Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
🎉 Released VidGear Stable v0.3.0 [#352]
📓 Complete Release Notes: https://abhitronix.github.io/vidgear/latest/changelog/#v030-2023-01-26 #### New Features ✨ - **WriteGear:** * Added support for user-defined and higher than 8-bit depth input frames pixel-format. + Added support for higher than 8-bit depth frames with datatypes of unsigned integer(`uint`) kind and element size `2`. + Added `dtype` parameter to internal `Preprocess` method for passing input frames datatype. + Implemented auto-calculation of input pixel-format based on number of channels in higher than 8-bit depth frames. + Added various known working pixel-formats(based on number of channels), supported by all prominent computer vision libraries. + Added support for up to 1-channel(`gray16-le/be`) to all the way up to 4-channels(`bgra64-le/be`) in input frames. + Added endianness little(`le`) or big(`be`) at the suffix of pixel-format based on byte-order of input frames datatypes. + Extended support for higher RGB 8-bit depth frames through RGB mode. * Added support for user-defined custom input pixel-format. + Added new `-input_pixfmt` attribute to `output_params` dictionary parameter for easily specifying custom input pixel-format. + Added newly implemented `get_supported_pixfmts` method import for verifying user-defined input pixel-format against Installed FFmpeg supported pixel-formats. Unsupported values will be discarded. + Implemented runtime datatype validation check, such that all input frames must have same datatype. * Added support for Context Managers for proper handling of resources via `with` statement for allocating and releasing resources precisely. (Suggested by @sueskind) + Implement the `__enter__()` and `__exit__()` methods. + Added `__enter__` method that returns reference to the WriteGear Class. + Added `__exit__` method that automatically executes `close()` for performing the cleanup operations and handling exception gracefully. - **StreamGear:** * Added support for Context Managers for proper handling of resources via `with` statement for allocating and releasing resources precisely. (Suggested by @sueskind) + Implement the `__enter__()` and `__exit__()` methods. + Added `__enter__` method that returns reference to the StreamGear Class. + Added `__exit__` method that automatically executes `close()` for performing the cleanup operations and handling exception gracefully. - **WebGear:** * Added way to completely disable Data-Files Auto-Generation WorkFlow. + Added new `skip_generate_webdata` boolean optional attribute(`False` by default) to completely disable Data-Files Auto-Generation WorkFlow. + This flag enables only `/video` route for disabled Data-Files Auto-Generation WorkFlow. + Implemented JSONResponse as placeholder response instead of Index, `404` and `500` HTML pages, when workflow is disabled. _(Note: Index HTML page will throw `404` status code.)_ + Added necessary imports. - **Helper:** * Added more robust implementation of validate_audio method. + Added new more robust regex pattern for extracting audio-samplerate. + Added new `validate_audio` method for calculating accurate bitrate(in kbps) from audio samplerate, channels, bit-depth values. + Implemented new patterns and logic for accurately extracting audio channels and bit-depth from given metadata. * Added support for Linux video device path _(such as `/dev/video0`)_. - **Maintenance:** * Logging current vidgear version when vidgear APIs are called, not at import. + Added `logcurr_vidgear_ver` helper function to facilitate logging current vidgear version, when called within a API. + Implemented `ver_is_logged` global variable in helper to log version only once, which can modifiable with `logcurr_vidgear_ver` method only. Followed recommendation given in official python docs: https://docs.python.org/3/faq/programming.html#how-do-i-share-global-variables-across-modules + Current version can only be logged by VidGear APIs with the logging turned on _(i.e. `logging=True`)_. - **Docs:** * Added new WriteGear Bonus Example: + Added "Using WriteGear's Compression Mode with `v4l2loopback` Virtual Cameras bonus python example. + Added related prerequisites and dependencies for creating `v4l2loopback` Virtual Cameras on Linux machines. + Added both With/Without-Audio cases for "Using WriteGear's Compression Mode for YouTube-Live Streaming". * Added `content.code.copy` and `content.tabs.link` features. * Added docs related to `skip_generate_webdata` optional attribute. * Added feedback features to mkdocs.yml. * Added `404.html` static template to `mkdocs.yml`. - **CI:** * Added v4l2loopback support for testing `/dev/video0` device on Linux machines. * Added test cases for newer implementation of `validate_audio` method. * Added `test_skip_generate_webdata` to test `skip_generate_webdata` optional attribute. * Added tests for user-defined and higher than 8-bit depth input frames pixel-format. #### Updates/Improvements ⚡️ - WriteGear: * Completely revamped code structure and comments. + Updated comments, description, and logging messages to more sensible and developer friendly. + Implemented operator short-circuiting to cleanup code as much as possible. + Renamed `startFFmpeg_Process` internal class method to `start_FFProcess`. + Renamed `Preprocess` internal class method to `PreprocessFFParams`. + Renamed `startCV_Process` internal class method to `start_CVProcess`. + Renamed `initiate` internal class parameter to `initiate_process`. + Renamed `force_termination` internal class parameter to `forced_termination`. + Enabled `output_params` parameters logging in both modes. + Improved `compression` and `logging` parameters boolean value handling. + Impelemented `stdout` closing to cleanup pipeline before terminating. - Helper: * Updated `validate_audio` method with improved and more robust regex patterns for identifying audio bitrate in ay audio file. - Setup.py: * Bumped version to `0.3.0`. * Replaced `>=` comparsion operator with more flexible `~=`. * Replaced `distutils.version.LooseVersion` with `pkg_resources.parse_version`. - Maintenance: * Replaced depreciated `LooseVersion` with `parse_version`. * Updated `Retry` package to be imported from `requests.adapters`. * Moved terminal and python code text area to Question GitHub Form Schema. * Removed unnecessary imports. * Removed redundant code. * Improved logging messages. * Updated code comments. * Updated method descriptions. * Refactored code. * Increased coverage. - Bash Script: * Updated FFmpeg Static Binaries links to latest date/version tag to `12-07-2022`. * Removed depreciated binaries download links and code. - Docs: * Replaced all `raw.githubusercontent.com` GIF URLs with `user-images.githubusercontent.com`. * Reformatted `custom.css` and added missing comments. * Updated sponsor block. * Enabled Code Highlights. * Updated announcement bar. * Reduced `webgear_rtc.gif` size. * Updated Zenodo badge and the BibTeX entry. - CI: * Added more flexible formats to `return_testvideo_path` function. * Updated `test_write` test for higher than 8-bit depth input frames pixel-format in WriteGear's Compression Mode. * Updated `actions/checkout` to `v3`. * Updated `actions/setup-python` to `v4`. * Updated `codecov/codecov-action` to `v3`. * Moved `test_colorspaces` test to CamGear tests. * Added deffcode library import. * Re-stuctured yaml code. #### Breaking Updates/Changes 💥 - WriteGear: * Renamed `output_filename` string parameter to `output`. + Since WriteGear API accepts all sorts of streams _(such as valid filename/path/URL)_ for encoding, thereby changing parameter name to `output` will be more true to its purpose. + Renaming `output_filename` to `output` in WriteGear API will also help user to not accidentally assume WriteGear supports only encoding of video files. * It matches the `output` parameter in StreamGear which basically does the same thing. * Renamed `cmd` parameter in `execute_ffmpeg_cmd()` class method to more sensible `command`. * `ValueError` will be raised if datatype of input frames mismatches Writegear API #### Bug-fixes 🐛 - Camgear: * Fixed `CamGear.read()` blocked unnecessarily. + 💬 When `THREADED_QUEUE_MODE` is enabled `CamGear.read()` blocks for an excessive duration when attempting to read past the end of a stream. + Added `None` frame to the queue at the end to signal we're done. + Added `terminate` Event check before continuing. * Fixed deadlock on exit. + 💬 The deadlock is due to `self.__queue.get(timeout=self.__thread_timeout)` line in `read()` method, which still waits for timeout(thread_timeout) to happen when main `update()` thread was already terminated on exit and queue was empty. Since there was no way to signal queue that stream is already ended, the blocking `queue.get()` keeps on waiting until timeout occurs. + The solution was to signal `queue.get()` that stream is already ended by putting `None` in queue on exiting the main `update()` thread. - ScreenGear: * Fixed `ScreenGear.read()` blocked during cold startup. + 💬 During startup, `ScreenGear.read()` doesn't checks if queue is empty before continuing. - WriteGear: * Fixed gstpipeline_mode not activating when wrongly assuming `output` value as valid path. * Fixed name 'compression' is not defined bug. * Fixed `AttributeError`. - Helper: * Fixed `fltp` keyword in regex pattern causing non-ftlp streams to be not recognized. * Fixed response.headers returning `content-length` as Nonetype since it may not necessarily have the Content-Legth header set. + Reason: The response from gitlab.com contains a Transfer-Encoding field as `'Transfer-Encoding': 'chunked'`, which means data is sent in a series of chunks, so the Content-Length header is emitted. More info: https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Transfer-Encoding#Directives * Fixed Linux video device paths still not working. + Moved `helper.py` logic to WriteGear and StreamGear APIs resp. * Fixed KeyError for empty metadata. - Setup: * Pinned `pyzmq==24.0.1` working version. * Removed redundant patch for the issue. - Maintaince: * Fixed missing pkg name `import_dependency_safe` functions calls. - Bash Script: * Fixed gstreamer installation. - CI: * Fixed missing v4l2loopback apt dependency on Linux envs. * Added fix for RTCPeerConnection fails to create RTCDtlsTransport (Related issue: aiortc/aiortc#804) + Pinned `cryptography==38.0.4` in dependencies. * Pinned Linux image to `ubuntu-20.04` in github actions. * Fixed No module named 'httpx' bug. + Added `httpx` library import. * Fixed F821 undefined name bug. * Fixed Gstreamer bug. - Docs: * Fixed hyperlinks to new GitHub's form schemas. * Fixed non-rendering images in README.md + Replaced all relative image/gifs paths with absolute URLs in README.md. * Fixed badges/shields#8671 badge issue in README.md * Fixed GitLab CDN links throwing blocked by CORS policy bug. + Replaced gitlab GitHack CDN links with with bitbucket. * Fixed DASH playback failing by setting the `maxAttempts` to Infinity. * Removed `x-sign` glow-text effect CSS. * Fixed several typos (suggested by @timgates42) * Fixed coverage badge.
- Loading branch information
