Decode HTML entities in CSL text fields

.changeset/silly-phones-warn.md

@@ -0,0 +1,5 @@
+---
+'citation-js-utils': patch
+---
+
+Decode HTML entities in CrossRef citation data

packages/citation-js-utils/package.json

@@ -38,10 +38,12 @@
     "@citation-js/core": "^0.7.18",
     "@citation-js/plugin-bibtex": "^0.7.18",
     "@citation-js/plugin-csl": "^0.7.18",
+    "he": "^1.2.0",
     "doi-utils": "^2.0.5",
     "sanitize-html": "^2.7.0"
   },
   "devDependencies": {
+    "@types/he": "^1.2.0",
     "@types/sanitize-html": "^2.6.2"
   }
 }

packages/citation-js-utils/src/index.ts

@@ -2,6 +2,7 @@ import { Cite, plugins } from '@citation-js/core';
 import { doi as doiUtils } from 'doi-utils';
 import { clean as cleanCSL } from '@citation-js/core/lib/plugins/input/csl.js';
 import sanitizeHtml from 'sanitize-html';
+import he from 'he';
 
 import '@citation-js/plugin-bibtex';
 import '@citation-js/plugin-csl';
@@ -267,7 +268,8 @@ export function getCitationRenderers(data: CSL[]): CitationRenderer {
       if (!c.id) {
         c.id = formatLabel(c);
       }
-      // Trim the titles, DOIs, etc. on load
+      // Trim the titles, DOIs, etc. on load; decode HTML entities from
+      // CrossRef metadata.
       [
         'title',
         'note',
@@ -279,7 +281,7 @@ export function getCitationRenderers(data: CSL[]): CitationRenderer {
         'DOI',
         'ISSN',
       ].forEach((tag) => {
-        if (c[tag] && typeof c[tag] === 'string') c[tag] = c[tag].trim();
+        if (c[tag] && typeof c[tag] === 'string') c[tag] = he.decode(c[tag].trim());
       });
       // Trim the DOIs and URLs (these are encoded) on load
       if (c.URL) c.URL = c.URL.replace(/^(%20)*/, '').replace(/(%20)*$/, '');

packages/citation-js-utils/tests/basic.spec.ts

@@ -79,6 +79,26 @@ describe('yearFromCitation', () => {
   });
 });
 
+describe('HTML entity decoding in CSL fields', () => {
+  it('decodes HTML entities in title and container-title from CrossRef CSL-JSON', () => {
+    const csl = [
+      {
+        id: 'test2024',
+        type: 'article-journal',
+        title: 'Signal & Image Processing & More',
+        'container-title': 'Computers & Geosciences',
+        author: [{ family: 'Smith', given: 'J.' }],
+        issued: { 'date-parts': [[2024]] },
+        DOI: '10.1234/test',
+      },
+    ];
+    const renderers = getCitationRenderers(csl);
+    const cite = renderers['test2024'];
+    expect(cite.cite.title).toEqual('Signal & Image Processing & More');
+    expect(cite.cite['container-title']).toEqual('Computers & Geosciences');
+  });
+});
+
 describe('firstNonDoiUrl', () => {
   it('no url returns undefined', async () => {
     expect(firstNonDoiUrl('my citation', 'abc123')).toEqual(undefined);