Store ScreenJSON in DynamoDB

From the CLI

screenjson convert -i screenplay.fdx \
  --db dynamodb \
  --db-collection screenplays \
  --blob-region us-west-2

Recommended schema

Attribute	Type	Notes
id (partition key)	String	The ScreenJSON document UUID
version (sort key)	String	Semver, sorts deterministically
title	Map	Language-keyed title
genre	List<String>
themes	List<String>
document	Map	The nested document object
analysis	Map	Optional, consider a sibling table for size

Global secondary indexes

• By author — projects authors[].id so you can list every script by a writer across the catalogue.
• By genre — sparse index on the first element of genre, useful for faceted browsing.
• By scene count — computed at write time, stored as a number, used for “shortest / longest script” queries.

Size discipline

A feature-length ScreenJSON document with analysis populated will often exceed DynamoDB’s 400 KB item limit. Split analysis off into a sibling table keyed by id, and treat it as a join on read.

Table:  screenplays         Attributes: id, version, document, …
Table:  screenplays_analysis  Attributes: id, embeddings, passages, summaries

Next

• How-to: Store ScreenJSON in MongoDB
• How-to: Store ScreenJSON in Elasticsearch