The benefit of this is that it make it possible to convert text cell contents
vectors into text cells with a method (see next commit). Casualties include
having to call `.into()` on vectors everywhere, which I'm not convinced is a
bad thing.
Because, strictly speaking, it's not a length, it's a width!
Also, re-order some struct constructors so that they're no longer
order-dependent (it's no longer the case that a value will be borrowed for one
field then consumed in another, meaning they have to be ordered in a certain
way to compile. Now the value is just worked out beforehand and the fields can
be specified in any order)
By removing the `File#file_name_width` method, we can make the file module
have no dependency on the output module -- in other words, the model (file)
and the view (output) are now separate again!
This commit introduces the `output::cell::DisplayWidth` struct, which
encapsulates the Unicode *display width* of a string in a struct that makes it
less easily confused with the *length* of a string.
The use of this type means that it's now harder to accidentally use a string's
length-in-bytes as its width. I've fixed at least one case in the code where
this was being done!
The only casualty is that it introduces a dependency on the output module from
the file module, which will be removed next commit.
A recent change to ansi-term [1] means that `ANSIString`s can now hold either
owned *or* borrowed data (Rust calls this the Cow type). This means that we
can delay formatting ANSIStrings into ANSI-control-code-formatted strings
until it's absolutely necessary. The process for doing this was:
1. Replace the `Cell` type with a `TextCell` type that holds a vector of
`ANSIString` values instead of a formatted string. It still does the
width tracking.
2. Rework the details module's `render` functions to emit values of this
type.
3. Similarly, rework the functions that produce cells containing filenames
to use a `File` value's `name` field, which is an owned `String` that
can now be re-used.
4. Update the printing, formatting, and width-calculating code in the
details and grid-details views to produce a table by adding vectors
together instead of adding strings together, delaying the formatting as
long as it can.
This results in fewer allocations (as fewer `String` values are produced), and
makes the API tidier (as fewer `String` values are being passed around without
having their contents specified).
This also paves the way to Windows support, or at least support for
non-ANSI terminals: by delaying the time until strings are formatted,
it'll now be easier to change *how* they are formatted.
Casualties include:
- Bump to ansi_term v0.7.1, which impls `PartialEq` and `Debug` on
`ANSIString`.
- The grid_details and lines views now need to take a vector of files, rather
than a borrowed slice, so the filename cells produced now own the filename
strings that get taken from files.
- Fixed the signature of `File#link_target` to specify that the
file produced refers to the same directory, rather than some phantom
directory with the same lifetime as the file. (This was wrong from the
start, but it broke nothing until now)
References:
[1]: ansi-term@f6a6579ba8174de1cae64d181ec04af32ba2a4f0
src/feature/xattr.rs:6:5: 6:22 warning: unused import, #[warn(unused_imports)] on by default
src/feature/xattr.rs:6 use std::ffi::CString;
^~~~~~~~~~~~~~~~~
The buggy code assumed that, if output isn't to a terminal, then the only view available is the Lines view. This is incorrect, as the Details view doesn't require a set width either, so check for --long even when there's no set width.
One of those two date formats was re-compiled before any date was displayed. Now they are compiled only the first time they're used, and cached versions are used thereafter, resulting in a speedup.
Instead of using the getopts crate’s dynamically-generated usage string, use a more static one:
- The options are organised by category now
- You can use `--help --long` to display only the ones that pertain to `--long`
- They’re aligned in a table sort of way
It could be generated statically, because all the options to change it are determined at compile time, but they’re not, yet...
Technically speaking, picking which timestamp to show for a file is a function of an output module, rather than the file itself. This also means that the `output::column` and `file` modules are now completely separate.
By taking the file names as a mutable vector, we can avoid having to allocate a new one when it’s empty. The recent changes to Options::getopts have made it more obvious that we could move the same vector out of getopts’s matches, instead of cloning it there.
This cleans up the options module, moving the structs that were *only* in use for the columns view out of it.
The new OptionSet trait is used to add the ‘deduce’ methods that used to be present on the values.
Changes to the way ANSIStrings work mean we need to dereference the strings before putting them in an ANSIString. There's more that can be done here, but this gets it to compile for now.
this commit fixes below type mismatch error:
```
src/output/details.rs:585:49: 585:60 error: mismatched types:
expected `i64`,
found `i32`
(expected i64,
found i32) [E0308]
src/output/details.rs:585 let date = self.tz.at(LocalDateTime::at(timestamp.0));
^~~~~~~~~~~
src/output/details.rs:585:49: 585:60 help: run `rustc --explain E0308` to see a detailed explanation
error: aborting due to previous error
Could not compile `exa`.
