nushell/crates/nu-protocol/src/engine/engine_state.rs

1225 lines
39 KiB
Rust
Raw Normal View History

2021-09-02 20:21:37 +02:00
use super::Command;
use crate::{
ast::Block, BlockId, DeclId, Example, Overlay, OverlayId, PipelineData, ShellError, Signature,
Span, Type, Value, VarId,
};
2021-07-23 23:19:30 +02:00
use core::panic;
2021-10-28 06:13:10 +02:00
use std::{
collections::HashMap,
sync::{atomic::AtomicBool, Arc},
};
2021-06-30 03:42:56 +02:00
#[cfg(feature = "plugin")]
use std::path::PathBuf;
// Tells whether a decl etc. is visible or not
#[derive(Debug, Clone)]
struct Visibility {
decl_ids: HashMap<DeclId, bool>,
}
impl Visibility {
fn new() -> Self {
Visibility {
decl_ids: HashMap::new(),
}
}
fn is_decl_id_visible(&self, decl_id: &DeclId) -> bool {
*self.decl_ids.get(decl_id).unwrap_or(&true) // by default it's visible
}
fn hide_decl_id(&mut self, decl_id: &DeclId) {
self.decl_ids.insert(*decl_id, false);
}
fn use_decl_id(&mut self, decl_id: &DeclId) {
self.decl_ids.insert(*decl_id, true);
}
fn merge_with(&mut self, other: Visibility) {
// overwrite own values with the other
self.decl_ids.extend(other.decl_ids);
// self.env_var_ids.extend(other.env_var_ids);
}
fn append(&mut self, other: &Visibility) {
// take new values from other but keep own values
for (decl_id, visible) in other.decl_ids.iter() {
if !self.decl_ids.contains_key(decl_id) {
self.decl_ids.insert(*decl_id, *visible);
}
}
}
}
2021-10-25 06:01:02 +02:00
#[derive(Debug, Clone)]
2021-09-06 04:20:02 +02:00
pub struct ScopeFrame {
2021-10-05 04:46:24 +02:00
pub vars: HashMap<Vec<u8>, VarId>,
predecls: HashMap<Vec<u8>, DeclId>, // temporary storage for predeclarations
2021-11-02 04:08:05 +01:00
pub decls: HashMap<Vec<u8>, DeclId>,
pub aliases: HashMap<Vec<u8>, Vec<Span>>,
pub env_vars: HashMap<Vec<u8>, BlockId>,
pub overlays: HashMap<Vec<u8>, OverlayId>,
visibility: Visibility,
2021-06-30 03:42:56 +02:00
}
impl ScopeFrame {
pub fn new() -> Self {
Self {
vars: HashMap::new(),
predecls: HashMap::new(),
2021-07-01 08:09:55 +02:00
decls: HashMap::new(),
2021-08-09 02:19:07 +02:00
aliases: HashMap::new(),
env_vars: HashMap::new(),
overlays: HashMap::new(),
visibility: Visibility::new(),
2021-06-30 03:42:56 +02:00
}
}
2021-09-06 04:20:02 +02:00
pub fn get_var(&self, var_name: &[u8]) -> Option<&VarId> {
self.vars.get(var_name)
}
}
impl Default for ScopeFrame {
fn default() -> Self {
Self::new()
}
2021-06-30 03:42:56 +02:00
}
/// The core global engine state. This includes all global definitions as well as any global state that
/// will persist for the whole session.
///
/// Declarations, variables, blocks, and other forms of data are held in the global state and referenced
/// elsewhere using their IDs. These IDs are simply their index into the global state. This allows us to
/// more easily handle creating blocks, binding variables and callsites, and more, because each of these
/// will refer to the corresponding IDs rather than their definitions directly. At runtime, this means
/// less copying and smaller structures.
///
/// Note that the runtime stack is not part of this global state. Runtime stacks are handled differently,
/// but they also rely on using IDs rather than full definitions.
///
/// A note on implementation:
///
/// Much of the global definitions are built on the Bodil's 'im' crate. This gives us a way of working with
/// lists of definitions in a way that is very cheap to access, while also allowing us to update them at
/// key points in time (often, the transition between parsing and evaluation).
///
/// Over the last two years we tried a few different approaches to global state like this. I'll list them
/// here for posterity, so we can more easily know how we got here:
///
/// * `Rc` - Rc is cheap, but not thread-safe. The moment we wanted to work with external processes, we
/// needed a way send to stdin/stdout. In Rust, the current practice is to spawn a thread to handle both.
/// These threads would need access to the global state, as they'll need to process data as it streams out
/// of the data pipeline. Because Rc isn't thread-safe, this breaks.
///
/// * `Arc` - Arc is the thread-safe version of the above. Often Arc is used in combination with a Mutex or
/// RwLock, but you can use Arc by itself. We did this a few places in the original Nushell. This *can* work
/// but because of Arc's nature of not allowing mutation if there's a second copy of the Arc around, this
/// ultimately becomes limiting.
///
/// * `Arc` + `Mutex/RwLock` - the standard practice for thread-safe containers. Unfortunately, this would
/// have meant we would incur a lock penalty every time we needed to access any declaration or block. As we
/// would be reading far more often than writing, it made sense to explore solutions that favor large amounts
/// of reads.
