feat(filters): Add number of occurrences, close #54

dataesr · Jan 12, 2024 · 94d23cc · 94d23cc
1 parent f8b8f58
commit 94d23cc
Show file tree

Hide file tree

Showing 6 changed files with 45 additions and 40 deletions.
diff --git a/client/src/pages/datasetsTab.jsx b/client/src/pages/datasetsTab.jsx
@@ -25,9 +25,9 @@ export default function DatasetsTab({ datasets, publishers, selectedDatasets, se
 
   useEffect(() => {
     setFilteredDatasets(datasets);
-    setFilteredPublishers(publishers);
-    setFilteredYears(years);
-    setFilteredTypes(types);
+    setFilteredPublishers(Object.keys(publishers));
+    setFilteredYears(Object.keys(years));
+    setFilteredTypes(Object.keys(types));
   }, [datasets, publishers, types, years]);
 
   useEffect(() => {
@@ -40,7 +40,7 @@ export default function DatasetsTab({ datasets, publishers, selectedDatasets, se
         && filteredPublishers.includes(dataset.publisher)
         && filteredStatus.includes(dataset.status)
         && filteredTypes.includes(dataset.type)
-        && filteredYears.includes(Number(dataset.year)));
+        && filteredYears.includes(dataset.year));
       setFilteredDatasets(filteredDatasetsTmp);
     }, 500);
     setTimer(timerTmp);
@@ -142,11 +142,11 @@ export default function DatasetsTab({ datasets, publishers, selectedDatasets, se
             hint="Filter publications on selected years"
             legend="Years"
           >
-            {years.map((year) => (
+            {Object.keys(years).sort().reverse().map((year) => (
               <Checkbox
                 checked={filteredYears.includes(year)}
                 key={year}
-                label={year.toString()}
+                label={`${year} (${years[year]})`}
                 onChange={() => onYearsChange(year)}
                 size="sm"
               />
@@ -156,11 +156,11 @@ export default function DatasetsTab({ datasets, publishers, selectedDatasets, se
             hint="Filter publications on selected types"
             legend="Types"
           >
-            {types.map((type) => (
+            {Object.keys(types).map((type) => (
               <Checkbox
                 checked={filteredTypes.includes(type)}
                 key={type}
-                label={type.toString()}
+                label={`${type} (${types[type]})`}
                 onChange={() => onTypesChange(type)}
                 size="sm"
               />
@@ -170,11 +170,11 @@ export default function DatasetsTab({ datasets, publishers, selectedDatasets, se
             hint="Filter publications on selected publishers"
             legend="Publishers"
           >
-            {publishers.map((publisher) => (
+            {Object.keys(publishers).map((publisher) => (
               <Checkbox
                 checked={filteredPublishers.includes(publisher)}
                 key={publisher}
-                label={publisher.toString()}
+                label={`${publisher} (${publishers[publisher]})`}
                 onChange={() => onPublishersChange(publisher)}
                 size="sm"
               />
@@ -209,7 +209,7 @@ DatasetsTab.propTypes = {
     status: PropTypes.string.isRequired,
     type: PropTypes.string.isRequired,
   })).isRequired,
-  publishers: PropTypes.arrayOf(PropTypes.string).isRequired,
+  publishers: PropTypes.object.isRequired,
   selectedDatasets: PropTypes.arrayOf(PropTypes.shape({
     affiliations: PropTypes.arrayOf(PropTypes.string).isRequired,
     allIds: PropTypes.arrayOf(PropTypes.object).isRequired,
@@ -222,6 +222,6 @@ DatasetsTab.propTypes = {
   })).isRequired,
   setSelectedDatasets: PropTypes.func.isRequired,
   tagDatasets: PropTypes.func.isRequired,
-  types: PropTypes.arrayOf(PropTypes.string).isRequired,
-  years: PropTypes.arrayOf(PropTypes.number).isRequired,
+  types: PropTypes.object.isRequired,
+  years: PropTypes.object.isRequired,
 };
diff --git a/client/src/pages/index.jsx b/client/src/pages/index.jsx
@@ -149,7 +149,6 @@ export default function Home() {
         </Row>
       </Container>
       <Container className="fr-mx-5w" as="section" fluid>
-        {/* TODO add Datasets ? */}
         <Actions
           allAffiliations={allAffiliations}
           allPublications={allPublications}

