2015-05-16 15:10:58 +00:00
|
|
|
|
//! Files, and methods and fields to access their metadata.
|
|
|
|
|
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
use std::fs;
|
2016-10-29 19:27:23 +00:00
|
|
|
|
use std::io::Error as IOError;
|
2015-12-15 21:47:37 +00:00
|
|
|
|
use std::io::Result as IOResult;
|
2015-05-03 11:46:05 +00:00
|
|
|
|
use std::os::unix::fs::{MetadataExt, PermissionsExt};
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
use std::path::{Path, PathBuf};
|
2014-05-04 20:33:14 +00:00
|
|
|
|
|
2016-04-16 17:59:25 +00:00
|
|
|
|
use fs::dir::Dir;
|
|
|
|
|
use fs::fields as f;
|
2014-06-16 23:27:05 +00:00
|
|
|
|
|
2016-06-14 03:14:36 +00:00
|
|
|
|
#[cfg(any(target_os = "macos", target_os = "linux"))]
|
|
|
|
|
use std::os::unix::fs::FileTypeExt;
|
2015-11-04 14:56:37 +00:00
|
|
|
|
|
2017-05-01 21:43:28 +00:00
|
|
|
|
|
2017-05-03 09:05:16 +00:00
|
|
|
|
#[allow(trivial_numeric_casts)]
|
2015-11-04 14:56:37 +00:00
|
|
|
|
mod modes {
|
2017-05-03 09:05:16 +00:00
|
|
|
|
use libc;
|
|
|
|
|
|
|
|
|
|
pub type Mode = u32;
|
|
|
|
|
// The `libc::mode_t` type’s actual type varies, but the value returned
|
|
|
|
|
// from `metadata.permissions().mode()` is always `u32`.
|
|
|
|
|
|
|
|
|
|
pub const USER_READ: Mode = libc::S_IRUSR as Mode;
|
|
|
|
|
pub const USER_WRITE: Mode = libc::S_IWUSR as Mode;
|
|
|
|
|
pub const USER_EXECUTE: Mode = libc::S_IXUSR as Mode;
|
|
|
|
|
pub const GROUP_READ: Mode = libc::S_IRGRP as Mode;
|
|
|
|
|
pub const GROUP_WRITE: Mode = libc::S_IWGRP as Mode;
|
|
|
|
|
pub const GROUP_EXECUTE: Mode = libc::S_IXGRP as Mode;
|
|
|
|
|
pub const OTHER_READ: Mode = libc::S_IROTH as Mode;
|
|
|
|
|
pub const OTHER_WRITE: Mode = libc::S_IWOTH as Mode;
|
|
|
|
|
pub const OTHER_EXECUTE: Mode = libc::S_IXOTH as Mode;
|
2015-11-04 14:56:37 +00:00
|
|
|
|
}
|
2015-11-04 08:51:34 +00:00
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
|
2015-01-24 12:38:05 +00:00
|
|
|
|
/// A **File** is a wrapper around one of Rust's Path objects, along with
|
|
|
|
|
/// associated data about the file.
|
|
|
|
|
///
|
|
|
|
|
/// Each file is definitely going to have its filename displayed at least
|
2015-05-16 17:16:35 +00:00
|
|
|
|
/// once, have its file extension extracted at least once, and have its metadata
|
2015-01-24 12:38:05 +00:00
|
|
|
|
/// information queried at least once, so it makes sense to do all this at the
|
|
|
|
|
/// start and hold on to all the information.
|
2015-05-12 14:38:12 +00:00
|
|
|
|
pub struct File<'dir> {
|
2015-05-16 15:10:58 +00:00
|
|
|
|
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
/// The filename portion of this file's path, including the extension.
|
|
|
|
|
///
|
|
|
|
|
/// This is used to compare against certain filenames (such as checking if
|
|
|
|
|
/// it’s “Makefile” or something) and to highlight only the filename in
|
|
|
|
|
/// colour when displaying the path.
|
2015-08-25 14:04:15 +00:00
|
|
|
|
pub name: String,
|
2015-05-16 15:10:58 +00:00
|
|
|
|
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
/// The file’s name’s extension, if present, extracted from the name.
