Start quotes at the beginning of the file name

Author: Leonidas-from-XIV

src/fname.mli

@@ -1,5 +1 @@
-type t =
-  | DevNull
-  | Filename of string
-
-val filename : string -> t
+val filename : string -> string

src/fname.mll

@@ -1,34 +1,119 @@
 {
-type t =
-  | DevNull
-  | Filename of string
+type token =
+  | Quote
+  | Text of string
+  | Tab
 }
 
-rule quoted_fn acc = parse
-  | '"' { Buffer.contents acc }
-  | [^ '"']+ as s {
-    Buffer.add_string acc s;
-    quoted_fn acc lexbuf
+rule quoted_fn tokens = parse
+  | '"' {
+    (* after a quoted filename it might continue unquoted *)
+    let tokens = fn tokens lexbuf in
+    Quote :: tokens
   }
-  | _ { quoted_fn acc lexbuf }
+  | '\\' {
+    let token = escape lexbuf in
+    let tokens = quoted_fn tokens lexbuf in
+    token :: tokens
+  }
+  | [^ '"' '\\']+ as s {
+    let token = Text s in
+    (* consume the end quote or the escape *)
+    let tokens = quoted_fn tokens lexbuf in
+    token :: tokens
+  }
+  | eof { tokens }
 
-and unquoted_fn acc = parse
-  | [^ '\t']+ as s {
-    Buffer.add_string acc s;
-    unquoted_fn acc lexbuf
+and escape = parse
+  | '"' { Quote }
+  | _ as c {
+    (* ignore all other escapes, reconstruct the escape as text *)
+    Text (Printf.sprintf {|\%c|} c)
   }
-  | '\t' { Buffer.contents acc }
-  | eof { Buffer.contents acc }
 
-and fn acc = parse
-  | "/dev/null" { DevNull }
-  | '"' { Filename (quoted_fn acc lexbuf) }
-  | _ as c { 
-    Buffer.add_char acc c;
-    Filename (unquoted_fn acc lexbuf) }
+and unquoted_fn tokens = parse
+  | [^ '\t']+ as s {
+    let token = Text s in
+    token :: tokens
+  }
+  | '\t' { (Tab :: tokens) }
+  | eof { tokens }
 
+and fn tokens = parse
+  | '"' {
+    let tokens = quoted_fn tokens lexbuf in
+    Quote :: tokens
+  }
+  | '\t' {
+    let tokens = fn tokens lexbuf in
+    Tab :: tokens
+  }
+  | _ as c {
+    let token = Text (String.make 1 c) in
+    let tokens = unquoted_fn tokens lexbuf in
+    token :: tokens
+  }
+  | eof { tokens }
 {
+
+let lsplit2 ~f xs =
+  let rec loop acc = function
+    | [] -> List.rev acc, []
+    | x::xs when f x -> (List.rev acc), xs
+    | x::xs -> loop (x::acc) xs
+  in
+  loop [] xs
+
+let rec ends_with e = function
+  | [] -> false
+  | [x] -> e = x
+  | _::xs -> ends_with e xs
+
+let starts_with e = function
+  | [] -> false
+  | x::_ -> e = x
+
+let remove_start_end = function
+  | [] -> []
+  | _::_::[] -> []
+  | _::xs -> (
+      let rec remove_last acc = function
+        | []
+        | [_] -> List.rev acc
+        | x::xs -> remove_last (x::acc) xs
+      in
+      remove_last [] xs)
+
+let rec lstrip = function
+  | [] -> []
+  | (Text s)::xs when String.equal String.empty (String.trim s) -> lstrip xs
+  | otherwise -> otherwise
+
+let rstrip xs = xs |> List.rev |> lstrip |> List.rev
+
+let strip xs = xs |> lstrip |> rstrip
+
+let is_quoted_filename tokens =
+  starts_with Quote tokens && ends_with Quote tokens
+
+let unquote_filename tokens =
+  match is_quoted_filename tokens with
+  | true -> remove_start_end tokens
+  | false -> tokens
+
 let filename s =
   let lexbuf = Lexing.from_string s in
-  fn (Buffer.create 16) lexbuf
+  let tokens = fn [] lexbuf in
+  (* read the first bit until a tab, if available *)
+  let is_tab = function | Tab -> true | _ -> false in
+  let (tokens, _) = lsplit2 ~f:is_tab tokens in
+  let chunks = tokens
+    |> strip
+    |> unquote_filename
+    |> List.map (function
+      | Text s -> s
+      | Quote -> {|"|}
+      | Tab -> "\t")
+  in
+  String.concat "" chunks
 }

