From 03a8589b5146da90c2057d2a680eaf654e99bd5c Mon Sep 17 00:00:00 2001
From: maxgfr <25312957+maxgfr@users.noreply.github.com>
Date: Tue, 3 Dec 2024 15:19:31 +0100
Subject: [PATCH] fix: glossary

---
 .../src/workers/glossary/__mocks__/glossaryData.ts        | 7 +++++++
 .../workers/glossary/__tests__/addGlossaryContent.test.ts | 8 ++++++++
 .../src/workers/glossary/explodeGlossaryTerms.ts          | 4 ++--
 3 files changed, 17 insertions(+), 2 deletions(-)

diff --git a/targets/export-elasticsearch/src/workers/glossary/__mocks__/glossaryData.ts b/targets/export-elasticsearch/src/workers/glossary/__mocks__/glossaryData.ts
index 36236e50c..f4946c1ee 100644
--- a/targets/export-elasticsearch/src/workers/glossary/__mocks__/glossaryData.ts
+++ b/targets/export-elasticsearch/src/workers/glossary/__mocks__/glossaryData.ts
@@ -1,4 +1,11 @@
 export const glossaryData = [
+  {
+    abbreviations: ["ZC"],
+    definition: "Definition de la zone commerciale",
+    term: "Zone commerciale",
+    variants: [],
+    references: [],
+  },
   {
     abbreviations: [],
     definition:
diff --git a/targets/export-elasticsearch/src/workers/glossary/__tests__/addGlossaryContent.test.ts b/targets/export-elasticsearch/src/workers/glossary/__tests__/addGlossaryContent.test.ts
index ffd454bcf..ca655b3b6 100644
--- a/targets/export-elasticsearch/src/workers/glossary/__tests__/addGlossaryContent.test.ts
+++ b/targets/export-elasticsearch/src/workers/glossary/__tests__/addGlossaryContent.test.ts
@@ -542,6 +542,14 @@ describe("Glossary", () => {
           `<webcomponent-tooltip content="word">ward</webcomponent-tooltip>`
         );
       });
+
+      test("should work with a composed name in tag with a non breaking space", async () => {
+        const htmlContent = "<p>Une zone commerciale&nbsp;</p>";
+        expect(await addGlossaryContent(glossaryData, htmlContent)).toEqual(
+          `<p>Une <webcomponent-tooltip content="Definition%20de%20la%20zone%20commerciale">zone commerciale</webcomponent-tooltip>&nbsp;</p>`
+        );
+      });
+
       test("should match a variant with plural", async () => {
         const htmlContent = `wards`;
         expect(
diff --git a/targets/export-elasticsearch/src/workers/glossary/explodeGlossaryTerms.ts b/targets/export-elasticsearch/src/workers/glossary/explodeGlossaryTerms.ts
index 9bbac238b..b5e1623f5 100644
--- a/targets/export-elasticsearch/src/workers/glossary/explodeGlossaryTerms.ts
+++ b/targets/export-elasticsearch/src/workers/glossary/explodeGlossaryTerms.ts
@@ -3,8 +3,8 @@ import { GlossaryTerms, Glossary, Term } from "@socialgouv/cdtn-types";
 const conventionMatchers =
   "[Cc]onventions? [Cc]ollectives?|[Aa]ccords? de [Bb]ranches?|[Dd]ispositions? [Cc]onventionnelles?";
 
-const startWordBreaks = `(?<=^| |\\.|,|'|>|\\()`;
-const endWordBreaks = `(?= |\\.|,|'|$|<|\\))`;
+const startWordBreaks = `(?<=^| |\\.|,|'|>|\\(|&nbsp;)`;
+const endWordBreaks = `(?= |\\.|,|'|$|<|\\)|&nbsp;)`;
 
 const startAnchorOmit = `(?<!<span class="(?:sub-)?title">[^<]*)`;
 const endAnchorOmit = `(?![^<]*(?:</a>|</summary>|</strong>.*</summary>|</h[1-6]>))`;