In some rare cases, the global predeclarations would clash, for example:
> module spam { export def foo [] { "foo" } }; def foo [] { "bar" }
In the example, the `foo [] { "bar" }` would get predeclared first, then
the predeclaration would be overwritten and consumed by `foo [] {"foo"}`
inside the module, then when parsing the actual `foo [] { "bar" }`, it
would not find its predeclaration.
* plugin: basic from_mp4 implementation
This patch introduces a very basic implementation of from_mp4, with only
a few bits of meta-data available. The rest of the available meta-data
(which is more than half left), will be included in a later patch
* Mp4: Almost all track metadata is implemented
Only meta-data that is not implemented is duration, facing some weird
issue I am going to check on later
* Mp4: All meta-data fields implemented
All meta-data fields that can be retrieved are now retrieved, with the
exception of duration for both tracks and the entire file itself because
there is still an issue. However, that will be fixed in the upcoming
patches
* fix: UntaggedValue::duration() serializes correctly now
Previous to this patch, there was an issue where when you would use
UntaggedValue::duration() it would result in an invalid JSONRPC
resulting string when using the protocol. This patch fixes this issue
* Mp4: Duration fixed for file and tracks
* plugins: Add plugin extra to src/plugins
* Mp4: Replace unwrap() with expect()
* Fix: Remove test mp4 file
We've relied on `clap` for building our cli app bootstrapping that figures out the positionals, flags, and other convenient facilities. Nu has been capable of solving this problem for quite some time. Given this and much more reasons (including the build time caused by `clap`) we start here working with our own.
* add query json plugin for experimentation
* add some error handling
* closer but Kind::Array is still horked
* unravel the table so the output looks right
* clippy
* added the ability to use gjson modifiers
With the current code it is possible to attach custom commands from
a custom binary, but only for interactive mode. This change makes
it possible to also customize the evaluation context for commands
and scripts.
* Revert "History, more test coverage improvements, and refactorings. (#3217)"
This reverts commit 8fc8fc89aa.
* Add tests
* Refactor .nu-env
* Change logic of Config write to logic of read()
* Fix reload always appends to old vars
* Fix reload always takes last_modified of global config
* Add reload_config in evaluation context
* Reload config after writing to it in cfg set / cfg set_into
* Add --no-history to cli options
* Use --no-history in tests
* Add comment about maybe_print_errors
* Get ctrl_exit var from context.global_config
* Use context.global_config in command "config"
* Add Readme in engine how env vars are now handled
* Update docs from autoenv command
* Move history_path from engine to nu_data
* Move load history out of if
* No let before return
* Add import for indexmap
Improvements overall to Nu. Also among the changes here, we can also be more confident towards incorporating `3041`. End to end tests for checking envs properly exported to externals is not added here (since it's in the other PR)
A few things added in this PR (probably forgetting some too)
* no writes happen to history during test runs.
* environment syncing end to end coverage added.
* clean up / refactorings few areas.
* testing API for finer control (can write tests passing more than one pipeline)
* can pass environment variables in tests that nu will inherit when running.
* No longer needed.
* no longer under a module. No need to use super.
* Playground infraestructure (tests, etc) additions.
A few things to note:
* Nu can be started with a custom configuration file (`nu --config-file /path/to/sample_config.toml`). Useful for mocking the configuration on test runs.
* When given a custom configuration file Nu will save any changes to the file supplied appropiately.
* The `$nu.config-path` variable either shows the default configuration file (or the custom one, if given)
* We can now run end to end tests with finer grained control (currently, since this is baseline work, standard out) This will allow to check things like exit status, assert the contents with a format, etc)
* Remove (for another PR)
* move commands, futures.rs, script.rs, utils
* move over maybe_print_errors
* add nu_command crate references to nu_cli
* in commands.rs open up to pub mod from pub(crate)
* nu-cli, nu-command, and nu tests are now passing
* cargo fmt
* clean up nu-cli/src/prelude.rs
* code cleanup
* for some reason lex.rs was not formatted, may be causing my error
* remove mod completion from lib.rs which was not being used along with quickcheck macros
* add in allow unused imports
* comment out one failing external test; comment out one failing internal test
* revert commenting out failing tests; something else might be going on; someone with a windows machine should check and see what is going on with these failing windows tests
* Update Cargo.toml
Extend the optional features to nu-command
Co-authored-by: Jonathan Turner <jonathandturner@users.noreply.github.com>
* Begin allowing comments and multiline scripts.
