Merge pull request #14297 from supabase/fix/search-custom-slugs

fix(embeddings): custom heading slugs

.github/workflows/search.yml

@@ -9,7 +9,12 @@ on:
       - 'supabase/migrations/**'
       - 'apps/docs/**'
       - 'spec/**'
-  workflow_dispatch: {}
+  workflow_dispatch:
+    inputs:
+      refresh:
+        description: 'Refresh all pages'
+        required: false
+        type: boolean
   schedule:
     - cron: '0 0 * * *'
 
@@ -46,6 +51,12 @@ jobs:
       - name: Run migrations
         run: npx supabase db push
 
+      - name: Update embeddings
+        working-directory: ./apps/docs
+        if: ${{ !inputs.refresh }}
+        run: npm run embeddings
+
       - name: Refresh embeddings
         working-directory: ./apps/docs
-        run: npm run embeddings
+        if: ${{ inputs.refresh }}
+        run: npm run embeddings:refresh

apps/docs/scripts/search/sources/markdown.ts

@@ -96,6 +96,23 @@ export function splitTreeBy(tree: Root, predicate: (node: Content) => boolean) {
   }, [])
 }
 
+/**
+ * Parses a markdown heading which can optionally
+ * contain a custom anchor in the format:
+ *
+ * ```markdown
+ * ### My Heading [#my-custom-anchor]
+ * ```
+ */
+export function parseHeading(heading: string): { heading: string; customAnchor?: string } {
+  const match = heading.match(/(.*) *\[#(.*)\]/)
+  if (match) {
+    const [, heading, customAnchor] = match
+    return { heading, customAnchor }
+  }
+  return { heading }
+}
+
 /**
  * Processes MDX content for search indexing.
  * It extracts metadata, strips it of all JSX,
@@ -140,12 +157,20 @@ export function processMdxForSearch(content: string): ProcessedMdx {
 
   const sections = sectionTrees.map((tree) => {
     const [firstNode] = tree.children
+    const content = toMarkdown(tree)
 
-    const heading = firstNode.type === 'heading' ? toString(firstNode) : undefined
-    const slug = heading ? slugger.slug(heading) : undefined
+    const rawHeading: string = firstNode.type === 'heading' ? toString(firstNode) : undefined
+
+    if (!rawHeading) {
+      return { content }
+    }
+
+    const { heading, customAnchor } = parseHeading(rawHeading)
+
+    const slug = slugger.slug(customAnchor ?? heading)
 
     return {
-      content: toMarkdown(tree),
+      content,
       heading,
       slug,
     }