* todo:
* refactor: remove usage of singleton in cmd-cache
* refactor: pass aliases/env to commands_cache on creation
reduce the usage of singletons
* fix: setting up mockups for tests
* feat: add $XONSH_CACHE_DIR
* refactor: use cache path to stire cmds-cache
* chore: todo
* feat: efficient cache commands per path and modified time
* refactor: move aliases to commands cache
now XonshSession is not needed to setup cmds-cache
* refactor: update tests
* fix: type annotation
* refactor: init aliases at commands-cache
* test: update failing tests
* fix: handle paths mtime changing
* docs: add news item
* refactor: remove $COMMANDS_CACHE_SIZE_WARNING
* fix: loading on windows fails because of setup sequence
* fix: failing tests
* Add XONSH_HISTORY_IGNORE_REGEX support
* Add changelog text
* Add cleanup notes for XONSH_HISTORY_IGNORE_REGEX
* Rename should_append to is_ignored
* Change is_ignored to a History method
* Remove unused re imports
* Docstring formatting cleanup, Black compliance
* Move XONSH_HISTORY_IGNORE_REGEX to History class
* Clarify with parenthesis
* Formatting compliance
* Add validate_ignore_regex check
* Remove validation comment
* Remove XONSH_HISTORY_IGNORE_REGEX test comments
* Add __init__ chaining to DummyHistory
* Add broken test for XONSH_IGNORE_REGEX
* Remove extra test variants
* Fix unit test environment variable name
* Add test for bad regex error output
* Remove unneeded __init__ chaining
* Add test_is_ignore
* Add is_regex and associated tests
* Add doc for XONSH_HISTORY_IGNORE_REGEX
* Change validate_ignore_regex to use is_regex
* Compile regex once and use a cached property
* Let cached_property handle saving regex
* Update terminal title before executing subprocess
This eliminates the need to use `pause_call_resume()`, which causes issues
when it sends SIGSTOP to a process that is in the middle of a syscall
(for example, the `swaymsg` command fails when it gets stopped during
a `recv` syscall).
Additionally, this approach is more flexible, as it allows captured
commands and callable aliases to also be updated as expected in the
terminal's title.
* Clear prompt cache in readline shell
* Add type annotation
* Use reset_key() to only reset current_job
* Convert into PromptField class
* Support PromptField with value of None
Allows {current_job:{} | } to work as expected when current_job is a
PromptField.
* Add tests
* Fixing bash completion bug when prefix started with '>', '<', or ':'
* Fixing code formatting
* More code formatting fixes
* Fixing test
* Refactor to fix bug by improving completion parser
* Supporting '<' and '>>' in completion parser
* Support for all IO redirect tokens
* Make job control task queue thread-local
This allows callable aliases to maintain a separate task queue from the
main thread.
* Make job control dictionary thread-local
Use XSH.all_jobs for the main thread and a separate dictionary for other
threads. This allows callable aliases to keep track of their jobs
separate from the main thread.
* Implement use_main_jobs() context manager
This allows threaded commands like jobs, disown, and bg to handle the
main thread's job control
* Run commands in the same process group if in a thread
* Fix tests that use jobs and tasks
* Add tests for subprocess call inside alias
* Add news
* Remove type declarations for _jobs_thread_local
Fixes the mypy error: "Type cannot be declared in assignment to non-self attribute"
Co-authored-by: Noorhteen Raja NJ <jnoortheen@gmail.com>
* Update test_pip_completer.py
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* feat: add support for f-glob strings
Move xonsh_pathsearch() into the BaseParser class because it needs to use self._set_error()
Parametrize 6 backtick tests in test_parser.py into test_backtick() (and add cases for f-glob strings)
* add news
* docs: update tutorial
* fix news file
* Change ![] to return an object for background command
* Fix `fg` for commands started in the background
Fixes#4595
This issue was caused by `CommandPipeline.term_pgid` not being set for commands started in the background.
* feat: add function to make event registration from function signature
* docs: add xontrib special functions description
* feat: handle xontribs with special functions
also reduce usage of XSH singleton
* fix: missing XSH
for now import singleton
* docs: fix .rst format
* fix: failing tests
* feat: implement primitive xontrib-unload and xontrib-reload
* chore: give explicit name
* docs: update doc
* add test for importing empty .xsh file
* test: empty lines do not get appended to history
prompt-toolkit needs its own test outside of test_base_shell.py because it uses a custom _push() method
* fix: do not append empty/comment-only input to history
Adds a compile_empty_tree argument to Execer.compile()
By default, the argument is `True`, and `compile()` returns a compiled `pass` statement for comment-only input.
When the argument is `False`, `compile()` returns `None` for comment-only input.
The base shell and prompt-toolkit shell use `compile_empty_tree = False` so that they get `None` as the compiled code and don't append the command to the history.
* add news
* fix tests
* We can't be sure there's a python binary
Debian doesn't currently ship a /usr/bin/python binary, unless a user
specifically installs python-is-python3.
However, we can safely assume that a python3 binary exists.
* Use shutil.which, as suggested in review
* Fix tab-completion trailing space for command arguments that use equals sign
* Add test for completion of path ending with equal sign
* Move test_path_completers.py to tests/completers
* Restore _quote_paths() to original behaviour
* Apply trailing equal sign fix to _bash_quote_paths()
Use _bash_quote_paths instead of _quote_paths in complete_from_bash()
* Do not need to pass quote_paths=_bash_quote_paths
* refactor: use stdout passed from function
better handling of errors
* test: no need to skip these on windows
* test: load local plugin
* test: update vox-runin
* test: speedup vox tests
we dont need to test venv, only calls made to it via subprocess
* test: pip.exe on windows
fix: qa error, it is not a class method
still B019 is raised from bugbear
fix: failing vox tests in windows
fix: unc check on windows on ci
fix: vox failing on windows
get_binary_path already appends .exe
* Fix regex globbing for paths that contain special regex characters
* Skip regex globbing asterisk character tests on Windows
Asterisks are forbidden in Windows filenames. Plus signs are allowed in Windows filenames, so the plus sign tests run properly.
* feat: initial match statement support
* refactor assignment exprs for <3.8 compatibility; sort imports
* remove match extensions
* bump qa-workflow python version, minor alterations for qa-pipeline
* add mypy exception
* update mypy version
* Bump workflow main python version to 3.10
* remove outdated note
Co-authored-by: Alexander Firbas <alexander.firbas@gmail.com>
