Fix non-identchars appearing in backend idents

compiler/catala_utils/string.ml

@@ -17,6 +17,13 @@
 include Stdlib.String
 
 let to_ascii : string -> string = Ubase.from_utf8
+
+let to_id s =
+  to_ascii s
+  |> map (function
+       | ('a' .. 'z' | 'A' .. 'Z' | '0' .. '9') as c -> c
+       | _ -> '_')
+
 let is_uppercase_ascii = function 'A' .. 'Z' -> true | _ -> false
 
 let begins_with_uppercase (s : string) : bool =
@@ -29,7 +36,7 @@ let begins_with_uppercase (s : string) : bool =
 let to_snake_case (s : string) : string =
   let out = Buffer.create (2 * length s) in
   s
-  |> to_ascii
+  |> to_id
   |> iteri (fun i c ->
          if is_uppercase_ascii c && 0 <> i && get s (i - 1) <> '_' then
            Buffer.add_char out '_';
@@ -40,7 +47,7 @@ let to_camel_case (s : string) : string =
   let last_was_underscore = ref true in
   let out = Buffer.create (length s) in
   s
-  |> to_ascii
+  |> to_id
   |> iter (function
        | '_' -> last_was_underscore := true
        | c ->

compiler/catala_utils/string.mli

@@ -29,6 +29,10 @@ val to_ascii : string -> string
 (** Removes all non-ASCII diacritics from a string by converting them to their
     base letter in the Latin alphabet. *)
 
+val to_id : string -> string
+(** Like [to_ascii], but in addition replaces any non-alphanumeric character by
+    [_] *)
+
 val is_uppercase_ascii : char -> bool
 (** [is_uppercase c] returns if [c] is in the set ['A'...'Z']. *)
 

compiler/plugins/api_web.ml

@@ -37,7 +37,7 @@ module To_jsoo = struct
      other modules: here everything is flattened in the current namespace *)
   let format_struct_name ppf name =
     StructName.to_string name
-    |> String.to_ascii
+    |> String.to_id
     |> String.uncapitalize_ascii
     |> String.map (function '.' -> '_' | c -> c)
     |> Format.pp_print_string ppf

compiler/scalc/to_c.ml

@@ -81,8 +81,8 @@ let renaming =
     let id = f id |> Re.replace_string module_sep_re ~by:"_" in
     String.concat "__" (pfx @ [id])
   in
-  let cap s = String.to_ascii s |> String.capitalize_ascii in
-  let uncap s = String.to_ascii s |> String.uncapitalize_ascii in
+  let cap s = String.to_id s |> String.capitalize_ascii in
+  let uncap s = String.to_id s |> String.uncapitalize_ascii in
   let upper s = String.to_snake_case s |> String.uppercase_ascii in
   Renaming.program ()
     ~reserved:c_keywords

compiler/shared_ast/renaming.ml

@@ -471,8 +471,8 @@ let process_type_ident
       ctx_enums = EnumName.Map.add new_name ctx_constrs tctx.ctx_enums;
     }
 
-let cap s = String.to_ascii s |> String.capitalize_ascii
-let uncap s = String.to_ascii s |> String.uncapitalize_ascii
+let cap s = String.to_id s |> String.capitalize_ascii
+let uncap s = String.to_id s |> String.uncapitalize_ascii
 
 (* Todo? - handle separate namespaces ? (e.g. allow a field and var to have the
    same name for backends that support it) - register module names as reserved