Makes C-c print "^C" and continue prompting on a new line.
Also fixes a problem where an interrupted get_line() would need more
read()'s than required to update the display.
This does not work perfectly (just like every other shell...), if the
running program handles the signal (SIGINT in this case) and quits
cleanly, the shell cannot detect the interruption.
This is the case with our `sleep(1)`.
According to the linux waitid manpage, the value of wstatus is
unspecified if wait() returns 0, so we should not assume that any
value it holds is correct (including the exit code).
This is only applicable to the Lagom build.
This commit fixes job control by putting children in their own process
group, and proxying TTY signals to active jobs.
This also cleans up the code around builtin_disown a bit to use
the newer job interfaces.
This commit also completely reworks the execution, highlighting and
completion model to work with the new AST.
New additions:
- $(...) stdout captures
- fd>&fd redirections
- fd>&- redirections (close fd)
- read-write redirections (<> path)
- completely event-based execution
- the weird idea of allowing the user to redirect the shell's own fds
- variables in strings
- local variables
- minimal list support
- adding hyperlinks to all paths that exist
This commit makes LibLine handle interrupts (as reported via
interrupted() and resized()) again.
There is a little catch with the shell:
```
$ ls |
pipe> <C-c> (prompt stays here until a key is pressed)
```
* In some cases, we can first call sigaction()/signal(), then *not* pledge
sigaction.
* In other cases, we pledge sigaction at first, call sigaction()/signal()
second, then pledge again, this time without sigaction.
* In yet other cases, we keep the sigaction pledge. I suppose these could all be
migrated to drop it or not pledge it at all, if somebody is interested in
doing that.
LibLine should ultimately not care about what a "token" means in the
context of its user, so force the user to split the buffer itself.
This also allows the users to pick up contextual clues as well, since
they have to lex the line themselves.
This commit pacthes Shell and the JS repl to better handle completions,
so certain wrong behaviours are now corrected as well:
- JS repl can now complete "Object . getOw<tab>"
- Shell can now complete "echo | ca<tab>" and paths inside strings
In the case of a syntax error the shell parser prints an error message
to stderr and returns an empty Vector<Command> - in that case we
shouldn't try to determine whether or not we can continue parsing but
abort immediately - is_complete() expects that *something* was parsed
successfully.
Fixes#2251.
This fixes the little issue with Shell not allowing cancellation of
commands once they were in continuation mode
```
$ ls '
$ # No matter what we do here, we cannot escape 'ls'
```
This patchset adds some metadata to Parser::parse() which allows the
Shell to ask for the rest of a command, if it is not complete.
A command is considered complete if it has no trailing pipe, or
unterminated string.
This patchset fixes incorrect handling of escaped tokens (`a\ b`) in
Shell autocompletion and LibLine.
The users of LibLine can now choose between two token splitting modes,
either taking into account escapes, or ignoring them.
It was impractical to return a RefPtr<File> since that left us no way
to extract the error string. This is usually needed for the UI, so the
old static open() got basically no use.
These strings would be applied when inserted into the buffer, but are
not shown as part of the suggestion.
This commit also patches up Userland/js and Shell to use this
functionality
