syntax highlighting with syntect

5 files changed, 40 insertions, 107 deletions

diff --git a/code/src/html.rs b/code/src/html.rs
index d310ad3..23ac6c8 100644
--- a/code/src/html.rs
+++ b/code/src/html.rs
@@ -1,7 +1,10 @@
 use crate::{
-    article_metadata, file_history, get_articles, markdown::{render::blocks_to_html, self}, ArticleMeta,
+    article_metadata, file_history, get_articles,
+    markdown::{self, render::blocks_to_html},
+    ArticleMeta,
 };
 use laby::{frag, html, iter, li, ul, Render};
+use pest::Parser;
 use std::fs::read_to_string;
 
 pub fn scaffold(title: String, body: impl Render) -> impl Render {
@@ -25,6 +28,14 @@ pub fn scaffold(title: String, body: impl Render) -> impl Render {
 }
 
 pub fn article(path: String) -> impl Render {
+    // match parsers::markdown::Grammar::parse(
+    //     parsers::markdown::Rule::file,
+    //     &read_to_string(&path).unwrap(),
+    // ) {
+    //     Ok(ast) => eprintln!("{ast:#?}"),
+    //     Err(e) => panic!("{e}"),
+    // }
+
     scaffold(
         article_metadata(path.clone().into()).title,
         frag!(
diff --git a/code/src/markdown/render.rs b/code/src/markdown/render.rs
index f6b7686..7962561 100644
--- a/code/src/markdown/render.rs
+++ b/code/src/markdown/render.rs
@@ -34,11 +34,11 @@ pub fn blocks_to_html(blocks: Vec<Block>) -> String {
             Block::Paragraph(p) => format!("<p>{}</p>", span_to_html(p)),
             Block::Blockquote(q) => format!("<quote>{}</quote>", blocks_to_html(q)),
             Block::CodeBlock(syntax, content) => {
-                if let Some(s) = &syntax {
-                    format!("<pre>{}</pre>", syntax_highlight(s, &content))
-                } else {
-                    format!("<pre>{}</pre>", escape(&content))
-                }
+                format!(
+                    "<pre>{}</pre>",
+                    syntax_highlight(&syntax.unwrap_or(String::from("")), &content)
+                        .unwrap_or_else(|| escape(&content))
+                )
             }
             Block::OrderedList(els) => format!(
                 "<ol>{}</ol>",
diff --git a/code/src/syntax_highlight/grammar.rs b/code/src/syntax_highlight/grammar.rs
deleted file mode 100644
index 0cecac9..0000000
--- a/code/src/syntax_highlight/grammar.rs
+++ /dev/null
@@ -1,67 +0,0 @@
-const CONST: &str = "[\\. \\(,=^]([A-Z][A-Z0-9_]*)[\\. ,\\)=$]";
-const TYPE: &str = "[\\. \\(,=^]([A-Z][a-zA-Z0-9_]*)[\\. ,\\)\\(=$]";
-const IDENT: &str = "[\\. \\(,=^]([a-z][a-zA-Z0-9_]*)[\\. ,\\)=$]";
-const FUNC: &str = "[\\. \\(,=^]([a-z][A-Z0-9_]*)\\(";
-
-pub fn grammar_for(syntax: &str) -> &'static [(&'static str, &'static [&'static str])] {
-    match syntax {
-        "rs" | "rust" => &[
-            (
-                "keyword",
-                &[
-                    "fn", "pub", "async", "return", "if", "else", "let", "for", "in", "while",
-                    "loop", "impl", "for", "trait", "struct", "enum", "dyn",
-                ],
-            ),
-            (
-                "type",
-                &[
-                    TYPE, "bool", "usize", "u8", "u16", "u32", "u64", "u128", "i8", "i16", "i32",
-                    "i64", "i128", "isize", "f32", "f64",
-                ],
-            ),
-            ("constant", &[CONST]),
-            ("comment", &["(?m)(//.*)$", "(?ms)/\\*.*?\\*/"]),
-            ("macro", &["[\\. ,=^]([a-z_][A-Za-z0-9_]*!)[\\. ,=$]"]),
-            ("identifier", &[IDENT]),
-            ("function", &[FUNC]),
-            (
-                "literal",
-                &["\".*?\"", "(0x|0o|0b)[0-9a-f]+", "\\d+", "true", "false"],
-            ),
-        ],
-        "py" | "python" => &[
-            ("type", &["bytes", "bool", "int", "str", "float"]),
-            (
-                "keyword",
-                &[
-                    "from", "for", "def", "if", "else", "elif", "while", "with", "in", "assert",
-                    "global", "nonlocal", "as", "import",
-                ],
-            ),
-            ("comment", &["(?m)(#.*)$"]),
-            ("constant", &[CONST]),
-            ("identifier", &[IDENT]),
-            (
-                "literal",
-                &[
-                    "b?f?\".*?\"",
-                    "b?f?'.*?'",
-                    "(0x|0o|0b)[0-9a-f]+",
-                    "\\d+",
-                    "true",
-                    "false",
-                ],
-            ),
-        ],
-        "tree" => &[("keyword", &["[├─└│]+"])],
-        // makefile doesnt really match the token-kinds, i'll just use something that looks goo
-        "makefile" | "mk" => &[
-            ("comment", &["(?m)(#.*)$"]),
-            ("literal", &[".+: "]),
-            ("macro", &["\\$\\(\\w+\\)"]),
-            ("type", &["\\$@", "\\$<"]),
-        ],
-        _ => &[],
-    }
-}
diff --git a/code/src/syntax_highlight/mod.rs b/code/src/syntax_highlight/mod.rs
index 62a2536..86b689a 100644
--- a/code/src/syntax_highlight/mod.rs
+++ b/code/src/syntax_highlight/mod.rs
@@ -1,27 +1,29 @@
-pub mod grammar;
-pub mod theme;
 
