Add more languages and experimental block keywords

lib/matchers/languages.ml

@@ -1,9 +1,27 @@
+let experimental = true
+
+module Text = struct
+  module Syntax = struct
+    let user_defined_delimiters = []
+    let escapable_string_literals = []
+
+    let escape_char =
+      '\\'
+
+    let raw_string_literals = []
+
+    let comment_parser = []
+  end
+
+  include Matcher.Make(Syntax)
+end
+
 module Dyck = struct
   module Syntax = struct
     let user_defined_delimiters =
-      [ ("(", ")")
-      ; ("{", "}")
-      ; ("[", "]")
+      [ "(", ")"
+      ; "{", "}"
+      ; "[", "]"
       ]
 
     let escapable_string_literals = []
@@ -20,6 +38,8 @@ module Dyck = struct
   include Matcher.Make(Syntax)
 end
 
+module Json = Dyck
+
 module Latex = struct
   module Syntax = struct
     open Types
@@ -27,7 +47,7 @@ module Latex = struct
 
     let user_defined_delimiters =
       Dyck.Syntax.user_defined_delimiters @
-      [ ({|\if|}, {|\fi|})
+      [ {|\if|}, {|\fi|}
       ]
 
     let comment_parser =
@@ -105,8 +125,8 @@ module Bash = struct
 
     let user_defined_delimiters =
       Dyck.Syntax.user_defined_delimiters @
-      [ ("if", "fi")
-      ; ("case", "esac")
+      [ "if ", "fi"
+      ; "case ", "esac"
       ]
 
     let comment_parser =
@@ -121,6 +141,24 @@ module Ruby = struct
     open Types
     include Generic.Syntax
 
+    let user_defined_delimiters =
+      Generic.Syntax.user_defined_delimiters
+      @ if experimental then
+        [ "class", "end"
+        ; "def", "end"
+        ; "do", "end"
+        ; "if", "end"
+        ; "case", "end"
+        ; "unless", "end"
+        ; "while", "end"
+        ; "until", "end"
+        ; "for", "end"
+        ; "begin", "end"
+        ; "module", "end"
+        ]
+      else
+        []
+
     let raw_string_literals =
       [ ({|"|}, {|"|})
       ]
@@ -139,6 +177,19 @@ module Elixir = struct
     open Types
     include Generic.Syntax
 
+    let user_defined_delimiters =
+      Generic.Syntax.user_defined_delimiters
+      @ if experimental then
+        [ "fn", "end"
+        ; "do", "end"
+        ; "case", "end"
+        ; "cond", "end"
+        ; "if", "end"
+        ; "<", ">"
+        ]
+      else
+        []
+
     let raw_string_literals =
       [ ({|"""|}, {|"""|})
       ]
@@ -202,6 +253,16 @@ module Erlang = struct
     open Types
     include Generic.Syntax
 
+    let user_defined_delimiters =
+      Generic.Syntax.user_defined_delimiters
+      @ if experimental then
+        [ "fun", "end"
+        ; "case", "end"
+        ; "if", "end"
+        ]
+      else
+        []
+
     let comment_parser =
       [ Until_newline "%"
       ]
@@ -224,6 +285,8 @@ module C = struct
   include Matcher.Make(Syntax)
 end
 
+module Csharp = C
+
 module Java = C
 
 module CSS = C
@@ -310,6 +373,17 @@ module OCaml = struct
     open Types
     include Generic.Syntax
 
+    let user_defined_delimiters =
+      Generic.Syntax.user_defined_delimiters
+      @ if experimental then
+        [ "begin", "end"
+        ; "struct", "end"
+        ; "sig", "end"
+        ]
+      else
+        []
+
+
     (* Override ' as escapable string literal, since
        these can be used in typing *)
     let escapable_string_literals =
@@ -328,6 +402,8 @@ module OCaml = struct
   include Matcher.Make(Syntax)
 end
 
+module Fsharp = OCaml
+
 (** Follow Free Pascal that allows nested comments, although Rosetta takes the opposite view. *)
 module Pascal = struct
   module Syntax = struct
@@ -409,17 +485,24 @@ let select_with_extension extension : (module Types.Matcher.S) =
   match extension with
   | ".c" | ".h" | ".cc" | ".cpp" | ".hpp" -> (module C)
   | ".clj" -> (module Clojure)
+  | ".cs" -> (module Csharp)
   | ".css" -> (module CSS)
   | ".dart" -> (module Dart)
   | ".elm" -> (module Elm)
   | ".erl" -> (module Erlang)
   | ".ex" -> (module Elixir)
+  | ".f" | ".for" | ".f90"
+  | ".f95" | ".f03" | ".f08" | ".F" | ".F90" -> (module Fortran)
+  | ".fsx" -> (module Fsharp)
   | ".html" | ".xml" -> (module Html)
   | ".hs" -> (module Haskell)
   | ".go" -> (module Go)
   | ".java" -> (module Java)
+  | ".jl" -> (module Julia)
   | ".js" | ".ts" -> (module Javascript)
+  | ".json" -> (module Json)
   | ".ml" | ".mli" -> (module OCaml)
+  | ".pas" -> (module Pascal)
   | ".php" -> (module Php)
   | ".py" -> (module Python)
   | ".rb" -> (module Ruby)
@@ -430,4 +513,5 @@ let select_with_extension extension : (module Types.Matcher.S) =
   | ".sh" -> (module Bash)
   | ".swift" -> (module Swift)
   | ".tex" | ".bib" -> (module Latex)
+  | ".txt" -> (module Text)
   | _ -> (module Generic)

lib/matchers/matcher.ml

@@ -8,6 +8,7 @@ open Location
 open Types
 
 let configuration_ref = ref (Configuration.create ())
+let weaken_delimiter_hole_matching = false
 
