|
| 1 | +/* |
| 2 | + * Copyright OpenSearch Contributors |
| 3 | + * SPDX-License-Identifier: Apache-2.0 |
| 4 | + */ |
| 5 | +package org.opensearch.neuralsearch.mapper; |
| 6 | + |
| 7 | +import lombok.Getter; |
| 8 | +import lombok.Setter; |
| 9 | +import org.opensearch.core.xcontent.XContentBuilder; |
| 10 | +import org.opensearch.index.mapper.BinaryFieldMapper; |
| 11 | +import org.opensearch.index.mapper.KeywordFieldMapper; |
| 12 | +import org.opensearch.index.mapper.MappedFieldType; |
| 13 | +import org.opensearch.index.mapper.Mapper; |
| 14 | +import org.opensearch.index.mapper.MapperParsingException; |
| 15 | +import org.opensearch.index.mapper.MatchOnlyTextFieldMapper; |
| 16 | +import org.opensearch.index.mapper.ParametrizedFieldMapper; |
| 17 | +import org.opensearch.index.mapper.ParseContext; |
| 18 | +import org.opensearch.index.mapper.TextFieldMapper; |
| 19 | +import org.opensearch.index.mapper.TokenCountFieldMapper; |
| 20 | +import org.opensearch.index.mapper.WildcardFieldMapper; |
| 21 | +import org.opensearch.neuralsearch.constants.MappingConstants; |
| 22 | +import org.opensearch.neuralsearch.mapper.semanticfieldtypes.SemanticFieldTypeFactory; |
| 23 | +import org.opensearch.neuralsearch.mapper.semanticfieldtypes.SemanticParameters; |
| 24 | + |
| 25 | +import java.io.IOException; |
| 26 | +import java.util.HashMap; |
| 27 | +import java.util.List; |
| 28 | +import java.util.Map; |
| 29 | +import java.util.Set; |
| 30 | + |
| 31 | +import static org.opensearch.neuralsearch.constants.SemanticFieldConstants.MODEL_ID; |
| 32 | +import static org.opensearch.neuralsearch.constants.SemanticFieldConstants.RAW_FIELD_TYPE; |
| 33 | +import static org.opensearch.neuralsearch.constants.SemanticFieldConstants.SEARCH_MODEL_ID; |
| 34 | +import static org.opensearch.neuralsearch.constants.SemanticFieldConstants.SEMANTIC_INFO_FIELD_NAME; |
| 35 | + |
| 36 | +/** |
| 37 | + * FieldMapper for the semantic field. It will hold a delegate field mapper to delegate the data parsing and query work |
| 38 | + * based on the raw_field_type. |
| 39 | + */ |
| 40 | +public class SemanticFieldMapper extends ParametrizedFieldMapper { |
| 41 | + public static final String CONTENT_TYPE = "semantic"; |
| 42 | + private final SemanticParameters semanticParameters; |
| 43 | + |
| 44 | + @Setter |
| 45 | + @Getter |
| 46 | + private ParametrizedFieldMapper delegateFieldMapper; |
| 47 | + |
| 48 | + protected SemanticFieldMapper( |
| 49 | + String simpleName, |
| 50 | + MappedFieldType mappedFieldType, |
| 51 | + MultiFields multiFields, |
| 52 | + CopyTo copyTo, |
| 53 | + ParametrizedFieldMapper delegateFieldMapper, |
| 54 | + SemanticParameters semanticParameters |
| 55 | + ) { |
| 56 | + super(simpleName, mappedFieldType, multiFields, copyTo); |
| 57 | + this.delegateFieldMapper = delegateFieldMapper; |
| 58 | + this.semanticParameters = semanticParameters; |
| 59 | + } |
| 60 | + |
| 61 | + @Override |
| 62 | + public Builder getMergeBuilder() { |
| 63 | + Builder semanticFieldMapperBuilder = (Builder) new Builder(simpleName(), SemanticFieldTypeFactory.getInstance()).init(this); |
| 64 | + ParametrizedFieldMapper.Builder delegateBuilder = delegateFieldMapper.getMergeBuilder(); |
| 65 | + semanticFieldMapperBuilder.setDelegateBuilder(delegateBuilder); |
| 66 | + return semanticFieldMapperBuilder; |
| 67 | + } |
| 68 | + |
| 69 | + @Override |
| 70 | + public final ParametrizedFieldMapper merge(Mapper mergeWith) { |
| 71 | + if (mergeWith instanceof SemanticFieldMapper) { |
| 72 | + try { |
| 73 | + delegateFieldMapper = delegateFieldMapper.merge(((SemanticFieldMapper) mergeWith).delegateFieldMapper); |
| 74 | + } catch (IllegalArgumentException e) { |
| 75 | + String err = "Failed to update the mapper [" |
| 76 | + + this.name() |
| 77 | + + "] because failed to update the delegate " |
| 78 | + + "mapper for the raw_field_type " |
| 79 | + + this.semanticParameters.getRawFieldType() |
| 80 | + + ". " |
| 81 | + + e.getMessage(); |
| 82 | + throw new IllegalArgumentException(err, e); |
| 83 | + } |