-use crate::{markdown::render::escape, syntax_highlight::theme::theme};
-use grammar::grammar_for;
-use synoptic::{Highlighter, Token};
+use crate::markdown::render::escape;
+use lazy_static::lazy_static;
+use syntect::easy::HighlightLines;
+use syntect::highlighting::{Style, ThemeSet};
+use syntect::parsing::SyntaxSet;
+use syntect::util::{as_24_bit_terminal_escaped, LinesWithEndings};
 
-pub fn syntax_highlight(lang: &str, source: &str) -> String {
-    let mut h = Highlighter::new();
-    for (kind, regex) in grammar_for(lang) {
-        h.join(regex, kind).unwrap();
-    }
-    let highlighting = h.run(source);
-    let mut out = String::new();
+lazy_static! {
+    static ref PS: SyntaxSet = SyntaxSet::load_defaults_newlines();
+    static ref TS: ThemeSet = ThemeSet::load_defaults();
+}
+
+pub fn syntax_highlight(lang: &str, source: &str) -> Option<String> {
+    let syntax = PS.find_syntax_by_extension(lang)?;
+    let mut h = HighlightLines::new(syntax, &TS.themes["Solarized (dark)"]);
 
-    for (_c, row) in highlighting.iter().enumerate() {
-        for tok in row {
-            match tok {
-                Token::Start(kind) => out += &format!("<span style=\"color:{}\">", theme(kind)),
-                Token::Text(text) => out += &escape(text),
-                Token::End(_kind) => out += "</span>",
-            }
+    let mut o = String::new();
+    for line in LinesWithEndings::from(source) {
+        let ranges: Vec<(Style, &str)> = h.highlight_line(line, &PS).unwrap();
+        for (style, span) in ranges {
+            o += &format!(
+                "<span style=\"color: #{:02x}{:02x}{:02x}\">{}</span>",
+                style.foreground.r, style.foreground.g, style.foreground.b, span
+            );
         }
-        out += "\n"
     }
-    out
+    return Some(o);
 }
diff --git a/code/src/syntax_highlight/theme.rs b/code/src/syntax_highlight/theme.rs
deleted file mode 100644
index 2dfd10d..0000000
--- a/code/src/syntax_highlight/theme.rs
+++ /dev/null
@@ -1,13 +0,0 @@
-pub fn theme(kind: &str) -> &'static str {
-    match kind {
-        "keyword" => "#9999ff",
-        "macro" => "#ff2863",
-        "literal" => "#26a6ff",
-        "function" => "#26ffbc",
-        "types" => "#26ff34",
-        "identifier" => "#ccff26",
-        "constant" => "#ff7c26",
-        "comment" => "#6e6e6e",
-        _ => "#ff00ff",
-    }
-}