//! Tree-sitter syntax highlighting: one style byte per content byte, filled by //! running each grammar's highlights.scm query (slurped at build time into the //! ts_queries options module). Grammar set is tiered: `zig`; `minimal` (c, cpp, //! zig); and `full`, which adds ~23 languages lazily on first use. const std = @import("std"); const config = @import("pardes_config"); const tracy = @import("tracy.zig"); pub const enabled = config.syntax_highlighting; const zig_grammar = config.syntax_zig_grammar; const minimal_grammars = config.syntax_minimal_grammars; const full_grammars = config.syntax_full_grammars; const ts = if (enabled) @import("tree-sitter") else struct { pub const Language = opaque {}; pub const Query = opaque {}; }; const ts_queries = if (enabled) @import("ts_queries") else struct {}; extern fn tree_sitter_ada() callconv(.c) *const ts.Language; extern fn tree_sitter_bash() callconv(.c) *const ts.Language; extern fn tree_sitter_c() callconv(.c) *const ts.Language; extern fn tree_sitter_c_sharp() callconv(.c) *const ts.Language; extern fn tree_sitter_clojure() callconv(.c) *const ts.Language; extern fn tree_sitter_cpp() callconv(.c) *const ts.Language; extern fn tree_sitter_css() callconv(.c) *const ts.Language; extern fn tree_sitter_elixir() callconv(.c) *const ts.Language; extern fn tree_sitter_erlang() callconv(.c) *const ts.Language; extern fn tree_sitter_fortran() callconv(.c) *const ts.Language; extern fn tree_sitter_go() callconv(.c) *const ts.Language; extern fn tree_sitter_haskell() callconv(.c) *const ts.Language; extern fn tree_sitter_html() callconv(.c) *const ts.Language; extern fn tree_sitter_java() callconv(.c) *const ts.Language; extern fn tree_sitter_javascript() callconv(.c) *const ts.Language; extern fn tree_sitter_json() callconv(.c) *const ts.Language; extern fn tree_sitter_kotlin() callconv(.c) *const ts.Language; extern fn tree_sitter_ocaml() callconv(.c) *const ts.Language; extern fn tree_sitter_pascal() callconv(.c) *const ts.Language; extern fn tree_sitter_php() callconv(.c) *const ts.Language; extern fn tree_sitter_powershell() callconv(.c) *const ts.Language; extern fn tree_sitter_python() callconv(.c) *const ts.Language; extern fn tree_sitter_ruby() callconv(.c) *const ts.Language; extern fn tree_sitter_rust() callconv(.c) *const ts.Language; extern fn tree_sitter_scala() callconv(.c) *const ts.Language; extern fn tree_sitter_zig() callconv(.c) *const ts.Language; pub const Syn = enum(u8) { none, keyword, string, number, comment }; const Spec = struct { exts: []const []const u8, language: *const fn () callconv(.c) *const ts.Language, query_src: []const u8, compiled_query: ?*ts.Query = null, }; var zig_specs = if (zig_grammar) [_]Spec{ .{ .exts = &.{ ".zig", ".zon" }, .language = tree_sitter_zig, .query_src = ts_queries.zig_highlights }, } else [_]Spec{}; var minimal_specs = if (minimal_grammars) [_]Spec{ .{ .exts = &.{ ".c", ".h" }, .language = tree_sitter_c, .query_src = ts_queries.c_highlights }, .{ .exts = &.{ ".cpp", ".cc", ".cxx", ".hpp", ".hh", ".hxx" }, .language = tree_sitter_cpp, .query_src = ts_queries.cpp_highlights }, } else [_]Spec{}; var full_specs = if (full_grammars) [_]Spec{ .{ .exts = &.{ ".adb", ".ads", ".ada" }, .language = tree_sitter_ada, .query_src = ts_queries.ada_highlights }, .{ .exts = &.{ ".sh", ".bash", ".zsh" }, .language = tree_sitter_bash, .query_src = ts_queries.bash_highlights }, .{ .exts = &.{ ".cs", ".csx" }, .language = tree_sitter_c_sharp, .query_src = ts_queries.c_sharp_highlights }, .{ .exts = &.{ ".clj", ".cljs", ".cljc", ".edn" }, .language = tree_sitter_clojure, .query_src = ts_queries.clojure_highlights }, .{ .exts = &.{".css"}, .language = tree_sitter_css, .query_src = ts_queries.css_highlights }, .{ .exts = &.{ ".ex", ".exs" }, .language = tree_sitter_elixir, .query_src = ts_queries.elixir_highlights }, .{ .exts = &.{ ".erl", ".hrl" }, .language = tree_sitter_erlang, .query_src = ts_queries.erlang_highlights }, .{ .exts = &.{ ".f", ".for", ".ftn", ".f90", ".f95", ".f03", ".f08" }, .language = tree_sitter_fortran, .query_src = ts_queries.fortran_highlights }, .{ .exts = &.{".go"}, .language = tree_sitter_go, .query_src = ts_queries.go_highlights }, .{ .exts = &.{ ".hs", ".lhs" }, .language = tree_sitter_haskell, .query_src = ts_queries.haskell_highlights }, .{ .exts = &.{ ".html", ".htm" }, .language = tree_sitter_html, .query_src = ts_queries.html_highlights }, .{ .exts = &.{".java"}, .language = tree_sitter_java, .query_src = ts_queries.java_highlights }, .