```
The `unused_results` lint was complaining that the results of inserting into a `MockUsers` object weren't being inspected. These are mock users, so all that would be returned is `None` to indicate that they weren't already in the table -- they're fine to ignore! So, suppress the warnings for those two testing modules.
This commit removes the threadpool in `main.rs` that stats each command-line argument separately, and replaces it with a *scoped* threadpool in `options/details.rs` that builds the table in parallel! Running this on my machine halves the execution time when tree-ing my entire home directory (which isn't exactly a common occurrence, but it's the only way to give exa a large running time)
The statting will be added back in parallel at a later stage. This was facilitated by the previous changes to recursion that made it easier to deal with.
There's a lot of large sweeping architectural changes. Here's a smattering of them:
- In `main.rs`, the files are now passed around as vectors of files rather than array slices of files. This is because `File`s aren't `Clone`, and the `Vec` is necessary to give away ownership of the files at the appropriate point.
- In the details view, files are now sorted *all* the time, rather than obeying the command-line order. As they're run in parallel, they have no guaranteed order anyway, so we *have* to sort them again. (I'm not sure if this should be the intended behaviour or not!) This means that the `Details` struct has to have the filter *all* the time, not only while recursing, so it's been moved out of the `recurse` field.
- We use `scoped_threadpool` over `threadpool`, a recent addition. It's only safely used on Nightly, which we're using anyway, so that's OK!
- Removed a bunch of out-of-date comments.
This also fixes#77, mainly by accident :)
Had to thread the value in at display-time to get it to only query the attributes once!
This isn't the nicest way to do it, but this *is* a bit of an edge-case (it's the only thing where a column depends on something that gets calculated later)
This prints three separate groups of child nodes: firstly the xattrs, then the errors, then any file children. It's done this way to only check for the 'last' child when necessary.
This does a similar thing that we did with the xattrs, except with the nested files: it removes the 'this' field on File, and replaces it with a method (to_dir) that has the same effect.
This means we get to remove a bunch of 'recurse' fields and parameters that really had no business being there! Now the table doesn't need to know whether it's going to need to list files recursively or not.
This changes the way extended attributes (xattrs) are printed. Before, they were artificially printed out on their own line both in lines mode *and* details mode, which looked a bit weird. Now, they are additional 'child nodes' of that item that get printed alongside errors.
All this allows all the 'extra info' that is going to be present for very few entries to be consolidated and listed in the same way, without resorting to extra printlns.
As a great side-effect, it allows taking out some of the more redundant code in the Table impl -- it is now *always* going to be in create-child-nodes mode, as *any* file now can, not only when we have the --tree flag in use.
Also, it now actually displays errors when failing to read the extended attributes, such as if the user doesn't have permission to read them.
The extended attribute flag has been temporarily disabled while I work out the best way to do it!
Now we have one Ur-module that contains functionality common to both supported platforms.
The benefits of doing it this way are that:
1. It doesn't implement a dummy interface - rather, there will be less code generated when the feature is not present;
2. The code shared between them can be kept in sync. The other two modules were something like 80% the same.
When tree mode is active, this will print out errors as another form of child node in the tree, instead of in one big block before any output.
The 'this' field now holds the io::Result of the readdir call, rather than only a *successful* result.
This is part of work to make the flow of files more iterator-able, rather than going in and out of vectors. Here, a Dir returns an iterator of files, rather than a pre-filled vector.
For now, this removes the ability for error messages to be displayed. Will be added in later though!
This is very slow (see #28) at the moment, so there's an option to switch off repo discovery. However, they were still always being queried. Now, if there's no Git option in the flags, it won't try to discover a repo.
These replace `init()` and `tail()` which are deprecated in favour of these.
In fact, it's a good thing they're deprecated, because part of the path_prefix code involved working around a call to init() that would panic otherwise - doing the same check with an `Option` is much more ergonomic.
Previously, each time it tried to render a table (to check its width), it both re-queried the filesystem and re-formatted the values into coloured strings.
These values are now calculated only once before the table is drawn, and are used repeatedly throughout.
Although it looks as though there's more `clone()`ing going on than before, it used to be recalculating things and storing them as vectors anyway, so the memory would still be used in any case.
This commit adds --grid, which, when used with --long, will split the details into multiple columns. Currently this is just 2 columns, but in the future it will be based on the width of the terminal.
In order to do this, I had to do two things:
1. Add a `links` parameter to the filename function, which disables the printing of the arrow and link target in the details view. When this is active, the columns get way too large, and it becomes not worth it.