* clippy
* Finish moving to groups. Test pass
* Keep going
* WIP
* WIP
* BROKEN WIP
* WIP
* WIP
* Fix more tests
* WIP: alias starts working
* Broken WIP
* Broken WIP
* Variables begin to work
* captures start working
* A little better but needs fixed scope
* Shorthand env setting
* Update main merge
* Broken WIP
* WIP
* custom command parsing
* Custom commands start working
* Fix coloring and parsing of block
* Almost there
* Add some tests
* Add more param types
* Bump version
* Fix benchmark
* Fix stuff
* first step of making selector
* wip
* wip tests working
* probably good enough for a first pass
* oops, missed something.
* and something else...
* grrrr version errors
We introduce the `plugin` nu sub command (`nu plugin`) with basic plugin
loading support. We can choose to load plugins from a directory. Originally
introduced to make integration tests faster (by not loading any plugins on startup at all)
but `nu plugin --load some_path ; test_pipeline_that_uses_plugins_just_loaded` does not see it.
Therefore, a `nu_with_plugins!` macro for tests was introduced on top of nu`s `--skip-plugins`
switch executable which is set to true when running the integration tests that use the `nu!` macro now..
This changeset contains everything that a separate binary needs to
register its own commands (including the new help function). It is
very possible that this commit misses other pub use exports, but
the contained ones work for our use cases so far.
This improves incremental build time when working on what was previously
the root package. For example, previously all plugins would be rebuilt
with a change to `src/commands/classified/external.rs`, but now only
`nu-cli` will have to be rebuilt (and anything that depends on it).
In particular, one thing that we can't (properly) do before this commit
is consuming an infinite input stream. For example:
```
yes | grep y | head -n10
```
will give 10 "y"s in most shells, but blocks indefinitely in nu. This PR
resolves that by doing blocking I/O in threads, and reducing the `await`
calls we currently have in our pipeline code.
* Create a function to create an empty directory entry
* Print an empty directory entry if permission is denied
* Fix rustfmt whitespace issues.
* Made metadata optional for `dir_entry_dict`.
Removed `empty_dir_entry_dict` as its not needed anymore.
* typo fixes
* Change signature to take in short-hand flags
* update help information
* Parse short-hand flags as their long counterparts
* lints
* Modified a couple tests to use shorthand flags
* Add block size to du
* Change blocks to physical size
* Use path instead of strings for file/directory names
* Why don't I just use paths instead of strings anyway?
* shorten physical size and apparent size to physical and apparent resp.
* Refactor pipeline ahead of block changes. Add '-c' commandline option
* Update pipelining an error value
* Fmt
* Clippy
* Add stdin redirect for -c flag
* Add stdin redirect for -c flag
* Fixed mv not throwing error when the source path was invalid
* Fixed failing test
* Fixed another lint error
* Fix $PATH conflicts in .gitpod.Dockerfile (#1349)
- Use the correct user for gitpod Dockerfile.
- Remove unneeded packages (curl, rustc) from gitpod Dockerfile.
* Added test to check for the error
* Fixed linting error
* Fixed mv not moving files on Windows. (#1342)
Move files correctly in windows.
* Fixed mv not throwing error when the source path was invalid
* Fixed failing test
* Fixed another lint error
* Added test to check for the error
* Fixed linting error
* Changed error message
* Typo and fixed test
Co-authored-by: Sean Hellum <seanhellum45@gmail.com>
* Upgrade futures, async-stream, and futures_codec
These were the last three dependencies on futures-preview. `nu` itself
is now fully dependent on `futures@0.3`, as opposed to `futures-preview`
alpha.
Because the update to `futures` from `0.3.0-alpha.19` to `0.3.0` removed
the `Stream` implementation of `VecDeque` ([changelog][changelog]), most
commands that convert a `VecDeque` to an `OutputStream` broke and had to
be fixed.
The current solution is to now convert `VecDeque`s to a `Stream` via
`futures::stream::iter`. However, it may be useful for `futures` to
create an `IntoStream` trait, implemented on the `std::collections` (or
really any `IntoIterator`). If something like this happends, it may be
worthwhile to update the trait implementations on `OutputStream` and
refactor these commands again.
While upgrading `futures_codec`, we remove a custom implementation of
`LinesCodec`, as one has been added to the library. There's also a small
refactor to make the stream output more idiomatic.
[changelog]: https://github.com/rust-lang/futures-rs/blob/master/CHANGELOG.md#030---2019-11-5
* Upgrade sys & ps plugin dependencies
They were previously dependent on `futures-preview`, and `nu_plugin_ps`
was dependent on an old version of `futures-timer`.
