Optimize LSP snippet parsing

This commit is contained in:
Urgau 2023-02-07 20:15:39 +01:00 committed by Blaž Hrastnik
parent 9c12e0fb76
commit e973b71c83
4 changed files with 57 additions and 19 deletions

1
Cargo.lock generated
View File

@ -1141,7 +1141,6 @@ dependencies = [
"helix-loader",
"log",
"lsp-types",
"once_cell",
"serde",
"serde_json",
"thiserror",

View File

@ -27,4 +27,3 @@ thiserror = "1.0"
tokio = { version = "1.26", features = ["rt", "rt-multi-thread", "io-util", "io-std", "time", "process", "macros", "fs", "parking_lot", "sync"] }
tokio-stream = "0.1.12"
which = "4.4"
once_cell = "1.15"

View File

@ -50,14 +50,11 @@ pub struct Snippet<'a> {
elements: Vec<SnippetElement<'a>>,
}
pub fn parse<'a>(s: &'a str) -> Result<Snippet<'a>> {
pub fn parse(s: &str) -> Result<Snippet<'_>> {
parser::parse(s).map_err(|rest| anyhow!("Failed to parse snippet. Remaining input: {}", rest))
}
mod parser {
use helix_core::regex;
use once_cell::sync::Lazy;
use helix_parsec::*;
use super::{CaseChange, FormatItem, Regex, Snippet, SnippetElement};
@ -86,17 +83,34 @@ mod parser {
else ::= text
*/
static DIGIT: Lazy<regex::Regex> = Lazy::new(|| regex::Regex::new(r"^[0-9]+").unwrap());
static VARIABLE: Lazy<regex::Regex> =
Lazy::new(|| regex::Regex::new(r"^[_a-zA-Z][_a-zA-Z0-9]*").unwrap());
static TEXT: Lazy<regex::Regex> = Lazy::new(|| regex::Regex::new(r"^[^\$]+").unwrap());
fn var<'a>() -> impl Parser<'a, Output = &'a str> {
pattern(&VARIABLE)
// var = [_a-zA-Z][_a-zA-Z0-9]*
move |input: &'a str| match input
.char_indices()
.take_while(|(p, c)| {
*c == '_'
|| if *p == 0 {
c.is_ascii_alphabetic()
} else {
c.is_ascii_alphanumeric()
}
})
.last()
{
Some((index, c)) if index >= 1 => {
let index = index + c.len_utf8();
Ok((&input[index..], &input[0..index]))
}
_ => Err(input),
}
}
fn text<'a>() -> impl Parser<'a, Output = &'a str> {
take_while(|c| c != '$')
}
fn digit<'a>() -> impl Parser<'a, Output = usize> {
filter_map(pattern(&DIGIT), |s| s.parse().ok())
filter_map(take_while(|c| c.is_ascii_digit()), |s| s.parse().ok())
}
fn case_change<'a>() -> impl Parser<'a, Output = CaseChange> {
@ -152,7 +166,7 @@ fn format<'a>() -> impl Parser<'a, Output = FormatItem<'a>> {
|seq| { Conditional(seq.1, None, Some(seq.4)) }
),
// Any text
map(pattern(&TEXT), Text),
map(text(), Text),
)
}
@ -245,12 +259,9 @@ fn variable<'a>() -> impl Parser<'a, Output = SnippetElement<'a>> {
)
}
fn text<'a>() -> impl Parser<'a, Output = SnippetElement<'a>> {
map(pattern(&TEXT), SnippetElement::Text)
}
fn anything<'a>() -> impl Parser<'a, Output = SnippetElement<'a>> {
choice!(tabstop(), placeholder(), choice(), variable(), text())
let text = map(text(), SnippetElement::Text);
choice!(tabstop(), placeholder(), choice(), variable(), text)
}
fn snippet<'a>() -> impl Parser<'a, Output = Snippet<'a>> {

View File

@ -157,6 +157,35 @@ pub fn take_until<'a, F>(pattern: F) -> impl Parser<'a, Output = &'a str>
}
}
/// A parser which matches all values until the specified pattern no longer match.
///
/// This parser only ever fails if the input has a length of zero.
///
/// # Examples
///
/// ```
/// use helix_parsec::{take_while, Parser};
/// let parser = take_while(|c| c == '1');
/// assert_eq!(Ok(("2", "11")), parser.parse("112"));
/// assert_eq!(Err("22"), parser.parse("22"));
/// ```
pub fn take_while<'a, F>(pattern: F) -> impl Parser<'a, Output = &'a str>
where
F: Fn(char) -> bool,
{
move |input: &'a str| match input
.char_indices()
.take_while(|(_p, c)| pattern(*c))
.last()
{
Some((index, c)) => {
let index = index + c.len_utf8();
Ok((&input[index..], &input[0..index]))
}
_ => Err(input),
}
}
// Variadic parser combinators
/// A parser combinator which matches a sequence of parsers in an all-or-nothing fashion.