mdxt/container/codefence/
syntect.rs

1use crate::escape::escape_htmls;
2use crate::utils::{from_v32, inclusive_split, into_v32};
3use lazy_static::lazy_static;
4use syntect::easy::HighlightLines;
5use syntect::highlighting::{Color, Theme, ThemeSet};
6use syntect::parsing::SyntaxSet;
7
8lazy_static! {
9    pub static ref SYNTAX_SET: SyntaxSet = {
10        let default_set = SyntaxSet::load_defaults_newlines();
11        let mut syntax_set_builder = default_set.into_builder();
12
13        if let Err(e) = syntax_set_builder.add_from_folder("./extra_syntaxes", true) {
14            println!("Error while reading `./extra_syntaxes`: {e:?}");
15        }
16
17        syntax_set_builder.build()
18    };
19    pub static ref THEME: Theme = {
20        let theme_set = ThemeSet::load_defaults();
21
22        theme_set.themes["base16-eighties.dark"].clone()
23    };
24}
25
26pub fn highlight_syntax(content: &[u32], language: &[u32], class_prefix: &str) -> Vec<Vec<u32>> {
27
28    #[cfg(test)]
29    assert!(is_syntax_available(language));
30
31    // it assumes that the given language is available
32    let syntax_reference = SYNTAX_SET.find_syntax_by_token(&from_v32(language)).unwrap();
33    let mut highlighter = HighlightLines::new(syntax_reference, &THEME);
34    let mut result = vec![];
35
36    // it needs `\n` characters to highlight syntax properly (eg: without `\n`, single line comments don't work)
37    for line_u32 in inclusive_split(&content, '\n' as u32).into_iter() {
38        let line_u32 = line_u32.to_vec();
39
40        let curr_line = &from_v32(&line_u32);
41        let mut curr_stack = vec![];
42
43        match highlighter.highlight_line(curr_line, &SYNTAX_SET) {
44            Ok(styled_line) => {
45
46                for (style, content) in styled_line.into_iter() {
47                    let curr_stack_len = curr_stack.len();
48
49                    if curr_stack.is_empty() {
50                        curr_stack.push((style.foreground.clone(), content.to_string()));
51                    }
52
53                    else if curr_stack.last().unwrap().0 == style.foreground {
54                        curr_stack[curr_stack_len - 1].1 = format!("{}{content}", curr_stack.last().unwrap().1);
55                    }
56
57                    else if content.chars().all(|c| c == ' ') {
58                        curr_stack[curr_stack_len - 1].1 = format!("{}{content}", curr_stack.last().unwrap().1);
59                    }
60
61                    else {
62                        curr_stack.push((style.foreground.clone(), content.to_string()));
63                    }
64
65                }
66
67                result.push(curr_stack.iter().map(
68                    |(color, content)|
69                    classify_style_to_css(color, content, class_prefix)
70                ).collect::<Vec<Vec<u32>>>().concat());
71            }
72            Err(_) => {
73                result.push(classify_style_to_css(&Color::WHITE, curr_line, class_prefix));
74            }
75        }
76
77    }
78
79    result
80}
81
82pub fn is_syntax_available(language: &[u32]) -> bool {
83    SYNTAX_SET.find_syntax_by_token(&from_v32(language)).is_some()
84}
85
86fn classify_style_to_css(color: &Color, content: &str, class_prefix: &str) -> Vec<u32> {
87
88    // convert syntect's palette to its own
89    let color = match color {
90        Color { r: 211, g: 208, b: 200, .. } => "white",
91        Color { r: 45, g: 45, b: 45, .. } => "dark",
92        Color { r: 242, g: 119, b: 122, .. } => "red",
93        Color { r: 116, g: 115, b: 105, .. } => "gray",
94        Color { r: 204, g: 153, b: 204, .. } => "violet",
95        Color { r: 102, g: 153, b: 204, .. } => "aqua",
96        Color { r: 249, g: 145, b: 87, .. } => "gold",
97        Color { r: 153, g: 204, b: 153, .. } => "green",
98        Color { r: 102, g: 204, b: 204, .. } => "emerald",
99        Color { r: 210, g: 123, b: 83, .. } => "pink",
100        Color { r: 255, g: 204, b: 102, .. } => "grassgreen",
101        Color { r, g, b, .. } => if cfg!(test) {
102            panic!("Uninitialized Color: (r: {r} g: {g} b: {b})")
103        } else {
104            "white"
105        }
106    };
107
108    let content_v32 = into_v32(content);
109
110    // though we've included `\n` characters for the sake of proper syntax highlighting,
111    // the characters have to be erased before it's rendered to html.
112    // otherwise the result html will have redundant `\n`s.
113
114    // it doesn't touch empty pieces
115    if content_v32.is_empty() {
116        vec![]
117    }
118
119    // it doesn't touch empty pieces
120    else if content_v32.iter().all(|c| *c == ' ' as u32 || *c == '\n' as u32) {
121        content_v32.into_iter().filter(|c| *c != '\n' as u32).collect()
122    }
123
124    else {
125        vec![
126            into_v32(&format!("<span class=\"{class_prefix}color-{color}\">")),
127            escape_htmls(&content_v32).into_iter().filter(|c| *c != '\n' as u32).collect(),
128            vec![60, 47, 115, 112, 97, 110, 62],  // into_v32("</span>")
129        ].concat()
130    }
131
132}
mdxt/container/codefence/syntect.rs

mdxt/container/codefence/
syntect.rs
