| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Most terminal emulators seem to support setting their title with a
special control sequence: "\e]0;<title>\a". Or in Rust:
"\x1b]0;<title>\x07". This patch adds some code that updates the
terminal emulator's title with the $(pwd), when at the prompt, or the
command that was issued by the user (max of 32 characters for now).
There are no configuration options as of yet, and also it does not
appear to work with tmux.
Notes:
Out of the box, this should be disabled by default, and only enabled
with a configuration option. Maybe some other options to tune it as well
(such as max length to display, etc).
|
| |
|
|
|
|
|
|
| |
This patch changes the check for and removal of a second slash by
enabling it only after the user as initiated completion (i.e. presed Tab
or Shift+Tab). This makes it possible to type two forward slashes (for
instance, when typing or pasting in a URL) without having to go back and
edit the buffer after the fact.
|
| |
|
|
|
|
|
|
|
|
|
| |
This patch updates the codepaths of the arrow keys and backspace keys to
calculate the width correctly when moving around in the buffer. See
commit ac5152886fed ("Workaround for faulty unicode_width values") for
more details, or you can also read the large comment at the top of the
new width! macro (which was added to reduce some redundant code, since
we need to properly compute character width in 4 different places).
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
|
| |
After a bit of analysis work, I believe the unicode_width library is
returning incorrect widths for certain characters (i.e. returning a
width of 0, when the width should be 1). This patch adds a workaround
for that in the comp() function, alongside a long comment documentating
the issue with more resources and tools.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Always append a newline to the end of the buffer. This ensures that the
shell will not quit if the user simply presses the return key without
any other input.
Fix a disrepency between the buffer length, and text width in the comp
function. The length of the buffer is the number of indices, whereas the
width of the buffer is the summed width of each character within the
buffer. We need the length when operating on the buffer's indicies, but
the width when dealing with what actually gets outputted to the shell.
In this case, the the width was being used where the length (bpos)
should've been, and was causing an index out of bounds error for any
strings that contained a character wider than 1. This rectifies the
issue, as well as refactors some code that was (appropriately) using the
width.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch rewrites much of the code in src/buffer.rs to (utlimately),
be less buggy. It also changed getchar() to have proper support for
UTF-8 characters. The autocomplete function was also enhanced to support
completions with filenames that have spaces in their paths. It handles
these by placing a backslash ('\') before each space in the filename.
There is not yet any completion support with quote ('), nor double-quote
characters ("). The buffer is still navigable with arrow keys, so
arbitrary deletions and insertions are still possible. Deletions and
insertions with multi-width UTF-8 characters work as expected.
Signed-off-by: Rory Dudley <rory@netc.lu>
Notes:
The complection function only works if the cursor is at the end of the
buffer. Pressing tab anywhere else will result in functionally a no op.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, switching between Tab and Shift+Tab while cycling through
autocomplete options could temporarily mess up the order of items. This
patch fixes that flaw by adding two new variables to help keep track of
the state:
- 'last', the last key that was recorded by getchar(),
- and 'length', which keeps track of the length of the last buffer
generated by the autocomplete() function.
These variables are then checked first whenever the user uses Tab or
Shift+Tab. The last know buffer 'length' is used to deal with overflow
in the case of Shift+Tab.
Finally, the logic for processing the autocomplete directory was moved
below the code for handling the actual user input (i.e. appending to the
getline() buffer). This is because the last character in the buffer
(i.e. the last character the user typed) is important for properly
updating the autcomplete working directory.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
| |
This patch makes it so that the autcomplete() function automatically
inserts the backslash character (`\`) into words with spaces when
cylcing through the options.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
| |
If, for whatever reason, the autocomplete() functions fails, don't panic
and quit. Instead, return an empty string.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch expands upon the last one, by providing a filepath
autcomplete routine, which works more similarly to that of zsh or bash.
