Import source: app/backend/src/lib/link-engine

Import the existing local project snapshot. This commit organizes existing files; it does not represent newly developed functionality.

app/backend/src/lib/link-engine.ts

@@ -0,0 +1,70 @@
+import { PrismaClient } from '@prisma/client';
+import * as cheerio from 'cheerio';
+
+const prisma = new PrismaClient();
+
+/**
+ * Autolink glossary terms in the provided HTML content using cheerio.
+ * Safely traverses text nodes and wraps matching terms in anchor tags,
+ * avoiding already linked text, code blocks, and headings.
+ */
+export async function autolinkGlossaryTerms(htmlContent: string): Promise<string> {
+  if (!htmlContent) return htmlContent;
+
+  const glossaryTerms = await prisma.article.findMany({
+    where: { type: 'glossary', status: 'منتشرشده' },
+    select: { title: true, slug: true },
+  });
+
+  if (glossaryTerms.length === 0) return htmlContent;
+
+  // Sort by length descending to match longest phrases first
+  const sortedTerms = glossaryTerms.sort((a, b) => b.title.length - a.title.length);
+  
+  const $ = cheerio.load(htmlContent, null, false);
+
+  // Exclude elements where we shouldn't inject links
+  const excludedTags = ['a', 'pre', 'code', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'script', 'style'];
+
+  // Find all text nodes that are not inside excluded tags
+  const textNodes: any[] = [];
+  $('*').contents().each(function() {
+    if (this.type === 'text') {
+      let parent = $(this).parent();
+      let shouldExclude = false;
+      while (parent.length > 0 && parent[0].name) {
+        if (excludedTags.includes(parent[0].name.toLowerCase())) {
+          shouldExclude = true;
+          break;
+        }
+        parent = parent.parent();
+      }
+      if (!shouldExclude) {
+        textNodes.push(this);
+      }
+    }
+  });
+
+  for (const textNode of textNodes) {
+    let nodeText = $(textNode).text();
+    let replaced = false;
+
+    for (const term of sortedTerms) {
+      const escapedTerm = term.title.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+      // Using word boundaries; since Farsi spaces/zwnj might be tricky, we can use a simpler regex
+      // that matches the phrase separated by whitespace or punctuation.
+      const regex = new RegExp(`(?<=^|[\\s،؛.!?()»«])(${escapedTerm})(?=[\\s،؛.!?()»«]|$)`, 'gi');
+      
+      if (regex.test(nodeText)) {
+        nodeText = nodeText.replace(regex, `<a href="/glossary/${term.slug}" class="glossary-link" title="${term.title}">$1</a>`);
+        replaced = true;
+      }
+    }
+
+    if (replaced) {
+      $(textNode).replaceWith(nodeText);
+    }
+  }
+
+  return $.html();
+}