///
/// * `im` - the `im` crate was ultimately chosen because it has some very nice properties: it gives the
/// ability to cheaply clone these structures, which is nice as EngineState may need to be cloned a fair bit
/// to follow ownership rules for closures and iterators. It also is cheap to access. Favoring reads here fits
/// more closely to what we need with Nushell. And, of course, it's still thread-safe, so we get the same
/// benefits as above.
///
2021-11-02 04:08:05 +01:00
#[derive(Clone)]
pub struct EngineState {
files: im::Vector<(String, usize, usize)>,
file_contents: im::Vector<(Vec<u8>, usize, usize)>,
vars: im::Vector<Type>,
decls: im::Vector<Box<dyn Command + 'static>>,
blocks: im::Vector<Block>,
overlays: im::Vector<Overlay>,
2021-11-02 04:08:05 +01:00
pub scope: im::Vector<ScopeFrame>,
pub ctrlc: Option<Arc<AtomicBool>>,
#[cfg(feature = "plugin")]
pub plugin_signatures: Option<PathBuf>,
2021-06-30 03:42:56 +02:00
}
2021-11-02 04:08:05 +01:00
pub const NU_VARIABLE_ID: usize = 0;
pub const SCOPE_VARIABLE_ID: usize = 1;
pub const IN_VARIABLE_ID: usize = 2;
pub const CONFIG_VARIABLE_ID: usize = 3;
2021-11-02 04:08:05 +01:00
2021-09-02 10:25:22 +02:00
impl EngineState {
2021-06-30 03:42:56 +02:00
pub fn new() -> Self {
2021-07-01 08:09:55 +02:00
Self {
2021-10-25 06:01:02 +02:00
files: im::vector![],
file_contents: im::vector![],
vars: im::vector![Type::Unknown, Type::Unknown, Type::Unknown, Type::Unknown],
2021-10-25 06:01:02 +02:00
decls: im::vector![],
blocks: im::vector![],
overlays: im::vector![],
2021-10-25 06:01:02 +02:00
scope: im::vector![ScopeFrame::new()],
2021-10-28 06:13:10 +02:00
ctrlc: None,
#[cfg(feature = "plugin")]
plugin_signatures: None,
2021-07-01 08:09:55 +02:00
}
2021-06-30 03:42:56 +02:00
}
/// Merges a `StateDelta` onto the current state. These deltas come from a system, like the parser, that
/// creates a new set of definitions and visible symbols in the current scope. We make this transactional
/// as there are times when we want to run the parser and immediately throw away the results (namely:
/// syntax highlighting and completions).
///
/// When we want to preserve what the parser has created, we can take its output (the `StateDelta`) and
/// use this function to merge it into the global state.
pub fn merge_delta(&mut self, mut delta: StateDelta) -> Result<(), ShellError> {
2021-06-30 03:42:56 +02:00
// Take the mutable reference and extend the permanent state from the working set
self.files.extend(delta.files);
self.file_contents.extend(delta.file_contents);
self.decls.extend(delta.decls);
self.vars.extend(delta.vars);
self.blocks.extend(delta.blocks);
self.overlays.extend(delta.overlays);
if let Some(last) = self.scope.back_mut() {
let first = delta.scope.remove(0);
for item in first.decls.into_iter() {
last.decls.insert(item.0, item.1);
}
for item in first.vars.into_iter() {
last.vars.insert(item.0, item.1);
2021-07-17 08:31:34 +02:00
}
2021-08-09 09:53:06 +02:00
for item in first.aliases.into_iter() {
last.aliases.insert(item.0, item.1);
}
for item in first.overlays.into_iter() {
last.overlays.insert(item.0, item.1);
2021-09-26 12:25:52 +02:00
}
last.visibility.merge_with(first.visibility);
#[cfg(feature = "plugin")]
if delta.plugins_changed {
let result = self.update_plugin_file();
if result.is_ok() {
delta.plugins_changed = false;
}
return result;
}
}
Ok(())
}
#[cfg(feature = "plugin")]
pub fn update_plugin_file(&self) -> Result<(), ShellError> {
use std::io::Write;
// Updating the signatures plugin file with the added signatures
self.plugin_signatures
.as_ref()
.ok_or_else(|| ShellError::PluginFailedToLoad("Plugin file not found".into()))
.and_then(|plugin_path| {
// Always create the file, which will erase previous signatures
std::fs::File::create(plugin_path.as_path())
.map_err(|err| ShellError::PluginFailedToLoad(err.to_string()))
})
.and_then(|mut plugin_file| {
2021-12-03 00:11:25 +01:00
// Plugin definitions with parsed signature
self.plugin_decls().try_for_each(|decl| {
2021-12-03 00:11:25 +01:00
// A successful plugin registration already includes the plugin filename
// No need to check the None option
let path = decl.is_plugin().expect("plugin should have file name");
let file_name = path.to_str().expect("path should be a str");
2021-12-04 13:38:21 +01:00
serde_json::to_string_pretty(&decl.signature())
.map(|signature| format!("register {} {}\n\n", file_name, signature))
2021-12-04 13:38:21 +01:00
.map_err(|err| ShellError::PluginFailedToLoad(err.to_string()))
.and_then(|line| {
plugin_file
.write_all(line.as_bytes())
.map_err(|err| ShellError::PluginFailedToLoad(err.to_string()))
})
})
})
2021-06-30 03:42:56 +02:00
}
pub fn num_files(&self) -> usize {
self.files.len()
}
2021-07-01 08:09:55 +02:00
pub fn num_vars(&self) -> usize {
self.vars.len()
}
pub fn num_decls(&self) -> usize {
self.decls.len()