 let debug = false
 
@@ -230,18 +231,32 @@ module Make (Syntax : Syntax.S) = struct
       |>> fun result -> (String.concat @@ [from] @ result @ [until])
     in
     let between_nested_delims p =
-      (match left_delimiter, right_delimiter with
-       | Some left_delimiter, Some right_delimiter -> [ (left_delimiter, right_delimiter) ]
-       | _ -> Syntax.user_defined_delimiters)
+      let trigger_nested_parsing_prefix =
+        if weaken_delimiter_hole_matching then
+          match left_delimiter, right_delimiter with
+          | Some left_delimiter, Some right_delimiter ->
+            [ (left_delimiter, right_delimiter) ]
+          | _ -> Syntax.user_defined_delimiters
+        else
+          Syntax.user_defined_delimiters
+      in
+      trigger_nested_parsing_prefix
       |> List.map ~f:fst
       |> List.map ~f:(between_nested_delims p)
       |> choice
     in
     (* applies looser delimiter constraints for matching *)
     let reserved =
-      (match left_delimiter, right_delimiter with
-       | Some left_delimiter, Some right_delimiter -> [ (left_delimiter, right_delimiter) ]
-       | _ -> Syntax.user_defined_delimiters)
+      let trigger_nested_parsing_prefix =
+        if weaken_delimiter_hole_matching then
+          match left_delimiter, right_delimiter with
+          | Some left_delimiter, Some right_delimiter ->
+            [ (left_delimiter, right_delimiter) ]
+          | _ -> Syntax.user_defined_delimiters
+        else
+          Syntax.user_defined_delimiters
+      in
+      trigger_nested_parsing_prefix
       |> List.concat_map ~f:(fun (from, until) -> [from; until])
       |> List.map ~f:string
       |> choice

lib/parsers/delimiters.ml

@@ -1,18 +1,4 @@
 open MParser
 
-(** Significant, potentially nested delimiters. *)
-
-let parens' p =
-  char '(' >> p << char ')'
-
-let braces' p =
-  char '{' >> p << char '}'
-
-let brackets' p =
-  char '<' >> p << char '>'
-
-let squares' p =
-  char '[' >> p << char ']'
-
 let between p from until =
   string from >> p << string until

test/test_bash.ml

@@ -45,9 +45,7 @@ let%expect_test "custom_long_delimiters" =
   let rewrite_template = {|case nuked blocks esac|} in
 
   run_bash source match_template rewrite_template;
-  [%expect_exact {|
-      case nuked blocks esac
-    |}]
+  [%expect_exact {|case nuked blocks esac|}]
 
 let%expect_test "custom_long_delimiters_doesn't_work_in_go" =
   let source =

test/test_special_matcher_cases.ml

@@ -38,3 +38,100 @@ let%expect_test "parse_ocaml_apostrophe_ok" =
 
   run (module Matchers.OCaml) source match_template rewrite_template;
   [%expect_exact {|'a, int |}]
+
+let%expect_test "strict_nested_matching" =
+  let source = {|({})|} in
+  let match_template = {|(:[1])|} in
+  let rewrite_template = {|:[1]|} in
+
+  run (module Matchers.Dyck) source match_template rewrite_template;
+  [%expect_exact {|{}|}]
+
+let%expect_test "strict_nested_matching" =
+  let source = {|(})|} in
+  let match_template = {|(:[1])|} in
+  let rewrite_template = {|:[1]|} in
+
+  run (module Matchers.Dyck) source match_template rewrite_template;
+  [%expect_exact {|No matches.|}]
+
+
+
+(*
+let%expect_test "ocaml_blocks" =
+  let source = {|
+    module M : sig
+        type t
+    end = struct
+       type t = int
+
+       module Nested_M = struct
+         type r = int
+       end
+    end
+|}
+  in
+  let match_template = {|struct :[1] end|} in
+  let rewrite_template = {|struct <bye> end|} in
+
+  run (module Matchers.OCaml) source match_template rewrite_template;
+  [%expect_exact {|No matches.|}]
+*)
+
+let%expect_test "ocaml_complex_blocks_with_same_end" =
+  let source = {|
+    begin
+    match x with
+    | _ ->
+        let module M = struct type t end
+        begin
+        begin
+        match y with
+        | _ -> ()
+        end
+        end
+    end
+|}
+  in
+  let match_template = {|begin :[1] end|} in
+  let rewrite_template = {|-Body->:[1]<-Body-|} in
+
+  run (module Matchers.OCaml) source match_template rewrite_template;
+  [%expect_exact {|
+    -Body->match x with
+    | _ ->
+        let module M = struct type t end
+        begin
+        begin
+        match y with
+        | _ -> ()
+        end
+        end<-Body-
+|}]
+
+(* FIXME(#35): "before" triggers "for" block *)
+let%expect_test "ruby_blocks" =
+  let source = {|
+class ActionController::Base
+  before_filter :generate_css_from_less
+
+  def generate_css_from_less
+    Less::More.generate_all
+  end
+end
+|}
+  in
+  let match_template = {|class :[1] end|} in
+  let rewrite_template = {|-Block->:[1]<-Block-|} in
+
+  run (module Matchers.Ruby) source match_template rewrite_template;
+  [%expect_exact {|No matches.|}]
+
+
+let%expect_test "erlang_blocks" =
+  let source = {|Big =  fun(X) -> if X > 10 -> true; true -> false end end.|} in
+  let match_template = {|fun(:[1]) :[rest] end|} in
+  let rewrite_template = {|-Block->:[rest]<-Block-|} in
+
+  run (module Matchers.Erlang) source match_template rewrite_template;
+  [%expect_exact {|Big =  -Block->-> if X > 10 -> true; true -> false end<-Block-.|}]