* refactor(directory): Introduce `logical-path` argument which allows a shell to explicitly specify both a logical and physical filesystem path
Fix `directory::module` to consume both path and logical-path (if provided). The "logical" path is preferred when rendering the "display path", while the "physical" path is used to resolve the "read only" flag. Repo- and home-directory contraction behavior is maintained, based on the logical path if it is set, or the physical path if it is not.
The custom "get_current_dir" logic has been removed entirely, and the `directory` module now relies on `context.current_dir` / `context.logical_dir` entirely.
Changes have been made to `init/starship.ps1` to work with this new flag:
- Calculate and pass "physical" and "logical" paths explicitly (as other shells do not pass `--logical-path` that they fall back to rendering the physical path)
- Moved the "powershell provider prefix" cleanup code to the PowerShell script - this code _should_ now support any kind of powershell path prefix.
* fix(powershell): Fix an issue with trailing backslashes on file paths causing command line parsing issues.
This is a bit of a footgun!
The work-around chosen is to append a trailing space when a path string ends with a backslash, and then trim any extra whitespace away in the Context constructor.
Other alternatives considered and rejected:
1. Always trim trailing backslashes as the filesystem generally doesn't need them.
2. Escape trailing backslashes with another backslash. This proved complex as PS only quotes string args when the string includes some whitespace, and other backslashes within the string apparently don't need to be escaped.
* fix(powershell): Use Invoke-Native pattern for safely invoking native executables with strings which may contain characters which need to be escaped carefully.
* fix(context): Remove superfluous argument trims
These were in place to clean up extra whitespace sometimes injected by starship.ps1::prompt, and are no longer required with the new Invoke-Native helper in place.
* refactor(directory): Clean up the semantics of `logical_dir` defaulting it to `current_dir` but overridable by the `--logical-dir` flag.
- Restore `use_logical_path` config flag.
- Always attempt to contract repo paths from the `current_dir`.
* fix(directory) :Use logical_dir for contracting the home directory
This keeps the two calls to contract_path in sync.
* fix(directory): Remove test script
* refactor(directory): Convert current_dir to canonical filesystem path when use_logical_path = false
- This requires some clean-up to remove the extended-path prefix on Windows
- The configured logical_dir is ignored entirely in this mode - we calculate a new logical_dir by cleaning up the physical_dir path for display.
- Test coverage
* fix(directory): Use AsRef style for passing Path arguments
* fix(directory): Strip the windows extended-path prefix from the display string later in the render process
* fix(docs): Update docs/config/README.md for use_logical_path
* refactor(context): Populate `current_dir` from `--path` or `std::env::current_dir`, populate `logical_dir` from `--logical-path` or the `PWD` env var
- `current_dir` is always canonicalized
- On Windows, `current_dir` will have an extended-path prefix
- `logical_dir` is now always set
- `directory::module` now just selects between `current_dir` and `logical_dir` when picking which path to render
- Test coverage
* fix(directory): Fix path comparison operations in directory to ignore differences between path prefixes
- Added PathExt extension trait which adds `normalised_equals`, `normalised_starts_with` and `without_prefix`
* fix(path): Add test coverage for PathExt on *nix
* fix(directory): Test coverage for `contract_repo_path`, `contract_path` with variations of verbatim and non-verbatim paths
* fix(directory): Update path-slash to latest
This fixes the issue with the trailing character of some Windows paths being truncated, e.g. `\\server\share` and `C:`
* fix(powershell): Improve UTF8 output handling, argument encoding
- Use `ProcessStartInfo` to launch native executable, replacing manual UTF8 output encoding handling
- If we detect we're on PWSH6+ use the new `System.Diagnostics.ProcessStartInfo.ArgumentList` parameter, otherwise manually escape the argument string
- Move `Get-Cwd` and `Invoke-Native` into the prompt function scope so that they don't leak into the user's shell scope
* fix(path): Make PathExt methods no-ops on *nix
* fix(path): Cargo fmt
* fix(powershell): Remove typo ';'. Fix variable assignment lint.
This makes it possible to configure when the python module is shown
based on the contents of a directory. This should make it possible to
be a lot more granular when configuring the module.