It works for:
- Any paths containing the current directory (`.`)
- Any paths containing the parent directory (`..`)
- Any paths containing the root directory (`/`)
- Any paths that don't contain any explicit directory (defaults to the
current directory, i.e. `.`)
In order to achieve this, the pwd (present working directory) was broken
out from the autocomplete() function, and is now recomputed every time a
character is typed (in the getline() function). pwd always starts off as
the current directory, but may change based off of user input. If the
current directory cannot be read (likely due to a permissions issue),
dwvsh defaults to the user's home directory for now.
The filepath autocomplete also works for paths with mutliple directories
now (i.e. /etc/mail/xxx), will still allow you to autocomplete files and
directories beginning with 'xxx' inside the '/etc/mail' directory.
It is also possible to use the tilda character (`~`) with the filepath
autocomplete now (it denotes the user's home directory, for instance,
autocomplete works for '~/.config/').
Finally, some logic for Shift+Tab was added to autocomplete. It works
the same as Tab, except runs backwards through the files and directories
(as opposed to forwards).
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
| |
This patch adds a fairly rudimentary form of autocomplete. For starters,
it only works for filepaths, not for programs. Additionally, it
currently only works for the present working directory. You cannot yet
autocomplete even one more level deep.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
|
| |
When at the prompt, ICANON and ECHO should be turned off, since all the
output logic is handled via getline()/getchar(). However, they need to
be turned back on (alongside ECHOE, which allows for backspace
characters to work properly with ICANON), in case other programs need to
get user input, their input will echo to the terminal.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The getchar() function was changed so that it is able to detect some
ANSI escape sequences. To better handle this, getchar() now returns a
value from the Key enum, indicating whether or not an escape sequence
was found. Currently, the only escape sequences the function deals with
are arrow keys.
Handling of the left and right arrow keys were added to the getline()
function, in order to allow a user to go back and edit their command
inplace. Up and down arrow keys are also handled, but they are just
ignored for now (i.e. they do not move the cursor around anymore). The
local 'pos' variable became an Arc<Mutex<usize>>, since it needs to be
reset to 0 if ctrl-c is pressed (the handler for which is outside the
scope of getline()).
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
| |
Keep track of the cursor position in getline(), this way it is not
possible to backspace the prompt.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Added the termios crate to facilitate the changing of certain terminal
options. It is a wrapper around the termios C library, so 'man 3
termios' for more details.
Added the custom getchar() function, with retrieves characters from
STDIN as they are typed by the user (as opposed to waiting for a
newline, like io::stdin().read_line()). This is necessary, since keys
like <tab> and <up> have special functionality, which needs to be acted
on before command submission.
Added the custom getline() function, which uses getchar() to read
characters as they are typed. The getline() function contains the logic
for the various key presses. For most characters, we simply push the
byte to a buffer, and print it out to the screen (since getline()
assumes ECHO is off).
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
| |
This patch adds the 'Notes' rune (`=`), which allows for passing
environment variables, inline, to a binary, like so:
EDITOR=vim sudo visudo
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
| |
There is no reason to return an option for clause(), since it makes us
perform a match twice, and since a blank vector is perfectly acceptable
in all cases where the verse at hand may not have a clause.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
| |
This patch adds the '\' character as a new rune, Rune::Special. This is
for escaping special characters. Also works in strings (", ').
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
| |
Ignore patch files generated by 'git format-patch'.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
| |
Accidentally sourced 'dist/etc/dwvshrc' twice, instead of sourcing
'dist/etc/linuxrc' for Linux distros.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
| |
Do not run certain commands, except on specific operating systems, when
calling the global run command file. This allows users to avoid having
unaliases and unsets in their user-defined rc file (~/.dwvshrc).
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
|
| |
Rewrites the compose::env() function to be more modular, concerning the
run command files to read in. Also, the rrr() function now takes a
vector of files to read in. If a file is missing, even the global rc
file, it is skipped, instead of an error being thrown, and the shell
exiting.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
|
|
| |
This patch adds the following metadata fields to Cargo.toml, so that the
package can be published on https://crates.io:
- homepage: Our main website
- documentation: Also our main website (no docs for now, but they will
end up being hosted here in the near future)
- repository: A link to the git server with the source
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
|
| |
It appears that some older versions of the Rust compiler do not like the
double colon syntax when setting environment variables for the build
process. Until the latest cargo release will only accept the double
colon syntax, use a single colon when setting environment variables in
build.rs.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
| |
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
| |
This patch licences the repository under the BSD 3-Clause license
(https://choosealicense.com/licenses/bsd-3-clause/).