diff --git a/client/src/pages/publicationsTab.jsx b/client/src/pages/publicationsTab.jsx
@@ -25,9 +25,9 @@ export default function PublicationsTab({ publications, publishers, selectedPubl
 
   useEffect(() => {
     setFilteredPublications(publications);
-    setFilteredPublishers(publishers);
-    setFilteredYears(years);
-    setFilteredTypes(types);
+    setFilteredPublishers(Object.keys(publishers));
+    setFilteredYears(Object.keys(years));
+    setFilteredTypes(Object.keys(types));
   }, [publications, publishers, types, years]);
 
   useEffect(() => {
@@ -142,11 +142,11 @@ export default function PublicationsTab({ publications, publishers, selectedPubl
             hint="Filter publications on selected years"
             legend="Years"
           >
-            {years.map((year) => (
+            {Object.keys(years).sort().reverse().map((year) => (
               <Checkbox
                 checked={filteredYears.includes(year)}
                 key={year}
-                label={year.toString()}
+                label={`${year} (${years[year]})`}
                 onChange={() => onYearsChange(year)}
                 size="sm"
               />
@@ -156,11 +156,11 @@ export default function PublicationsTab({ publications, publishers, selectedPubl
             hint="Filter publications on selected types"
             legend="Types"
           >
-            {types.map((type) => (
+            {Object.keys(types).map((type) => (
               <Checkbox
                 checked={filteredTypes.includes(type)}
                 key={type}
-                label={type.toString()}
+                label={`${type} (${types[type]})`}
                 onChange={() => onTypesChange(type)}
                 size="sm"
               />
@@ -170,11 +170,11 @@ export default function PublicationsTab({ publications, publishers, selectedPubl
             hint="Filter publications on selected publishers"
             legend="Publishers"
           >
-            {publishers.map((publisher) => (
+            {Object.keys(publishers).map((publisher) => (
               <Checkbox
                 checked={filteredPublishers.includes(publisher)}
                 key={publisher}
-                label={publisher.toString()}
+                label={`${publisher} (${publishers[publisher]})`}
                 onChange={() => onPublishersChange(publisher)}
                 size="sm"
               />
@@ -209,7 +209,7 @@ PublicationsTab.propTypes = {
     status: PropTypes.string.isRequired,
     type: PropTypes.string.isRequired,
   })).isRequired,
-  publishers: PropTypes.arrayOf(PropTypes.string).isRequired,
+  publishers: PropTypes.object.isRequired,
   selectedPublications: PropTypes.arrayOf(PropTypes.shape({
     affiliations: PropTypes.arrayOf(PropTypes.string).isRequired,
     allIds: PropTypes.arrayOf(PropTypes.object).isRequired,
@@ -222,6 +222,6 @@ PublicationsTab.propTypes = {
   })).isRequired,
   setSelectedPublications: PropTypes.func.isRequired,
   tagPublications: PropTypes.func.isRequired,
-  types: PropTypes.arrayOf(PropTypes.string).isRequired,
-  years: PropTypes.arrayOf(PropTypes.number).isRequired,
+  types: PropTypes.object.isRequired,
+  years: PropTypes.object.isRequired,
 };
diff --git a/server/src/routes/works.routes.js b/server/src/routes/works.routes.js
@@ -1,6 +1,6 @@
 import express from 'express';
 
-import { range } from '../utils/utils';
+import { range, countUniqueValues } from '../utils/utils';
 import { deduplicateWorks, getFosmWorks, getOpenAlexPublications, groupByAffiliations } from '../utils/works';
 import webSocketServer from '../webSocketServer';
 