}
2021-07-16 08:24:46 +02:00
pub fn num_blocks(&self) -> usize {
self.blocks.len()
}
pub fn num_overlays(&self) -> usize {
self.overlays.len()
}
2021-07-23 23:19:30 +02:00
pub fn print_vars(&self) {
for var in self.vars.iter().enumerate() {
println!("var{}: {:?}", var.0, var.1);
}
}
pub fn print_decls(&self) {
for decl in self.decls.iter().enumerate() {
2021-09-02 10:25:22 +02:00
println!("decl{}: {:?}", decl.0, decl.1.signature());
2021-07-23 23:19:30 +02:00
}
}
pub fn print_blocks(&self) {
for block in self.blocks.iter().enumerate() {
println!("block{}: {:?}", block.0, block.1);
}
}
2021-09-25 18:28:15 +02:00
pub fn print_contents(&self) {
2021-10-25 06:01:02 +02:00
for (contents, _, _) in self.file_contents.iter() {
2021-10-25 18:58:58 +02:00
let string = String::from_utf8_lossy(contents);
2021-10-25 06:01:02 +02:00
println!("{}", string);
}
2021-09-25 18:28:15 +02:00
}
2021-07-23 23:46:55 +02:00
pub fn find_decl(&self, name: &[u8]) -> Option<DeclId> {
let mut visibility: Visibility = Visibility::new();
2021-07-23 23:46:55 +02:00
for scope in self.scope.iter().rev() {
visibility.append(&scope.visibility);
2021-07-23 23:46:55 +02:00
if let Some(decl_id) = scope.decls.get(name) {
if visibility.is_decl_id_visible(decl_id) {
return Some(*decl_id);
}
2021-07-23 23:46:55 +02:00
}
}
None
}
#[cfg(feature = "plugin")]
pub fn plugin_decls(&self) -> impl Iterator<Item = &Box<dyn Command + 'static>> {
let mut unique_plugin_decls = HashMap::new();
// Make sure there are no duplicate decls: Newer one overwrites the older one
for decl in self.decls.iter().filter(|d| d.is_plugin().is_some()) {
unique_plugin_decls.insert(decl.name(), decl);
}
let mut plugin_decls: Vec<(&str, &Box<dyn Command>)> =
unique_plugin_decls.into_iter().collect();
// Sort the plugins by name so we don't end up with a random plugin file each time
plugin_decls.sort_by(|a, b| a.0.cmp(b.0));
plugin_decls.into_iter().map(|(_, decl)| decl)
}
pub fn find_overlay(&self, name: &[u8]) -> Option<OverlayId> {
for scope in self.scope.iter().rev() {
if let Some(overlay_id) = scope.overlays.get(name) {
return Some(*overlay_id);
}
}
None
}
2021-09-10 00:09:40 +02:00
pub fn find_commands_by_prefix(&self, name: &[u8]) -> Vec<Vec<u8>> {
let mut output = vec![];
for scope in self.scope.iter().rev() {
for decl in &scope.decls {
if decl.0.starts_with(name) {
output.push(decl.0.clone());
}
}
}
output
}
2021-09-12 17:34:43 +02:00
pub fn get_span_contents(&self, span: &Span) -> &[u8] {
2021-10-25 06:01:02 +02:00
for (contents, start, finish) in &self.file_contents {
2021-10-25 18:58:58 +02:00
if span.start >= *start && span.end <= *finish {
2021-10-25 06:01:02 +02:00
return &contents[(span.start - start)..(span.end - start)];
}
}
panic!("internal error: span missing in file contents cache")
2021-09-10 00:09:40 +02:00
}
2021-07-23 07:14:49 +02:00
pub fn get_var(&self, var_id: VarId) -> &Type {
self.vars
.get(var_id)
.expect("internal error: missing variable")
2021-07-01 08:09:55 +02:00
}
2021-09-04 09:59:38 +02:00
#[allow(clippy::borrowed_box)]
2021-09-02 10:25:22 +02:00
pub fn get_decl(&self, decl_id: DeclId) -> &Box<dyn Command> {
2021-07-23 07:14:49 +02:00
self.decls
.get(decl_id)
.expect("internal error: missing declaration")
2021-07-01 08:09:55 +02:00
}
#[allow(clippy::borrowed_box)]
pub fn get_decl_with_input(&self, decl_id: DeclId, input: &PipelineData) -> &Box<dyn Command> {
let decl = self.get_decl(decl_id);
match input {
PipelineData::Stream(..) => decl,
PipelineData::Value(value, ..) => match value {
Value::CustomValue { val, .. } => {
// This filter works because the custom definitions were declared
// before the default nushell declarations. This means that the custom
// declarations that get overridden by the default declarations can only
// be accessed if the input value has the required category
let decls = self
.decls
.iter()
.enumerate()
.filter(|(_, decl_inner)| {
decl.name() == decl_inner.name()
&& decl_inner.signature().category == val.category()
})
.map(|(index, _)| index)
.collect::<Vec<usize>>();
match decls.first() {
Some(index) => self.get_decl(*index),
None => decl,
}
}
_ => decl,
},
}
}
/// Get all IDs of all commands within scope, sorted by the commads' names
pub fn get_decl_ids_sorted(&self, include_hidden: bool) -> impl Iterator<Item = DeclId> {
let mut decls_map = HashMap::new();
2021-10-01 23:53:13 +02:00
for frame in &self.scope {
let frame_decls = if include_hidden {
frame.decls.clone()
} else {
frame
.decls
.clone()
.into_iter()
.filter(|(_, id)| frame.visibility.is_decl_id_visible(id))
.collect()
};
decls_map.extend(frame_decls);
2021-10-01 23:53:13 +02:00
}
let mut decls: Vec<(Vec<u8>, DeclId)> = decls_map.into_iter().collect();
decls.sort_by(|a, b| a.0.cmp(&b.0));
decls.into_iter().map(|(_, id)| id)
2021-10-01 23:53:13 +02:00
}
/// Get signatures of all commands within scope.