|
|
|
|
|
///
|
|
|
|
|
/// This is queried many times over, so it’s worth caching it.
|
2015-08-25 14:04:15 +00:00
|
|
|
|
pub ext: Option<String>,
|
2015-05-16 15:10:58 +00:00
|
|
|
|
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
/// The path that begat this file.
|
|
|
|
|
///
|
|
|
|
|
/// Even though the file's name is extracted, the path needs to be kept
|
|
|
|
|
/// around, as certain operations involve looking up the file's absolute
|
|
|
|
|
/// location (such as the Git status, or searching for compiled files).
|
2015-08-25 14:04:15 +00:00
|
|
|
|
pub path: PathBuf,
|
2015-05-16 15:10:58 +00:00
|
|
|
|
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
/// A cached `metadata` call for this file.
|
|
|
|
|
///
|
|
|
|
|
/// This too is queried multiple times, and is *not* cached by the OS, as
|
|
|
|
|
/// it could easily change between invocations - but exa is so short-lived
|
|
|
|
|
/// it's better to just cache it.
|
2015-08-25 14:04:15 +00:00
|
|
|
|
pub metadata: fs::Metadata,
|
2015-05-16 15:10:58 +00:00
|
|
|
|
|
|
|
|
|
/// A reference to the directory that contains this file, if present.
|
|
|
|
|
///
|
|
|
|
|
/// Filenames that get passed in on the command-line directly will have no
|
|
|
|
|
/// parent directory reference - although they technically have one on the
|
|
|
|
|
/// filesystem, we'll never need to look at it, so it'll be `None`.
|
|
|
|
|
/// However, *directories* that get passed in will produce files that
|
|
|
|
|
/// contain a reference to it, which is used in certain operations (such
|
|
|
|
|
/// as looking up a file's Git status).
|
2015-08-25 14:04:15 +00:00
|
|
|
|
pub dir: Option<&'dir Dir>,
|
2014-05-04 20:33:14 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-05-12 14:38:12 +00:00
|
|
|
|
impl<'dir> File<'dir> {
|
2015-09-03 17:48:53 +00:00
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// Create a new `File` object from the given `Path`, inside the given
|
|
|
|
|
/// `Dir`, if appropriate.
|
2015-01-24 12:38:05 +00:00
|
|
|
|
///
|
2015-05-03 12:10:25 +00:00
|
|
|
|
/// This uses `symlink_metadata` instead of `metadata`, which doesn't
|
|
|
|
|
/// follow symbolic links.
|
2015-12-15 21:47:37 +00:00
|
|
|
|
pub fn from_path(path: &Path, parent: Option<&'dir Dir>) -> IOResult<File<'dir>> {
|
2015-08-26 08:35:11 +00:00
|
|
|
|
fs::symlink_metadata(path).map(|metadata| File::with_metadata(metadata, path, parent))
|
2014-11-25 01:27:26 +00:00
|
|
|
|
}
|
2014-11-25 20:50:23 +00:00
|
|
|
|
|
2015-05-16 17:16:35 +00:00
|
|
|
|
/// Create a new File object from the given metadata result, and other data.
|
2015-08-26 08:35:11 +00:00
|
|
|
|
pub fn with_metadata(metadata: fs::Metadata, path: &Path, parent: Option<&'dir Dir>) -> File<'dir> {
|
2017-04-25 22:33:57 +00:00
|
|
|
|
let filename = match path.components().next_back() {
|
|
|
|
|
Some(comp) => comp.as_os_str().to_string_lossy().to_string(),
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
None => String::new(),
|
|
|
|
|
};
|
2014-06-21 18:39:27 +00:00
|
|
|
|
|
2014-12-12 11:17:55 +00:00
|
|
|
|
File {
|
2015-09-03 17:48:53 +00:00
|
|
|
|
path: path.to_path_buf(),
|
|
|
|
|
dir: parent,
|
|
|
|
|
metadata: metadata,
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
ext: ext(path),
|
|
|
|
|
name: filename,
|
2014-12-12 11:17:55 +00:00
|
|
|
|
}
|
2014-05-04 20:33:14 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// Whether this file is a directory on the filesystem.
