Set the console to use UTF-8 on Windows

master_changes.md

@@ -17,6 +17,7 @@ users)
 
 ## Global CLI
   * Fix a typo in the variable description returned by "opam var" [#5961 @jmid]
+  * Out-of-the-box UTF-8 paged --help on Windows [#5970 @kit-ty-kate]
 
 ## Plugins
 
@@ -115,6 +116,7 @@ users)
 ## Internal
 
 ## Internal: Windows
+  * Set the console to use UTF-8 on Windows using SetConsoleCP and SetConsoleOutputCP [#5970 @kit-ty-kate]
 
 ## Test
 

src/client/opamCliMain.ml

@@ -483,9 +483,11 @@ let json_out () =
         (Printexc.to_string e)
 
 let main () =
-  if Sys.win32 then
+  if Sys.win32 then begin
     (* Disable the critical error handling dialog *)
     ignore (OpamStubs.setErrorMode (1 lor OpamStubs.getErrorMode ()));
+    OpamStubs.setConsoleToUTF8 ();
+  end;
   OpamStd.Sys.at_exit (fun () ->
       flush_all_noerror ();
       if OpamClientConfig.(!r.print_stats) then (

src/core/opamStubs.dummy.ml

@@ -41,3 +41,4 @@ let win_create_process _ _ _ _ _ = that's_a_no_no
 let getConsoleWindowClass = that's_a_no_no
 let setErrorMode = that's_a_no_no
 let getErrorMode = that's_a_no_no
+let setConsoleToUTF8 = that's_a_no_no

src/core/opamStubs.mli

@@ -142,3 +142,6 @@ val setErrorMode : int -> int
 
 val getErrorMode : unit -> int
 (** Windows only. Directly wraps GetErrorMode. *)
+
+val setConsoleToUTF8 : unit -> unit
+(** Windows only. Directly wraps SetConsoleOutputCP(CP_UTF8). *)

src/stubs/win32/opamWin32Stubs.ml

@@ -39,3 +39,4 @@ external getConsoleAlias : string -> string -> string = "OPAMW_GetConsoleAlias"
 external getConsoleWindowClass : unit -> string option = "OPAMW_GetConsoleWindowClass"
 external setErrorMode : int -> int = "OPAMW_SetErrorMode"
 external getErrorMode : unit -> int = "OPAMW_GetErrorMode"
+external setConsoleToUTF8 : unit -> unit = "OPAMW_SetConsoleToUTF8"

src/stubs/win32/opamWindows.c

@@ -28,6 +28,7 @@
 #include <TlHelp32.h>
 #include <Knownfolders.h>
 #include <Objbase.h>
+#include <WinCon.h>
 
 #include <stdio.h>
 
@@ -795,3 +796,11 @@ CAMLprim value OPAMW_GetErrorMode(value mode)
 {
   return Val_int(GetErrorMode());
 }
+
+CAMLprim value OPAMW_SetConsoleToUTF8(value _unit) {
+  /* NOTE: Setting Input (SetConsoleCP) is necessary for more.com
+   * called by cmdliner to correctly output UTF-8 characters */
+  SetConsoleCP(CP_UTF8);
+  SetConsoleOutputCP(CP_UTF8);
+  return Val_unit;
+}