* Remove dependency on futures-timer from nu
* Update Cargo.lock
* Fix formatting
* Revert fmt regressions
CI is still on 1.40.0, but the latest rustfmt v1.41.0 has changes to the
`val @ pattern` syntax, causing the linting job to fail.
* Fix clippy warnings
* Add ability to have numbers in plugin name. Plugin must start with alphabetic char
* remove the first character as alphabetic requirement
* Update cli.rs
Going ahead and changing to plus to prevent issue notryanb found
* Update cli.rs
Co-authored-by: Jonathan Turner <jonathandturner@users.noreply.github.com>
* Adding kill command, unclean code
* Removing old comments
* Added quiet option, supports variable number of ids
* Made it per_item_command, calling commands directly without the shell
Moves the state changes for setting and removing environment variables
into the context's host as opposed to calling `std::env::*` directly
from anywhere else.
Introduced FakeHost to prevent environemnt state changes leaking
between unit tests and cause random test failures.
* `ls` will return error if no files/folders match path/pattern
* Revert changes to src/data/files.rs
* Add a name_only flag to dir_entry_dict
Add name_only flag to indicate if the caller only cares about filenames
or wants the whole path
* Update ls changes from feedback
* Little cleanup
* Resolve merge conflicts
* lints
* Add `--with-symlink-targets` option for the `ls` command that displays a new column for the target files of symlinks
* Fix clippy warning
Co-authored-by: Jonathan Turner <jonathandturner@users.noreply.github.com>
* compute directory sizes from contained files and directories
* De-lint
* Revert "De-lint"
This reverts commit 9df9fc07d777014fef8f5749a84b4e52e1ee652a.
* Revert "compute directory sizes from contained files and directories"
This reverts commit d43583e9aa20438bd613f78a36e641c9fd48cae3.
* Nu du command
* Nu du for you
* Add async support
* Lints
* so much bug fixing
* Added attributes to from-xml command
* Added attributes as their own rows
* Removed unneccesary lifetime declarations
* from-xml now has children and attributes side by side
* Fixed tests and linting
* Fixed lint-problem
* Switch to using `shell`
Switch to using the shell for subprocess to enable more natural shelling out.
* Update external.rs
* This is a test with .shell() for external
* El pollo loco's PR
* co co co
* Attempt to fix windows
* Fmt
* Less is more?
Co-authored-by: Andrés N. Robalino <andres@androbtech.com>
Restructure and streamline token expansion
The purpose of this commit is to streamline the token expansion code, by
removing aspects of the code that are no longer relevant, removing
pointless duplication, and eliminating the need to pass the same
arguments to `expand_syntax`.
The first big-picture change in this commit is that instead of a handful
of `expand_` functions, which take a TokensIterator and ExpandContext, a
smaller number of methods on the `TokensIterator` do the same job.
The second big-picture change in this commit is fully eliminating the
coloring traits, making coloring a responsibility of the base expansion
implementations. This also means that the coloring tracer is merged into
the expansion tracer, so you can follow a single expansion and see how
the expansion process produced colored tokens.
One side effect of this change is that the expander itself is marginally
more error-correcting. The error correction works by switching from
structured expansion to `BackoffColoringMode` when an unexpected token
is found, which guarantees that all spans of the source are colored, but
may not be the most optimal error recovery strategy.
That said, because `BackoffColoringMode` only extends as far as a
closing delimiter (`)`, `]`, `}`) or pipe (`|`), it does result in
fairly granular correction strategy.
The current code still produces an `Err` (plus a complete list of
colored shapes) from the parsing process if any errors are encountered,
but this could easily be addressed now that the underlying expansion is
error-correcting.
This commit also colors any spans that are syntax errors in red, and
causes the parser to include some additional information about what
tokens were expected at any given point where an error was encountered,
so that completions and hinting could be more robust in the future.
Co-authored-by: Jonathan Turner <jonathandturner@users.noreply.github.com>
Co-authored-by: Andrés N. Robalino <andres@androbtech.com>
Also, this commit makes `ls` a per-item command.
A command that processes things item by item may still take some time to stream
out the results from a single item. For example, `ls` on a directory with a lot
of files could be interrupted in the middle of showing all of these files.
* WIP --help works for PerItemCommands.