pub fn get_signatures(&self, include_hidden: bool) -> Vec<Signature> {
self.get_decl_ids_sorted(include_hidden)
.map(|id| {
let decl = self.get_decl(id);
2021-10-09 03:02:01 +02:00
let mut signature = (*decl).signature();
signature.usage = decl.usage().to_string();
signature.extra_usage = decl.extra_usage().to_string();
signature
})
.collect()
}
2021-10-09 03:02:01 +02:00
/// Get signatures of all commands within scope.
///
/// In addition to signatures, it returns whether each command is:
/// a) a plugin
/// b) custom
pub fn get_signatures_with_examples(
&self,
include_hidden: bool,
) -> Vec<(Signature, Vec<Example>, bool, bool)> {
self.get_decl_ids_sorted(include_hidden)
.map(|id| {
let decl = self.get_decl(id);
let mut signature = (*decl).signature();
signature.usage = decl.usage().to_string();
signature.extra_usage = decl.extra_usage().to_string();
(
signature,
decl.examples(),
decl.is_plugin().is_some(),
decl.get_block_id().is_some(),
)
})
.collect()
2021-10-09 03:02:01 +02:00
}
2021-07-23 07:14:49 +02:00
pub fn get_block(&self, block_id: BlockId) -> &Block {
self.blocks
.get(block_id)
.expect("internal error: missing block")
2021-07-22 21:50:59 +02:00
}
pub fn get_overlay(&self, overlay_id: OverlayId) -> &Overlay {
self.overlays
.get(overlay_id)
.expect("internal error: missing overlay")
}
pub fn next_span_start(&self) -> usize {
2021-10-25 18:58:58 +02:00
if let Some((_, _, last)) = self.file_contents.last() {
*last
} else {
0
}
}
2021-10-25 06:01:02 +02:00
pub fn files(&self) -> impl Iterator<Item = &(String, usize, usize)> {
self.files.iter()
}
pub fn get_filename(&self, file_id: usize) -> String {
for file in self.files.iter().enumerate() {
if file.0 == file_id {
return file.1 .0.clone();
}
}
"<unknown>".into()
}
pub fn get_file_source(&self, file_id: usize) -> String {
for file in self.files.iter().enumerate() {
if file.0 == file_id {
2021-10-25 06:01:02 +02:00
let contents = self.get_span_contents(&Span {
start: file.1 .1,
end: file.1 .2,
});
let output = String::from_utf8_lossy(contents).to_string();
return output;
}
}
"<unknown>".into()
}
2021-07-02 00:54:04 +02:00
#[allow(unused)]
2021-06-30 03:42:56 +02:00
pub(crate) fn add_file(&mut self, filename: String, contents: Vec<u8>) -> usize {
let next_span_start = self.next_span_start();
2021-10-25 06:01:02 +02:00
let next_span_end = next_span_start + contents.len();
2021-06-30 03:42:56 +02:00
2021-10-25 06:01:02 +02:00
self.file_contents
.push_back((contents, next_span_start, next_span_end));
2021-10-25 06:01:02 +02:00
self.files
.push_back((filename, next_span_start, next_span_end));
self.num_files() - 1
}
2021-06-30 03:42:56 +02:00
}
2021-11-02 04:08:05 +01:00
impl Default for EngineState {
fn default() -> Self {
Self::new()
}
}
/// A temporary extension to the global state. This handles bridging between the global state and the
/// additional declarations and scope changes that are not yet part of the global scope.
///
/// This working set is created by the parser as a way of handling declarations and scope changes that
/// may later be merged or dropped (and not merged) depending on the needs of the code calling the parser.
2021-09-02 10:25:22 +02:00
pub struct StateWorkingSet<'a> {
pub permanent_state: &'a EngineState,
pub delta: StateDelta,
}
/// A delta (or change set) between the current global state and a possible future global state. Deltas
/// can be applied to the global state to update it to contain both previous state and the state held
/// within the delta.