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
pub fn is_directory(&self) -> bool {
|
2015-05-16 17:16:35 +00:00
|
|
|
|
self.metadata.is_dir()
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-09-03 17:48:53 +00:00
|
|
|
|
/// If this file is a directory on the filesystem, then clone its
|
|
|
|
|
/// `PathBuf` for use in one of our own `Dir` objects, and read a list of
|
|
|
|
|
/// its contents.
|
|
|
|
|
///
|
|
|
|
|
/// Returns an IO error upon failure, but this shouldn't be used to check
|
|
|
|
|
/// if a `File` is a directory or not! For that, just use `is_directory()`.
|
2015-12-15 21:47:37 +00:00
|
|
|
|
pub fn to_dir(&self, scan_for_git: bool) -> IOResult<Dir> {
|
2015-09-03 09:51:59 +00:00
|
|
|
|
Dir::read_dir(&*self.path, scan_for_git)
|
2015-08-26 08:35:11 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// Whether this file is a regular file on the filesystem - that is, not a
|
|
|
|
|
/// directory, a link, or anything else treated specially.
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
pub fn is_file(&self) -> bool {
|
2015-05-16 17:16:35 +00:00
|
|
|
|
self.metadata.is_file()
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// Whether this file is both a regular file *and* executable for the
|
|
|
|
|
/// current user. Executable files have different semantics than
|
|
|
|
|
/// executable directories, and so should be highlighted differently.
|
2015-05-07 21:27:33 +00:00
|
|
|
|
pub fn is_executable_file(&self) -> bool {
|
2015-11-04 14:56:37 +00:00
|
|
|
|
let bit = modes::USER_EXECUTE;
|
2015-05-16 17:16:35 +00:00
|
|
|
|
self.is_file() && (self.metadata.permissions().mode() & bit) == bit
|
2015-05-07 21:27:33 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// Whether this file is a symlink on the filesystem.
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
pub fn is_link(&self) -> bool {
|
2015-05-16 17:16:35 +00:00
|
|
|
|
self.metadata.file_type().is_symlink()
|
2016-10-29 19:27:23 +00:00
|
|
|
|
}
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// Whether this file is a dotfile, based on its name. In Unix, file names
|
|
|
|
|
/// beginning with a dot represent system or configuration files, and
|
|
|
|
|
/// should be hidden by default.
|
2014-05-05 09:51:24 +00:00
|
|
|
|
pub fn is_dotfile(&self) -> bool {
|
2017-03-31 16:10:18 +00:00
|
|
|
|
self.name.starts_with('.')
|
2014-05-05 09:51:24 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-01-24 17:26:26 +00:00
|
|
|
|
/// Assuming the current file is a symlink, follows the link and
|
|
|
|
|
/// returns a File object from the path the link points to.
|
|
|
|
|
///
|
|
|
|
|
/// If statting the file fails (usually because the file on the
|
2016-06-11 15:54:06 +00:00
|
|
|
|
/// other end doesn't exist), returns the path to the file
|
2015-01-24 17:26:26 +00:00
|
|
|
|
/// that should be there.
|
2016-10-29 19:27:23 +00:00
|
|
|
|
pub fn link_target(&self) -> FileTarget<'dir> {
|
2015-05-11 22:28:01 +00:00
|
|
|
|
let path = match fs::read_link(&self.path) {
|
|
|
|
|
Ok(path) => path,
|
2016-10-29 19:27:23 +00:00
|
|
|
|
Err(e) => return FileTarget::Err(e),
|
2015-05-11 22:28:01 +00:00
|
|
|
|
};
|
|
|
|
|
|
2017-04-25 23:06:26 +00:00
|
|
|
|
let (metadata, ext) = {
|
|
|
|
|
let target_path_ = match self.dir {
|
|
|
|
|
Some(dir) if dir.path != Path::new(".") => Some(dir.join(&*path)),
|
|
|
|
|
_ => None
|
|
|
|
|
};
|
|
|
|
|
let target_path = target_path_.as_ref().unwrap_or(&path);
|
|
|
|
|
// Use plain `metadata` instead of `symlink_metadata` - we *want* to follow links.