* De-linting
* Add more comments (#1228)
* Add some more docs
* More docs
* More docs
* More docs (#1229)
* Add some more docs
* More docs
* More docs
* Add more docs
* External commands: wrap values that contain spaces in quotes (#1214) (#1220)
* External commands: wrap values that contain spaces in quotes (#1214)
* Add fn's argument_contains_whitespace & add_quotes (#1214)
* Fix formatting with cargo fmt
* Don't wrap argument in quotes when $it is already quoted (#1214)
* Implement --help for internal commands
* Externals now spawn independently. (#1230)
This commit changes the way we shell out externals when using the `"$it"` argument. Also pipes per row to an external's stdin if no `"$it"` argument is present for external commands.
Further separation of logic (preparing the external's command arguments, getting the data for piping, emitting values, spawning processes) will give us a better idea for lower level details regarding external commands until we can find the right abstractions for making them more generic and unify within the pipeline calling logic of Nu internal's and external's.
* Poll externals quicker. (#1231)
* WIP --help works for PerItemCommands.
* De-linting
* Implement --help for internal commands
* Make having --help the default
* Update test to include new default switch
Co-authored-by: Jonathan Turner <jonathandturner@users.noreply.github.com>
Co-authored-by: Koenraad Verheyden <mail@koenraadverheyden.com>
Co-authored-by: Andrés N. Robalino <andres@androbtech.com>
This commit changes the way we shell out externals when using the `"$it"` argument. Also pipes per row to an external's stdin if no `"$it"` argument is present for external commands.
Further separation of logic (preparing the external's command arguments, getting the data for piping, emitting values, spawning processes) will give us a better idea for lower level details regarding external commands until we can find the right abstractions for making them more generic and unify within the pipeline calling logic of Nu internal's and external's.
* Put a sample_data.ods file for testing
This is a copy of the sample_data.xlsx file but in ods format
* Add the from-ods command
Most of the work was doing `rg xlsx` and then copy/paste with light editing
* Add tests for the from-ods command
* Fix failing test
The problem was improper filename sorting in the test `prepares_and_decorates_filesystem_source_files`
* Detect built-in commands passed as args to `which`
This expands the built-in `which` command to detect nushell commands
that may have the same name as a binary in the path.
* Allow which to interpret multiple arguments
Previously, it would discard any argument besides the first. This allows
`which` to process multiple arguments. It also makes the output a stream
of rows.
* Use map to build the output
* Add boolean column for builtins
* Use macros for entry creation shortcuts
* Process command args and use async_stream
In order to use `ichwh`, I'll need to use async_stream. But in order to
avoid lifetime errors with that, I have to process the command args
before using them. I'll admit I don't fully understand what is going on
with the `args.process(...)` function, but it works.
* Use `ichwh` for path searching
This commit transitions from `which` to `ichwh`. The path search is now
done asynchronously.
* Enable the `--all` flag on `which`
* Make `which` respect external commands
Escaped commands passed to wich (e.g., `which "^ls"`), are now searched
before builtins.
* Fix clippy warnings
This commit resolves two warnings from clippy, in light of #1142.
* Update Cargo.lock to get new `ichwh` version
`ichwh@0.2.1` has support for local paths.
* Add documentation for command
* Clippy fixes
* Finish converting to use clippy
* fix warnings in new master
* fix windows
* fix windows
Co-authored-by: Artem Vorotnikov <artem@vorotnikov.me>
* start playing with ways to use the uniq command
* WIP
* Got uniq working, but still need to figure out args issue and add tests
* Add some tests for uniq
* fmt
* remove commented out code
* Add documentation and some additional tests showing uniq values and rows. Also removed args TODO
* add changes that didn't get committed
* whoops, I didn't save the docs correctly...
* fmt
* Add a test for uniq with nested json
* Add another test
* Fix unique-ness when json keys are out of order and make the test json more complicated
* Manifests check. Ignore doctests for now.
* We continue with refactorings towards the separation of concerns between
crates. `nu_plugin_inc` and `nu_plugin_str` common test helpers usage
has been refactored into `nu-plugin` value test helpers.
Inc also uses the new API for integration tests.
This provides a huge performance boost for pipelines that end in an
external command. Rough testing shows an improvement from roughly 400ms
to 30ms when `cat`-ing a large file.
Add tests for ~tilde expansion:
- test that "~" is expanded (no more "~" in output)
- ensure that "1~1" is not expanded to "1/home/user1" as it was
before
Fixes#972
Note: the first test does not check the literal expansion because
the path on Windows is expanded as a Linux path, but the correct
expansion may come for free once `shellexpand` will use the `dirs`
crate too (https://github.com/netvl/shellexpand/issues/3).
