Add fixes to parse aliases and externals with Ruby.

Add grammar of TypeScript.

.ocp-indent

@@ -0,0 +1,22 @@
+# See https://github.com/OCamlPro/ocp-indent/blob/master/.ocp-indent for more
+
+# Indent for clauses inside a pattern-match (after the arrow):
+#    match foo with
+#    | _ ->
+#    ^^^^bar
+# the default is 2, which aligns the pattern and the expression
+match_clause = 4
+
+# When nesting expressions on the same line, their indentation are in
+# some cases stacked, so that it remains correct if you close them one
+# at a line. This may lead to large indents in complex code though, so
+# this parameter can be used to set a maximum value. Note that it only
+# affects indentation after function arrows and opening parens at end
+# of line.
+#
+# for example (left: `none`; right: `4`)
+#    let f = g (h (i (fun x ->     #    let f = g (h (i (fun x ->
+#          x)                      #        x)
+#        )                         #      )
+#      )                           #    )
+max_indent = 2

scripts/install-tree-sitter

@@ -10,5 +10,7 @@ project_root=$(git rev-parse --show-toplevel)
 # The tree-sitter executable will be in 'node_modules/.bin'.
 (
   cd "$project_root"
-  npm install tree-sitter tree-sitter-cli
+
+  # tree-sitter-javascript is needed to generate grammar.json for typescript.
+  npm install tree-sitter tree-sitter-cli tree-sitter-javascript
 )

src/lib/AST_grammar.ml

@@ -45,6 +45,8 @@ let rec translate (x : Tree_sitter_t.rule_body) =
   | PREC_LEFT (_opt_prio, x) -> translate x
   | PREC_RIGHT (_opt_prio, x) -> translate x
   | FIELD (_name, x) -> translate x (* TODO not sure about ignoring this *)
+  | ALIAS alias ->
+      translate alias.content (* TODO probably good to not ignore *)
   | IMMEDIATE_TOKEN x -> translate x (* TODO check what this is *)
   | TOKEN x -> translate x
 

src/lib/Json_rule_adapter.ml

@@ -41,6 +41,7 @@ let normalize_variant_object fields =
     | "PREC_DYNAMIC" -> Some (`List [get "value"; get "content"])
     | "PREC_LEFT" -> Some (`List [opt "value"; get "content"])
     | "PREC_RIGHT" -> Some (`List [opt "value"; get "content"])
+    | "ALIAS" -> Some (`Assoc fields)
     | "FIELD" -> Some (`List [get "name"; get "content"])
     | "IMMEDIATE_TOKEN" -> Some (get "content")
     | "TOKEN" -> Some (get "content")

src/lib/Tree_sitter.atd

@@ -12,7 +12,7 @@ type grammar = {
   ~extras: rule_body list; (* tokens (can only be STRING or PATTERN) *)
   ~inline: ident list; (* rule names *)
   ~conflicts: ident list list; (* list of lists of rule names *)
-  ~externals: ident list; (* list of token names *)
+  ~externals: rule_body list; (* list of token names? *)
   ?word: ident option; (* token name *)
   ~supertypes: ident list; (* rule names *)
 }
@@ -30,8 +30,15 @@ type rule_body = [
   | PREC_DYNAMIC of (int * rule_body)
   | PREC_LEFT of (int nullable * rule_body)
   | PREC_RIGHT of (int nullable * rule_body)
+  | ALIAS of alias
   | FIELD of (ident * rule_body)
   | IMMEDIATE_TOKEN of rule_body (* presumably must contain a token *)
   | TOKEN of rule_body
 ] <json adapter.ocaml="Json_rule_adapter"> (* see Json_rule_adapter.ml *)
   <ocaml repr="classic">
+
+type alias = {
+  value: ident;
+  named: bool;
+  content: rule_body;
+}

tests/typescript/README.md

@@ -0,0 +1,5 @@
+The `grammar.js` for typescript was assembled from two files from the 
+tree-sitter-typescript repo. It's special because another dialect called 'tsx' 
+is also supported.
+
+Additionally, we have to install tree-sitter-javascript.