransformProvider { fieldType = field.getDataType(); } else if (exp instanceof SummaryExpression) { - Field field = schema.getSummaryField(fieldName); + SummaryField field = schema.getSummaryField(fieldName); if (field == null) { // Use document field if summary field is not found var sdField = schema.getConcreteField(fieldName); |