|
|
|
|
|
(fs::metadata(&target_path), ext(&target_path))
|
2015-05-11 22:28:01 +00:00
|
|
|
|
};
|
|
|
|
|
|
2017-04-25 23:06:26 +00:00
|
|
|
|
let filename = match path.components().next_back() {
|
2017-04-25 22:33:57 +00:00
|
|
|
|
Some(comp) => comp.as_os_str().to_string_lossy().to_string(),
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
None => String::new(),
|
|
|
|
|
};
|
2014-11-25 20:50:23 +00:00
|
|
|
|
|
2017-04-25 23:06:26 +00:00
|
|
|
|
if let Ok(metadata) = metadata {
|
2016-10-29 19:27:23 +00:00
|
|
|
|
FileTarget::Ok(File {
|
2017-04-25 23:06:26 +00:00
|
|
|
|
path: path,
|
2015-09-03 17:48:53 +00:00
|
|
|
|
dir: self.dir,
|
|
|
|
|
metadata: metadata,
|
2017-04-25 23:06:26 +00:00
|
|
|
|
ext: ext,
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
name: filename,
|
2015-01-24 17:26:26 +00:00
|
|
|
|
})
|
|
|
|
|
}
|
|
|
|
|
else {
|
2017-04-25 23:06:26 +00:00
|
|
|
|
FileTarget::Broken(path)
|
2014-06-21 18:39:27 +00:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// This file's number of hard links.
|
2015-01-24 12:38:05 +00:00
|
|
|
|
///
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// It also reports whether this is both a regular file, and a file with
|
|
|
|
|
/// multiple links. This is important, because a file with multiple links
|
|
|
|
|
/// is uncommon, while you can come across directories and other types
|
|
|
|
|
/// with multiple links much more often. Thus, it should get highlighted
|
|
|
|
|
/// more attentively.
|
2015-05-12 02:33:40 +00:00
|
|
|
|
pub fn links(&self) -> f::Links {
|
2015-06-16 23:49:29 +00:00
|
|
|
|
let count = self.metadata.nlink();
|
2015-05-11 22:28:01 +00:00
|
|
|
|
|
2015-05-12 02:33:40 +00:00
|
|
|
|
f::Links {
|
2015-05-11 22:28:01 +00:00
|
|
|
|
count: count,
|
|
|
|
|
multiple: self.is_file() && count > 1,
|
|
|
|
|
}
|
2015-01-24 12:38:05 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// This file's inode.
|
2015-05-12 02:33:40 +00:00
|
|
|
|
pub fn inode(&self) -> f::Inode {
|
2015-06-16 23:49:29 +00:00
|
|
|
|
f::Inode(self.metadata.ino())
|
2015-01-24 12:38:05 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// This file's number of filesystem blocks.
|
|
|
|
|
///
|
|
|
|
|
/// (Not the size of each block, which we don't actually report on)
|
2015-05-12 02:33:40 +00:00
|
|
|
|
pub fn blocks(&self) -> f::Blocks {
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
if self.is_file() || self.is_link() {
|
2015-06-16 23:49:29 +00:00
|
|
|
|
f::Blocks::Some(self.metadata.blocks())
|
2015-01-24 12:38:05 +00:00
|
|
|
|
}
|
|
|
|
|
else {
|
2015-05-12 02:33:40 +00:00
|
|
|
|
f::Blocks::None
|
2015-01-24 12:38:05 +00:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// The ID of the user that own this file.
|
2015-05-12 02:33:40 +00:00
|
|
|
|
pub fn user(&self) -> f::User {
|
2015-06-16 23:49:29 +00:00
|
|
|
|
f::User(self.metadata.uid())
|
2015-01-24 12:38:05 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// The ID of the group that owns this file.
