EDINET Replay: a reproducible, provenance-preserving extraction toolkit for Japan's EDINET corporate filings

EDINET Replay retrieves official filing packages from Japan's Financial Services Agency EDINET system, preserves the original source artifacts, resolves XBRL and Inline XBRL through Arelle, and exports faithful, provenance-preserving facts with traceable references back to the source filing. It targets faithful reproduction rather than financial interpretation.

This release is beta (0.1.0-beta.1). The vertical path fetch → extract → validate works for resolved XBRL against hash-pinned FSA taxonomies. A multi-standard β corpus pins expected canonical hashes for one JP GAAP, one IFRS, and one US GAAP annual report. Inline XBRL presentation provenance is not yet implemented. See CHANGELOG.md for the full scope and known limitations.