| 84 | + } |
| 85 | + return super.merge(mergeWith); |
| 86 | + } |
| 87 | + |
| 88 | + @Override |
| 89 | + protected void parseCreateField(ParseContext context) throws IOException { |
| 90 | + delegateFieldMapper.parse(context); |
| 91 | + } |
| 92 | + |
| 93 | + @Override |
| 94 | + protected String contentType() { |
| 95 | + return CONTENT_TYPE; |
| 96 | + } |
| 97 | + |
| 98 | + public static class Builder extends ParametrizedFieldMapper.Builder { |
| 99 | + @Getter |
| 100 | + protected final Parameter<String> modelId = Parameter.stringParam( |
| 101 | + MODEL_ID, |
| 102 | + true, |
| 103 | + m -> ((SemanticFieldMapper) m).semanticParameters.getModelId(), |
| 104 | + null |
| 105 | + ); |
| 106 | + @Getter |
| 107 | + protected final Parameter<String> searchModelId = Parameter.stringParam( |
| 108 | + SEARCH_MODEL_ID, |
| 109 | + true, |
| 110 | + m -> ((SemanticFieldMapper) m).semanticParameters.getSearchModelId(), |
| 111 | + null |
| 112 | + ); |
| 113 | + @Getter |
| 114 | + protected final Parameter<String> rawFieldType = Parameter.stringParam( |
| 115 | + RAW_FIELD_TYPE, |
| 116 | + false, |
| 117 | + m -> ((SemanticFieldMapper) m).semanticParameters.getRawFieldType(), |
| 118 | + TextFieldMapper.CONTENT_TYPE |
| 119 | + ); |
| 120 | + @Getter |
| 121 | + protected final Parameter<String> semanticInfoFieldName = Parameter.stringParam( |
| 122 | + SEMANTIC_INFO_FIELD_NAME, |
| 123 | + false, |
| 124 | + m -> ((SemanticFieldMapper) m).semanticParameters.getSemanticInfoFieldName(), |
| 125 | + null |
| 126 | + ); |
| 127 | + |
| 128 | + @Setter |
| 129 | + protected ParametrizedFieldMapper.Builder delegateBuilder; |
| 130 | + private final SemanticFieldTypeFactory semanticFieldTypeFactory; |
| 131 | + |
| 132 | + protected Builder(String name, SemanticFieldTypeFactory semanticFieldTypeFactory) { |
| 133 | + super(name); |
| 134 | + this.semanticFieldTypeFactory = semanticFieldTypeFactory; |
| 135 | + } |
| 136 | + |
| 137 | + @Override |
| 138 | + protected List<Parameter<?>> getParameters() { |
| 139 | + return List.of(modelId, searchModelId, rawFieldType, semanticInfoFieldName); |
| 140 | + } |
| 141 | + |
| 142 | + @Override |
| 143 | + public SemanticFieldMapper build(BuilderContext context) { |
| 144 | + final ParametrizedFieldMapper delegateMapper = delegateBuilder.build(context); |
| 145 | + |
| 146 | + final SemanticParameters semanticParameters = this.getSemanticParameters(); |
| 147 | + final MappedFieldType semanticFieldType = semanticFieldTypeFactory.createSemanticFieldType( |
| 148 | + delegateMapper, |
| 149 | + rawFieldType.getValue(), |
| 150 | + semanticParameters |
| 151 | + ); |
| 152 | + |
| 153 | + return new SemanticFieldMapper( |
| 154 | + name, |
| 155 | + semanticFieldType, |
| 156 | + multiFieldsBuilder.build(this, context), |
| 157 | + copyTo.build(), |
| 158 | + delegateMapper, |
| 159 | + semanticParameters |
| 160 | + ); |
| 161 | + } |
| 162 | + |
| 163 | + public SemanticParameters getSemanticParameters() { |
| 164 | + return new SemanticParameters( |
| 165 | + modelId.getValue(), |
| 166 | + searchModelId.getValue(), |
| 167 | + rawFieldType.getValue(), |
| 168 | + semanticInfoFieldName.getValue() |
| 169 | + ); |
| 170 | + } |
| 171 | + } |
| 172 | + |
| 173 | + public static class TypeParser implements Mapper.TypeParser { |
| 174 | + |
| 175 | + private final static Set<String> SUPPORTED_RAW_FIELD_TYPE = Set.of( |
| 176 | + TextFieldMapper.CONTENT_TYPE, |
| 177 | + KeywordFieldMapper.CONTENT_TYPE, |
| 178 | + MatchOnlyTextFieldMapper.CONTENT_TYPE, |
| 179 | + WildcardFieldMapper.CONTENT_TYPE, |
| 180 | + TokenCountFieldMapper.CONTENT_TYPE, |
| 181 | + BinaryFieldMapper.CONTENT_TYPE |
| 182 | + ); |
| 183 | + |
| 184 | + @Override |
| 185 | + public Builder parse(String name, Map<String, Object> node, ParserContext parserContext) throws MapperParsingException { |
| 186 | + final String rawFieldType = (String) node.getOrDefault(RAW_FIELD_TYPE, TextFieldMapper.CONTENT_TYPE); |
| 187 | + |
| 188 | + validateRawFieldType(rawFieldType); |
| 189 | + |