2021-09-02 10:25:22 +02:00
pub struct StateDelta {
files: Vec<(String, usize, usize)>,
2021-10-25 06:01:02 +02:00
pub(crate) file_contents: Vec<(Vec<u8>, usize, usize)>,
vars: Vec<Type>, // indexed by VarId
decls: Vec<Box<dyn Command>>, // indexed by DeclId
blocks: Vec<Block>, // indexed by BlockId
overlays: Vec<Overlay>, // indexed by OverlayId
2021-09-06 04:20:02 +02:00
pub scope: Vec<ScopeFrame>,
#[cfg(feature = "plugin")]
plugins_changed: bool, // marks whether plugin file should be updated
2021-07-01 08:09:55 +02:00
}
2021-09-02 10:25:22 +02:00
impl StateDelta {
pub fn num_files(&self) -> usize {
self.files.len()
}
pub fn num_decls(&self) -> usize {
self.decls.len()
}
pub fn num_blocks(&self) -> usize {
self.blocks.len()
}
pub fn num_overlays(&self) -> usize {
self.overlays.len()
}
pub fn enter_scope(&mut self) {
self.scope.push(ScopeFrame::new());
}
pub fn exit_scope(&mut self) {
self.scope.pop();
}
}
2021-09-02 10:25:22 +02:00
impl<'a> StateWorkingSet<'a> {
pub fn new(permanent_state: &'a EngineState) -> Self {
2021-06-30 03:42:56 +02:00
Self {
2021-09-02 10:25:22 +02:00
delta: StateDelta {
files: vec![],
file_contents: vec![],
vars: vec![],
decls: vec![],
blocks: vec![],
overlays: vec![],
scope: vec![ScopeFrame::new()],
#[cfg(feature = "plugin")]
plugins_changed: false,
},
2021-06-30 03:42:56 +02:00
permanent_state,
}
}
pub fn num_files(&self) -> usize {
self.delta.num_files() + self.permanent_state.num_files()
2021-06-30 03:42:56 +02:00
}
2021-07-16 08:24:46 +02:00
pub fn num_decls(&self) -> usize {
self.delta.num_decls() + self.permanent_state.num_decls()
2021-07-16 08:24:46 +02:00
}
pub fn num_blocks(&self) -> usize {
self.delta.num_blocks() + self.permanent_state.num_blocks()
2021-07-16 08:24:46 +02:00
}
pub fn num_overlays(&self) -> usize {
self.delta.num_overlays() + self.permanent_state.num_overlays()
}
2021-09-02 10:25:22 +02:00
pub fn add_decl(&mut self, decl: Box<dyn Command>) -> DeclId {
let name = decl.name().as_bytes().to_vec();
2021-07-16 08:24:46 +02:00
self.delta.decls.push(decl);
2021-07-16 08:24:46 +02:00
let decl_id = self.num_decls() - 1;
2021-07-02 00:40:08 +02:00
let scope_frame = self
.delta
2021-07-02 00:40:08 +02:00
.scope
.last_mut()
.expect("internal error: missing required scope frame");
2021-07-02 00:40:08 +02:00
scope_frame.decls.insert(name, decl_id);
scope_frame.visibility.use_decl_id(&decl_id);
2021-07-02 00:40:08 +02:00
decl_id
}
pub fn use_decls(&mut self, decls: Vec<(Vec<u8>, DeclId)>) {
let scope_frame = self
.delta
.scope
.last_mut()
.expect("internal error: missing required scope frame");
for (name, decl_id) in decls {
scope_frame.decls.insert(name, decl_id);
scope_frame.visibility.use_decl_id(&decl_id);
}
}
pub fn add_predecl(&mut self, decl: Box<dyn Command>) -> Option<DeclId> {
let name = decl.name().as_bytes().to_vec();
self.delta.decls.push(decl);
let decl_id = self.num_decls() - 1;
let scope_frame = self
.delta
.scope
.last_mut()
.expect("internal error: missing required scope frame");
scope_frame.predecls.insert(name, decl_id)
}
#[cfg(feature = "plugin")]
pub fn mark_plugins_file_dirty(&mut self) {
self.delta.plugins_changed = true;
}
pub fn merge_predecl(&mut self, name: &[u8]) -> Option<DeclId> {
let scope_frame = self
.delta
.scope
.last_mut()
.expect("internal error: missing required scope frame");
if let Some(decl_id) = scope_frame.predecls.remove(name) {
scope_frame.decls.insert(name.into(), decl_id);
scope_frame.visibility.use_decl_id(&decl_id);
return Some(decl_id);
}
None
}
pub fn hide_decl(&mut self, name: &[u8]) -> Option<DeclId> {
let mut visibility: Visibility = Visibility::new();
// Since we can mutate scope frames in delta, remove the id directly
for scope in self.delta.scope.iter_mut().rev() {
visibility.append(&scope.visibility);
if let Some(decl_id) = scope.decls.remove(name) {
return Some(decl_id);
}
}
// We cannot mutate the permanent state => store the information in the current scope frame
let last_scope_frame = self
.delta
.scope
.last_mut()
.expect("internal error: missing required scope frame");
for scope in self.permanent_state.scope.iter().rev() {
visibility.append(&scope.visibility);
if let Some(decl_id) = scope.decls.get(name) {
if visibility.is_decl_id_visible(decl_id) {
2021-10-04 19:33:27 +02:00
// Hide decl only if it's not already hidden
last_scope_frame.visibility.hide_decl_id(decl_id);
return Some(*decl_id);
}
}
}
None
}
pub fn hide_decls(&mut self, decls: &[(Vec<u8>, DeclId)]) {
for decl in decls.iter() {
self.hide_decl(&decl.0); // let's assume no errors
}
}
