fix: exclude link-only text blocks from search results; refs #180

src/services/unbody/dataTypes/TextBlock.dataType.ts

@@ -22,12 +22,14 @@ export const TextBlockDataType: UnbodyDataTypeConfig<
 
     if (text.length > 0 || html.length > 0) {
       const isLink =
-        /^<p[^>]*><span[^>]*><a[^>]*href="(https):\/\/[^ "]+"[^>]*>.*<\/a><\/span><\/p>$/.test(
+        /^<p[^>]*>(<span[^>]*>[\s]*<\/span>)*<span[^>]*><a[^>]*href="(https):\/\/[^ "]+"[^>]*>.*<\/a><\/span>(<span[^>]*>(\s|(<br(\/)?>))*<\/span>)*<\/p>$/.test(
           html,
         )
       const isIframe = /<iframe[^>]*>(?:<\/iframe>|[^]*?<\/iframe>)/.test(text)
 
       if (isLink) {
+        labels.push(LPE.Post.ContentBlockLabels.LinkOnly)
+
         const youtube = html.match(
           /(https?\:\/\/)?((www\.)?youtube\.com|youtu\.?be)\/[^ "]+/gi,
         )

src/types/lpe.types.ts

@@ -62,6 +62,7 @@ export namespace LPE {
       Paragraph: 'paragraph',
       CoverImage: 'cover_image',
       Embed: 'embed',
+      LinkOnly: 'link_only',
       YoutubeEmbed: 'youtube_embed',
       SimplecastEmbed: 'simplecast_embed',
     } as const

src/utils/search.utils.ts

@@ -13,7 +13,8 @@ export const searchBlocksBasicFilter = (
   if (block.type === LPE.ContentTypes.Text) {
     return (
       !isTitle(block.data as LPE.Post.TextBlock) &&
-      isLongEnough(block.data as LPE.Post.TextBlock)
+      isLongEnough(block.data as LPE.Post.TextBlock) &&
+      !block.data.labels.includes('link_only')
     )
   } else {
     const isPodcastImage = block.data.document.type === LPE.PostTypes.Podcast