An argument substitution utility intended to make working on projects with deeply nested directories a bit easier.
"Lax" stands for "Locate Arguments and Execute."
Given a binary and an "@" pattern "@foo", Lax will will find the file "foo" and replace "@foo" with "foo"'s full path, then execute the binary with the new arguments
lax echo @foo
-> echo ./foobar/foo
Multiple "@" patterns are possible:
lax stat @foo @bar @baz
-> stat ./foobar/foo ./foobar/target/bar ./foobar/src/baz
Mixing and matching "@" patterns and normal arguments is also possible:
lax cat -n @foo bar @baz
-> cat -n ./foobar/foo bar ./foobar/src/baz
Globbing is fully supported via globset.
lax echo @*.rs
-> echo ./some/directory/main.rs
**
is also supported, but is treated specially. The portion of the pattern
before the first /**/
(note the surrounding slashes) is the search entry
point. That is, given a pattern like @foo/**/bar
, Lax will look in the
directory ./foo
for a file that matches bar
. Any following **
are handled
normally:
$ lax echo @foo/**/bar/**/baz
foo/bizz/bazz/bar/beez/baz
Will look in directory ./foo
for a path that matches bar/**/baz
. That is, any
entity baz
that is a descendent of directory bar
.
Making use of the search entry point can speed up searches if you know which top-level
subdirectory your query is in, but you don't want to cd
into it for whatever
reason. It can also be used to specify a path outside your directory.
You can search from the git(or svn) root by using a %
:
lax vim @%README.md
If there are multiple files matching the given name, Lax will prompt you to choose. However, you can also specify which file you'd want ahead of time:
$ lax echo @*.rs^1 # Select the first match
a.rs
$ lax echo @*.rs^2 # Select the second match
b.rs
$ lax echo @*.rs^l # Select the last match. You can also use '-1'
d.rs
$ lax echo @*.rs^-2 # Select the second to last match
c.rs
$ lax echo @*.rs^1,3 # Select the first and third match
a.rs c.rs
$ lax echo @*.rs^a # Select all matches
a.rs b.rs c.rs d.rs
$ lax echo @*.rs^/[ab] # Select with regex
a.rs b.rs
Now you know the full syntax for "@" patterns:
@[%][SEARCH_ENTRY_POINT/**/]GLOB_PATTERN[^SELECTOR[,SELECTOR]...]
Where SEARCH_ENTRY_POINT
is a directory, GLOB_PATTERN
is a glob pattern,
and SELECTOR
is [-n..-1|1..n|'a'|'l'|/regex]
# Escape the initial '@' symbol
$ lax echo \\@
@
$ lax echo '\@'
@
# Only match *directories* by adding a forward slash
$ lax echo @foo/
./foo/
# You can also use command-line options to achieve a similar effect
$ lax -d echo @foo
./foo/
# Or only look for files
$ lax -f echo @foo
./tests/foobar/foo
# Or transform a file to its parent
$ lax -fD echo @foo
./tests/foobar
# We also have the ability to specify fallback binaries. This will use `cowsay`
# if it's installed, otherwise it will fallback to `echo`
$ lax 'cowsay|echo' hello
_______
< hello >
-------
\ ^__^
\ (oo)\_______
(__)\ )\/\
||----w |
|| ||
In your .bashrc
, you can write alias vim="lax vim"
From now on, you can just write:
vim @foo
-> vim ./foobar/foo
This makes working on projects with deep directories, like U-Boot and Yocto,
easier. If the compiler complains about an error in
my_stupid_little_c_file.c
, you can:
vim @*stupid*file.c
You might try to use Lax to do things like:
lax cd @some_deep_nested_subdirectory
But you will quickly discover this does not work as intended, as cd is affecting the current directory of its environment, which belongs to a child process of the shell.
Instead, you can add this to your .bashrc
:
cd(){
local args;
if ! args=$(lax -pd -- ${@}); then
return 1
fi
command cd ${args}
}
The -p
flag tells Lax to not execute anything, but simply transform arguments
and print them to stdout. The -d
flag tells it to only match with directories,
as cd
has no interest in files. You could also use the -D
flag if you want
to match files, but cd
to their parent directory, instead.
cargo install --git https://github.com/Property404/lax
or for a more optimized build:
cargo install --git https://github.com/Property404/lax --profile lto
MIT or Apache-2.0