{ .exts = &.{ ".js", ".jsx", ".mjs", ".cjs" }, .language = tree_sitter_javascript, .query_src = ts_queries.javascript_highlights }, .{ .exts = &.{".json"}, .language = tree_sitter_json, .query_src = ts_queries.json_highlights }, .{ .exts = &.{ ".kt", ".kts" }, .language = tree_sitter_kotlin, .query_src = ts_queries.kotlin_highlights }, .{ .exts = &.{ ".ml", ".mli" }, .language = tree_sitter_ocaml, .query_src = ts_queries.ocaml_highlights }, .{ .exts = &.{ ".pas", ".pp", ".p" }, .language = tree_sitter_pascal, .query_src = ts_queries.pascal_highlights }, .{ .exts = &.{ ".php", ".phtml", ".php3", ".php4", ".php5" }, .language = tree_sitter_php, .query_src = ts_queries.php_highlights }, .{ .exts = &.{ ".ps1", ".psm1", ".psd1" }, .language = tree_sitter_powershell, .query_src = ts_queries.powershell_highlights }, .{ .exts = &.{ ".py", ".pyw" }, .language = tree_sitter_python, .query_src = ts_queries.python_highlights }, .{ .exts = &.{ ".rb", ".rake" }, .language = tree_sitter_ruby, .query_src = ts_queries.ruby_highlights }, .{ .exts = &.{".rs"}, .language = tree_sitter_rust, .query_src = ts_queries.rust_highlights }, .{ .exts = &.{ ".scala", ".sc" }, .language = tree_sitter_scala, .query_src = ts_queries.scala_highlights }, } else [_]Spec{}; const Selected = struct { lang: *const ts.Language, query: *ts.Query }; // NOTE: don't lang.destroy() — the tree_sitter_*() languages are static // singletons reused on every open; destroying one use-after-frees the next. fn ensure(spec: *Spec) !Selected { const lang = spec.language(); if (spec.compiled_query) |query| return .{ .lang = lang, .query = query }; var error_offset: u32 = 0; const query = try ts.Query.create(lang, spec.query_src, &error_offset); spec.compiled_query = query; return .{ .lang = lang, .query = query }; } fn forExt(ext: []const u8) !?Selected { for (&zig_specs) |*spec| { for (spec.exts) |choice| { if (std.ascii.eqlIgnoreCase(ext, choice)) return try ensure(spec); } } for (&minimal_specs) |*spec| { for (spec.exts) |choice| { if (std.ascii.eqlIgnoreCase(ext, choice)) return try ensure(spec); } } for (&full_specs) |*spec| { for (spec.exts) |choice| { if (std.ascii.eqlIgnoreCase(ext, choice)) return try ensure(spec); } } return null; } // Queries compile on first use (`ensure`), never at startup. Pre-compiling the // compact tier in Pardes.init cost EVERY boot ~120ms of ts_query__perform_analysis // (55% of a Debug startup) to save ~40ms on the first .zig/.c/.cpp open — a pane // of prose or a shell paid for a language it never opened. Grammar availability // is unchanged; only the timing moved. fn synFor(name: []const u8) Syn { for ([_]struct { []const u8, Syn }{ .{ "comment", .comment }, .{ "string", .string }, .{ "character", .string }, .{ "number", .number }, .{ "float", .number }, .{ "boolean", .number }, .{ "keyword", .keyword }, .{ "include", .keyword }, .{ "conditional", .keyword }, .{ "repeat", .keyword }, }) |m| { if (std.mem.indexOf(u8, name, m[0]) != null) return m[1]; } return .none; } pub fn highlightFile(gpa: std.mem.Allocator, path: []const u8, content: []const u8) ![]u8 { return highlightFileRange(gpa, path, content, 0, content.len); } /// One Syn byte per content byte in [start, end). Caller frees. pub fn highlightFileRange(gpa: std.mem.Allocator, path: []const u8, content: []const u8, start_byte_raw: usize, end_byte_raw: usize) ![]u8 { const tz = tracy.zone(@src(), "highlightFileRange"); defer tz.end(); if (!enabled) return &.{}; const ext = std.fs.path.extension(path); const selected = (forExt(ext) catch return &.{}) orelse return &.{}; const start_byte = @min(start_byte_raw, content.len); const end_byte = @max(start_byte, @min(end_byte_raw, content.len)); const source = content[start_byte..end_byte]; const styles = try gpa.alloc(u8, source.len); errdefer gpa.free(styles); @memset(styles, 0); const parser = ts.Parser.create(); defer parser.destroy(); parser.setLanguage(selected.lang) catch return styles; const tree = parser.parseString(source, null) orelse return styles; defer tree.destroy(); const cursor = ts.QueryCursor.create(); defer cursor.destroy(); cursor.exec(selected.query, tree.rootNode()); while (cursor.nextMatch()) |match| { for (match.captures) |cap| { const syn = synFor(selected.query.captureNameForId(cap.index) orelse ""); if (syn == .none) continue; var b: usize = cap.node.startByte(); const end = @min(@as(usize, cap.node.endByte()), styles.len); while (b < end) : (b += 1) styles[b] = @intFromEnum(syn); } } return styles; }