Use bat's syntect syntax set

Cargo.toml

@@ -8,6 +8,7 @@ version = "0.2.1"
 edition = "2021"
 
 [dependencies]
+bincode = "1.3"
 clap = { version = "4.4", features = ["derive", "string"] }
 comrak = { version = "0.19", default-features = false }
 crossterm = { version = "0.27", features = ["serde"] }

flake.nix

@@ -26,6 +26,7 @@
           ".cargo"
           "src"
           "themes"
+          "syntaxes"
         ];
 
         buildSrc = flakeboxLib.filterSubPaths {

src/markdown/elements.rs

@@ -160,39 +160,55 @@ pub(crate) struct Code {
 /// The language of a piece of code.
 #[derive(Clone, Debug, PartialEq, Eq, EnumIter)]
 pub(crate) enum CodeLanguage {
+    Ada,
     Asp,
+    Awk,
     Bash,
     BatchFile,
     C,
+    CMake,
+    Crontab,
     CSharp,
     Clojure,
     Cpp,
     Css,
     DLang,
+    Docker,
+    Dotenv,
+    Elixir,
+    Elm,
     Erlang,
     Go,
     Haskell,
     Html,
     Java,
     JavaScript,
     Json,
+    Kotlin,
     Latex,
     Lua,
     Makefile,
     Markdown,
     OCaml,
     Perl,
     Php,
+    Protobuf,
+    Puppet,
     Python,
     R,
     Rust,
     Scala,
     Shell(String),
     Sql,
+    Swift,
+    Svelte,
+    Terraform,
     TypeScript,
     Unknown,
     Xml,
     Yaml,
+    Vue,
+    Zig,
 }
 
 impl CodeLanguage {

src/markdown/parse.rs

@@ -127,37 +127,53 @@ impl<'a> MarkdownParser<'a> {
         let info = block.info.as_str();
         let mut tokens = info.split(' ');
         let language = match tokens.next().unwrap_or("") {
+            "ada" => Ada,
             "asp" => Asp,
+            "awk" => Awk,
             "c" => C,
+            "cmake" => CMake,
+            "crontab" => Crontab,
             "csharp" => CSharp,
             "clojure" => Clojure,
             "cpp" | "c++" => Cpp,
             "css" => Css,
             "d" => DLang,
+            "docker" => Docker,
+            "dotenv" => Dotenv,
+            "elixir" => Elixir,
+            "elm" => Elm,
             "erlang" => Erlang,
             "go" => Go,
             "haskell" => Haskell,
             "html" => Html,
             "java" => Java,
             "javascript" | "js" => JavaScript,
             "json" => Json,
+            "kotlin" => Kotlin,
             "latex" => Latex,
             "lua" => Lua,
             "make" => Makefile,
             "markdown" => Markdown,
             "ocaml" => OCaml,
             "perl" => Perl,
             "php" => Php,
+            "protobuf" => Protobuf,
+            "puppet" => Puppet,
             "python" => Python,
             "r" => R,
             "rust" => Rust,
             "scala" => Scala,
             "shell" => Shell("sh".into()),
             interpreter @ ("bash" | "sh" | "zsh" | "fish") => Shell(interpreter.into()),
             "sql" => Sql,
+            "svelte" => Svelte,
+            "swift" => Swift,
+            "terraform" => Terraform,
             "typescript" | "ts" => TypeScript,
             "xml" => Xml,
             "yaml" => Yaml,
+            "vue" => Vue,
+            "zig" => Zig,
             _ => Unknown,
         };
         let flags = CodeFlags { execute: tokens.any(|token| token == "+exec") };

src/render/highlighting.rs

@@ -7,7 +7,10 @@ use syntect::{
     util::{as_24_bit_terminal_escaped, LinesWithEndings},
 };
 
-static SYNTAX_SET: Lazy<SyntaxSet> = Lazy::new(SyntaxSet::load_defaults_newlines);
+static SYNTAX_SET: Lazy<SyntaxSet> = Lazy::new(|| {
+    let contents = include_bytes!("../../syntaxes/syntaxes.bin");
+    bincode::deserialize(contents).expect("syntaxes are broken")
+});
 static THEMES: Lazy<ThemeSet> = Lazy::new(ThemeSet::load_defaults);
 
 /// A code highlighter.
@@ -43,40 +46,56 @@ impl CodeHighlighter {
     fn language_extension(language: &CodeLanguage) -> &'static str {
         use CodeLanguage::*;
         match language {
+            Ada => "adb",
             Asp => "asa",
+            Awk => "awk",
             Bash => "bash",
             BatchFile => "bat",
             C => "c",
+            CMake => "cmake",
             CSharp => "cs",
             Clojure => "clj",
             Cpp => "cpp",
+            Crontab => "crontab",
             Css => "css",
             DLang => "d",
+            Docker => "Dockerfile",
+            Dotenv => "env",
+            Elixir => "ex",
+            Elm => "elm",
             Erlang => "erl",
             Go => "go",
             Haskell => "hs",
             Html => "html",
             Java => "java",
             JavaScript => "js",
             Json => "json",
+            Kotlin => "kt",
             Latex => "tex",
             Lua => "lua",
             Makefile => "make",
             Markdown => "md",
             OCaml => "ml",
             Perl => "pl",
             Php => "php",
+            Protobuf => "proto",
+            Puppet => "pp",
             Python => "py",
             R => "r",
             Rust => "rs",
             Scala => "scala",
             Shell(_) => "sh",
             Sql => "sql",
-            TypeScript => "js",
-            Xml => "xml",
-            Yaml => "yaml",
+            Swift => "swift",
+            Svelte => "svelte",
+            Terraform => "tf",
+            TypeScript => "ts",
             // default to plain text so we get the same look&feel
             Unknown => "txt",
+            Vue => "vue",
+            Xml => "xml",
+            Yaml => "yaml",
+            Zig => "zig",
         }
     }
 }

syntaxes/syntaxes.git-hash

@@ -0,0 +1 @@
+3d87b25b190e0990e0e75a2ab8f994d6c277d263

syntaxes/update.sh

@@ -0,0 +1,27 @@
+#!/bin/bash
+
+set -e
+
+if [ $# -ne 1 ]
+then
+  echo "Usage: $0 <bat-git-hash>"
+  exit 1
+fi
+
+script_path=$(realpath "$0")
+script_dir=$(dirname "$script_path")
+git_hash=$1
+clone_path=$(mktemp -d)
+output_file="$script_dir/syntaxes.bin"
+output_tag="$script_dir/syntaxes.git-hash"
+
+echo "Cloning repo @ ${git_hash} into '$clone_path'"
+git clone https://github.com/sharkdp/bat.git $clone_path
+cd $clone_path
+git reset --hard $git_hash
+
+cp assets/syntaxes.bin $output_file
+echo $git_hash > $output_tag
+
+echo "syntaxes file copied to '$output_file'"
+