Merge pull request #651 from geonetwork/multilang-search

Multilingual search

conf/default.toml

@@ -12,8 +12,9 @@ geonetwork4_api_url = "/geonetwork/srv/api"
 proxy_path = ""
 # This optional parameter defines, in which language metadata should be queried in elasticsearch.
 # Use ISO 639-2/B (https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) format to indicate the language of the metadata.
+# Setting to "current" will use the current language of the User Interface.
 # If not indicated, a wildcard is used and no language preference is applied for the search.
-# metadata_language = "fre"
+# metadata_language = "current"
 # This optional URL should point to the login page that allows authentication to the datahub.
 # If not indicated, the default geonetwork login page is used.
 # The following three placeholders can be part of this URL:

libs/api/repository/src/lib/gn4/elasticsearch/constant.ts

@@ -21,11 +21,12 @@ export const ES_SOURCE_SUMMARY = [
   'qualityScore',
 ]
 
-export const ES_QUERY_STRING_FIELDS = [
-  'resourceTitleObject.${searchLang}^5',
-  'tag.${searchLang}^4',
-  'resourceAbstractObject.${searchLang}^3',
-  'lineageObject.${searchLang}^2',
-  'any.${searchLang}',
-  'uuid',
-]
+export type EsQueryFieldsPriorityType = Record<string, number>
+export const ES_QUERY_FIELDS_PRIORITY = {
+  'resourceTitleObject.${searchLang}': 5,
+  'tag.${searchLang}': 4,
+  'resourceAbstractObject.${searchLang}': 3,
+  'lineageObject.${searchLang}': 2,
+  'any.${searchLang}': 1,
+  uuid: 1,
+}

libs/api/repository/src/lib/gn4/elasticsearch/elasticsearch.service.spec.ts

@@ -1,13 +1,32 @@
 import { ElasticsearchService } from './elasticsearch.service'
 import { ES_FIXTURE_AGGS_RESPONSE } from '@geonetwork-ui/common/fixtures'
-import { EsSearchParams } from '../types/elasticsearch.model'
+import { LangService } from '@geonetwork-ui/util/i18n'
+import { EsSearchParams } from '@geonetwork-ui/api/metadata-converter'
+import { TestBed } from '@angular/core/testing'
+import { METADATA_LANGUAGE } from '../../metadata-language'
+
+class LangServiceMock {
+  iso3 = 'eng'
+}
 
 describe('ElasticsearchService', () => {
   let service: ElasticsearchService
   let searchFilters
 
   beforeEach(() => {
-    service = new ElasticsearchService('fre')
+    TestBed.configureTestingModule({
+      providers: [
+        {
+          provide: LangService,
+          useClass: LangServiceMock,
+        },
+        {
+          provide: METADATA_LANGUAGE,
+          useValue: 'fre',
+        },
+      ],
+    })
+    service = TestBed.inject(ElasticsearchService)
   })
 
   it('should be created', () => {
@@ -356,6 +375,72 @@ describe('ElasticsearchService', () => {
     })
   })
 
+  describe('#injectLangInQueryStringFields - Search language', () => {
+    let queryStringFields = { 'resourceTitleObject.${searchLang}': 1 }
+    describe('When no lang from config', () => {
+      beforeEach(() => {
+        service['metadataLang'] = undefined
+      })
+      it('use * wildcard', () => {
+        expect(
+          service['injectLangInQueryStringFields'](queryStringFields)[0].split(
+            '.'
+          )[1]
+        ).toEqual('*')
+      })
+    })
+    describe('When one lang in config', () => {
+      beforeEach(() => {
+        service['metadataLang'] = 'fre'
+      })
+      it('search in the config language', () => {
+        expect(
+          service['injectLangInQueryStringFields'](queryStringFields)[0].split(
+            '.'
+          )[1]
+        ).toEqual('langfre')
+      })
+    })
+    describe('When "current" language from config"', () => {
+      beforeEach(() => {
+        service['metadataLang'] = 'current'
+        service['lang3'] = 'eng'
+      })
+      it('search in the UI language', () => {
+        expect(
+          service['injectLangInQueryStringFields'](queryStringFields)[0].split(
+            '.'
+          )[1]
+        ).toEqual('langeng^11')
+      })
+      it('add * fallback with low priority', () => {
+        queryStringFields = {
+          'resourceTitleObject.${searchLang}': 5,
+          'tag.${searchLang}': 4,
+          'resourceAbstractObject.${searchLang}': 3,
+          'lineageObject.${searchLang}': 2,
+          'any.${searchLang}': 1,
+          uuid: 1,
+        }
+        expect(
+          service['injectLangInQueryStringFields'](queryStringFields)
+        ).toEqual([
+          'resourceTitleObject.langeng^15',
+          'resourceTitleObject.*^5',
+          'tag.langeng^14',
+          'tag.*^4',
+          'resourceAbstractObject.langeng^13',
+          'resourceAbstractObject.*^3',
+          'lineageObject.langeng^12',
+          'lineageObject.*^2',
+          'any.langeng^11',
+          'any.*',
+          'uuid',
+        ])
+      })
+    })
+  })
+
   describe('#buildAutocompletePayload', () => {
     describe('given an autocomplete config', () => {
       it('returns the search payload', () => {
@@ -373,9 +458,9 @@ describe('ElasticsearchService', () => {
                 {
                   multi_match: {
                     fields: [
-                      'resourceTitleObject.langfre',
-                      'resourceAbstractObject.langfre',
-                      'tag',
+                      'resourceTitleObject.langfre^4',
+                      'resourceAbstractObject.langfre^3',
+                      'tag^2',
                       'resourceIdentifier',
                     ],
                     query: 'blarg',

