Refactor file type check (lazy and only 1024 bytes) (#30)

* Add new binary check
* Add new file type type and use
* Simplify Filec.t
* Fix: use with to make sure it gets closed

lib/fs/filec.ml

@@ -1,21 +1,31 @@
-type t =
-  | Binary
-  | Text of {
-      lines : Pretty.Doc.t array;
-      offset : int;
-    }
+type t = {
+  lines : Pretty.Doc.t array;
+  offset : int;
+}
 
-(* Regex to used to determine if bat outputs a binary file warning. This is a
-   bit of a fragile approach, but there is no robust way to determine if a file
-   is binary or not. Improve this if it becomes a problem.
-*)
-let binary_file_pattern = Str.regexp ".*\\[bat warning\\].*Binary.*content*."
+type file_type =
+  | Binary
+  | Text
 
 let binary_file_warning =
   [| Pretty.Doc.str "This file is binary and cannot be displayed" |]
 
-let has_binary_warning contents =
-  Str.string_match binary_file_pattern contents 0
+let open_file_bin num_bytes path =
+  let buffer = Bytes.create num_bytes in
+  let n =
+    In_channel.with_open_bin path (fun ic -> input ic buffer 0 num_bytes)
+  in
+  Bytes.sub buffer 0 n
+
+let has_zero_bytes buffer =
+  let rec has_null = function
+    | i when i = Bytes.length buffer -> false
+    | i when Bytes.get buffer i = '\x00' -> true
+    | i -> has_null (i + 1)
+  in
+  has_null 0
+
+let is_likely_binary path = path |> open_file_bin 1024 |> has_zero_bytes
 
 let bat_cmd =
   {|bat --style=numbers,changes --color=always --italic-text=always --paging=never --terminal-width=80 |}
@@ -24,25 +34,19 @@ let read_file_raw path = Shell.proc_stdout (bat_cmd ^ path)
 
 (* Reads file contents using 'bat' to have pretty syntax highlighting *)
 let read path =
-  let contents = read_file_raw path in
-  if has_binary_warning contents then Binary
+  if is_likely_binary path then { lines = binary_file_warning; offset = 0 }
   else
     let lines =
-      contents
+      path
+      |> read_file_raw
       |> String.split_on_char '\n'
       |> List.map Pretty.Doc.str
       |> Array.of_list
     in
-    Text { lines; offset = 0 }
-
-let offset = function
-  | Text { offset; _ } -> offset
-  | Binary -> 0
+    { lines; offset = 0 }
 
-(* Returns the lines of the file contents *)
-let lines = function
-  | Text { lines; _ } -> lines
-  | Binary -> binary_file_warning
+(* Returns file type based on contents *)
+let type_of_path path = if is_likely_binary path then Binary else Text
 
 (* Returns the number of lines in the file contents *)
-let length filec = filec |> lines |> Array.length
+let length filec = filec.lines |> Array.length

lib/fs/filec.mli

@@ -1,20 +1,19 @@
 (** File contents as an array of lines, where each line is wrapped into a
     document (for rendering efficiency) *)
-type t =
+type t = {
+  lines : Pretty.Doc.t array;
+  offset : int;
+}
+
+type file_type =
   | Binary
-  | Text of {
-      lines : Pretty.Doc.t array;
-      offset : int;
-    }
+  | Text
 
 (** Reads file contents using 'bat' to have pretty syntax highlighting **)
 val read : string -> t
 
-(** Returns offset based on file type of contents **)
-val offset : t -> int
-
 (** Returns the len of file contents **)
 val length : t -> int
 
-(** Returns the lines of file contents **)
-val lines : t -> Pretty.Doc.t array
+(** Returns file_type based on the first 1024 bytes of the file **)
+val type_of_path : string -> file_type

lib/fs/fs.ml

@@ -1,7 +1,7 @@
 module Filec = Filec
 
 type tree =
-  | File of string * Filec.t Lazy.t
+  | File of string * Filec.t Lazy.t * Filec.file_type Lazy.t
   | Dir of string * tree array Lazy.t
 
 type dir_cursor = {
@@ -15,7 +15,7 @@ type cursor =
 
 (* Extracts the file name from a tree node *)
 let file_name = function
-  | File (name, _) -> name
+  | File (name, _, _) -> name
   | Dir (name, _) -> name
 
