Instead of having the option do nothing at runtime on unsupported
platforms, it is now only available on the systems that support it in
the first place.
This adds a `--same-file-system` CLI option that instructs the walker to
not cross file system boundaries.
Due to the fact that the corresponding option of the `ignore` crate's
`WalkBuilder` does not support platforms other than Unix and Windows,
the option does nothing on platforms other than those.
Resolves#507
This is just a split commit, refraining from renaming too much.
The drop(tx) call is no longer necessary, as the first sender
is dropped at the end of spawn_senders(..)
This PR splits `internals.rs` into the `internal` module with multiple
files, and moves `FdOptions` into `opts.rs`.
The main motivation behind this is to move logic for constructing
`FdOptions` out of the main function and more readable and easier to
understand in the `opts` module. The goal will eventually to be able
to write `FdOptions::from(matches)` and have the options constructed.
Previously, the constants defined in `src/exit_codes` weren't being
used, and the constants for exit codes were being redefined in the
`internals` module.
This PR removes the exit code consts and instead uses an enum defined in
`src/exit_codes`. This centralizes the definitions of exit codes making
them easier to modify and keep track of.
Add a new `empty`/`e` type to search for empty files and/or directories.
To search for both empty files and directories, use one of the
following:
fd --type empty
fd -te
fd --type empty --type file --type directory
To search for empty files, use
fd --type empty --type file
fd -te -tf
To search for empty directories, use
fd --type empty --type directory
fd -te -td
closes#273
* Adding support for multiple paths. (panic)
- Started adding multiple file support
- fd panics with multiple files right now
* Moved the ctrlc handler to main.
- Moved the ctrlc handler to main so we can search multiple files
* Tests now allow custom directory setup
- TestEnv::new() now takes two arguments, the directories to create and
the files to create inside those directories.
* rust-fmt changes
* rust-fmt changes
* Moving code around, no need to do everything in one big loop
- PathDisplay was never actually used for anything, removed it during refactor of main
- Removed redundant logic for absolute paths
- Moved code placed needlessly inside a loop in the last commit outside of that loop.
* Moving code around, no need to do everything in one big loop
- PathDisplay was never actually used for anything, removed it during refactor of main
- Removed redundant logic for absolute paths
- Moved code placed needlessly inside a loop in the last commit outside of that loop.
* Removed commented code in testenv
* Refactored walk::scan to accept the path buffer vector. Using the ParallelWalker allows for multithreaded searching of multiple directories
* Moved ctrlc handler back into walker, it is only called once from main now.
* Moved the colored output check back to it's original place
* Removing shell-escape, not sure how it got added...
* Added test for `fd 'a.foo' test1` to show that a.foo is only found in the test1 and not the test2 direcotry
* Removing side effect from walk::scan, `dir_vec` is no longer a mutable reference and an iterator is being used instead.
* Running rustfmt to format code correctly
Previously, we were using the relative path to extract the search
string. For the current directory, the returned relative path was equal
to `""`. This is why the current directory did not show in the search
results (see #81).
This commit also changes the way that `--full-path` works, which was
previously working on relative paths. It seems more useful to search the
absolute path, though. Otherwise, search results could change just by
calling fd (with a given search path) from a different directory.