This includes a breaking change since we are removing the
`scan_for_pyfiles` configuration option in favour of setting the
`detect_extensions` to an empty array.
* fix(windows): don't inherit stdin when executing commands
On Windows, inheriting stdin from starship might lead to leaking the
console reference to the command we're executing. `id.exe` supplied with
Git has been observed to disable the ENABLE_VIRTUAL_TERMINAL_PROCESSING
console flag if it inherits stdin -- leading to Windows Terminal not
processing ANSI escape sequences.
This change fixes#2254 by explicitly disabling stdin inheritance.
The fix was suggested by David Knaack.
* fix(username): don't call `id -u` on Windows
This was done to check if user is root by comparing the UID to 0. Windows
does not have a concept of UID 0 anyway, so it's pointless to call `id.exe`
(which is installed with MSYS2 or Git, for example).
* feat: add support for elvish shell
* improve doc
* elvish 0.15 is out
* fix example init
* update systax for 0.15 stable
* udpate second init example too
* remove warning from swift module
* add warning to status module docs
* prefix elvish version with v
In #1897 we replaced a 'wc -l' with a bash-native job counter, but
subsequently discovered that bash on MacOS folds '<<<' output into
a single line, preventing line counting.
A for loop works around that problem, is still bash-native, and works
on Linux as well.
While we're at it, also removed the need for command substitution and
an echo by doing the work directly on NUM_JOBS.
Fixes#2241.
I simplified the code in the git status module by moving everything from RwLock<_> to OnceCell<_>. I think this should also get rid of any remaining race conditions that remained after #1777.
We currently invoke `starship time` in a sub-shell to compute time, which is non-performant. By using $EPOCHREALTIME,
which is an inbuilt env-var in ZSH5, we can get better performance. This commit only targets ZSH5+ and maintains the
old behaviour.
* perf(zig): evaluate version lazily
* fix(zig): update format string; update tests
* refact(zig): remove redundant clone and put everything to do with version eval into match statement
* tiny optimization
Co-authored-by: David Knaack <davidkna@users.noreply.github.com>
Co-authored-by: Moritz Vetter <mv@3yourmind.com>
Co-authored-by: David Knaack <davidkna@users.noreply.github.com>
* perf(perl): evaluate version lazily
* fix(perl): update format string; update tests
* refact(perl): run rustfmt
Co-authored-by: Moritz Vetter <mv@3yourmind.com>
* perf(utils): Add timeout to `utils::exec_cmd`
This adds a timeout to any command executed using the `utils::exec_cmd`.
The initial time limit is hard coded to 500ms but if required we can
make this configurable. Have also switched the tests to be a bit more
granular on which systems they are ignored.
* Terminate the processes if they timeout
* fix(nodejs): update format string
* test(nodejs): adjust tests to new format strings
* fix(nodejs): use once_cell's Lazy to implement hassleless lazy execution
* chore(nodejs): run rustfmt
Co-authored-by: Moritz Vetter <mv@3yourmind.com>
* perf(dart): evaluate version lazily
* fix(dart): update format string
* fix: use suggested format string
Co-authored-by: Moritz Vetter <mv@3yourmind.com>
* perf(java): evaluate version lazily
* fix(java): update format string
* fix: use suggested format string
Co-authored-by: Moritz Vetter <mv@3yourmind.com>
* perf(elm): evaluate version lazily
* fix(elm): update format string
* fix: use suggested format string
Co-authored-by: Moritz Vetter <mv@3yourmind.com>
* fix(dotnet): update dotnet format string
* fix(dotnet): update erlang format string
* fix(dotnet): update golang format string
* fix(dotnet): update helm format string
* fix(dotnet): update julia format string
* fix(dotnet): update nim format string
* fix(dotnet): update ruby format string
* fix(dotnet): update rust format string
* test: update formatted strings in unit tests
* Use suggested format strings
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
Co-authored-by: Moritz Vetter <mv@3yourmind.com>
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
* perf(python): evaluate version lazily
* fix(python): update format string
* fix: use suggested format strings
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
Co-authored-by: Moritz Vetter <mv@3yourmind.com>
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
* perf(crystal): Lazily evaluate version command
This updates the module to lazily execute the `crystal --version`
command only when the `version` variable is used in the format string.