2. Change the `print_table` function from actually printing the table to stdout to returning a list of `Cells` based on the table. This list then gets its width measured to calculate the width of the resulting table.
Finally! The benefit of having all the field-rendering code (in details.rs) separate from the value-getting code (in file.rs) is that rendering them can be tested again.
Colours are now disabled when output is not to a terminal. Fixes#53!
This required some internal restructuring - colours are now in their own object that gets passed around everywhere it's needed.
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.
This module provides feature-specific implementations, and also dummy implementations for when they aren't supported by the system or OS.
Doing it this way limits all the #[cfg(feature)] annotations, as we can now just include the module or not.
In cases where symlink targets were more than a single directory down,
exa did not print the '/' targets when separating directories, resulting
in the following output:
symlink => dirAdirBdirC/file
Instead of
symlink => dirA/dirB/dirC/file
By adding a '/' character after each component of the filename, this
error is fixed.
The old implementation blindly assumed that a symlink target would have a directory compoment, which the current directory, parent directory, and root directory technically don't have.
Fixes#20.
The logic of the previous version wasn't correct. Also, presuming
natural ordering of full filenames is still reasonable when the
extensions are identical.
The challenge is that the paths returned from libgit2's status listing
are from the perspective of the Git repository and thus effectively
relative to the working tree root, while the other paths we're
manipulating are (potentially) relative to our current working
directory. So, if those two aren't identical (if running from outside
the working tree, or from a subdirectory), the paths won't match up.
A reasonably reliable way around this is to resolve both types of paths
to absolute paths before comparing them. This fixes#15 at a basic
level, anyway.
What still doesn't work: referring to the working tree or one of its
descendants via a symlink. For that, we'd probably need to fully resolve
symlinks in the file path.
(The unwrap_or()'s are messy and will probably just result in missing
status information, but then, what information could you hope to get
without having both a current working directory and a Git working tree?)
Filenames with characters that need multiple bytes in UTF-8, and/or
control characters like combining diacritics, would break the grid.
`StrExt::width` seems to do the right thing, and there's conveniently a
function here (which was otherwise unused) to call that.
These were completely missing because I forgot about them... I added sorting functionality before I added date functionality, but it turns out I didn't even need the datetime library to do this!
However, this implementation feels incomplete. If you sort by the accessed date, it still displays the modified date, so the dates still look out of order. If I were to change the date column for the user, though, then there would still be no header to signify that the column has changed. I'm not sure how to proceed with this, but this is still missing functionality, so in it goes.
This has been mostly done with changes in the datetime crate's suddenly
supporting locales.
It's still important that the user's locale is touched only once and
cached from that point on, so a struct in output::details has been made
public, along with that module. This will change later as that object
gains more and more uses thoughout the codes.
Use the `locale` crate as a dependency to read in the set
thousands-separator character, and pass this to the file size column,
which uses it to add the separators in.
en_GB uses ","
fr_FR uses "" and just displays the numbers in one go.
Add options -u, -U, and -m as shorthand options for displaying different
types of timestamp, not just one. It's possible to have more than one by
specifying more than one of these shorthands, but *not* when used with
--time, as this is only meant to display one at a time.
Using the datetime crate, add an extra column to the --long view that
prints out the modified, accessed, or created timestamp for each file.
Also, let the user pick which one they want to see based on the --time
command-line option.
- Turn the views and main program loop into structs, rather than just as one gigantic function
- Separate views into their own files
The addition of the git column and the tree view meant that a lot of functions now just took extra arguments that didn't seem to fit. For example, it didn't really work to have only one 'view' method that printed out everything, as the different view options now all take different parameters.
FileName was always a special-cased column, as it was assumed to be the last column in the output. Now, it's explicitly marked as such. This allows the hash marks to be placed before the filename, rather than at the start of the line.
There's still a lot to do, but this is actually *something*. The tree hierarchy is displayed using hashes at the start of a line. I want to have it just before the filename, but this will need some changes to the way that columns are handled.
Instead of stripping the ANSI formatting characters from our strings, work out the length without them and use that. This is per-column, but most of them are simple (just the same number of characters in the non-coloured string).
Sometimes, this is really simple: for example, trwxrwxrwx permissions strings are always going to be ten characters long, and the strings that get returned are chock full of ANSI escape codes.
This should have a small benefit on performance.
This makes it more like ls. The --long (-l) argument has been added to get at the old behaviour, and the --links argument is now on -H. I can't crib this behaviour from ls because it shows it by default.
TODO: The terminal size is currently assumed to be 80, and it uses the string length, rather than the width.