src/patch.ml

@@ -247,6 +247,7 @@ let pp ~git ppf {operation; hunks; mine_no_nl; their_no_nl} =
 let pp_list ~git ppf diffs =
   List.iter (Format.fprintf ppf "%a" (pp ~git)) diffs
 
+(* TODO: remove this and let users decide the prefix level they want *)
 let process_git_prefix ~git ~prefix s =
   if git && String.is_prefix ~prefix s then
     String.slice ~start:(String.length prefix) s
@@ -256,8 +257,8 @@ let process_git_prefix ~git ~prefix s =
 let operation_of_strings git mine their =
   let get_filename_opt s =
     match Fname.filename s with
-    | DevNull -> None
-    | Filename fn -> Some fn
+    | "/dev/null" -> None
+    | fn -> Some fn
   in
   match get_filename_opt mine, get_filename_opt their with
   | None, Some b ->

test/test.ml

@@ -846,39 +846,39 @@ let busybox_diff_spaces =
   operations [Patch.Edit ("a/foo bar", "b/foo bar")] busybox_diff_spaces
 
 let unified_diff_quotes = {|\
---- "foo \"bar\" baz"	2024-09-27 11:06:42.612922437 +0200
-+++ "foo bar \"baz\""	2024-09-27 11:09:48.325541553 +0200
+--- "foo bar \"baz\""	2024-09-27 11:09:48.325541553 +0200
++++ "\"foo\" bar baz"	2024-09-27 11:06:42.612922437 +0200
 @@ -1 +1 @@
--This is wrong.
-+This is right.
+-This is right.
++This is wrong.
 |}
 
 let unified_diff_quotes =
-  operations [Patch.Edit ({|foo "bar" baz|}, {|foo bar "baz"|})] unified_diff_quotes
+  operations [Patch.Edit ({|foo bar "baz"|}, {|"foo" bar baz|})] unified_diff_quotes
 
 let git_diff_quotes = {|\
-diff --git "a/foo \"bar\" baz" "b/foo bar \"baz\""
-index ef00db3..88adca3 100644
---- "a/foo \"bar\" baz"
-+++ "b/foo bar \"baz\""
+diff --git "a/foo bar \"baz\"" "b/\"foo\" bar baz"
+index 88adca3..ef00db3 100644
+--- "a/foo bar \"baz\"" 
++++ "b/\"foo\" bar baz" 
 @@ -1 +1 @@
--This is wrong.
-+This is right.
+-This is right.
++This is wrong.
 |}
 
 let git_diff_quotes =
-  operations [Patch.Edit ({|foo "bar" baz|}, {|foo bar "baz"|})] git_diff_quotes
+  operations [Patch.Edit ({|foo bar "baz"|}, {|"foo" bar baz|})] git_diff_quotes
 
 let busybox_diff_quotes = {|\
---- foo "bar" baz
-+++ foo bar "baz"
+--- foo bar "baz"
++++ "foo" bar baz
 @@ -1 +1 @@
--This is wrong.
-+This is right.
+-This is right.
++This is wrong.
 |}
 
 let busybox_diff_quotes =
-  operations [Patch.Edit ({|foo "bar" baz|}, {|foo bar "baz"|})] busybox_diff_quotes
+  operations [Patch.Edit ({|foo bar "baz"|}, {|"foo" bar baz|})] busybox_diff_quotes
 
 let dev_null_like = {|\
 --- /dev/null_but_actually_not