* perf(crystal): Update format string
Update format string to handle the `$version` variable not being set.
* feat: Add support for cygwin/msys2/git-bash evironment
* Update src/init/mod.rs
Co-authored-by: David Knaack <davidkna@users.noreply.github.com>
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
Co-authored-by: David Knaack <davidkna@users.noreply.github.com>
* fix(user): Fix username detection on Windows
This switches Windows to use the env var `USERNAME` rather than `USER`
for getting the username of the current user. I have also done some minor
refactoring to simplify some of the code and make the checks to see if
the user is root or another user a bit more robust.
* Update src/modules/username.rs
Co-authored-by: David Knaack <davidkna@users.noreply.github.com>
Co-authored-by: David Knaack <davidkna@users.noreply.github.com>
Have added support to the `utils::exec_cmd` to allow it to execute
commands that are not `.exe` on Windows. Have also added a timer to
measure how long a command took to execute.
I was surprised to see these (unprefixed) variables being set in my
shell. I think it's better to have them in the STARSHIP_ "namespace".
(Actually, STATUS got the new name STARSHIP_CMD_STATUS.)
* refactor: convert some vecs to static arrays and slices
* refactor(openstack): lazy yaml file reading, skip files without expected structure, and avoid panics for some edge cases
* feat(git_branch): Show remote name in addition to remote branch
* feat(git_branch): Fix table indentation in config README
* feat(git_branch): Use a different method to fetch remote information
* feat(git_branch): Fix the Clippy issue regarding string constant
* fix(git_commit): show last created tag on current commit
* style(git_commit): fixed linter & format alerts on PR checks
* test(git_commit): added 'test_latest_tag_shown_with_tag_enabled'
* test(git_commit): add a second between tags on same commit in 'test_latest_tag_shown_with_tag_enabled'
* feat: Try harder to guess if inside ssh
* Add test for SSH_CLIENT
* Update documentation on checking ssh connection
* Update docs/config/README.md
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
Changes the behaviour of a bg:none color string so that
it causes the background to have no color instead of
decoloring the entire string like it did before.
Update the python module to try multiple python binaries when
determining the version. With the new logic if starship doesn't find
`python` on the `PATH`, which is the default for some Linux Distros, it
will fallback to `python3` and then `python2`.
* feat(rust): Support new rust-toolchain format
* Match file parsing with rustup and update link
* Use cargo to deserialize the rust-toolchain file
* Filter empty channel strings after extraction
* Use the option value instead of rewrapping
* feat(git_branch): add remote branch name if different than local branch
* feat(git_branch): Implement a more customizable remote branch
* feat(git_branch): Use more explicit API function name
* feat(git_branch): Remove forgotten draft documentation
* feat(git_branch): Set less verbose defaults
* feat(git_branch): Handle case to always display remote
* feat(git_branch): Fix error in rebase operation
* feat(git_branch): add 'only_attached' config bool
This adds a new boolean, `only_attached`, to the configuration for
the `git_branch` module. This mirrors the `only_detached` config
value for the `git_commit` module: setting the value to true causes
the module to suppress its output when the user is not on a branch.
This allows users to have either a branch name or a commit hash in
their prompt, as opposed to having either a branch name or the
overly-wordy "HEAD (sha1)".
* Fix formatting
* Add CMakeCache.txt to cmake module
* Remove trailing whitespace
* Apply fixes by @andytom
* Add CMakeCache.txt to docs
* Revert documentation for languages other than en
* [f]Use_global_dollar_hook_for_lastExitCodeForPrompt Switched to use the dollar hook, as source of truth for exit code of las command, included comments to clarify.
* fix: Adjusted accordingly to comments on PR.
* fix: Moved last exit code handling inside if to reuse that variable.
Co-authored-by: Marcos Quezada Perez <marcos.quezadaperez@peakwork.com>
Have refactored the kubernetes module to better support stacked
kubeconfigs. Rather than trying to grab both the current-context and
namespace on a single pass we now do two passes. These changes should
fix the problems where the current-context is defined in one file but
the context and the namespace are defined in a different file.
Have switched to use a monotonic clock for calculating the timeout when
indexing the current directory for the context to avoid any issues with
calculating the timeout when the systems clock might change.
