This adds support for zstd decoding using the python package zstandard.
This is similar to how it is implemented in urllib3. I also chose the
optional installation option httpx[zstd] to mimic the same option in
urllib3.
zstd decoding is similar to brotli, but in benchmarks it is supposed to
be even faster. The zstd compression is described in RFC 8878.
See https://github.com/encode/httpx/discussions/1986
Co-authored-by: Kamil Monicz <kamil@monicz.dev>
Inline the compressed Brotli samples in tests to make them independent
of Brotli implementation. This makes it possible to run the test suite
both against Brotli and brotlicffi.
Fixes#2906
* Replace quadratic algo in LineDecoder
Leading to enormous speedups when doing things such as
Response(...).iter_lines() as described on issue #2422
* Update httpx/_decoders.py
* Update _decoders.py
Handle text ending in `\r` more gracefully.
Return as much content as possible.
* Update test_decoders.py
* Update _decoders.py
* Update _decoders.py
* Update _decoders.py
* Update httpx/_decoders.py
Co-authored-by: cdeler <serj.krotov@gmail.com>
* Update _decoders.py
---------
Co-authored-by: Tom Christie <tom@tomchristie.com>
Co-authored-by: cdeler <serj.krotov@gmail.com>
* replace pytest-asyncio with anyio
* remove pytest-trio also
* Update setup.cfg
* use anyio.Lock in test_auth
Co-authored-by: Tom Christie <tom@tomchristie.com>
* Typing: enable disallow_untyped_calls
Only the test suite needed adjusting to add type hints.
* Update setup.cfg
Co-authored-by: Tom Christie <tom@tomchristie.com>
* Add Response(..., default_encoding=...)
* Add tests for Response(..., default_encoding=...)
* Add Client(..., default_encoding=...)
* Switch default encoding to 'utf-8' instead of 'autodetect'
* Make charset_normalizer an optional dependancy, not a mandatory one.
* Documentation
* Use callable for default_encoding
* Update tests for new charset autodetection API
* Update docs for new charset autodetection API
* Update requirements
* Drop charset_normalizer from requirements
* Use either brotli (recommended for CPython) or brotlicffi (Recommended for PyPy and others)
* Add comments in places where we switch behaviour depending on brotli/brotlicffi
Co-authored-by: Florimond Manca <florimond.manca@gmail.com>
* Added httpx.BaseTransport and httpx.AsyncBaseTransport
* Test coverage and default transports to calling .close on __exit__
* BaseTransport documentation
* Use 'handle_request' for the transport API.
* Docs tweaks
* Docs tweaks
* Minor docstring tweak
* Transport API docs
* Drop 'Optional' on Transport API
* Docs tweaks
* Tweak CHANGELOG
* Drop erronous example.py
* Push httpcore exception wrapping out of client into transport (#1524)
* Push httpcore exception wrapping out of client into transport
* Include close/aclose extensions in docstring
* Comment about the request property on RequestError exceptions
* Extensions reason_phrase and http_version as bytes (#1526)
* Extensions reason_phrase and http_version as bytes
* Update BaseTransport docstring
* Neaten up our try...except structure for ensuring responses (#1525)
* Fix CHANGELOG typo
Co-authored-by: Florimond Manca <florimond.manca@gmail.com>
* Fix CHANGELOG typo
Co-authored-by: Florimond Manca <florimond.manca@gmail.com>
* stream: Iterator[bytes] -> stream: Iterable[bytes]
* Use proper bytestream interfaces when calling into httpcore
* Grungy typing workaround due to httpcore using Iterator instead of Iterable in bytestream types
* Update docs/advanced.md
Co-authored-by: Florimond Manca <florimond.manca@gmail.com>
* Consistent typing imports across tranports
* Update docs/advanced.md
Co-authored-by: Florimond Manca <florimond.manca@gmail.com>
Co-authored-by: Florimond Manca <florimond.manca@gmail.com>
* Support iter_raw(chunk_size=...) and aiter_raw(chunk_size=...)
* Unit tests for ByteChunker
* Support iter_bytes(chunk_size=...)
* Add TextChunker
* Support iter_text(chunk_size=...)
* Fix merge with master
Co-authored-by: Florimond Manca <florimond.manca@gmail.com>
* Internal refactoring to swap auth/redirects ordering
* Drop chardet for charset detection
* Drop chardet in favour of simpler charset autodetection
* Revert unintentionally included changes
* Update test case
* Refactor to prefer different decoding style
* Update text decoding docs/docstrings
* Resolve typo
* Update docs/quickstart.md
Co-authored-by: Florimond Manca <florimond.manca@gmail.com>
Co-authored-by: Florimond Manca <florimond.manca@gmail.com>
* Fixes#1033 - Ensure that text that spans 3 invocations before newline is handled - don't clobber the buffer between invocations that haven't seen any lines.
This seems like a one character fix + the test.
* Update the tests.
* Undo formatting/indentation.
* Update long comment.
* Fix trailing cr line decoding
Co-authored-by: Sheridan C Rawlins <scr@verizonmedia.com>
Fixes#1033 - Ensure that text that spans 3 invocations before newline is handled - don't clobber the buffer between invocations that haven't seen any lines.
* Bump coverage
* Tests for iterative text decoding with 'aiter_text'
* nocover on xfail exception cases
* nocover API that is pending deprecation
* Tweak test to removed uncovered line
* Ingest request body in RedirectBodyUnavailable test case
* Support thread-pooled dispatch
* Add ConcurrencyBackend.run
* Initial work towards support byte-iterators on sync request data
* Test case for byte iterator content
* byte iterator support for RequestData
* Add BaseResponse
* Bridge sync/async data in SyncResponse
* Add BaseClient
* SyncResponse -> Response
* Tweaking type annotation
* Distinct classes for Request, AsyncRequest
* Tweak is_streaming, content in BaseRequest
* Stream handling moves to client
* Handle mediating to AsyncResponse from a standard sync Dispatcher class
* Working on thread-pooled dispatcher
* Support threaded dispatch, inc. streaming requests/responses
* Increase test coverage
* Coverage and tweaks
* Include Accept and User-Agent headers by default