2021-07-16 22:26:40 +02:00
pub fn add_block(&mut self, block: Block) -> BlockId {
self.delta.blocks.push(block);
2021-07-16 08:24:46 +02:00
self.num_blocks() - 1
}
pub fn add_env_var(&mut self, name_span: Span, block: Block) -> BlockId {
self.delta.blocks.push(block);
let block_id = self.num_blocks() - 1;
let name = self.get_span_contents(name_span).to_vec();
let scope_frame = self
.delta
.scope
.last_mut()
.expect("internal error: missing required scope frame");
scope_frame.env_vars.insert(name, block_id);
block_id
}
pub fn add_overlay(&mut self, name: &str, overlay: Overlay) -> OverlayId {
let name = name.as_bytes().to_vec();
self.delta.overlays.push(overlay);
let overlay_id = self.num_overlays() - 1;
let scope_frame = self
.delta
.scope
.last_mut()
.expect("internal error: missing required scope frame");
scope_frame.overlays.insert(name, overlay_id);
overlay_id
}
pub fn next_span_start(&self) -> usize {
2021-10-25 18:58:58 +02:00
let permanent_span_start = self.permanent_state.next_span_start();
if let Some((_, _, last)) = self.delta.file_contents.last() {
*last
2021-10-25 18:58:58 +02:00
} else {
permanent_span_start
}
}
2021-07-22 22:45:23 +02:00
pub fn global_span_offset(&self) -> usize {
self.permanent_state.next_span_start()
}
pub fn files(&'a self) -> impl Iterator<Item = &(String, usize, usize)> {
self.permanent_state.files().chain(self.delta.files.iter())
}
pub fn get_filename(&self, file_id: usize) -> String {
for file in self.files().enumerate() {
if file.0 == file_id {
return file.1 .0.clone();
}
}
"<unknown>".into()
}
pub fn get_file_source(&self, file_id: usize) -> String {
for file in self.files().enumerate() {
if file.0 == file_id {
let output = String::from_utf8_lossy(self.get_span_contents(Span {
start: file.1 .1,
end: file.1 .2,
}))
.to_string();
return output;
}
}
"<unknown>".into()
}
pub fn add_file(&mut self, filename: String, contents: &[u8]) -> usize {
let next_span_start = self.next_span_start();
2021-10-25 06:01:02 +02:00
let next_span_end = next_span_start + contents.len();
2021-10-25 06:01:02 +02:00
self.delta
.file_contents
.push((contents.to_vec(), next_span_start, next_span_end));
self.delta
.files
.push((filename, next_span_start, next_span_end));
2021-06-30 03:42:56 +02:00
self.num_files() - 1
}
pub fn get_span_contents(&self, span: Span) -> &[u8] {
let permanent_end = self.permanent_state.next_span_start();
if permanent_end <= span.start {
2021-10-25 06:01:02 +02:00
for (contents, start, finish) in &self.delta.file_contents {
2021-10-25 18:58:58 +02:00
if (span.start >= *start) && (span.end <= *finish) {
return &contents[(span.start - start)..(span.end - start)];
2021-10-25 06:01:02 +02:00
}
}
2021-07-02 09:15:30 +02:00
} else {
2021-10-25 06:01:02 +02:00
return self.permanent_state.get_span_contents(&span);
2021-07-02 09:15:30 +02:00
}
2021-10-25 06:01:02 +02:00
panic!("internal error: missing span contents in file cache")
2021-07-02 09:15:30 +02:00
}
2021-06-30 03:42:56 +02:00
pub fn enter_scope(&mut self) {
self.delta.enter_scope();
2021-06-30 03:42:56 +02:00
}
pub fn exit_scope(&mut self) {
self.delta.exit_scope();
2021-06-30 03:42:56 +02:00
}
2021-07-01 08:09:55 +02:00
pub fn find_decl(&self, name: &[u8]) -> Option<DeclId> {
let mut visibility: Visibility = Visibility::new();
for scope in self.delta.scope.iter().rev() {
visibility.append(&scope.visibility);
if let Some(decl_id) = scope.predecls.get(name) {
return Some(*decl_id);
}
2021-07-08 00:55:46 +02:00
if let Some(decl_id) = scope.decls.get(name) {
return Some(*decl_id);
2021-07-01 08:09:55 +02:00
}
}
for scope in self.permanent_state.scope.iter().rev() {
visibility.append(&scope.visibility);
if let Some(decl_id) = scope.decls.get(name) {
if visibility.is_decl_id_visible(decl_id) {
return Some(*decl_id);
}
2021-07-17 08:31:34 +02:00
}
}
2021-07-01 08:09:55 +02:00
None
}
pub fn find_overlay(&self, name: &[u8]) -> Option<OverlayId> {
2021-09-26 12:25:52 +02:00
for scope in self.delta.scope.iter().rev() {
if let Some(overlay_id) = scope.overlays.get(name) {
return Some(*overlay_id);
2021-09-26 12:25:52 +02:00
}
}
for scope in self.permanent_state.scope.iter().rev() {
if let Some(overlay_id) = scope.overlays.get(name) {
return Some(*overlay_id);
2021-09-26 12:25:52 +02:00
}
}
None
}
2021-09-02 10:25:22 +02:00
// pub fn update_decl(&mut self, decl_id: usize, block: Option<BlockId>) {
// let decl = self.get_decl_mut(decl_id);
// decl.body = block;
// }
2021-07-31 06:04:42 +02:00
2021-07-17 21:34:43 +02:00
pub fn contains_decl_partial_match(&self, name: &[u8]) -> bool {