|
2015-05-12 02:33:40 +00:00
|
|
|
|
pub fn group(&self) -> f::Group {
|
2015-06-16 23:49:29 +00:00
|
|
|
|
f::Group(self.metadata.gid())
|
2015-01-24 12:38:05 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// This file's size, if it's a regular file.
|
2015-01-24 12:38:05 +00:00
|
|
|
|
///
|
|
|
|
|
/// For directories, no size is given. Although they do have a size on
|
|
|
|
|
/// some filesystems, I've never looked at one of those numbers and gained
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// any information from it. So it's going to be hidden instead.
|
2015-05-12 02:33:40 +00:00
|
|
|
|
pub fn size(&self) -> f::Size {
|
2015-02-26 14:05:26 +00:00
|
|
|
|
if self.is_directory() {
|
2015-05-12 02:33:40 +00:00
|
|
|
|
f::Size::None
|
2014-11-25 20:50:23 +00:00
|
|
|
|
}
|
|
|
|
|
else {
|
2015-05-16 17:16:35 +00:00
|
|
|
|
f::Size::Some(self.metadata.len())
|
2014-05-25 14:52:36 +00:00
|
|
|
|
}
|
2014-05-04 20:33:14 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-11-15 16:12:16 +00:00
|
|
|
|
pub fn modified_time(&self) -> f::Time {
|
|
|
|
|
f::Time(self.metadata.mtime())
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
pub fn created_time(&self) -> f::Time {
|
|
|
|
|
f::Time(self.metadata.ctime())
|
|
|
|
|
}
|
2015-02-09 18:14:05 +00:00
|
|
|
|
|
2015-11-15 16:12:16 +00:00
|
|
|
|
pub fn accessed_time(&self) -> f::Time {
|
|
|
|
|
f::Time(self.metadata.mtime())
|
2015-02-09 16:33:27 +00:00
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// This file's 'type'.
|
2015-01-24 12:38:05 +00:00
|
|
|
|
///
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// This is used in the leftmost column of the permissions column.
|
2015-01-24 12:38:05 +00:00
|
|
|
|
/// Although the file type can usually be guessed from the colour of the
|
|
|
|
|
/// file, `ls` puts this character there, so people will expect it.
|
2016-04-16 19:01:45 +00:00
|
|
|
|
pub fn type_char(&self) -> f::Type {
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
if self.is_file() {
|
2015-05-12 02:33:40 +00:00
|
|
|
|
f::Type::File
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
}
|
|
|
|
|
else if self.is_directory() {
|
2015-05-12 02:33:40 +00:00
|
|
|
|
f::Type::Directory
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
}
|
|
|
|
|
else if self.is_pipe() {
|
2015-05-12 02:33:40 +00:00
|
|
|
|
f::Type::Pipe
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
}
|
|
|
|
|
else if self.is_link() {
|
2015-05-12 02:33:40 +00:00
|
|
|
|
f::Type::Link
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
}
|
2016-06-14 03:14:36 +00:00
|
|
|
|
else if self.is_char_device() {
|
|
|
|
|
f::Type::CharDevice
|
|
|
|
|
}
|
|
|
|
|
else if self.is_block_device() {
|
|
|
|
|
f::Type::BlockDevice
|
|
|
|
|
}
|
|
|
|
|
else if self.is_socket() {
|
|
|
|
|
f::Type::Socket
|
|
|
|
|
}
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
else {
|
2015-05-12 02:33:40 +00:00
|
|
|
|
f::Type::Special
|
2014-05-04 20:33:14 +00:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// This file's permissions, with flags for each bit.
|
2015-09-03 17:48:53 +00:00
|
|
|
|
///
|
|
|
|
|
/// The extended-attribute '@' character that you see in here is in fact
|
|
|
|
|
/// added in later, to avoid querying the extended attributes more than
|
|
|
|
|
/// once. (Yes, it's a little hacky.)
