NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI
PyPI, #3472 most downloaded on PyPI
The world's fastest WARC parsing library written in Rust with bindings for Python.
Latest stable
2.0.1
Last release 18 days ago
04 Nov 2026
Release timing varies
gaps range from 9 days to 12 months
Most releases are documented
notes for 47 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
5 years old
91 releases · first in 2021
Drag across the ruler to compare, or tap a release to open it.
Fix a crash in Lexbor triggered by empty attribute values.
Switch source builds to Maturin.
noscript.css module to selectors.HTMLTree to HtmlTree.DOMNode and DOMCollection with typed nodes and collections, such as ElementNodeList.matches(selector) test elements themselves rather than their descendants.query_selector(selector) is not None for the previous descendant-search behaviour.Your coding agent can read these notes before it upgrades. Set up the MCP server →
Fix a memory-guard startup race.
NodeList collections from live LiveCollection views.node_value and text_content setters.warcinfo records and detaching HTTP payload decoders during in-place iteration.Fix slicing errors when accessing empty or whitespace-only header values.
This project publishes no release notes. Between v1.0.8 and v1.0.9 there were 11 commits, 10 of them substantive:
Improve record filtering performance, including native dispatch for built-in Python filter predicates.
WarcHeader enum for standard header names.WarcRead implementations to be Send.This project publishes no release notes. Between v1.0.7 and v1.0.8 there were 16 commits, 15 of them substantive:
Preserve original header bytes when writing unmodified records, allowing byte-identical serialisation.
WarcRecord::init_headers take a WarcRecordType directly instead of an Option.Unknown record-type flag from 512 to 32768.This project publishes no release notes. Between v1.0.6 and v1.0.7 there were 13 commits, 12 of them substantive:
Use a default buffer size of 1 MiB when iterating local files.
SharedWarcRecord to the warc::record module.This project publishes no release notes. Between v1.0.5 and v1.0.6 there were 10 commits, 9 of them substantive:
Python: Emit deprecation warnings when legacy functionality is used rather than imported.
This project publishes no release notes. Between v1.0.4 and v1.0.5 there were 24 commits, 23 of them substantive:
…and 3 more.
Python: Deprecate legacy stream shims and API aliases.
init_headers(record_urn=...) adding a duplicate urn: prefix.This project publishes no release notes. Between v1.0.3 and v1.0.4 there were 28 commits, 27 of them substantive:
…and 7 more.
Improve reading and seeking with larger buffers, cached positions, and buffered forward seeks.
This project publishes no release notes. Between v1.0.2 and v1.0.3 there were 17 commits, 16 of them substantive:
Fix a severe performance regression caused by repeatedly wrapping stream readers.
WarcRecord.from_bytes and WarcRecord.from_reader constructors.fsspec extra and fix handling of fsspec_args=False.This project publishes no release notes. Between v1.0.1 and v1.0.2 there were 46 commits, 45 of them substantive:
…and 25 more.
Fix compression auto-detection for Zstd archives with embedded dictionaries.
This project publishes no release notes. Between v1.0.0 and v1.0.1 there were 27 commits, 26 of them substantive:
…and 6 more.
Replace the default Cython backend with a Rust implementation and PyO3 Python bindings.
fastwarc.legacy for compatibility.Restore module-level enum exports when building with Cython 3.1.
fsspec support for remote archives, including S3.ArchiveIterator and stream constructors.Remove the deprecated distutils dependency from source builds.
fastwarc and resiliparse CLI commands.distutils dependency from source builds.post_meta extraction option to control whether article and blog metadata is retained.hidden_elements to retain content hidden by class names during main-content extraction.Fix macOS cross-compilation when building wheels.
Add minimal HTML output with preserve_formatting='minimal_html'.
preserve_formatting='minimal_html'.Include py.typed markers and correct missing or inaccurate type annotations.
py.typed markers and correct missing or inaccurate type annotations.Add Python type stubs for the public API.
Improve buffered reading performance and reduce allocation overhead.
Add Linux ARM64 wheels for Python 3.10 and newer.
Update source builds for newer Cython and RE2 versions.
Fix a memory leak when using CSS selectors.
Migrate package metadata and build configuration to pyproject.toml.
pyproject.toml.Fix Apple Silicon builds and compilation with the updated Lexbor API.
Support Python 3.11 and provide Apple Silicon wheels.
EncodingDetector.reset() and increase the default encoding-detection sample size from 4 KiB to 128 KiB.Improve plain-text extraction performance by reducing GIL locking.
No user-facing changes.
No user-facing changes.
Add WarcRecord.reason_phrase and parse HTTP status codes even when the reason phrase is missing.
WarcRecord.reason_phrase and parse HTTP status codes even when the reason phrase is missing.Decode HTTP deflate content with or without a zlib header.
GZipStream parameter deflate to zlib.Content-Encoding: none without a warning.Fix deflate decoding and ensure freezing a record refreshes its payload buffer.
Fix MemoryError when pickling records after HTTP content decoding.
MemoryError when pickling records after HTTP content decoding.Allow HTTP transfer decoding and content decoding to be enabled independently.
Add automatic HTTP content and transfer decoding for Gzip, deflate, and Brotli payloads.
Add WarcRecord.http_last_modified for the HTTP Last-Modified header.
WarcRecord.http_last_modified for the HTTP Last-Modified header.Fix ArchiveIterator.min_content_length filtering.
ArchiveIterator.min_content_length filtering.Move progress_loop from resiliparse.itertools to resiliparse.process_guard, retaining a deprecated alias.
progress_loop from resiliparse.itertools to resiliparse.process_guard, retaining a deprecated alias.Add the cli-benchmark extra and report missing optional CLI dependencies explicitly.
cli-benchmark extra and report missing optional CLI dependencies explicitly.Fix invalid memory access when extracting link URLs as plain text.
Add the all extra to install all optional dependencies.
all extra to install all optional dependencies.WarcRecord.http_date for the HTTP Date header and improve charset parsing.Add WarcRecord.record_date for the WARC-Date header.
WarcRecord.record_date for the WARC-Date header.Update Lexbor to fix a use-after-free bug in HTML parsing.
Add strict_mode=False to allow LF-only line endings in WARC and HTTP headers.
strict_mode=False to allow LF-only line endings in WARC and HTTP headers.Accept additional WARC 0.x and 1.x version strings.
Fix record stream positions when reading from an initial offset or after compression auto-detection.
Fix extraction incorrectly restricting the content root when main_content=False.
main_content=False.Improve main-content extraction precision and speed.
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
One column per quarter.