1 files changed, 43 insertions, 0 deletions
diff --git a/src/grammar_manifest.zig b/src/grammar_manifest.zig
index 0303af17..d6666729 100644
--- a/src/grammar_manifest.zig
+++ b/src/grammar_manifest.zig
@@ -2,6 +2,8 @@
 //! runtime syntax registry. Keep target symbols and query values out of this
 //! file: the build runner imports it as host code.
 
+const std = @import("std");
+
 pub const Tier = enum { zig, minimal, full };
 
 pub const Grammar = struct {
@@ -49,3 +51,44 @@ pub const all = [_]Grammar{
     .{ .name = "typst", .dep = "ts_typst", .exts = &.{ ".typ", ".typst" }, .tier = .full, .scanner = true, .query = "queries/typst/highlights.scm" },
     .{ .name = "zig", .dep = "ts_zig", .exts = &.{ ".zig", ".zon" }, .tier = .zig },
 };
+
+/// The names a language goes by besides its grammar's: a fence's `py`,
+/// a `Repl sh`.
+pub const aliases = [_]struct { []const u8, []const u8 }{
+    .{ "js", "javascript" },
+    .{ "jsx", "javascript" },
+    .{ "mjs", "javascript" },
+    .{ "py", "python" },
+    .{ "sh", "bash" },
+    .{ "shell", "bash" },
+    .{ "zsh", "bash" },
+    .{ "bash", "bash" },
+    .{ "rs", "rust" },
+    .{ "c++", "cpp" },
+    .{ "cxx", "cpp" },
+    .{ "cc", "cpp" },
+    .{ "cs", "c_sharp" },
+    .{ "kt", "kotlin" },
+    .{ "rb", "ruby" },
+    .{ "ml", "ocaml" },
+    .{ "hs", "haskell" },
+};
+
+/// The grammar a language name means: its own name or an alias, in any case.
+pub fn byName(name: []const u8) ?usize {
+    var canonical = name;
+    for (aliases) |a| if (std.ascii.eqlIgnoreCase(name, a[0])) {
+        canonical = a[1];
+        break;
+    };
+    for (all, 0..) |g, i| if (std.ascii.eqlIgnoreCase(canonical, g.name)) return i;
+    return null;
+}
+
+/// The grammar a path's extension selects, in any case.
+pub fn byPath(path: []const u8) ?usize {
+    const ext = std.fs.path.extension(path);
+    if (ext.len == 0) return null;
+    for (all, 0..) |g, i| for (g.exts) |e| if (std.ascii.eqlIgnoreCase(ext, e)) return i;
+    return null;
+}