Add 'line-endings' option to specify line endings used in the formatted output (#1703)

Author: nojb

.ocamlformat

@@ -3,3 +3,4 @@ break-cases = fit
 margin = 77
 parse-docstrings = true
 wrap-comments = true
+line-endings = lf

CHANGES.md

@@ -68,6 +68,9 @@
     - Break the line and reindent the cursor when pressing <ENTER>
   (#1639, #1685, @gpetiot) (#1687, @bcc32)
 
+  + Add 'line-endings=lf|crlf' option to specify the line endings used in the
+    formatted output. (#1703, @nojb)
+
 #### Internal
 
   + A script `tools/build-mingw64.sh` is provided to build a native Windows

bin/ocamlformat/main.ml

@@ -24,10 +24,19 @@ let format ?output_file ~kind ~input_name ~source conf opts =
     Translation_unit.parse_and_format kind ?output_file ~input_name ~source
       conf opts
 
+let write_all output_file ~data =
+  Out_channel.with_file ~binary:true output_file ~f:(fun oc ->
+      Out_channel.output_string oc data )
+
 let to_output_file output_file data =
   match output_file with
-  | None -> Out_channel.output_string Out_channel.stdout data
-  | Some output_file -> Out_channel.write_all output_file ~data
+  | None ->
+      Out_channel.flush Out_channel.stdout ;
+      Out_channel.set_binary_mode Out_channel.stdout true ;
+      Out_channel.output_string Out_channel.stdout data ;
+      Out_channel.flush Out_channel.stdout ;
+      Out_channel.set_binary_mode Out_channel.stdout false
+  | Some output_file -> write_all output_file ~data
 
 let source_from_file = function
   | Conf.Stdin -> In_channel.input_all In_channel.stdin
@@ -52,7 +61,7 @@ let run_action action opts =
         match result with
         | Ok formatted ->
             if not (String.equal formatted source) then
-              Out_channel.write_all input_file ~data:formatted ;
+              write_all input_file ~data:formatted ;
             Ok ()
         | Error e -> Error (fun () -> print_error conf opts e)
       in

lib/Conf.ml

@@ -51,6 +51,7 @@ type t =
   ; let_binding_indent: int
   ; let_binding_spacing: [`Compact | `Sparse | `Double_semicolon]
   ; let_module: [`Compact | `Sparse]
+  ; line_endings: [`Lf | `Crlf]
   ; margin: int
   ; match_indent: int
   ; match_indent_nested: [`Always | `Auto | `Never]
@@ -86,7 +87,7 @@ let warn_raw, collect_warnings =
   let delayed_warning_list = ref [] in
   let warn_ s =
     if !delay_warning then delayed_warning_list := s :: !delayed_warning_list
-    else Format.eprintf "%s" s
+    else Format.eprintf "%s%!" s
   in
   let collect_warnings f =
     let old_flag, old_list = (!delay_warning, !delayed_warning_list) in
@@ -830,6 +831,16 @@ module Formatting = struct
     let msg = concrete_syntax_preserved_msg in
     C.removed_option ~names ~version ~msg
 
+  let line_endings =
+    let doc = "Line endings used." in
+    let all =
+      [ ("lf", `Lf, "$(b,lf) uses Unix line endings.")
+      ; ("crlf", `Crlf, "$(b,crlf) uses Windows line endings.") ]
+    in
+    C.choice ~names:["line-endings"] ~all ~doc ~allow_inline:false ~section
+      (fun conf x -> {conf with line_endings= x})
+      (fun conf -> conf.line_endings)
+
   let margin =
     let docv = "COLS" in
     let doc = "Format code to fit within $(docv) columns." in
@@ -1437,6 +1448,7 @@ let ocamlformat_profile =
   ; let_binding_indent= 2
   ; let_binding_spacing= `Compact
   ; let_module= `Compact
+  ; line_endings= `Lf
   ; margin= 80
   ; match_indent= 0
   ; match_indent_nested= `Never
@@ -1508,6 +1520,7 @@ let conventional_profile =
   ; let_binding_indent= C.default Formatting.let_binding_indent
   ; let_binding_spacing= C.default Formatting.let_binding_spacing
   ; let_module= C.default Formatting.let_module
+  ; line_endings= C.default Formatting.line_endings
   ; margin= C.default Formatting.margin
   ; match_indent= C.default Formatting.match_indent
   ; match_indent_nested= C.default Formatting.match_indent_nested
@@ -1632,6 +1645,7 @@ let janestreet_profile =
   ; let_binding_indent= 2
   ; let_binding_spacing= `Double_semicolon
   ; let_module= `Sparse
+  ; line_endings= `Lf
   ; margin= 90
   ; match_indent= 0
   ; match_indent_nested= `Never

lib/Conf.mli

@@ -53,6 +53,7 @@ type t =
   ; let_binding_indent: int
   ; let_binding_spacing: [`Compact | `Sparse | `Double_semicolon]
   ; let_module: [`Compact | `Sparse]
+  ; line_endings: [`Lf | `Crlf]
   ; margin: int  (** Format code to fit within [margin] columns. *)
   ; match_indent: int
   ; match_indent_nested: [`Always | `Auto | `Never]

lib/Translation_unit.ml

@@ -412,6 +412,30 @@ let parse_result ?(f = Ast_passes.Ast0.Parse.ast) fragment conf ~source
   | exception exn -> Error (Error.Invalid_source {exn; input_name})
   | parsed -> Ok parsed
 
