ReaderOptions

Creates explicit bounded ingestion defaults: 64 MiB input, 50,000 chunks, 32 Mi characters, 100,000 blocks, 512 assets/128 MiB asset payload, and 2,000 nested inputs/256 MiB at depth 8. Format-specific security limits continue to apply. Tune these budgets to the host workload.

Optional defaults for file and folder selection. Explicit folder options replace this policy.

Requested derived representations. Unsupported handlers reject TextAndMetadata and Text before parsing. The ordinary rich projection is the default.

Optional aggregate resource budgets for a read, including nested inputs.

Opt-in native review, formula, visibility and inert package evidence. Null retains ordinary extraction.

Optional maximum input size in bytes enforced by OfficeDocumentReader when reading from a file or seekable stream. When null, a registered format handler may apply its documented default; otherwise no size limit is enforced.

Default encoding for plain-text input without a Unicode byte-order mark. Null selects UTF-8. A BOM takes precedence. Encoding settings are snapshotted per read.

Reject invalid byte sequences in plain-text input. When false, replace them with U+FFFD and emit a source decoding warning. Default: false.

OpenXML security: maximum characters allowed per part when opening OpenXML packages (best-effort). When null, the OpenXML SDK default is used.

Optional admission ceiling for one uncompressed part in Word, Excel and PowerPoint package loading. The native policy applies to ZIP parts and compound streams, including encrypted envelopes. Null retains the owner's default policy. This is distinct from compressed input and output character limits.

Optional admission ceiling for aggregate uncompressed part sizes in Word, Excel and PowerPoint packages. Native package declarations are checked before model construction; this is not a hard process-memory or total repeated decoding-work guarantee. Null retains the owner's default policy.

OpenXML security: maximum image asset entries emitted from an Office package. When null, no image asset entry limit is enforced.

Optional password used to open encrypted Office files when the selected reader engine supports decryption.

OpenXML security: maximum placements emitted for one image relationship. When null, no per-relationship placement limit is enforced.

OpenXML security: maximum bytes read from a single image part. When null, no individual image payload limit is enforced.

OpenXML security: maximum unique image payload bytes retained while extracting assets. Repeated placements of the same image part count once toward this limit. When null, no total image payload limit is enforced.

Maximum characters per emitted chunk (best-effort).

Maximum number of table rows included per table chunk (best-effort).

When true, computes source/chunk hashes for incremental indexing workflows. Default: true.

Controls extension/content detection during reads. The default inspects content only when the extension is unknown.

Maximum prefix bytes inspected when content detection runs. Default: 64 KiB.

Maximum ZIP entries inspected when classifying Office, Visio, EPUB, and ZIP containers. Default: 512.

Inspect container structure when classifying content during reads. Default: true.