forked from extern/nushell
b943cbedff
This pr - fixes https://github.com/nushell/nushell/issues/10143 - fixes https://github.com/nushell/nushell/issues/5559 # Description Current `lite_parse` does not handle multiple line comments and eols in pipeline. When parsing the following tokens: | `"abcdefg"` | ` \|` | `# foobar` | ` \n` | `split chars` | | ------------- | ------------- |------------- |------------- |------------- | | [Command] | [Pipe] | [Comment] | [Eol] | [Command] | | | | Last Token |Current Token | | `TokenContent::Eol` handler only checks if `last_token` is `Pipe` but it will be broken if there exist any other thing, e.g. extra `[Comment]` in this example. This pr make the following change: - While parsing `[Eol]`, try to find the last non-comment token as `last_token` - Comment is supposed as `[Comment]+` or `([Comment] [Eol])+` - `[Eol]+` is still parsed just like current nu (i.e. generates `nothing`). Notice that this pr is just a quick patch if more comment/eol related issue occures, `lite_parser` may need a rewrite. # User-Facing Changes Now the following pipeline works: ```bash 1 | # comment each { |it| $it + 2 } | # comment math sum ``` Comment will not end the pipeline in interactive mode: ```bash ❯ 1 | # comment (now enter multiple line mode instead of end) ▶▶ # foo ▶▶ 2 ``` # Tests + Formatting - 🟢 `toolkit fmt` - 🟢 `toolkit clippy` - 🟢 `toolkit test` - 🟢 `toolkit test stdlib` # After Submitting None --------- Co-authored-by: Horasal <horsal@horsal.dev>
486 lines
18 KiB
Rust
486 lines
18 KiB
Rust
/// Lite parsing converts a flat stream of tokens from the lexer to a syntax element structure that
|
|
/// can be parsed.
|
|
use crate::{Token, TokenContents};
|
|
|
|
use nu_protocol::{ast::Redirection, ParseError, Span};
|
|
|
|
#[derive(Debug)]
|
|
pub struct LiteCommand {
|
|
pub comments: Vec<Span>,
|
|
pub parts: Vec<Span>,
|
|
}
|
|
|
|
impl Default for LiteCommand {
|
|
fn default() -> Self {
|
|
Self::new()
|
|
}
|
|
}
|
|
|
|
impl LiteCommand {
|
|
pub fn new() -> Self {
|
|
Self {
|
|
comments: vec![],
|
|
parts: vec![],
|
|
}
|
|
}
|
|
|
|
pub fn push(&mut self, span: Span) {
|
|
self.parts.push(span);
|
|
}
|
|
|
|
pub fn is_empty(&self) -> bool {
|
|
self.parts.is_empty()
|
|
}
|
|
}
|
|
|
|
// Note: the Span is the span of the connector not the whole element
|
|
#[derive(Debug)]
|
|
pub enum LiteElement {
|
|
Command(Option<Span>, LiteCommand),
|
|
Redirection(Span, Redirection, LiteCommand),
|
|
// SeparateRedirection variant can only be generated by two different Redirection variant
|
|
SeparateRedirection {
|
|
out: (Span, LiteCommand),
|
|
err: (Span, LiteCommand),
|
|
},
|
|
// SameTargetRedirection variant can only be generated by Command with Redirection::OutAndErr
|
|
SameTargetRedirection {
|
|
cmd: (Option<Span>, LiteCommand),
|
|
redirection: (Span, LiteCommand),
|
|
},
|
|
}
|
|
|
|
#[derive(Debug, Default)]
|
|
pub struct LitePipeline {
|
|
pub commands: Vec<LiteElement>,
|
|
}
|
|
|
|
impl LitePipeline {
|
|
pub fn new() -> Self {
|
|
Self { commands: vec![] }
|
|
}
|
|
|
|
pub fn push(&mut self, element: LiteElement) {
|
|
self.commands.push(element);
|
|
}
|
|
|
|
pub fn insert(&mut self, index: usize, element: LiteElement) {
|
|
self.commands.insert(index, element);
|
|
}
|
|
|
|
pub fn is_empty(&self) -> bool {
|
|
self.commands.is_empty()
|
|
}
|
|
}
|
|
|
|
#[derive(Debug)]
|
|
pub struct LiteBlock {
|
|
pub block: Vec<LitePipeline>,
|
|
}
|
|
|
|
impl Default for LiteBlock {
|
|
fn default() -> Self {
|
|
Self::new()
|
|
}
|
|
}
|
|
|
|
impl LiteBlock {
|
|
pub fn new() -> Self {
|
|
Self { block: vec![] }
|
|
}
|
|
|
|
pub fn push(&mut self, mut pipeline: LitePipeline) {
|
|
// once we push `pipeline` to our block
|
|
// the block takes ownership of `pipeline`, which means that
|
|
// our `pipeline` is complete on collecting commands.
