Merge branch 'development' into dependabot/npm_and_yarn/node-polyfill…

…-webpack-plugin-4.0.0
huridocs · Jun 18, 2024 · 109f97c · 109f97c
2 parents 188040c + f81cc39
commit 109f97c
Show file tree

Hide file tree

Showing 29 changed files with 1,640 additions and 207 deletions.
diff --git a/app/api/services/informationextraction/InformationExtraction.ts b/app/api/services/informationextraction/InformationExtraction.ts
@@ -29,6 +29,7 @@ import {
   FileWithAggregation,
   getFilesForTraining,
   getFilesForSuggestions,
+  propertyTypeIsWithoutExtractedMetadata,
   propertyTypeIsSelectOrMultiSelect,
 } from 'api/services/informationextraction/getFiles';
 import { Suggestions } from 'api/suggestions/suggestions';
@@ -95,6 +96,16 @@ type MaterialsData =
   | TextSelectionMaterialsData
   | ValuesSelectionMaterialsData;
 
+async function fetchCandidates(property: PropertySchema) {
+  const defaultLanguageKey = (await settings.getDefaultLanguage()).key;
+  const query: { template?: ObjectId; language: string } = {
+    language: defaultLanguageKey,
+  };
+  if (property.content !== '') query.template = new ObjectId(property.content);
+  const candidates = await entities.getUnrestricted(query, ['title', 'sharedId']);
+  return candidates;
+}
+
 class InformationExtraction {
   static SERVICE_NAME = 'information_extraction';
 
@@ -142,9 +153,9 @@ class InformationExtraction {
 
     let data: MaterialsData = { ..._data, language_iso };
 
-    const isSelect = propertyTypeIsSelectOrMultiSelect(propertyType);
+    const noExtractedData = propertyTypeIsWithoutExtractedMetadata(propertyType);
 
-    if (!isSelect && propertyLabeledData) {
+    if (!noExtractedData && propertyLabeledData) {
       data = {
         ...data,
         label_text: propertyValue || propertyLabeledData?.selection?.text,
@@ -155,7 +166,7 @@ class InformationExtraction {
       };
     }
 
-    if (isSelect) {
+    if (noExtractedData) {
       if (!Array.isArray(propertyValue)) {
         throw new Error('Property value should be an array');
       }
@@ -184,7 +195,7 @@ class InformationExtraction {
         );
         const { propertyValue, propertyType } = file;
 
-        const missingData = propertyTypeIsSelectOrMultiSelect(propertyType)
+        const missingData = propertyTypeIsWithoutExtractedMetadata(propertyType)
           ? !propertyValue
           : type === 'labeled_data' && !propertyLabeledData;
 
@@ -383,15 +394,22 @@ class InformationExtraction {
 
     const params: TaskParameters = {
       id: extractorId.toString(),
-      multi_value: property.type === 'multiselect',
+      multi_value: property.type === 'multiselect' || property.type === 'relationship',
     };
 
-    if (property.type === 'select' || property.type === 'multiselect') {
+    if (propertyTypeIsSelectOrMultiSelect(property.type)) {
       const thesauri = await dictionatiesModel.getById(property.content);
 
       params.options =
         thesauri?.values?.map(value => ({ label: value.label, id: value.id as string })) || [];
     }
+    if (property.type === 'relationship') {
+      const candidates = await fetchCandidates(property);
+      params.options = candidates.map(candidate => ({
+        label: candidate.title || '',
+        id: candidate.sharedId || '',
+      }));
+    }
 
     await this.taskManager.startTask({
       task: 'create_model',

diff --git a/app/api/services/informationextraction/getFiles.ts b/app/api/services/informationextraction/getFiles.ts
@@ -45,9 +45,16 @@ type FileEnforcedNotUndefined = {
 };
 
 const selectProperties: Set<string> = new Set([propertyTypes.select, propertyTypes.multiselect]);
+const propertiesWithoutExtractedMetadata: Set<string> = new Set([
+  ...Array.from(selectProperties),
+  propertyTypes.relationship,
+]);
 
 const propertyTypeIsSelectOrMultiSelect = (type: string) => selectProperties.has(type);
 
+const propertyTypeIsWithoutExtractedMetadata = (type: string) =>
+  propertiesWithoutExtractedMetadata.has(type);
+
 async function getFilesWithAggregations(files: (FileType & FileEnforcedNotUndefined)[]) {
   const filesNames = files.filter(x => x.filename).map(x => x.filename);
 
@@ -98,7 +105,7 @@ async function fileQuery(
   propertyType: string,
   entitiesFromTrainingTemplatesIds: string[]
 ) {
-  const needsExtractedMetadata = !propertyTypeIsSelectOrMultiSelect(propertyType);
+  const needsExtractedMetadata = !propertyTypeIsWithoutExtractedMetadata(propertyType);
   const query: {
     type: string;
     filename: { $exists: Boolean };
@@ -125,7 +132,7 @@ function entityForTrainingQuery(
   const query: {
     [key: string]: { $in?: ObjectIdSchema[]; $exists?: Boolean; $ne?: any[] };
   } = { template: { $in: templates } };
-  if (propertyTypeIsSelectOrMultiSelect(propertyType)) {
+  if (propertyTypeIsWithoutExtractedMetadata(propertyType)) {
     query[`metadata.${property}`] = { $exists: true, $ne: [] };
   }
   return query;
@@ -162,8 +169,8 @@ async function getFilesForTraining(templates: ObjectIdSchema[], property: string
       return { ...file, propertyType };
     }
 
-    if (propertyTypeIsSelectOrMultiSelect(propertyType)) {
-      const propertyValue = (entity.metadata[property] || []).map(({ value, label }) => ({
+    if (propertyTypeIsWithoutExtractedMetadata(propertyType)) {
+      const propertyValue = (entity.metadata?.[property] || []).map(({ value, label }) => ({
         value: ensure<string>(value),
         label: ensure<string>(label),
       }));
@@ -223,5 +230,6 @@ export {
   getFilesForSuggestions,
   getSegmentedFilesIds,
   propertyTypeIsSelectOrMultiSelect,
+  propertyTypeIsWithoutExtractedMetadata,
 };
 export type { FileWithAggregation };
diff --git a/app/api/services/informationextraction/ixextractors.ts b/app/api/services/informationextraction/ixextractors.ts
@@ -8,9 +8,16 @@ import {
   createBlankSuggestionsForExtractor,
   createBlankSuggestionsForPartialExtractor,
 } from 'api/suggestions/blankSuggestions';
+import { Subset } from 'shared/tsUtils';
+import { PropertyTypeSchema } from 'shared/types/commonTypes';
 import { IXExtractorModel as model } from './IXExtractorModel';
 
-type AllowedPropertyTypes = 'title' | 'text' | 'numeric' | 'date' | 'select' | 'multiselect';
+type AllowedPropertyTypes =
+  | Subset<
+      PropertyTypeSchema,
+      'text' | 'numeric' | 'date' | 'select' | 'multiselect' | 'relationship'
+    >
+  | 'title';
 
 const ALLOWED_PROPERTY_TYPES: AllowedPropertyTypes[] = [
   'title',
@@ -19,6 +26,7 @@ const ALLOWED_PROPERTY_TYPES: AllowedPropertyTypes[] = [
   'date',
   'select',
   'multiselect',
+  'relationship',
 ];
 
 const allowedTypeSet = new Set<string>(ALLOWED_PROPERTY_TYPES);