This commit contains two improvements:
- Support for a Range syntax (and a corresponding Range value)
- Work towards a signature syntax
Implementing the Range syntax resulted in cleaning up how operators in
the core syntax works. There are now two kinds of infix operators
- tight operators (`.` and `..`)
- loose operators
Tight operators may not be interspersed (`$it.left..$it.right` is a
syntax error). Loose operators require whitespace on both sides of the
operator, and can be arbitrarily interspersed. Precedence is left to
right in the core syntax.
Note that delimited syntax (like `( ... )` or `[ ... ]`) is a single
token node in the core syntax. A single token node can be parsed from
beginning to end in a context-free manner.
The rule for `.` is `<token node>.<member>`. The rule for `..` is
`<token node>..<token node>`.
Loose operators all have the same syntactic rule: `<token
node><space><loose op><space><token node>`.
The second aspect of this pull request is the beginning of support for a
signature syntax. Before implementing signatures, a necessary
prerequisite is for the core syntax to support multi-line programs.
That work establishes a few things:
- `;` and newlines are handled in the core grammar, and both count as
"separators"
- line comments begin with `#` and continue until the end of the line
In this commit, multi-token productions in the core grammar can use
separators interchangably with spaces. However, I think we will
ultimately want a different rule preventing separators from occurring
before an infix operator, so that the end of a line is always
unambiguous. This would avoid gratuitous differences between modules and
repl usage.
We already effectively have this rule, because otherwise `x<newline> |
y` would be a single pipeline, but of course that wouldn't work.
Adds modules for internal, external, and dynamic commands, as well as
the pipeline functionality. These are exported as their old names from
the classified module so as to keep its "interface" the same.
`left =~ right` return true if left contains right, using Rust's
`String::contains`. `!~` is the negated version.
A new `apply_operator` function is added which decouples evaluation from
`Value::compare`. This returns a `Value` and opens the door to
implementing `+` for example, though it wouldn't be useful immediately.
The `operator!` macro had to be changed slightly as it would choke on
`~` in arguments.
This commit extracts five new crates:
- nu-source, which contains the core source-code handling logic in Nu,
including Text, Span, and also the pretty.rs-based debug logic
- nu-parser, which is the parser and expander logic
- nu-protocol, which is the bulk of the types and basic conveniences
used by plugins
- nu-errors, which contains ShellError, ParseError and error handling
conveniences
- nu-textview, which is the textview plugin extracted into a crate
One of the major consequences of this refactor is that it's no longer
possible to `impl X for Spanned<Y>` outside of the `nu-source` crate, so
a lot of types became more concrete (Value became a concrete type
instead of Spanned<Value>, for example).
This also turned a number of inherent methods in the main nu crate into
plain functions (impl Value {} became a bunch of functions in the
`value` namespace in `crate::data::value`).
`left =~ right` return true if left contains right, using Rust's
`String::contains`. `!~` is the negated version.
A new `apply_operator` function is added which decouples evaluation from
`Value::compare`. This returns a `Value` and opens the door to
implementing `+` for example, though it wouldn't be useful immediately.
The `operator!` macro had to be changed slightly as it would choke on
`~` in arguments.
After the previous commit, nushell uses PrettyDebug and
PrettyDebugWithSource for our pretty-printed display output.
PrettyDebug produces a structured `pretty.rs` document rather than
writing directly into a fmt::Formatter, and types that implement
`PrettyDebug` have a convenience `display` method that produces a string
(to be used in situations where `Display` is needed for compatibility
with other traits, or where simple rendering is appropriate).
This commit extracts Tag, Span, Text, as well as source-related debug
facilities into a new crate called nu_source.
This change is much bigger than one might have expected because the
previous code relied heavily on implementing inherent methods on
`Tagged<T>` and `Spanned<T>`, which is no longer possible.
As a result, this change creates more concrete types instead of using
`Tagged<T>`. One notable example: Tagged<Value> became Value, and Value
became UntaggedValue.
This change clarifies the intent of the code in many places, but it does
make it a big change.
Adds modules for internal, external, and dynamic commands, as well as
the pipeline functionality. These are exported as their old names from
the classified module so as to keep its "interface" the same.
fixes#969, admittedly without a --delimiter alias
moves from_structured_data.rs to from_delimited_data.rs to better
identify its scope and adds to_delimited_data.rs. Now csv and tsv both
use the same code, tsv passes in a fixed '\t' argument where csv passes
in the value of --separator
`save` attempts to convert input based on the target filename extension,
and expects a stream of text otherwise. However the error message is
unclear and provides little guidance, hopefully this is less confusing
to new users.