|
|
self.merge_redirections(&mut pipeline);
|
|
self.merge_cmd_with_outerr_redirection(&mut pipeline);
|
|
|
|
self.block.push(pipeline);
|
|
}
|
|
|
|
pub fn is_empty(&self) -> bool {
|
|
self.block.is_empty()
|
|
}
|
|
|
|
fn merge_cmd_with_outerr_redirection(&self, pipeline: &mut LitePipeline) {
|
|
let mut cmd_index = None;
|
|
let mut outerr_index = None;
|
|
for (index, cmd) in pipeline.commands.iter().enumerate() {
|
|
if let LiteElement::Command(..) = cmd {
|
|
cmd_index = Some(index);
|
|
}
|
|
if let LiteElement::Redirection(_span, Redirection::StdoutAndStderr, _target_cmd) = cmd
|
|
{
|
|
outerr_index = Some(index);
|
|
break;
|
|
}
|
|
}
|
|
if let (Some(cmd_index), Some(outerr_index)) = (cmd_index, outerr_index) {
|
|
// we can make sure that cmd_index is less than outerr_index.
|
|
let outerr_redirect = pipeline.commands.remove(outerr_index);
|
|
let cmd = pipeline.commands.remove(cmd_index);
|
|
// `outerr_redirect` and `cmd` should always be `LiteElement::Command` and `LiteElement::Redirection`
|
|
if let (
|
|
LiteElement::Command(cmd_span, lite_cmd),
|
|
LiteElement::Redirection(span, _, outerr_cmd),
|
|
) = (cmd, outerr_redirect)
|
|
{
|
|
pipeline.insert(
|
|
cmd_index,
|
|
LiteElement::SameTargetRedirection {
|
|
cmd: (cmd_span, lite_cmd),
|
|
redirection: (span, outerr_cmd),
|
|
},
|
|
)
|
|
}
|
|
}
|
|
}
|
|
|
|
fn merge_redirections(&self, pipeline: &mut LitePipeline) {
|
|
// In case our command may contains both stdout and stderr redirection.
|
|
// We pick them out and Combine them into one LiteElement::SeparateRedirection variant.
|
|
let mut stdout_index = None;
|
|
let mut stderr_index = None;
|
|
for (index, cmd) in pipeline.commands.iter().enumerate() {
|
|
if let LiteElement::Redirection(_span, redirection, _target_cmd) = cmd {
|
|
match *redirection {
|
|
Redirection::Stderr => stderr_index = Some(index),
|
|
Redirection::Stdout => stdout_index = Some(index),
|
|
Redirection::StdoutAndStderr => {}
|
|
}
|
|
}
|
|
}
|
|
|
|
if let (Some(out_indx), Some(err_indx)) = (stdout_index, stderr_index) {
|
|
let (out_redirect, err_redirect, new_indx) = {
|
|
// to avoid panic, we need to remove commands which have larger index first.
|
|
if out_indx > err_indx {
|
|
let out_redirect = pipeline.commands.remove(out_indx);
|
|
let err_redirect = pipeline.commands.remove(err_indx);
|
|
(out_redirect, err_redirect, err_indx)
|
|
} else {
|
|
let err_redirect = pipeline.commands.remove(err_indx);
|
|
let out_redirect = pipeline.commands.remove(out_indx);
|
|
(out_redirect, err_redirect, out_indx)
|
|
}
|
|
};
|
|
// `out_redirect` and `err_redirect` should always be `LiteElement::Redirection`
|
|
if let (
|
|
LiteElement::Redirection(out_span, _, out_command),
|
|
LiteElement::Redirection(err_span, _, err_command),
|
|
) = (out_redirect, err_redirect)
|
|
{
|
|
// using insert with specific index to keep original
|
|
// pipeline commands order.