@@ -62,17 +62,12 @@ router.route('/works')
         webSocketServer.broadcast('step_4');
         // Compute distinct types & years for facet
         console.time(`5. Facet ${options.affiliations}`);
-        // TODO chek if Set is optim
-        const publicationsYears = [...new Set(
-          publications.filter((publication) => !!publication?.year).map((publication) => Number(publication.year)),
-        )].sort((a, b) => b - a);
-        const datasetsYears = [...new Set(
-          datasets.filter((dataset) => !!dataset?.year).map((dataset) => Number(dataset.year)),
-        )].sort((a, b) => b - a);
-        const publicationsTypes = [...new Set(publications.map((publication) => publication?.type))];
-        const datasetsTypes = [...new Set(datasets.map((dataset) => dataset?.type))];
-        const publicationsPublishers = [...new Set(publications.map((publication) => publication?.publisher))];
-        const datasetsPublishers = [...new Set(datasets.map((dataset) => dataset?.publisher))];
+        const publicationsYears = countUniqueValues({ data: publications, field: 'year' });
+        const datasetsYears = countUniqueValues({ data: datasets, field: 'year' });
+        const publicationsTypes = countUniqueValues({ data: publications, field: 'type' });
+        const datasetsTypes = countUniqueValues({ data: datasets, field: 'type' });
+        const publicationsPublishers = countUniqueValues({ data: publications, field: 'publisher' });
+        const datasetsPublishers = countUniqueValues({ data: datasets, field: 'publisher' });
         console.timeEnd(`5. Facet ${options.affiliations}`);
         webSocketServer.broadcast('step_5');
         // Build and serialize response

diff --git a/server/src/utils/utils.js b/server/src/utils/utils.js
@@ -8,6 +8,16 @@ const cleanId = (id) => (
     : null
 );
 
+const countUniqueValues = ({ data = [], field }) => {
+  const map = data
+    .map((item) => item?.[field] ?? '')
+    .reduce((acc, curr) => {
+      acc[curr] = acc?.[curr] ? acc[curr] + 1 : 1;
+      return acc;
+    }, {});
+  return Object.fromEntries(Object.entries(map).sort(([, a], [, b]) => b - a));
+};
+
 // See https://stackoverflow.com/a/18391901
 const defaultDiacriticsRemovalMap = [
   { base: 'A', letters: '\u0041\u24B6\uFF21\u00C0\u00C1\u00C2\u1EA6\u1EA4\u1EAA\u1EA8\u00C3\u0100\u0102\u1EB0\u1EAE\u1EB4\u1EB2\u0226\u01E0\u00C4\u01DE\u1EA2\u00C5\u01FA\u01CD\u0200\u0202\u1EA0\u1EAC\u1EB6\u1E00\u0104\u023A\u2C6F' },
@@ -122,6 +132,7 @@ const range = (startYear, endYear = new Date().getFullYear()) => {
 
 export {
   cleanId,
+  countUniqueValues,
   range,
   removeDiacritics,
 };
diff --git a/server/src/utils/works.js b/server/src/utils/works.js
@@ -93,7 +93,7 @@ const getFosmWorksByYear = async ({ results = [], options, pit, searchAfter }) =
         publisher: result._source?.publisher_dissemination ?? result._source?.publisher ?? '',
         title: result._source.title,
         type: result._source?.genre_raw ?? result._source.genre,
-        year: result._source.year,
+        year: result?._source?.year?.toString() ?? '',
       })));
       if (hits.length > 0 && (Number(process.env.FOSM_MAX_SIZE) === 0 || results.length < Number(process.env.FOSM_MAX_SIZE))) {
         // eslint-disable-next-line no-param-reassign
@@ -204,7 +204,7 @@ const getOpenAlexPublicationsByYear = (options, cursor = '*', previousResponse =
         publisher: result?.primary_location?.source?.host_organization_name ?? '',
         title: result?.display_name,
         type: getTypeFromOpenAlex(result.type),
-        year: result?.publication_year,
+        year: result?.publication_year?.toString() ?? '',
       })));
       const nextCursor = response?.meta?.next_cursor;
       if (nextCursor && hits.length > 0 && (Number(process.env.OPENALEX_MAX_SIZE) === 0 || results.length < Number(process.env.OPENALEX_MAX_SIZE))) {