chroma-syntaxis/src/lib.rs

175 lines
4.6 KiB
Rust

use std::collections::HashMap;
use html_escape::encode_text;
use tree_sitter_highlight::{HighlightConfiguration, HighlightEvent, Highlighter};
pub mod languages;
fn get_hash_for_attrs(attrs: &[&str]) -> u64 {
use std::{
collections::hash_map::DefaultHasher,
hash::{Hash, Hasher},
};
let mut hasher = DefaultHasher::new();
attrs.hash(&mut hasher);
hasher.finish()
}
fn write_opening_tag(out: &mut String, attrs: &[&str]) {
out.push_str("<span class=\"");
for (i, attr) in attrs.iter().enumerate() {
if i != 0 {
out.push(' ');
}
out.push_str(&attr.replace('.', " "));
}
out.push_str("\">");
}
fn build_highlighted_html<'a>(
source: &'a str,
events: impl Iterator<Item = HighlightEvent> + 'a,
highlight_names: &[&str],
) -> String {
let mut highlight_attrs: Vec<&str> = Vec::new();
let mut out = String::new();
// Collapse adjacent identical attribute sets
let mut last_attrs: u64 = 0;
let mut tag_is_open = false;
for event in events {
match event {
HighlightEvent::Source { start, end } => {
let source_section = &source[start..end];
let attr_hash = get_hash_for_attrs(&highlight_attrs);
if last_attrs != attr_hash && tag_is_open {
out.push_str("</span>");
tag_is_open = false;
}
if !highlight_attrs.is_empty() && (!tag_is_open || last_attrs != attr_hash) {
write_opening_tag(&mut out, &highlight_attrs);
tag_is_open = true;
last_attrs = attr_hash;
}
out.push_str(&encode_text(source_section));
}
HighlightEvent::HighlightStart(highlight) => {
let capture_name = highlight_names[highlight.0];
highlight_attrs.push(capture_name);
}
HighlightEvent::HighlightEnd => {
highlight_attrs.pop();
}
}
}
if tag_is_open {
out.push_str("</span>");
}
out
}
pub struct SyntaxHighlighter<'a> {
languages: HashMap<String, HighlightConfiguration>,
highlight_names: &'a [&'a str],
}
impl SyntaxHighlighter<'_> {
pub fn new() -> Self {
let mut highlighter = Self {
languages: HashMap::new(),
highlight_names: languages::COMMON_HIGHLIGHT_NAMES,
};
languages::register_builtin_languages(&mut highlighter);
highlighter
}
}
impl Default for SyntaxHighlighter<'_> {
fn default() -> Self {
Self::new()
}
}
impl<'a> SyntaxHighlighter<'a> {
pub fn new_empty(highlight_names: &'a [&'a str]) -> Self {
Self {
languages: HashMap::new(),
highlight_names,
}
}
pub fn register(&mut self, lang: String, config: HighlightConfiguration) {
self.languages.insert(lang, config);
}
pub fn highlight(&self, lang: &str, source: &str) -> String {
let highlight_config = match self.languages.get(lang) {
Some(config) => config,
None => return source.to_string(),
};
let source_bytes = source.as_bytes();
let mut highlighter = Highlighter::new();
let highlight_result =
highlighter.highlight(highlight_config, source_bytes, None, |injected_lang| {
self.languages.get(injected_lang)
});
let events = match highlight_result {
Ok(events) => events,
Err(_) => return source.to_string(),
}
.filter_map(|e| e.ok());
build_highlighted_html(source, events, self.highlight_names)
}
}
#[cfg(test)]
mod tests {
use crate::*;
#[test]
fn highlight_rust() {
let highlighter = SyntaxHighlighter::new();
let source = include_str!("../samples/fizzbuzz.rs");
let expected_result = include_str!("../samples/fizzbuzz.rs.html");
assert_eq!(highlighter.highlight("rust", source), expected_result);
}
#[test]
fn highlight_js() {
let highlighter = SyntaxHighlighter::new();
let source = include_str!("../samples/fizzbuzz.js");
let expected_result = include_str!("../samples/fizzbuzz.js.html");
assert_eq!(highlighter.highlight("js", source), expected_result);
}
#[test]
fn highlight_python() {
let highlighter = SyntaxHighlighter::new();
let source = include_str!("../samples/fizzbuzz.py");
let expected_result = include_str!("../samples/fizzbuzz.py.html");
assert_eq!(highlighter.highlight("python", source), expected_result);
}
}