|
2015-05-12 02:33:40 +00:00
|
|
|
|
pub fn permissions(&self) -> f::Permissions {
|
2015-05-16 17:16:35 +00:00
|
|
|
|
let bits = self.metadata.permissions().mode();
|
2015-05-11 22:28:01 +00:00
|
|
|
|
let has_bit = |bit| { bits & bit == bit };
|
|
|
|
|
|
2015-05-12 02:33:40 +00:00
|
|
|
|
f::Permissions {
|
2015-11-04 14:56:37 +00:00
|
|
|
|
user_read: has_bit(modes::USER_READ),
|
|
|
|
|
user_write: has_bit(modes::USER_WRITE),
|
|
|
|
|
user_execute: has_bit(modes::USER_EXECUTE),
|
|
|
|
|
group_read: has_bit(modes::GROUP_READ),
|
|
|
|
|
group_write: has_bit(modes::GROUP_WRITE),
|
|
|
|
|
group_execute: has_bit(modes::GROUP_EXECUTE),
|
|
|
|
|
other_read: has_bit(modes::OTHER_READ),
|
|
|
|
|
other_write: has_bit(modes::OTHER_WRITE),
|
|
|
|
|
other_execute: has_bit(modes::OTHER_EXECUTE),
|
2014-05-26 10:50:46 +00:00
|
|
|
|
}
|
|
|
|
|
}
|
2015-01-24 12:38:05 +00:00
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// Whether this file's extension is any of the strings that get passed in.
|
|
|
|
|
///
|
|
|
|
|
/// This will always return `false` if the file has no extension.
|
2015-05-09 15:10:26 +00:00
|
|
|
|
pub fn extension_is_one_of(&self, choices: &[&str]) -> bool {
|
|
|
|
|
match self.ext {
|
|
|
|
|
Some(ref ext) => choices.contains(&&ext[..]),
|
|
|
|
|
None => false,
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// Whether this file's name, including extension, is any of the strings
|
|
|
|
|
/// that get passed in.
|
2015-05-09 15:10:26 +00:00
|
|
|
|
pub fn name_is_one_of(&self, choices: &[&str]) -> bool {
|
|
|
|
|
choices.contains(&&self.name[..])
|
|
|
|
|
}
|
|
|
|
|
|
2015-05-16 15:10:58 +00:00
|
|
|
|
/// This file's Git status as two flags: one for staged changes, and the
|
|
|
|
|
/// other for unstaged changes.
|
|
|
|
|
///
|
|
|
|
|
/// This requires looking at the `git` field of this file's parent
|
|
|
|
|
/// directory, so will not work if this file has just been passed in on
|
|
|
|
|
/// the command line.
|
2015-05-12 02:33:40 +00:00
|
|
|
|
pub fn git_status(&self) -> f::Git {
|
2017-05-01 21:43:28 +00:00
|
|
|
|
use std::env::current_dir;
|
|
|
|
|
|
2015-05-11 22:28:01 +00:00
|
|
|
|
match self.dir {
|
2015-05-12 02:33:40 +00:00
|
|
|
|
None => f::Git { staged: f::GitStatus::NotModified, unstaged: f::GitStatus::NotModified },
|
2015-04-03 22:14:49 +00:00
|
|
|
|
Some(d) => {
|
|
|
|
|
let cwd = match current_dir() {
|
|
|
|
|
Err(_) => Path::new(".").join(&self.path),
|
Use new io + path + fs libraries (LOTS OF CHANGES)
Exa now uses the new IO, Path, and Filesystem libraries that have been out for a while now.
Unfortunately, the new libraries don't *entirely* cover the range of the old libraries just yet: in particular, to become more cross-platform, the data in `UnstableFileStat` isn't available in the Unix `MetadataExt` yet. Much of this is contained in rust-lang/rfcs#1044 (which is due to be implemented in rust-lang/rust#14711), but it's not *entirely* there yet.
As such, this commits a serious loss of functionality: no symlink viewing, no hard links or blocks, or users or groups. Also, some of the code could now be optimised. I just wanted to commit this to sort out most of the 'teething problems' of having a different path system in advance.
Here's an example problem that took ages to fix for you, just because you read this far: when I first got exa to compile, it worked mostly fine, except calling `exa` by itself didn't list the current directory. I traced where the command-line options were being generated, to where files and directories were sorted, to where the threads were spawned... and the problem turned out to be that it was using the full path as the file name, rather than just the last component, and these paths happened to begin with `.`, so it thought they were dotfiles.