for scope in self.delta.scope.iter().rev() {
2021-07-17 21:34:43 +02:00
for decl in &scope.decls {
if decl.0.starts_with(name) {
return true;
}
}
}
for scope in self.permanent_state.scope.iter().rev() {
for decl in &scope.decls {
if decl.0.starts_with(name) {
return true;
2021-07-17 21:34:43 +02:00
}
}
}
false
}
2021-07-01 08:09:55 +02:00
pub fn next_var_id(&self) -> VarId {
let num_permanent_vars = self.permanent_state.num_vars();
num_permanent_vars + self.delta.vars.len()
2021-07-01 08:09:55 +02:00
}
pub fn find_variable(&self, name: &[u8]) -> Option<VarId> {
for scope in self.delta.scope.iter().rev() {
2021-07-08 00:55:46 +02:00
if let Some(var_id) = scope.vars.get(name) {
2021-07-01 08:09:55 +02:00
return Some(*var_id);
2021-06-30 03:42:56 +02:00
}
}
for scope in self.permanent_state.scope.iter().rev() {
if let Some(var_id) = scope.vars.get(name) {
return Some(*var_id);
2021-07-17 08:31:34 +02:00
}
}
2021-06-30 03:42:56 +02:00
None
}
2021-08-09 02:19:07 +02:00
pub fn find_alias(&self, name: &[u8]) -> Option<&[Span]> {
for scope in self.delta.scope.iter().rev() {
if let Some(spans) = scope.aliases.get(name) {
return Some(spans);
}
}
for scope in self.permanent_state.scope.iter().rev() {
if let Some(spans) = scope.aliases.get(name) {
return Some(spans);
}
}
None
}
2021-07-30 00:56:51 +02:00
pub fn add_variable(&mut self, mut name: Vec<u8>, ty: Type) -> VarId {
2021-07-01 08:09:55 +02:00
let next_id = self.next_var_id();
2021-07-30 00:56:51 +02:00
// correct name if necessary
if !name.starts_with(b"$") {
name.insert(0, b'$');
}
2021-07-01 02:01:04 +02:00
let last = self
.delta
2021-07-01 02:01:04 +02:00
.scope
.last_mut()
.expect("internal error: missing stack frame");
last.vars.insert(name, next_id);
2021-07-23 23:19:30 +02:00
self.delta.vars.push(ty);
2021-07-01 02:01:04 +02:00
next_id
}
2021-07-01 08:09:55 +02:00
2021-08-09 02:19:07 +02:00
pub fn add_alias(&mut self, name: Vec<u8>, replacement: Vec<Span>) {
let last = self
.delta
.scope
.last_mut()
.expect("internal error: missing stack frame");
last.aliases.insert(name, replacement);
}
2021-07-23 23:19:30 +02:00
pub fn set_variable_type(&mut self, var_id: VarId, ty: Type) {
let num_permanent_vars = self.permanent_state.num_vars();
if var_id < num_permanent_vars {
panic!("Internal error: attempted to set into permanent state from working set")
} else {
self.delta.vars[var_id - num_permanent_vars] = ty;
}
}
2021-07-23 07:14:49 +02:00
pub fn get_variable(&self, var_id: VarId) -> &Type {
let num_permanent_vars = self.permanent_state.num_vars();
if var_id < num_permanent_vars {
2021-07-22 21:50:59 +02:00
self.permanent_state.get_var(var_id)
2021-07-01 08:09:55 +02:00
} else {
2021-07-23 07:14:49 +02:00
self.delta
.vars
.get(var_id - num_permanent_vars)
.expect("internal error: missing variable")
2021-07-01 08:09:55 +02:00
}
}
2021-09-04 09:59:38 +02:00
#[allow(clippy::borrowed_box)]
2021-09-02 10:25:22 +02:00
pub fn get_decl(&self, decl_id: DeclId) -> &Box<dyn Command> {
let num_permanent_decls = self.permanent_state.num_decls();
if decl_id < num_permanent_decls {
2021-07-22 21:50:59 +02:00
self.permanent_state.get_decl(decl_id)
} else {
2021-07-23 07:14:49 +02:00
self.delta
.decls
.get(decl_id - num_permanent_decls)
.expect("internal error: missing declaration")
2021-07-22 21:50:59 +02:00
}
}
2021-09-02 10:25:22 +02:00
pub fn get_decl_mut(&mut self, decl_id: DeclId) -> &mut Box<dyn Command> {
2021-07-31 06:04:42 +02:00
let num_permanent_decls = self.permanent_state.num_decls();
if decl_id < num_permanent_decls {
panic!("internal error: can only mutate declarations in working set")
} else {
self.delta
.decls
.get_mut(decl_id - num_permanent_decls)
.expect("internal error: missing declaration")
}
}
2021-09-10 00:09:40 +02:00
pub fn find_commands_by_prefix(&self, name: &[u8]) -> Vec<Vec<u8>> {
let mut output = vec![];
for scope in self.delta.scope.iter().rev() {
for decl in &scope.decls {
if decl.0.starts_with(name) {
output.push(decl.0.clone());
}
}
}
let mut permanent = self.permanent_state.find_commands_by_prefix(name);
output.append(&mut permanent);
output
}
2021-07-23 07:14:49 +02:00
pub fn get_block(&self, block_id: BlockId) -> &Block {
2021-07-22 21:50:59 +02:00
let num_permanent_blocks = self.permanent_state.num_blocks();
if block_id < num_permanent_blocks {
self.permanent_state.get_block(block_id)
2021-07-01 08:09:55 +02:00
} else {
2021-07-23 07:14:49 +02:00
self.delta
.blocks
.get(block_id - num_permanent_blocks)
.expect("internal error: missing block")
2021-07-01 08:09:55 +02:00
}
}
2021-07-22 09:48:45 +02:00