It might be worthwhile to also add a hint about adding an extension,
though I'm not sure if it's possible to emit multiple diagnostics.
tables and able to work with them for data processing & viewing
purposes. At the moment, certain ways to process said tables we
are able to view a histogram of a given column.
As usage matures, we may find certain core commands that could
be used ergonomically when working with tables on Nu.
The functions for retrieving, replacing, and inserting values into values all assumed they get the complete
column path as regular tagged strings. This commit changes for these to accept a tagged values instead. Basically
it means we can have column paths containing strings and numbers (eg. package.authors.1)
Unfortunately, for the moment all members when parsed and deserialized for a command that expects column paths
of tagged values will get tagged values (encapsulating Members) as strings only.
This makes it impossible to determine whether package.authors.1 package.authors."1" (meaning the "number" 1) is
a string member or a number member and thus prevents to know and force the user that paths enclosed in double
quotes means "retrieve the column at this given table" and that numbers are for retrieving a particular row number
from a table.
This commit sets in place the infraestructure needed when integer members land, in the mean time the workaround
is to convert back to strings the tagged values passed from the column paths.
The original purpose of this PR was to modernize the external parser to
use the new Shape system.
This commit does include some of that change, but a more important
aspect of this change is an improvement to the expansion trace.
Previous commit 6a7c00ea adding trace infrastructure to the syntax coloring
feature. This commit adds tracing to the expander.
The bulk of that work, in addition to the tree builder logic, was an
overhaul of the formatter traits to make them more general purpose, and
more structured.
Some highlights:
- `ToDebug` was split into two traits (`ToDebug` and `DebugFormat`)
because implementations needed to become objects, but a convenience
method on `ToDebug` didn't qualify
- `DebugFormat`'s `fmt_debug` method now takes a `DebugFormatter` rather
than a standard formatter, and `DebugFormatter` has a new (but still
limited) facility for structured formatting.
- Implementations of `ExpandSyntax` need to produce output that
implements `DebugFormat`.
Unlike the highlighter changes, these changes are fairly focused in the
trace output, so these changes aren't behind a flag.
Adds new substr function to str plugin with tests and documentation
Function takes a start/end location as a string in the form "##,##", both sides of comma are optional, and
behaves like Rust's own index operator [##..##].
a joy. Fundamentally we embrace functional programming principles for
transforming the dataset from any format picked up by Nu. This table
processing "primitive" commands will build up and make pipelines
composable with data processing capabilities allowing us the valuate,
reduce, and map, the tables as far as even composing this declartively.
On this regard, `split-by` expects some table with grouped data and we
can use it further in interesting ways (Eg. collecting labels for
visualizing the data in charts and/or suit it for a particular chart
of our interest).
This commit should finish the `coloring_in_tokens` feature, which moves
the shape accumulator into the token stream. This allows rollbacks of
the token stream to also roll back any shapes that were added.
This commit also adds a much nicer syntax highlighter trace, which shows
all of the paths the highlighter took to arrive at a particular coloring
output. This change is fairly substantial, but really improves the
understandability of the flow. I intend to update the normal parser with
a similar tracing view.
In general, this change also fleshes out the concept of "atomic" token
stream operations.
A good next step would be to try to make the parser more
error-correcting, using the coloring infrastructure. A follow-up step
would involve merging the parser and highlighter shapes themselves.
The code still compiles, so this doesn't seem to break anything. That also means
it's not critical to fix it, but having dead code around isn't great either.
Previously it would split the last column on the first separator value found
between the start of the column and the end of the row. Changing this to using
everything from the start of the column to the end of the string makes it behave
more similarly to the other columns, making it less surprising.
The table parsing/creation logic has changed from treating every line the same
to processing each line in context of the column header's placement. Previously,
lines on separate rows would go towards the same column as long as they were the
same index based on separator alone. Now, each item's index is based on vertical
alignment to the column header.
This may seem brittle, but it solves the problem of some tables operating with
empty cells that would cause remaining values to be paired with the wrong
column.
Based on kubernetes output (get pods, events), the new method has shown to have
much greater success rates for parsing.
New tests are added to test for additional cases that might be trickier to
handle with the new logic.
Old tests are updated where their expectations are no longer expected to hold true.
For instance: previously, lines would be treated separately, allowing any index
offset between columns on different rows, as long as they had the same row index
as decided by a separator. When this is no longer the case, some things need to
be adjusted.