2015-04-23 12:00:34 +00:00
|
|
|
|
Ok(dir) => dir.join(&self.path),
|
2015-04-03 22:14:49 +00:00
|
|
|
|
};
|
|
|
|
|
|
2015-05-11 22:28:01 +00:00
|
|
|
|
d.git_status(&cwd, self.is_directory())
|
2015-04-03 22:14:49 +00:00
|
|
|
|
},
|
2015-05-11 22:28:01 +00:00
|
|
|
|
}
|
2015-01-27 15:01:17 +00:00
|
|
|
|
}
|
2015-01-24 12:38:05 +00:00
|
|
|
|
}
|
|
|
|
|
|
2016-06-14 03:14:36 +00:00
|
|
|
|
#[cfg(any(target_os = "macos", target_os = "linux"))]
|
|
|
|
|
impl<'dir> File<'dir> {
|
|
|
|
|
/// Whether this file is a named pipe on the filesystem.
|
|
|
|
|
pub fn is_pipe(&self) -> bool {
|
|
|
|
|
self.metadata.file_type().is_fifo()
|
|
|
|
|
}
|
2016-10-29 19:27:23 +00:00
|
|
|
|
|
2016-06-14 03:14:36 +00:00
|
|
|
|
/// Whether this file is a char device on the filesystem.
|
|
|
|
|
pub fn is_char_device(&self) -> bool {
|
|
|
|
|
self.metadata.file_type().is_char_device()
|
|
|
|
|
}
|
2016-10-29 19:27:23 +00:00
|
|
|
|
|
2016-06-14 03:14:36 +00:00
|
|
|
|
/// Whether this file is a block device on the filesystem.
|
|
|
|
|
pub fn is_block_device(&self) -> bool {
|
|
|
|
|
self.metadata.file_type().is_block_device()
|
|
|
|
|
}
|
2016-10-29 19:27:23 +00:00
|
|
|
|
|
2016-06-14 03:14:36 +00:00
|
|
|
|
/// Whether this file is a socket on the filesystem.
|
|
|
|
|
pub fn is_socket(&self) -> bool {
|
|
|
|
|
self.metadata.file_type().is_socket()
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[cfg(not(any(target_os = "macos", target_os = "linux")))]
|
|
|
|
|
impl<'dir> File<'dir> {
|
|
|
|
|
/// Whether this file is a named pipe on the filesystem.
|
|
|
|
|
pub fn is_pipe(&self) -> bool {
|
|
|
|
|
false
|
|
|
|
|
}
|
2016-10-29 19:27:23 +00:00
|
|
|
|
|
2016-06-14 03:14:36 +00:00
|
|
|
|
/// Whether this file is a char device on the filesystem.
|
|
|
|
|
pub fn is_char_device(&self) -> bool {
|
|
|
|
|
false
|
|
|
|
|
}
|
2016-10-29 19:27:23 +00:00
|
|
|
|
|
2016-06-14 03:14:36 +00:00
|
|
|
|
/// Whether this file is a block device on the filesystem.
|
|
|
|
|
pub fn is_block_device(&self) -> bool {
|
|
|
|
|
false
|
|
|
|
|
}
|
2016-10-29 19:27:23 +00:00
|
|
|
|
|
2016-06-14 03:14:36 +00:00
|
|
|
|
/// Whether this file is a socket on the filesystem.
|
|
|
|
|
pub fn is_socket(&self) -> bool {
|
|
|
|
|
false
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2016-03-31 22:13:15 +00:00
|
|
|
|
impl<'a> AsRef<File<'a>> for File<'a> {
|
|
|
|
|
fn as_ref(&self) -> &File<'a> {
|
2017-03-31 16:09:32 +00:00
|
|
|
|
self
|
2016-03-31 22:13:15 +00:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2015-02-23 11:32:35 +00:00
|
|
|
|
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
/// Extract an extension from a file path, if one is present, in lowercase.
|
2015-01-24 12:38:05 +00:00
|
|
|
|
///
|
|
|
|
|
/// The extension is the series of characters after the last dot. This
|
|
|
|
|
/// deliberately counts dotfiles, so the ".git" folder has the extension "git".
|
2015-01-26 01:16:19 +00:00
|
|
|
|
///
|
|
|
|
|
/// ASCII lowercasing is used because these extensions are only compared
|
|
|
|
|
/// against a pre-compiled list of extensions which are known to only exist
|
|
|
|
|
/// within ASCII, so it's alright.
