async_zip

Releases

1. ▸▾0.0.1922 Aug 2026

   Bumps to v0.0.19, which lands #192 , a rewrite of the seek reader. See CHANGELOG.md and MIGRATION.md .

   Bumps to v0.0.19, which lands #192, a rewrite of the seek reader. See CHANGELOG.md and MIGRATION.md.

   Open source →

   Additional notes2 sources agree

   Added

   • This changelog and MIGRATION.md.
   • Terminology and high-level ZIP concepts (archive, file, LF/LFH, CDR/CDRH, EOCDR, CEOCDR) documented in the [base] module, and used consistently across the new API.
   • Tracing as an optional dependency, enabled via the tracing feature. Included in full.
   • An initial re-write of the read module. This is still in active development, with no stream reader currently. The intent is to fully replace the existing read module with it. Improvements include a large set of configurable validation checks. This reader is intended to be in feature parity with the old read module. It lives at [base::read1] until then, and provides:
     • [base::read1::seek::ZipArchiveReader], which reads the central directory up-front and opens files by index without consuming the source reader ([file()], or [file_oneshot()]).
     • [base::read1::seek::ZipArchiveFactory], a user-supplied generator function which produces additional readers over the same archive for concurrent/parallel file reads.
     • [base::read1::seek::ZipArchiveInner], shareable archive metadata which can be reused across readers via [new_with_inner()] to avoid re-reading the central directory.
     • [base::read1::ZipFileReader], an [AsyncRead] over a single file which validates the CRC32 and uncompressed size on EOF.
     • [base::read1::ZipOptions], a set of configurable validations and limits, plus [ZipOptions::untrusted()] as a conservative starting point for untrusted archives.
     • ZIP64 support, including ZIP64-aware accessors on [LF], [CDR] and [CEOCDR].
   • The [spec] module is now public, exposing [spec::headers1] (primitive headers), [spec::constructs] (headers combined with their variable-length data), [spec::extra] (extra fields), [spec::ZipString] and the [spec::KnownSize] trait.
   • New [ZipError] variants covering header/CDR mismatches, configured limits, invalid offsets, missing ZIP64 records, and binary parse failures.

   Changed

   • The full feature now includes tracing.
   • The specification link in the README and crate docs now points at PKWARE's APPNOTE.TXT.

   Deprecated

   • [base::read::seek::ZipFileReader], [base::read::mem::ZipFileReader] and [tokio::read::fs::ZipFileReader] in favour of [base::read1].

   Removed

   Fixed

   Security

   • The new reader bounds what an archive can make it allocate or read on your behalf. Extra field block sizes, extra field counts, central directory size and file counts, and per-file compressed and uncompressed sizes are all checked against configured limits before anything is allocated. File names remain your responsibility to sanitise.