* fix: rust modules are compiled twice
The modules being declared both in the _library_ crate and the _binary_
crate made cargo compile everything twice:
1. for the library
2. for the binary
What happened was:
1. The library "starship" was compiled.
2. The binary "starship" (and all its modules) were compiled.
* fix: stop compiling every rust module twice
restrict visibility
* Run cargo fmt
* Add bug_report module
Co-authored-by: Matan Kushner <hello@matchai.dev>
Co-authored-by: Kevin Song <chips@ksong.dev>
When running the AWS module it will parse the AWS config found in
`~/.aws/config` to get the region. This means that tests that expect no
region to be set will fail if there exists a default profile with a
region set, which is probably true for most AWS users. To avoid this
have set the AWS tests that depend on the non-existance of a
`.aws/config` to be ignored.
Have added a little bit more context to the error messages that occur
when if starship is unable to setup the logger. This should hopefully
make it a bit easier to work out why starship can't setup the logger.
* ✨ Add --prompt rendering to python virtualenv
* Use pyvenv.cfg to find prompt
* Remove usage of VIRTUAL_ENV_PROMPT variable
Seeing how both venv and virtualenv set a pyvenv.cfg, this isn't needed.
Additionally pyvenv.cfg is set in earlier versions than VIRTUAL_ENV_PROMPT,
which at this moment is in the unrelased python 3.10
* Smarter result unwrapping thanks to clippy
Some users have commit.gpgSign set to true in their global git config,
causing tests which run `git commit` to fail if the configured
user.signingKey is not present (as is often the case with PGP smart
cards). Passing this flag overrides the global git configuration,
preventing git from attempting sign the specified commit. This change
has no effect on users who haven't set commit.gpgSign to true.
Have reduced the log level when we fail to execute a command since this
is happening a lot more that expected, for example if a user checks out
a repo that contains a `.js` file but they don't have node installed.
The tests for git-state assume the default branch is `master`.
In git 2.28 support for a global option, `init.defaultBranch`, to
change the default branch name which causes git-state tests to fail.
This creates a custom logger for the log crate which logs everything to a file (/tmp/starship/session_$STARSHIP_SESSION_KEY.log) and it logs everything above Warn to stderr, but only if the log file does not contain the line that should be logged resulting in an error or warning to be only logged at the first starship invocation after opening the shell.
The default `disabled: true` is actually only available within the module (when the config struct is used and not the user toml) but not all (the hg_branch) modules checked it there again.
Document this in all places and add the check (+ test) to the hg_branch module.
* add the exit code module
this allows to display more precisely the last command exit code
and to configure starship to not change the last charcter of the
prompt even in case of failure. It is disabled by default, because
it seems a bit redundant with the character module in its default
configuration.
* rename exit_code module to status
* Enforce a default disabled=true
In the outer places, we only check for the disabled flag in the config toml file, only when this is loaded into the real config struct, we see the default. And if the default is true, we have to abort at that place. For status and hg_branch that wasn't so. I also commented the rest
* fix spaces in markdown table for status module
* Add a tip that status module is disabled by default
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
* Remove unrelated changes for default disabled=true
Co-authored-by: Gaëtan Lehmann <gaetan.lehmann@gmail.com>
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
* feat: Add computational duration to all computed modules
This also means that in case we do some computations and these end up empty, we submit an empty module
* feat: Add timings subcommand
This outputs the timings of all computed modules, sorted by the duration it took to compute the module.
Useful for debugging why the prompt takes so long.
* feat: Add timings to explain output
* fix: Ensure that even empty custom modules get timings
* format main.rs
* feat: Only show interesting timings
* fix(tests): Change tests to look for empty string instead of None
* Use proper wording in timings help
* Revert "fix(tests): Change tests to look for empty string instead of None"
This reverts commit aca5bd1b03c48e1dee1b7ca91d66e2bda2d5a97c.