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
|
|
|
|
| |
The '--version' option prints out the program version, along with the
first 10 characters of the git commit hash, indicating at what commit
the program was built. This patch reduces the length of the commit hash
that gets embeded in the dwvsh binary to 7 characters, as that should be
more than enought to determine the correct commit.
Signed-off-by: Rory Dudley <rory@netc.lu>
|
| |
|
|
|
|
| |
The dwvsh binary may optionally take a filename as the last argument.
Instead of spawning an interactive shell, it will instead run a shell
program at the path specified.
|
| |
|
|
|
|
|
|
|
| |
When using IO operations from within a file, the channel would get
overriden before pushing a filename to the appropriate vector (op, ep,
or both). This is because Rune::Continue is used for newlines, and was
resetting the channel to None, before the push operation. Rune::Continue
is now broken out into its own match clause, that doesn't set the
channel before calling verse.add().
|
| |
|
|
|
|
|
| |
Add some basic logic for parsing commandline arguments. Also, use
build.rs to embed the program version (and git commit) during the
compile step. The program currently accepts the '--version' commandline
argument to print the version, then quit.
|
| |
|
|
|
|
|
|
| |
The code to find and replace environment variables was only searching
for alphanumeric characters. This means that the shell was unable to
recognize environment variables that used an underscore, for instance.
This patch allows the underscore character to be used when settings and
reading environment variables.
|
| |
|
|
| |
Add a job for Github workflows that tests building the documentation.
|
| |
|
|
|
| |
This patch update a ton of the documentation comments throughout the
codebase, refactoring some areas, and adding new comments to others.
|
| |
|
|
|
|
|
|
|
|
| |
Sometime when the switch to the new built-in command system was
happening, we lost the logic to force the capture the output of STDOUT,
mainly used for running internal poems (i.e. 'ls `ls`'). This patch adds
a new field to the Environment struct, called fc (force capture). It
gets set to true before running internal poems, and unset afterwards.
Finally, some checks were added to the incant!() macro to properly
handle STDOUT when fc is set.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit reworks spellcheck() so it is more verbose about what it
returns. It also re-introduces the use of spellcheck() in
Poem::recite().
Spellcheck now returns a value in the enum, Spelling:
FullPath -> Indicates a full path was specified as the verb
OnPath -> Indicates the verb is on the $PATH
BuiltIn -> Indicates the verb is a built-in command
None (Option) -> The verb does not exist
This commit also removes some defunct (commented-out) code.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, built-in commands were fairly primitive, merely outputting
STDOUT and STDERR with the print! macros. However, we need them to
behave like normal programs, that is:
- Acknowledge their verse's meter (forking, piping, etc.),
- Ability to capture STDOUT and STDERR (>, 2>),
- and Affect the currently running environment.
For these reasons, the anthology was reworked, and now contains the
Anthology struct, which mimics both std::process::{Child, Command}.
The AnthologyStdin helper struct was also created, for built-ins to
take input on STDIN, though no built-in is currently using it.
Each built-ins' incant functions were updated to return a
std::process::Output. It contains output from STDOUT, output from
STDERR, and the exit code of the "process".
A fix was also implemented for aliases, where the STDOUT and STDERR
vectors were not being copied to the newly constructed verse.
Notes:
There is some cleanup that needs to happen on this patch. For one, the
spellcheck function is no longer being used, so there is a generic OS
error if the program cannot be found in the $PATH. Also,
anthology::lookup gets called twice, which shouldn't need to happen.
|
| |
|
|
|
|
| |
Added a built-in which command (for MacOS and BSD), which can check
aliases, and other shell built-in commands, in addition to bins on the
$PATH.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Instead of returning true or false, if a given bin is found on the
$PATH, return the index of where the bin was found. This is useful, in
case we want to actually get the full path of the bin.
