Analyse multiple documents in batch, returning a map from document ID to
analysis result. Uses caching so documents appearing multiple times are
only analysed once.
When the input array contains multiple versions of the same document
(e.g. a metadata-only record and a content-enriched record sharing the
same dok_id), the batch is de-duplicated up-front using a "prefer
enriched" strategy: among entries that share the same stable document
base key, the version with fullText or fullContent wins. This
prevents order-dependent overwrites and ensures the richest available
version is analysed.
Analyse multiple documents in batch, returning a map from document ID to analysis result. Uses caching so documents appearing multiple times are only analysed once.
When the input array contains multiple versions of the same document (e.g. a metadata-only record and a content-enriched record sharing the same
dok_id), the batch is de-duplicated up-front using a "prefer enriched" strategy: among entries that share the same stable document base key, the version withfullTextorfullContentwins. This prevents order-dependent overwrites and ensures the richest available version is analysed.