Creating a field in Sanity for a blog and generating an excerpt, with discussion on previewing references and using a renderer component.

You definitely can auto-generate an excerpt from your blockContent field! The best approach is using the pt::text() function in your GROQ queries to extract plain text from Portable Text, which handles images, text blocks, and any other content automatically.

Query-time Extraction (Recommended)

Instead of storing the excerpt as a separate field, generate it when you query your posts:

*[_type == "post"] {
  title,
  body,
  "excerpt": pt::text(body)[0..96]
}

The pt::text() function extracts all plain text from your Portable Text blocks, automatically handling whether your post starts with an image or text—exactly what you need! Then you can use array slicing ([0..96]) to limit it to the first 96 characters.

For a word-based excerpt instead of character-based, you can combine pt::text() with the string::split() function:

*[_type == "post"] {
  title,
  body,
  "excerpt": array::join(string::split(pt::text(body), " ")[0..20], " ")
}

This takes the first 20 words instead of cutting mid-word. Here's what's happening:

1. pt::text(body) extracts plain text from your Portable Text
2. string::split(pt::text(body), " ") splits it into an array of words
3. [0..20] takes the first 20 words
4. array::join(..., " ") joins them back into a string with spaces

Schema-level initialValue (Alternative)

If you really want a stored excerpt field in your schema, you can use initialValue:

{
  name: "excerpt",
  title: "Excerpt",
  type: "string",
  initialValue: async ({document, getClient}) => {
    if (!document?.body) return '';
    
    const client = getClient({apiVersion: '2023-01-01'});
    const result = await client.fetch(
      `*[_id == $id][0]{"text": pt::text(body)}`,
      {id: document._id}
    );
    
    return result?.text?.substring(0, 96) || '';
  }
}

However, this approach is more complex and has downsides—the excerpt won't update automatically when the body changes, and you'd need to manually re-save documents to refresh it.

Why Query-time is Better

The query-time approach using pt::text() is simpler and more maintainable because:

• It always reflects the current content (no sync issues)
• Handles images, embeds, and text blocks automatically—no need to worry about post.body[0] being an image
• No need to update the excerpt when body changes
• Works with GROQ's string and array manipulation functions for flexible text processing

The pt::text() function is specifically designed for this use case—extracting searchable, readable text from rich Portable Text content for things like excerpts, search indexing, and SEO meta descriptions.