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
fn ext(path: &Path) -> Option<String> {
|
2017-05-01 21:43:28 +00:00
|
|
|
|
use std::ascii::AsciiExt;
|
|
|
|
|
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
let name = match path.file_name() {
|
|
|
|
|
Some(f) => f.to_string_lossy().to_string(),
|
|
|
|
|
None => return None,
|
|
|
|
|
};
|
|
|
|
|
|
2015-01-26 01:16:19 +00:00
|
|
|
|
name.rfind('.').map(|p| name[p+1..].to_ascii_lowercase())
|
2014-05-04 20:33:14 +00:00
|
|
|
|
}
|
2015-01-25 13:04:15 +00:00
|
|
|
|
|
2015-06-08 20:33:39 +00:00
|
|
|
|
|
2016-10-29 19:27:23 +00:00
|
|
|
|
/// The result of following a symlink.
|
|
|
|
|
pub enum FileTarget<'dir> {
|
|
|
|
|
|
|
|
|
|
/// The symlink pointed at a file that exists.
|
|
|
|
|
Ok(File<'dir>),
|
|
|
|
|
|
|
|
|
|
/// The symlink pointed at a file that does not exist. Holds the path
|
|
|
|
|
/// where the file would be, if it existed.
|
|
|
|
|
Broken(PathBuf),
|
|
|
|
|
|
|
|
|
|
/// There was an IO error when following the link. This can happen if the
|
|
|
|
|
/// file isn't a link to begin with, but also if, say, we don't have
|
|
|
|
|
/// permission to follow it.
|
|
|
|
|
Err(IOError),
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2015-05-16 17:33:08 +00:00
|
|
|
|
#[cfg(test)]
|
|
|
|
|
mod test {
|
|
|
|
|
use super::ext;
|
2015-11-04 10:07:08 +00:00
|
|
|
|
use std::path::Path;
|
2015-01-25 13:04:15 +00:00
|
|
|
|
|
2015-01-25 13:47:07 +00:00
|
|
|
|
#[test]
|
|
|
|
|
fn extension() {
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
assert_eq!(Some("dat".to_string()), ext(Path::new("fester.dat")))
|
2015-01-25 13:47:07 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
fn dotfile() {
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
assert_eq!(Some("vimrc".to_string()), ext(Path::new(".vimrc")))
|
2015-01-25 13:47:07 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
fn no_extension() {
|
Print the parent path for passed-in files
This commit changes all the views to accommodate printing each path's prefix, if it has one.
Previously, each file was stripped of its ancestry, leaving only its file name to be displayed. So running "exa /usr/bin/*" would display only filenames, while running "ls /usr/bin/*" would display each file prefixed with "/usr/bin/". But running "ls /usr/bin/" -- without the glob -- would run ls on just the directory, printing out the file names with no prefix or anything.
This functionality turned out to be useful in quite a few situations: firstly, if the user passes in files from different directories, it would be hard to tell where they came from (especially if they have the same name, such as find | xargs). Secondly, this also applied when following symlinks, making it unclear exactly which file a symlink would be pointing to.
The reason that it did it this way beforehand was that I didn't think of these use-cases, rather than for any technical reason; this new method should not have any drawbacks save making the output slightly wider in a few cases. Compatibility with ls is also a big plus.
Fixes #104, and relates to #88 and #92.
2016-04-11 18:10:55 +00:00
|
|
|
|
assert_eq!(None, ext(Path::new("jarlsberg")))
|
2015-11-04 10:07:08 +00:00
|
|
|
|
}
|
2015-01-25 13:04:15 +00:00
|
|
|
|
}
|