The benefit of this is that coloring can be made atomic alongside token
stream forwarding.
I put the feature behind a flag so I can continue to iterate on it
without possibly regressing existing functionality. It's a lot of places
where the flags have to go, but I expect it to be a short-lived flag,
and the flags are fully contained in the parser.
Previously, we would build a command that looked something like this:
<ex_cmd> "$it" "&&" "<ex_cmd>" "$it"
So that the "&&" and "<ex_cmd>" would also be arguments to the command,
instead of a chained command. This commit builds up a command string
that can be passed to an external shell.
`std::fs::metadata` will attempt to follow symlinks, which results in a
"No such file or directory" error if the path pointed to by the symlink
does not exist. This shouldn't prevent `ls` from succeeding, so we
ignore errors.
Also, switching to use of `symlink_metadata` means we get stat info on
the symlink itself, not what it points to. This means `ls` will now
include broken symlinks in its listing.
* Moves off of draining between filters. Instead, the sink will pull on the stream, and will drain element-wise. This moves the whole stream to being lazy.
* Adds ctrl-c support and connects it into some of the key points where we pull on the stream. If a ctrl-c is detect, we immediately halt pulling on the stream and return to the prompt.
* Moves away from having a SourceMap where anchor locations are stored. Now AnchorLocation is kept directly in the Tag.
* To make this possible, split tag and span. Span is largely used in the parser and is copyable. Tag is now no longer copyable.
This commit adds the ability to work on features behind a feature flag
that won't be included in normal builds of nu.
These features are not exposed as Cargo features, as they reflect
incomplete features that are not yet stable.
To create a feature, add it to `features.toml`:
```toml
[hintsv1]
description = "Adding hints based on error states in the highlighter"
enabled = false
```
Each feature in `features.toml` becomes a feature flag accessible to `cfg`:
```rs
println!("hintsv1 is enabled");
```
By default, features are enabled based on the value of the `enabled` field.
You can also enable a feature from the command line via the
`NUSHELL_ENABLE_FLAGS` environment variable:
```sh
$ NUSHELL_ENABLE_FLAGS=hintsv1 cargo run
```
You can enable all flags via `NUSHELL_ENABLE_ALL_FLAGS`.
This commit also updates the CI setup to run the build with all flags off and
with all flags on. It also extracts the linting test into its own
parallelizable test, which means it doesn't need to run together with every
other test anymore.
When working on a feature, you should also add tests behind the same flag. A
commit is mergable if all tests pass with and without the flag, allowing
incomplete commits to land on master as long as the incomplete code builds and
passes tests.
This commit replaces the previous naive coloring system with a coloring
system that is more aligned with the parser.
The main benefit of this change is that it allows us to use parsing
rules to decide how to color tokens.
For example, consider the following syntax:
```
$ ps | where cpu > 10
```
Ideally, we could color `cpu` like a column name and not a string,
because `cpu > 10` is a shorthand block syntax that expands to
`{ $it.cpu > 10 }`.
The way that we know that it's a shorthand block is that the `where`
command declares that its first parameter is a `SyntaxShape::Block`,
which allows the shorthand block form.
In order to accomplish this, we need to color the tokens in a way that
corresponds to their expanded semantics, which means that high-fidelity
coloring requires expansion.
This commit adds a `ColorSyntax` trait that corresponds to the
`ExpandExpression` trait. The semantics are fairly similar, with a few
differences.
First `ExpandExpression` consumes N tokens and returns a single
`hir::Expression`. `ColorSyntax` consumes N tokens and writes M
`FlatShape` tokens to the output.
Concretely, for syntax like `[1 2 3]`
- `ExpandExpression` takes a single token node and produces a single
`hir::Expression`
- `ColorSyntax` takes the same token node and emits 7 `FlatShape`s
(open delimiter, int, whitespace, int, whitespace, int, close
delimiter)
Second, `ColorSyntax` is more willing to plow through failures than
`ExpandExpression`.
In particular, consider syntax like
```
$ ps | where cpu >
```
In this case
- `ExpandExpression` will see that the `where` command is expecting a
block, see that it's not a literal block and try to parse it as a
shorthand block. It will successfully find a member followed by an
infix operator, but not a following expression. That means that the
entire pipeline part fails to parse and is a syntax error.
- `ColorSyntax` will also try to parse it as a shorthand block and
ultimately fail, but it will fall back to "backoff coloring mode",
which parsing any unidentified tokens in an unfallible, simple way. In
this case, `cpu` will color as a string and `>` will color as an
operator.
