Crate text_mirror 

Converts binary and rich files into plain text and builds a replica text-mirror tree beside a provenance manifest.

The manifest schema text-mirror/manifest@1 is the public interface. The manifest module holds its serde types. See docs/DESIGN.md for the architecture.

In-process converters cover text, Office and OpenDocument documents, and spreadsheets. Text-layer PDF and the media adapters run behind the sandbox runner in a jailed worker. A format no converter claims, or one whose adapter engine is not pinned yet, is inventoried by detect and recorded as unsupported, so the manifest always reports the true denominator.

Re-exports

pub use runner::protocol::bodies as runner_bodies;

Modules

bundle

Bundle packaging, verification, and merging.

convert

The converter boundary and the plain text passthrough.

detect

Format detection. Magic bytes first, extension as tiebreak.

hash

Streaming BLAKE3 file hashing and the dedup index.

manifest

Serde types for the manifest schema text-mirror/manifest@1, the append-only JSONL writer, and the shard reader.

mirror

Mirror path mapping and atomic writes.

pipeline

The end-to-end pipeline behind the CLI verbs.

report

JSON summary types the CLI verbs emit.

runner

The sandboxed subprocess runner behind every subprocess adapter.

segments

The <source>.segments.jsonl file beside each converted artifact.

walk

Deterministic traversal of a division root.

Enums

Error

Errors returned by this library.

Type Aliases

Result

The result type used across this library.