1 (** Registry of bundled TextMate grammars for syntax highlighting. 2 3 Loads grammars embedded at build time via [ocaml-crunch] and registers them 4 with a shared {!TmLanguage.t} instance. Language lookup is by the canonical 5 name string produced by {!Detect.detect}. 6 7 The {!languages} table is the single source of truth for which grammars are 8 shipped and what scope each language maps to. Both the registry loader and 9 {!scope_of_lang} derive from it. *) 10 11 (** Each entry is [(grammar_file, scope_name, aliases)] where [aliases] includes 12 the canonical name and any accepted variants. *) 13 let languages = 14 [ 15 ("c", "source.c", [ "c" ]); 16 ("clojure", "source.clojure", [ "clojure" ]); 17 ("cpp", "source.cpp", [ "cpp" ]); 18 ("csharp", "source.cs", [ "csharp" ]); 19 ("css", "source.css", [ "css" ]); 20 ("dart", "source.dart", [ "dart" ]); 21 ("diff", "source.diff", [ "diff" ]); 22 ("dockerfile", "source.dockerfile", [ "dockerfile" ]); 23 ("dune", "source.dune", [ "dune" ]); 24 ("elixir", "source.elixir", [ "elixir" ]); 25 ("erlang", "source.erlang", [ "erlang" ]); 26 ("fortran", "source.fortran.free", [ "fortran" ]); 27 ("go", "source.go", [ "go" ]); 28 ("groovy", "source.groovy", [ "groovy" ]); 29 ("haskell", "source.haskell", [ "haskell" ]); 30 ("html", "text.html.basic", [ "html" ]); 31 ("java", "source.java", [ "java" ]); 32 ("javascript", "source.js", [ "javascript" ]); 33 ("json", "source.json", [ "json" ]); 34 ("julia", "source.julia", [ "julia" ]); 35 ("kotlin", "source.kotlin", [ "kotlin" ]); 36 ("lisp", "source.lisp", [ "lisp" ]); 37 ("lua", "source.lua", [ "lua" ]); 38 ("makefile", "source.makefile", [ "makefile" ]); 39 ("markdown", "text.html.markdown", [ "markdown" ]); 40 ("nix", "source.nix", [ "nix" ]); 41 ("objective-c", "source.objc", [ "objective-c"; "objc" ]); 42 ("ocaml", "source.ocaml", [ "ocaml" ]); 43 ("ocamldoc", "source.ocaml.ocamldoc", [ "ocamldoc"; "mld" ]); 44 ("opam", "source.ocaml.opam", [ "opam" ]); 45 ("org", "source.org", [ "org" ]); 46 ("perl", "source.perl", [ "perl" ]); 47 ("php", "source.php", [ "php" ]); 48 ("powershell", "source.powershell", [ "powershell" ]); 49 ("protobuf", "source.proto", [ "protobuf"; "proto" ]); 50 ("python", "source.python", [ "python" ]); 51 ("r", "source.r", [ "r" ]); 52 ("ruby", "source.ruby", [ "ruby" ]); 53 ("rust", "source.rust", [ "rust" ]); 54 ("scala", "source.scala", [ "scala" ]); 55 ("shell", "source.shell", [ "bash"; "shell"; "sh" ]); 56 ("sql", "source.sql", [ "sql" ]); 57 ("swift", "source.swift", [ "swift" ]); 58 ("terraform", "source.hcl.terraform", [ "terraform"; "hcl" ]); 59 ("toml", "source.toml", [ "toml"; "ini" ]); 60 ("typescript", "source.ts", [ "typescript" ]); 61 ("vue", "text.html.vue", [ "vue" ]); 62 ("xml", "text.html.basic", [ "xml" ]); 63 ("yaml", "source.yaml", [ "yaml" ]); 64 ("zig", "source.zig", [ "zig" ]); 65 ] 66 67 (** The shared grammar registry, lazily initialised. *) 68 let registry : TmLanguage.t Lazy.t = 69 lazy 70 (let t = TmLanguage.create () in 71 let load filename = 72 match Grammar_data.read filename with 73 | None -> () 74 | Some data -> ( 75 try 76 let json = Yojson.Basic.from_string data in 77 let grammar = TmLanguage.of_yojson_exn json in 78 TmLanguage.add_grammar t grammar 79 with _ -> ()) 80 in 81 List.iter (fun (name, _, _) -> load (name ^ ".json")) languages; 82 t) 83 84 (** Scope lookup table, built once from {!languages}. *) 85 let scope_table : (string, string) Hashtbl.t = 86 let tbl = Hashtbl.create 128 in 87 List.iter 88 (fun (_, scope, aliases) -> 89 List.iter (fun alias -> Hashtbl.replace tbl alias scope) aliases) 90 languages; 91 tbl 92 93 (** Map from the canonical language name (as returned by {!Detect.detect}) to 94 the scopeName used by the grammar file. *) 95 let scope_of_lang name = Hashtbl.find_opt scope_table name 96