Finally, it's very important that coloring a pipeline infallibly colors
the entire string, doesn't fail, and doesn't get stuck in an infinite
loop.
In order to accomplish this, this PR separates `ColorSyntax`, which is
infallible from `FallibleColorSyntax`, which might fail. This allows the
type system to let us know if our coloring rules bottom out at at an
infallible rule.
It's not perfect: it's still possible for the coloring process to get
stuck or consume tokens non-atomically. I intend to reduce the
opportunity for those problems in a future commit. In the meantime, the
current system catches a number of mistakes (like trying to use a
fallible coloring rule in a loop without thinking about the possibility
that it will never terminate).
The main thrust of this (very large) commit is an overhaul of the
expansion system.
The parsing pipeline is:
- Lightly parse the source file for atoms, basic delimiters and pipeline
structure into a token tree
- Expand the token tree into a HIR (high-level intermediate
representation) based upon the baseline syntax rules for expressions
and the syntactic shape of commands.
Somewhat non-traditionally, nu doesn't have an AST at all. It goes
directly from the token tree, which doesn't represent many important
distinctions (like the difference between `hello` and `5KB`) directly
into a high-level representation that doesn't have a direct
correspondence to the source code.
At a high level, nu commands work like macros, in the sense that the
syntactic shape of the invocation of a command depends on the
definition of a command.
However, commands do not have the ability to perform unrestricted
expansions of the token tree. Instead, they describe their arguments in
terms of syntactic shapes, and the expander expands the token tree into
HIR based upon that definition.
For example, the `where` command says that it takes a block as its first
required argument, and the description of the block syntactic shape
expands the syntax `cpu > 10` into HIR that represents
`{ $it.cpu > 10 }`.
This commit overhauls that system so that the syntactic shapes are
described in terms of a few new traits (`ExpandSyntax` and
`ExpandExpression` are the primary ones) that are more composable than
the previous system.
The first big win of this new system is the addition of the `ColumnPath`
shape, which looks like `cpu."max ghz"` or `package.version`.
Previously, while a variable path could look like `$it.cpu."max ghz"`,
the tail of a variable path could not be easily reused in other
contexts. Now, that tail is its own syntactic shape, and it can be used
as part of a command's signature.
This cleans up commands like `inc`, `add` and `edit` as well as
shorthand blocks, which can now look like `| where cpu."max ghz" > 10`
This resolves a small integration issue that would make custom prompts problematic (if they are implemented). The approach was to use the highlighter implementation in Helper to insert colour codes to the prompt however it heavily relies on the prompt being in a specific format, ending with a `> ` sequence. However, this should really be the job of the prompt itself not the presentation layer.
For now, I've simply stripped off the additional `> ` characters and passed in just the prompt itself without slicing off the last two characters. I moved the `\x1b[m` control sequence to the prompt creation in `cli.rs` as this feels like the more logical home for controlling what the prompt looks like. I can think of better ways to do this in future but this should be a fine solution for now.
In future it would probably make sense to completely separate prompts (be it, internal or external) from this code so it can be configured as an isolated piece of code.
Kind of touches on #356 by integrating the Starship prompt directly into the shell.
Not finished yet and has surfaced a potential bug in rustyline anyway. It depends on https://github.com/starship/starship/pull/509 being merged so the Starship prompt can be used as a library.
I could have tackled #356 completely and implemented a full custom prompt feature but I felt this was a simpler approach given that Starship is both written in Rust so shelling out isn't necessary and it already has a bunch of useful features built in.
However, I would understand if it would be preferable to just scrap integrating Starship directly and instead implement a custom prompt system which would facilitate simply shelling out to Starship.
If possible matches are not found then check if the passed in `obj`
parameter is a `string` or a `path`, if so then return it. I am not
sure this is the right fix, but I figured I would make an attempt and
get a conversation started about it.
When the last command has an input value larger than the data its
operating on it would crash. Added a check to ensure there are enough
elements to take.
This feature allows a user to set `ctrlc_exit` to `true` or `false` in their config to override how multiple CTRL-C invocations are handled. Without this change pressing CTRL-C multiple times will exit nu. With this change applied the user can configure the behavior to behave like other shells where multiple invocations will essentially clear the line.
This fixes#457.
Bare words now represent literal file names, and globs are a different
syntax shape called "Pattern". This allows commands like `cp` to ask for
a pattern as a source and a literal file as a target.
This also means that attempting to pass a glob to a command that expects
a literal path will produce an error.