|
|
pipeline.insert(
|
|
new_indx,
|
|
LiteElement::SeparateRedirection {
|
|
out: (out_span, out_command),
|
|
err: (err_span, err_command),
|
|
},
|
|
)
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
fn last_non_comment_token(tokens: &[Token], cur_idx: usize) -> Option<TokenContents> {
|
|
let mut expect = TokenContents::Comment;
|
|
for token in tokens.iter().take(cur_idx).rev() {
|
|
// skip ([Comment]+ [Eol]) pair
|
|
match (token.contents, expect) {
|
|
(TokenContents::Comment, TokenContents::Comment)
|
|
| (TokenContents::Comment, TokenContents::Eol) => expect = TokenContents::Eol,
|
|
(TokenContents::Eol, TokenContents::Eol) => expect = TokenContents::Comment,
|
|
(token, _) => return Some(token),
|
|
}
|
|
}
|
|
None
|
|
}
|
|
|
|
pub fn lite_parse(tokens: &[Token]) -> (LiteBlock, Option<ParseError>) {
|
|
let mut block = LiteBlock::new();
|
|
let mut curr_pipeline = LitePipeline::new();
|
|
let mut curr_command = LiteCommand::new();
|
|
|
|
let mut last_token = TokenContents::Eol;
|
|
|
|
let mut last_connector = TokenContents::Pipe;
|
|
let mut last_connector_span: Option<Span> = None;
|
|
|
|
if tokens.is_empty() {
|
|
return (LiteBlock::new(), None);
|
|
}
|
|
|
|
let mut curr_comment: Option<Vec<Span>> = None;
|
|
|
|
let mut error = None;
|
|
|
|
for (idx, token) in tokens.iter().enumerate() {
|
|
match &token.contents {
|
|
TokenContents::PipePipe => {
|
|
error = error.or(Some(ParseError::ShellOrOr(token.span)));
|
|
curr_command.push(token.span);
|
|
last_token = TokenContents::Item;
|
|
}
|
|
TokenContents::Item => {
|
|
// If we have a comment, go ahead and attach it
|
|
if let Some(curr_comment) = curr_comment.take() {
|
|
curr_command.comments = curr_comment;
|
|
}
|
|
curr_command.push(token.span);
|
|
last_token = TokenContents::Item;
|
|
}
|
|
TokenContents::OutGreaterThan
|
|
| TokenContents::ErrGreaterThan
|
|
| TokenContents::OutErrGreaterThan => {
|
|
if !curr_command.is_empty() {
|
|
match last_connector {
|
|
TokenContents::OutGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span
|
|
.expect("internal error: redirection missing span information"),
|
|
Redirection::Stdout,
|
|
curr_command,
|
|
));
|
|
}
|
|
TokenContents::ErrGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span
|
|
.expect("internal error: redirection missing span information"),
|
|
Redirection::Stderr,
|
|
curr_command,
|
|
));
|
|
}
|
|
TokenContents::OutErrGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span
|
|
.expect("internal error: redirection missing span information"),
|
|
Redirection::StdoutAndStderr,
|
|
curr_command,
|
|
));
|
|
}
|
|
_ => {
|
|
curr_pipeline
|
|
.push(LiteElement::Command(last_connector_span, curr_command));
|
|
}
|
|
}
|
|
curr_command = LiteCommand::new();
|
|
}
|
|
last_token = token.contents;
|
|
last_connector = token.contents;
|
|
last_connector_span = Some(token.span);
|
|
}
|
|
TokenContents::Pipe => {
|
|
if !curr_command.is_empty() {
|
|
match last_connector {
|
|
TokenContents::OutGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
token.span,
|
|
Redirection::Stdout,
|
|
curr_command,
|
|
));
|
|
}
|
|
TokenContents::ErrGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
token.span,
|
|
Redirection::Stderr,
|
|
curr_command,
|
|
));
|
|
}
|
|
TokenContents::OutErrGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
token.span,
|
|
Redirection::StdoutAndStderr,
|
|
curr_command,
|
|
));
|
|
}
|
|
_ => {
|
|
curr_pipeline
|
|
.push(LiteElement::Command(last_connector_span, curr_command));
|
|
}
|
|
}
|
|
curr_command = LiteCommand::new();
|
|
}
|
|
last_token = TokenContents::Pipe;
|
|
last_connector = TokenContents::Pipe;
|
|
last_connector_span = Some(token.span);
|
|
}
|
|
TokenContents::Eol => {
|
|
// Handle `[Command] [Pipe] ([Comment] | [Eol])+ [Command]`
|
|
//
|
|
// `[Eol]` branch checks if previous token is `[Pipe]` to construct pipeline
|
|
// and so `[Comment] | [Eol]` should be ignore to make it work
|
|
let actual_token = last_non_comment_token(tokens, idx);
|
|
if actual_token != Some(TokenContents::Pipe)
|
|
