Get unique years from datetime field in Sanity GROQ query

Yes! You can extract unique years from datetime fields in GROQ. The key is to convert the datetime to a string first, then use string::split() to extract the year portion. Here's the working solution:

{
  "uniqueYears": array::unique(
    *[_type in ["event", "exhibition"] && defined(endDate)]
      .endDate
  ) | [string::split(string(@), "-")[0]]
}

How this works:

1. The filter *[_type in ["event", "exhibition"] && defined(endDate)] gets your documents
2. .endDate extracts just the endDate values from each document
3. array::unique() removes duplicate datetime values
4. The pipe operator | passes the unique datetimes to the next operation
5. The projection [...] applies a transformation to each element in the array
6. string() function converts each datetime to a string in RFC3339 format (like "2018-01-06T12:30:00.000Z")
7. string::split(string(@), "-") breaks the string at each "-" character, creating an array like ["2018", "01", "06T12:30:00.000Z"]
8. [0] grabs just the first element (the year)

This will give you:

{
  "uniqueYears": ["2018", "2019", "2020"]
}

Alternative approach - You can also apply the transformation before uniquifying:

{
  "uniqueYears": array::unique(
    *[_type in ["event", "exhibition"] && defined(endDate)]
      .[string::split(string(endDate), "-")[0]]
  )
}

Why your original approach didn't work:

1. The split() function needs the string:: namespace prefix in GROQ
2. It only works on strings, so your datetime endDate field needs to be converted first using string()
3. The syntax for applying transformations across arrays requires proper use of projections with []

The years will be returned as strings. If you need them as numbers for sorting or comparison, you'd need to handle that conversion in your application code since GROQ doesn't have a built-in string-to-number conversion function.