 (* A files comparison:
@@ -30,7 +30,7 @@ let order_files t1 t2 =
   | _, _ -> String.compare (file_name t1) (file_name t2)
 
 let rec sort_tree = function
-  | File (name, contents) -> File (name, contents)
+  | File (name, contents, ft) -> File (name, contents, ft)
   | Dir (name, (lazy children)) ->
       Array.sort order_files children;
       Dir (name, lazy (Array.map sort_tree children))
@@ -46,7 +46,11 @@ let rec to_tree path =
     in
     let dirname = Filename.basename path in
     Dir (dirname, children)
-  else File (Filename.basename path, lazy (Filec.read path))
+  else
+    File
+      ( Filename.basename path,
+        lazy (Filec.read path),
+        lazy (Filec.type_of_path path) )
 
 let read_tree path = path |> to_tree |> sort_tree
 let file_at cursor = cursor.files.(cursor.pos)
@@ -75,13 +79,10 @@ let move_dir_cursor move cursor =
   { cursor with pos = new_pos }
 
 let move_file_cursor move cursor =
-  match cursor with
-  | Filec.Binary -> cursor
-  | Filec.Text txt_cur ->
-      let len = Filec.length cursor in
-      let new_offset = Filec.offset cursor + move in
-      if new_offset < 0 || new_offset + span > len then cursor
-      else Filec.Text { txt_cur with offset = new_offset }
+  let len = Filec.length cursor in
+  let new_offset = cursor.offset + move in
+  if new_offset < 0 || new_offset + span > len then cursor
+  else { cursor with offset = new_offset }
 
 let go_move move zipper =
   let move_dir = move_dir_cursor move in
@@ -99,7 +100,7 @@ let go_next zipper =
   | Dir_cursor cursor -> (
       let next = file_at cursor in
       match next with
-      | File (_name, contents) ->
+      | File (_name, contents, _) ->
           {
             parents = cursor :: zipper.parents;
             current = File_cursor (Lazy.force contents);

lib/fs/fs.mli

@@ -4,7 +4,7 @@ module Filec = Filec
 
 (** A definition of a file tree. *)
 type tree =
-  | File of string * Filec.t Lazy.t
+  | File of string * Filec.t Lazy.t * Filec.file_type Lazy.t
   | Dir of string * tree array Lazy.t
 
 (** Return the name of a given tree node. *)

lib/tui/tui.ml

@@ -29,7 +29,7 @@ let read_root_tree ~root_dir_path =
   let tree = Fs.read_tree root_dir_path in
   let files =
     match tree with
-    | Fs.File (path, _) ->
+    | Fs.File (path, _, _) ->
         Printf.printf "Given path '%s' is not a directory!" path;
         exit 1
     | Fs.Dir (_, files) -> files

lib/tui/widget/code.ml

@@ -28,10 +28,10 @@ let pwd root_dir_path (fs : Fs.zipper) =
   Pretty.Doc.(fmt Style.directory full_path)
 
 let file_contents_to_doc ~(file_contents : Fs.Filec.t) =
-  let lines = Fs.Filec.lines file_contents in
+  let lines = file_contents.lines in
   let len_lines = Array.length lines in
   let span = 40 in
-  let offset = Fs.Filec.offset file_contents in
+  let offset = file_contents.offset in
 
   let contents_span = Extra.List.of_sub_array ~offset ~len:span lines in
 
@@ -58,9 +58,9 @@ let max_file_name_len files =
 let fmt_file ~max_name_len (tree : Fs.tree) =
   let pad = Extra.String.fill_right max_name_len in
   match tree with
-  | File (name, contents) -> (
-      match Lazy.force contents with
-      | Fs.Filec.Text _ -> file_char ^ " " ^ pad name
+  | File (name, _, file_type) -> (
+      match Lazy.force file_type with
+      | Fs.Filec.Text -> file_char ^ " " ^ pad name
       | Fs.Filec.Binary -> bin_char ^ " " ^ pad name)
   | Dir (name, (lazy children)) -> (
       match children with
@@ -197,7 +197,7 @@ let fs_to_view (fs : Fs.zipper) =
     | File_cursor contents, parent :: _ -> (parent, File_selected contents)
     | Dir_cursor cursor, _ -> (
         match Fs.file_at cursor with
-        | File (_, contents) -> (cursor, File_selected (Lazy.force contents))
+        | File (_, contents, _) -> (cursor, File_selected (Lazy.force contents))
         | Dir (_, children) ->
             ( cursor,
               Dir_selected