&& actual_token != Some(TokenContents::OutGreaterThan)
|
|
{
|
|
if !curr_command.is_empty() {
|
|
match last_connector {
|
|
TokenContents::OutGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span.expect(
|
|
"internal error: redirection missing span information",
|
|
),
|
|
Redirection::Stdout,
|
|
curr_command,
|
|
));
|
|
}
|
|
TokenContents::ErrGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span.expect(
|
|
"internal error: redirection missing span information",
|
|
),
|
|
Redirection::Stderr,
|
|
curr_command,
|
|
));
|
|
}
|
|
TokenContents::OutErrGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span.expect(
|
|
"internal error: redirection missing span information",
|
|
),
|
|
Redirection::StdoutAndStderr,
|
|
curr_command,
|
|
));
|
|
}
|
|
_ => {
|
|
curr_pipeline
|
|
.push(LiteElement::Command(last_connector_span, curr_command));
|
|
}
|
|
}
|
|
|
|
curr_command = LiteCommand::new();
|
|
}
|
|
|
|
if !curr_pipeline.is_empty() {
|
|
block.push(curr_pipeline);
|
|
|
|
curr_pipeline = LitePipeline::new();
|
|
last_connector = TokenContents::Pipe;
|
|
last_connector_span = None;
|
|
}
|
|
}
|
|
|
|
if last_token == TokenContents::Eol {
|
|
// Clear out the comment as we're entering a new comment
|
|
curr_comment = None;
|
|
}
|
|
|
|
last_token = TokenContents::Eol;
|
|
}
|
|
TokenContents::Semicolon => {
|
|
if !curr_command.is_empty() {
|
|
match last_connector {
|
|
TokenContents::OutGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span
|
|
.expect("internal error: redirection missing span information"),
|
|
Redirection::Stdout,
|
|
curr_command,
|
|
));
|
|
}
|
|
TokenContents::ErrGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span
|
|
.expect("internal error: redirection missing span information"),
|
|
Redirection::Stderr,
|
|
curr_command,
|
|
));
|
|
}
|
|
TokenContents::OutErrGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span
|
|
.expect("internal error: redirection missing span information"),
|
|
Redirection::StdoutAndStderr,
|
|
curr_command,
|
|
));
|
|
}
|
|
_ => {
|
|
curr_pipeline
|
|
.push(LiteElement::Command(last_connector_span, curr_command));
|
|
}
|
|
}
|
|
|
|
curr_command = LiteCommand::new();
|
|
}
|
|
|
|
if !curr_pipeline.is_empty() {
|
|
block.push(curr_pipeline);
|
|
|
|
curr_pipeline = LitePipeline::new();
|
|
last_connector = TokenContents::Pipe;
|
|
last_connector_span = None;
|
|
}
|
|
|
|
last_token = TokenContents::Semicolon;
|
|
}
|
|
TokenContents::Comment => {
|
|
// Comment is beside something
|
|
if last_token != TokenContents::Eol {
|
|
curr_command.comments.push(token.span);
|
|
curr_comment = None;
|
|
} else {
|
|
// Comment precedes something
|
|
if let Some(curr_comment) = &mut curr_comment {
|
|
curr_comment.push(token.span);
|
|
} else {
|
|
curr_comment = Some(vec![token.span]);
|
|
}
|
|
}
|
|
|
|
last_token = TokenContents::Comment;
|
|
}
|
|
}
|
|
}
|
|
|
|
if !curr_command.is_empty() {
|
|
match last_connector {
|
|
TokenContents::OutGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span
|
|
.expect("internal error: redirection missing span information"),
|
|
Redirection::Stdout,
|
|
curr_command,
|
|
));
|
|
}
|
|
TokenContents::ErrGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span
|
|
.expect("internal error: redirection missing span information"),
|
|
Redirection::Stderr,
|
|
curr_command,
|
|
));
|
|
}
|
|
TokenContents::OutErrGreaterThan => {
|
|
curr_pipeline.push(LiteElement::Redirection(
|
|
last_connector_span
|
|
.expect("internal error: redirection missing span information"),
|
|
Redirection::StdoutAndStderr,
|
|
curr_command,
|
|
));
|
|
}
|
|
_ => {
|
|
curr_pipeline.push(LiteElement::Command(last_connector_span, curr_command));
|
|
}
|
|
}
|
|
}
|
|
|
|
if !curr_pipeline.is_empty() {
|
|
block.push(curr_pipeline);
|
|
}
|
|
|
|
if last_non_comment_token(tokens, tokens.len()) == Some(TokenContents::Pipe) {
|
|
(
|
|
block,
|
|
Some(ParseError::UnexpectedEof(
|
|
"pipeline missing end".into(),
|
|
tokens[tokens.len() - 1].span,
|
|
)),
|
|
)
|
|
} else {
|
|
(block, error)
|
|
}
|
|
}
|