+let normalize_eol ~line_endings s =
+  let buf = Buffer.create (String.length s) in
+  let rec loop seen_cr i =
+    if i = String.length s then (
+      if seen_cr then Buffer.add_char buf '\r' ;
+      Buffer.contents buf )
+    else
+      match (s.[i], line_endings) with
+      | '\r', _ ->
+          if seen_cr then Buffer.add_char buf '\r' ;
+          loop true (i + 1)
+      | '\n', `Crlf ->
+          Buffer.add_string buf "\r\n" ;
+          loop false (i + 1)
+      | '\n', `Lf ->
+          Buffer.add_char buf '\n' ;
+          loop false (i + 1)
+      | c, _ ->
+          if seen_cr then Buffer.add_char buf '\r' ;
+          Buffer.add_char buf c ;
+          loop false (i + 1)
+  in
+  loop false 0
+
 let parse_and_format (type a b) (fg0 : a Ast_passes.Ast0.t)
     (fgN : b Ast_passes.Ast_final.t) ?output_file ~input_name ~source conf
     opts =
@@ -421,6 +445,8 @@ let parse_and_format (type a b) (fg0 : a Ast_passes.Ast0.t)
   let parsed = {parsed with ast= Ast_passes.run fg0 fgN parsed.ast} in
   format fg0 fgN ?output_file ~input_name ~prev_source:source ~parsed conf
     opts
+  >>= fun formatted ->
+  Ok (normalize_eol ~line_endings:conf.Conf.line_endings formatted)
 
 let parse_and_format = function
   | Syntax.Structure -> parse_and_format Structure Structure

ocamlformat-help.txt

@@ -272,6 +272,11 @@ OPTIONS (CODE FORMATTING STYLE)
            ... = and before the in if the module declaration does not fit on
            a single line. The default value is compact.
 
+       --line-endings={lf|crlf}
+           Line endings used. lf uses Unix line endings. crlf uses Windows
+           line endings. The default value is lf. Cannot be set in
+           attributes.
+
        -m COLS, --margin=COLS
            Format code to fit within COLS columns. The default value is 80.
            Cannot be set in attributes.

test/passing/dune.inc

@@ -671,6 +671,30 @@
  (package ocamlformat)
  (action (diff tests/compact_lists_arrays.ml compact_lists_arrays.ml.output)))
 
+(rule
+ (deps tests/.ocamlformat )
+ (package ocamlformat)
+ (action
+   (with-outputs-to crlf_to_crlf.ml.output
+     (run %{bin:ocamlformat} --line-endings=crlf %{dep:tests/crlf_to_crlf.ml}))))
+
+(rule
+ (alias runtest)
+ (package ocamlformat)
+ (action (diff tests/crlf_to_crlf.ml.ref crlf_to_crlf.ml.output)))
+
+(rule
+ (deps tests/.ocamlformat )
+ (package ocamlformat)
+ (action
+   (with-outputs-to crlf_to_lf.ml.output
+     (run %{bin:ocamlformat} --line-endings=lf %{dep:tests/crlf_to_lf.ml}))))
+
+(rule
+ (alias runtest)
+ (package ocamlformat)
+ (action (diff tests/crlf_to_lf.ml.ref crlf_to_lf.ml.output)))
+
 (rule
  (deps tests/.ocamlformat )
  (package ocamlformat)

test/passing/tests/crlf_to_crlf.ml

@@ -0,0 +1,41 @@
+let _ = {|
+foo
+
+ bar
+|}
+
+(** This is verbatim:
+
+    {v
+   o  o
+  /\  /\
+  /\  /\
+    v}
+
+    This is preformated code:
+
+    {[
+      let verbatim s =
+        s |> String.split_lines |> List.map ~f:String.strip
+        |> fun s -> list s "@," Fmt.str
+    ]} *)
+
+(** Lists:
+
+    list with short lines:
+
+    - x
+
+    list with long lines:
+
+    - xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx
+      xxx xxx xxx xxx xxx xxx
+
+    list with sub lists:
+
+    {ul
+     {- xxx
+
+        - a
+     }
+    } *)

test/passing/tests/crlf_to_crlf.ml.opts

@@ -0,0 +1 @@
+--line-endings=crlf