Merge pull request #2539 from openzim/remove-mini-citation

benoit74 · web-flow · commit fde9ab2f9ee9 · 2025-10-02T15:46:37.000+02:00
Remove any citation material from article content in nodet flavor
diff --git a/Changelog b/Changelog
@@ -42,7 +42,8 @@ Unreleased:
 * CHANGED: Rework file downloads to reduce pressure on servers and better handle throttling (@benoit74 #2377)
 * FIX: Automatically delete old Redis lists (@benoit74 #1906)
 * FIX: Also retry 524 HTTP errors - usually coming from Cloudflare handling of timeouts (@benoit74 #2526)
-# FIX: Remove any iframe in article content (@benoit74 #2537)
+* FIX: Remove any iframe in article content (@benoit74 #2537)
+* FIX: Remove any citation material from article content in nodet flavor (@benoit74 #2520)
 
 1.16.0:
 * CHANGED: ActionParse renderer is now the preferred one when available (@benoit74 #2183)
diff --git a/src/renderers/abstract.renderer.ts b/src/renderers/abstract.renderer.ts
@@ -688,6 +688,28 @@ export abstract class Renderer {
     }
   }
 
+  private removeCitations(parsoidDoc: DominoElement) {
+    // Remove all citation-related material
+
+    const sups: DominoElement[] = Array.from(parsoidDoc.getElementsByTagName('sup'))
+    for (const sup of sups) {
+      if (!(sup.getAttribute('class') || '').includes('reference')) {
+        continue
+      }
+      DU.deleteNode(sup)
+    }
+
+    const citeErrors: DominoElement[] = Array.from(parsoidDoc.getElementsByClassName('mw-ext-cite-error'))
+    for (const citeError of citeErrors) {
+      DU.deleteNode(citeError)
+    }
+
+    const references: DominoElement[] = Array.from(parsoidDoc.getElementsByClassName('mw-references-wrap'))
+    for (const reference of references) {
+      DU.deleteNode(reference)
+    }
+  }
+
   private clearLinkAndInputTags(parsoidDoc: DominoElement, filtersConfig: any, dump: Dump) {
     /* Don't need <link> and <input> tags */
     const nodesToDelete: Array<{ class?: string; tag?: string; filter?: (n: any) => boolean }> = [{ tag: 'link' }, { tag: 'input' }]
@@ -792,6 +814,10 @@ export abstract class Renderer {
   private applyOtherTreatments(parsoidDoc: DominoElement, dump: Dump, articleId: string) {
     this.removeIframeTags(parsoidDoc)
 
+    if (dump.nodet) {
+      this.removeCitations(parsoidDoc)
+    }
+
     const filtersConfig = config.filters
     this.clearLinkAndInputTags(parsoidDoc, filtersConfig, dump)
 