If a full or relative path is specified, the function will return the
length of the bins vector.
Notes:
This is primarily useful for the built-in 'which' command, so that we
can print out the full bin path without invoking /usr/bin/which.
|
| |
|
|
|
| |
In the global /etc/dwvshrc file, alias grep so that it always displays
with color by default.
|
| |
|
|
|
|
| |
Added documentation comments for the Poem::recite() function, detailing
what the purpose of the function is, as well as what operations it
performs.
|
| |
|
|
|
|
|
|
|
|
| |
Added definitions of fields for the Verse struct, and updated its
wording to accomodate for the type change of the couplet field.
Updated the wording of Verse::add() to make it more verbose, and also
added descriptions of the arguments, alongside some example usage.
Added documentation comments for the Verse::incant() function.
|
| |
|
|
|
| |
Added documentation comments for the new Runes that help with handling
file redirection of STDERR.
|
| |
|
|
|
| |
Update documentation comments for the Poem::read() function, as well as
the Poem::add() function (which is crucial for read()).
|
| |
|
|
|
| |
Added documentation comments for the Environment struct, which includes
uses cases, and defining the fields.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, when trying to add an alias that used the same name as its
command, an infinite recursion loop would occur, and the program would
panic after the call stack got too deep. For instance, something like:
alias grep='grep --color=always'
would cause it to panic.
This patch adds a new field to the Environment struct called 'cs' (for
call stack), which can be used to keep track of how many levels deep
into the Poem::read() function we are in. At the moment, it only allows
going two levels deep, but since it's just a u8 value, this could be
increased in the future. The above example now works correctly, but it
does mean that aliases within aliases are not possible, currently.
|
| |
|
|
|
| |
Only clear the 'out' vector when the verse is not a couplet, otherwise
text will not be piped to the next verse in the poem.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch overhauls the reading and reciting of verses, such that the
redirection of STDERR (in addition to STDOUT, which was already a
feature), is now possible.
Removed the 'stdout' argument from recite(), since it is no longer
needed with how incantations function.
A verse's couplet indicator is now a u8, instead of a bool, with certain
values corresponding to types of couplets, for instance:
ls | grep Ca | lolcat
^ ^ ^
| | 2: right side of a couplet
| 3: both sides of a couplet
1: left side of a couplet
Incantions are no longer hanlded in rune.rs, and the task macros have
been removed. Now, a verse incants itself, matching on its own meter to
determine how to handle the next verse.
The following runes were added to help with handling STDERR:
Write2 -> 2>
WriteAll -> &>
Addendum2 -> 2>>
AddendumAll -> &>>
The 'io' field in verse was changed from an Option<Rune>, to an array of
Runes, since a single verse might have multiple IO operations.
The following fields were added to Verse, to assist with handling
STDERR:
ip -> List of filenames to read into STDIN
op -> List of filenames to send STDOUT to
ep -> List of filenames to send STDERR to
Keep track of channels when reading a poem. Channels are relating to IO
operations. If channel is None, words get pushed to the verse's primary
stanza (i.e. the verb or the clause). If a channel is selected, words
are pushed to one of the aforementioned new fields in Verse.
Read -> ip
Write/Addedum -> op
Write2/Addedum2 -> ep
WriteAll/AddendumAll -> op and ep
Notes:
This commit also added tests for the new Runes.
|
| |
|
|
|
| |
The instance of &mut Chars is already an iterator, so we can remove the
call to into_iter().
|
| |
|
|
|
|
|
| |
This patch replaces the next! macro with a next() function. It serves
the same purpose, but instead of only checking for two different
runes (a fallback, or one that matches the peek), it can now take a list
of runes/characters to look ahead for, in addition to the fallback.
|
| |
|
|
|
| |
Removed a println!() that was used for debugging, and was accidentally
left in.
|