libs/api/repository/src/lib/gn4/elasticsearch/elasticsearch.service.ts

@@ -1,6 +1,10 @@
 import { Inject, Injectable, Optional } from '@angular/core'
 import { Geometry } from 'geojson'
-import { ES_QUERY_STRING_FIELDS, ES_SOURCE_SUMMARY } from './constant'
+import {
+  ES_QUERY_FIELDS_PRIORITY,
+  ES_SOURCE_SUMMARY,
+  EsQueryFieldsPriorityType,
+} from './constant'
 import {
   Aggregation,
   AggregationParams,
@@ -19,6 +23,7 @@ import {
   SortParams,
   TermsAggregationResult,
 } from '@geonetwork-ui/api/metadata-converter'
+import { LangService } from '@geonetwork-ui/util/i18n'
 
 @Injectable({
   providedIn: 'root',
@@ -27,8 +32,10 @@ export class ElasticsearchService {
   // runtime fields are computed using a Painless script
   // see: https://www.elastic.co/guide/en/elasticsearch/reference/current/runtime-mapping-fields.html
   private runtimeFields: Record<string, string> = {}
+  private lang3 = this.langService.iso3
 
   constructor(
+    private langService: LangService,
     @Optional() @Inject(METADATA_LANGUAGE) private metadataLang: string
   ) {}
 
@@ -168,13 +175,40 @@ export class ElasticsearchService {
   }
 
   private injectLangInQueryStringFields(
-    queryStringFields: string[],
-    lang: string
+    queryFieldsPriority: EsQueryFieldsPriorityType
   ) {
-    const queryLang = lang ? `lang${lang}` : `*`
-    return queryStringFields.map((field) => {
-      return field.replace(/\$\{searchLang\}/g, queryLang)
-    })
+    const queryLang = this.getQueryLang()
+    return Object.keys(queryFieldsPriority).reduce((query, field) => {
+      const multiLangRegExp = /\$\{searchLang\}/g
+      const isMultilangField = multiLangRegExp.test(field)
+      const fieldPriority = queryFieldsPriority[field]
+      return [
+        ...query,
+        ...(this.isCurrentSearchLang() && isMultilangField
+          ? [
+              `${field.replace(multiLangRegExp, queryLang)}^${
+                fieldPriority + 10
+              }`,
+              field.replace(multiLangRegExp, '*') +
+                (fieldPriority > 1 ? `^${fieldPriority}` : ''),
+            ]
+          : [
+              field.replace(multiLangRegExp, queryLang) +
+                (fieldPriority > 1 ? `^${fieldPriority}` : ''),
+            ]),
+      ]
+    }, [])
+  }
+
+  private getQueryLang(): string {
+    if (this.metadataLang) {
+      return this.isCurrentSearchLang()
+        ? `lang${this.lang3}`
+        : `lang${this.metadataLang}`
+    } else return '*'
+  }
+  private isCurrentSearchLang() {
+    return this.metadataLang === 'current'
   }
 
   private buildPayloadQuery(
@@ -203,10 +237,7 @@ export class ElasticsearchService {
         query_string: {
           query: this.escapeSpecialCharacters(any),
           default_operator: 'AND',
-          fields: this.injectLangInQueryStringFields(
-            ES_QUERY_STRING_FIELDS,
-            this.metadataLang
-          ),
+          fields: this.injectLangInQueryStringFields(ES_QUERY_FIELDS_PRIORITY),
         },
       })
     }
@@ -290,15 +321,12 @@ export class ElasticsearchService {
               multi_match: {
                 query,
                 type: 'bool_prefix',
-                fields: this.injectLangInQueryStringFields(
-                  [
-                    'resourceTitleObject.${searchLang}',
-                    'resourceAbstractObject.${searchLang}',
-                    'tag',
-                    'resourceIdentifier',
-                  ],
-                  this.metadataLang
-                ),
+                fields: this.injectLangInQueryStringFields({
+                  'resourceTitleObject.${searchLang}': 4,
+                  'resourceAbstractObject.${searchLang}': 3,
+                  tag: 2,
+                  resourceIdentifier: 1,
+                }),
               },
             },
           ],