| 190 | + final ParametrizedFieldMapper.TypeParser typeParser = (ParametrizedFieldMapper.TypeParser) parserContext.typeParser( |
| 191 | + rawFieldType |
| 192 | + ); |
| 193 | + final Builder semanticFieldMapperBuilder = new Builder(name, SemanticFieldTypeFactory.getInstance()); |
| 194 | + |
| 195 | + // semantic field mapper builder parse semantic fields |
| 196 | + Map<String, Object> semanticConfig = extractSemanticConfig(node, semanticFieldMapperBuilder.getParameters(), rawFieldType); |
| 197 | + semanticFieldMapperBuilder.parse(name, parserContext, semanticConfig); |
| 198 | + |
| 199 | + // delegate field mapper builder parse remaining fields |
| 200 | + ParametrizedFieldMapper.Builder delegateBuilder = typeParser.parse(name, node, parserContext); |
| 201 | + semanticFieldMapperBuilder.setDelegateBuilder(delegateBuilder); |
| 202 | + |
| 203 | + return semanticFieldMapperBuilder; |
| 204 | + } |
| 205 | + |
| 206 | + private void validateRawFieldType(final String rawFieldType) { |
| 207 | + if (rawFieldType == null || !SUPPORTED_RAW_FIELD_TYPE.contains(rawFieldType)) { |
| 208 | + throw new IllegalArgumentException( |
| 209 | + RAW_FIELD_TYPE |
| 210 | + + ": [" |
| 211 | + + rawFieldType |
| 212 | + + "] is not supported. It " |
| 213 | + + "should be one of [" |
| 214 | + + String.join(", ", SUPPORTED_RAW_FIELD_TYPE) |
| 215 | + + "]" |
| 216 | + ); |
| 217 | + } |
| 218 | + } |
| 219 | + |
| 220 | + /** |
| 221 | + * In this function we will extract all the parameters defined in the semantic field mapper builder and parse it |
| 222 | + * later. The remaining parameters will be processed by the type parser of the raw field type. Here we cannot |
| 223 | + * pass the parameters defined by semantic field to the delegate type parser of the raw field type because it |
| 224 | + * cannot recognize them. |
| 225 | + * @param node field config |
| 226 | + * @param parameters parameters for semantic field |
| 227 | + * @param rawFieldType field type of the raw data |
| 228 | + * @return semantic field config |
| 229 | + */ |
| 230 | + private Map<String, Object> extractSemanticConfig(Map<String, Object> node, List<Parameter<?>> parameters, String rawFieldType) { |
| 231 | + final Map<String, Object> semanticConfig = new HashMap<>(); |
| 232 | + for (Parameter<?> parameter : parameters) { |
| 233 | + Object config = node.get(parameter.name); |
| 234 | + if (config != null) { |
| 235 | + semanticConfig.put(parameter.name, config); |
| 236 | + node.remove(parameter.name); |
| 237 | + } |
| 238 | + } |
| 239 | + semanticConfig.put(MappingConstants.TYPE, SemanticFieldMapper.CONTENT_TYPE); |
| 240 | + node.put(MappingConstants.TYPE, rawFieldType); |
| 241 | + return semanticConfig; |
| 242 | + } |
| 243 | + } |
| 244 | + |
| 245 | + @Override |
| 246 | + protected void doXContentBody(XContentBuilder builder, boolean includeDefaults, Params params) throws IOException { |
| 247 | + builder.field(MappingConstants.TYPE, contentType()); |
| 248 | + |
| 249 | + // semantic parameters |
| 250 | + final List<Parameter<?>> parameters = getMergeBuilder().getParameters(); |
| 251 | + for (Parameter<?> parameter : parameters) { |
| 252 | + // By default, we will not return the default value. But raw_field_type is useful info to let users know how |
| 253 | + // we will handle the raw data. So we explicitly return it even it is using the default value. |
| 254 | + if (RAW_FIELD_TYPE.equals(parameter.name)) { |
| 255 | + parameter.toXContent(builder, true); |
| 256 | + } else { |
| 257 | + parameter.toXContent(builder, includeDefaults); |
| 258 | + } |
| 259 | + } |
| 260 | + |
| 261 | + // non-semantic parameters |
| 262 | + // semantic field mapper itself does not handle multi fields or copy to. The delegate field mapper will handle it. |
| 263 | + delegateFieldMapper.multiFields().toXContent(builder, params); |
| 264 | + delegateFieldMapper.copyTo().toXContent(builder, params); |
| 265 | + delegateFieldMapper.getMergeBuilder().toXContent(builder, includeDefaults); |
| 266 | + } |
| 267 | +} |
0 commit comments