pub fn get_overlay(&self, overlay_id: OverlayId) -> &Overlay {
let num_permanent_overlays = self.permanent_state.num_overlays();
if overlay_id < num_permanent_overlays {
self.permanent_state.get_overlay(overlay_id)
} else {
self.delta
.overlays
.get(overlay_id - num_permanent_overlays)
.expect("internal error: missing overlay")
}
}
pub fn get_block_mut(&mut self, block_id: BlockId) -> &mut Block {
let num_permanent_blocks = self.permanent_state.num_blocks();
if block_id < num_permanent_blocks {
panic!("Attempt to mutate a block that is in the permanent (immutable) state")
} else {
self.delta
.blocks
.get_mut(block_id - num_permanent_blocks)
.expect("internal error: missing block")
}
}
2021-09-02 10:25:22 +02:00
pub fn render(self) -> StateDelta {
2021-07-22 09:48:45 +02:00
self.delta
}
2021-07-01 02:01:04 +02:00
}
2021-06-30 03:42:56 +02:00
impl<'a> miette::SourceCode for &StateWorkingSet<'a> {
fn read_span<'b>(
&'b self,
span: &miette::SourceSpan,
context_lines_before: usize,
context_lines_after: usize,
) -> Result<Box<dyn miette::SpanContents + 'b>, miette::MietteError> {
let debugging = std::env::var("MIETTE_DEBUG").is_ok();
if debugging {
let finding_span = "Finding span in StateWorkingSet";
dbg!(finding_span, span);
}
for (filename, start, end) in self.files() {
if debugging {
dbg!(&filename, start, end);
}
if span.offset() >= *start && span.offset() + span.len() <= *end {
if debugging {
let found_file = "Found matching file";
dbg!(found_file);
}
let our_span = Span {
start: *start,
end: *end,
};
// We need to move to a local span because we're only reading
// the specific file contents via self.get_span_contents.
let local_span = (span.offset() - *start, span.len()).into();
if debugging {
dbg!(&local_span);
}
let span_contents = self.get_span_contents(our_span);
if debugging {
dbg!(String::from_utf8_lossy(span_contents));
}
let span_contents = span_contents.read_span(
&local_span,
context_lines_before,
context_lines_after,
)?;
let content_span = span_contents.span();
// Back to "global" indexing
let retranslated = (content_span.offset() + start, content_span.len()).into();
if debugging {
dbg!(&retranslated);
}
2021-09-21 06:03:06 +02:00
let data = span_contents.data();
2021-09-21 06:03:06 +02:00
if filename == "<cli>" {
if debugging {
let success_cli = "Successfully read CLI span";
dbg!(success_cli, String::from_utf8_lossy(data));
}
2021-09-21 06:03:06 +02:00
return Ok(Box::new(miette::MietteSpanContents::new(
data,
2021-09-21 06:03:06 +02:00
retranslated,
span_contents.line(),
span_contents.column(),
span_contents.line_count(),
)));
} else {
if debugging {
let success_file = "Successfully read file span";
dbg!(success_file);
}
2021-09-21 06:03:06 +02:00
return Ok(Box::new(miette::MietteSpanContents::new_named(
filename.clone(),
data,
2021-09-21 06:03:06 +02:00
retranslated,
span_contents.line(),
span_contents.column(),
span_contents.line_count(),
)));
}
2021-09-02 10:25:22 +02:00
}
}
Err(miette::MietteError::OutOfBounds)
2021-09-02 10:25:22 +02:00
}
}
2021-06-30 03:42:56 +02:00
#[cfg(test)]
2021-09-02 10:25:22 +02:00
mod engine_state_tests {
2021-06-30 03:42:56 +02:00
use super::*;
#[test]
fn add_file_gives_id() {
2021-09-02 10:25:22 +02:00
let engine_state = EngineState::new();
let mut engine_state = StateWorkingSet::new(&engine_state);
let id = engine_state.add_file("test.nu".into(), &[]);
2021-06-30 03:42:56 +02:00
assert_eq!(id, 0);
}
#[test]
fn add_file_gives_id_including_parent() {
2021-09-02 10:25:22 +02:00
let mut engine_state = EngineState::new();
let parent_id = engine_state.add_file("test.nu".into(), vec![]);
2021-06-30 03:42:56 +02:00
2021-09-02 10:25:22 +02:00
let mut working_set = StateWorkingSet::new(&engine_state);
let working_set_id = working_set.add_file("child.nu".into(), &[]);
2021-06-30 03:42:56 +02:00
assert_eq!(parent_id, 0);
assert_eq!(working_set_id, 1);
}
#[test]
fn merge_states() -> Result<(), ShellError> {
2021-09-02 10:25:22 +02:00
let mut engine_state = EngineState::new();
engine_state.add_file("test.nu".into(), vec![]);
2021-06-30 03:42:56 +02:00
let delta = {
2021-09-02 10:25:22 +02:00
let mut working_set = StateWorkingSet::new(&engine_state);
working_set.add_file("child.nu".into(), &[]);
2021-07-22 09:48:45 +02:00
working_set.render()
};
2021-06-30 03:42:56 +02:00
engine_state.merge_delta(delta)?;
2021-06-30 03:42:56 +02:00
2021-09-02 10:25:22 +02:00
assert_eq!(engine_state.num_files(), 2);
assert_eq!(&engine_state.files[0].0, "test.nu");
assert_eq!(&engine_state.files[1].0, "child.nu");
Ok(())
2021-06-30 03:42:56 +02:00
}
}