* fix(tests): Returning None in case the module produced an empty string
* fix: Ensure that linebreaks (and space) make a module not-empty
* Make cargo clippy happy
* Make Module.duration a proper Duration
* Only return a module if we would report it
* Change to cleaner way to return None for empty modules
* Avoid unnecessary module creation
* Simplify a string comparison
* Add timings to trace
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
Have restored the `pyenv_prefix` option to the python module. This is
added as a new variable that will only be shown if `pyenv` is being used
to determine the version of python that is being used.
Previously the prompt function used in PowerShell would overwrite the $LASTEXITCODE and $? automatic variables that were set by the previous command run the user in the shell. This results in surprising behavior for the user if they inspect those variables looking for the result of the command they last ran.
This fixes the bug reported here: https://github.com/starship/starship/issues/1051
And goes further to also propagate the $? automatic variable which is not mentioned in that bug.
Previously, all modules would have prefixes, which lead to the first
module having a dangling prefix. This change ensures that the first
few modules would instead have suffixes so that we don't start or
end with a prefix or suffix.
* docs: Clarify that commands will be passed in on stdin
* docs: Clearer instruction how to include individual custom modules
* docs: Include link to #1252 in docs for custom modules
That issue is used to share custom modules.
* docs: Remove reference to prompt_order
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
Co-authored-by: Thomas O'Donnell <andytom@users.noreply.github.com>
Convert-Path in the powershell prompt script works with -Path,
which interprets paths as wildcard patterns.
Not all valid paths are also valid wildcard patterns, possibly causing
the prompt to error (eg []*).
Replace it with -LiteralPath that makes Convert-Path use the path as-is.
Often it is handy to get notified when the execution of a command finished.
This commit includes notify-rust in order to generate a desktop notification
when a command execution finished.
* fix(explain): align table correctly
* iterate over lines directly
* calculate desc_width with the actual space available
* custom unicode-aware textwrapping
* fix clippy error
* better width estimination
* explain +6
* move padding width into a constant
* initial commit of support for shlvl
* documentation for shlvl
* use a symbol instead
* test coverage for shlvl
* actually disable when the config says to
* fix docs
* tweak defaults some
* refactor from pr comments
* redisable
* return early to avoid indenting
* make default suffix empty space
* fixing tests after suffix change
* updating docs for format
* making shlvl compatible with formatting
* adding variables table for shlvl
* removing extra line
* doc clarity
* Add gitignore for Emacs backup files.
* Add package version support for Maven pom.xml.
* Update docs with Maven package version support.
* Fix for clippy.
Do not try to analyze if the current process can write network location
on Windows. There's no way on Windows to tell if we can write a network
location because it's not being controlled by the OS itself. Thus now
the lock symbol is never shown on network locations.
This PR introduces a new unsafe call.
* refactor(java): parse version using regex
Mock java version retrieval & extend java module test suite with
rendering tests.
* chore: remove nom crate
* fix(java): support parsing version from both stdout & stderr
* fix(java): fix java command mock
* refactor(java): simplify version regex
David Knaack
Benjamin Fox
Thomas O'Donnell
KOVACS Krisztian
Rich Lafferty
Furkan Türkal
Rodrigo Suárez
Gautham Warrier
Rust大闸蟹
Moritz Vetter
Alexandru Macovei
Hanif Ariffin
Ben Iofel
Kevin Song
Zoritle
Glandos
Hugues Morisset
Nicola Corti
dependabot-preview[bot]
花蛄
Bjørn Forsman
Daniel Beckwith
Sagittarius-a
Miguel Ángel Melón Pérez
yangsheng6810
quentm74
t-mangoe
Cody Bloemhard
Harald Hoyer
Dominik Nakamura
Zane Shelby
Gabriel de Perthuis
Aaron Gable
Masashi Aso
Adham
Jamie Kirkpatrick
Shu Kutsuzawa
Christoph Schlosser
Mauricio Teixeira
marcos-quezada
jRimbault
SarthakSingh31
Matan Kushner
Tilmann Meyer
Tomer
Jeremy Hilliker
Logan Magee
David Barnett
aspen
Jan Katins
Joshua Poehls
Dario Vladović
jRimbault
Martin Åkesson
Marc Schreiber
Daniel A. White
Stéphane Benoit
Takahiro Tsuruda
Jason Wang
Steve Smith
Alexey Chernyshov
Simão Mata
LOU Xun
Batuhan Apaydın