From c0c72b8ec4813833a476df69a3d7182224f6ab28 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Fri, 22 Dec 2023 19:33:10 +0000
Subject: [PATCH] Plumbing for path_to_jbig2enc

---
 cpdfcommand.ml | 18 +++++++++++++-----
 cpdfimage.ml   | 12 +++++++++---
 cpdfimage.mli  |  2 +-
 3 files changed, 23 insertions(+), 9 deletions(-)

diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index cd5e0bd..dd47c07 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -469,6 +469,7 @@ type args =
    mutable path_to_im : string;
    mutable path_to_p2p : string;
    mutable path_to_convert : string;
+   mutable path_to_jbig2enc : string;
    mutable frombox : string option;
    mutable tobox : string option;
    mutable mediabox_if_missing : bool;
@@ -589,10 +590,11 @@ let args =
    dashrange = "all";
    outline = false;
    linewidth = 1.0;
-   path_to_ghostscript = "";
-   path_to_im = "";
-   path_to_p2p = "";
-   path_to_convert = "";
+   path_to_ghostscript = "gs";
+   path_to_im = "magick";
+   path_to_p2p = "pnm2png";
+   path_to_convert = "convert";
+   path_to_jbig2enc = "jbig2enc";
    frombox = None;
    tobox = None;
    mediabox_if_missing = false;
@@ -1528,6 +1530,9 @@ let setimpath p =
 let setconvertpath p =
   args.path_to_convert <- p
 
+let setjbig2encpath p =
+  args.path_to_jbig2enc <- p
+
 let setp2ppath p =
   args.path_to_p2p <- p
 
@@ -2710,6 +2715,9 @@ and specs =
    ("-convert",
      Arg.String setconvertpath,
      " Path to convert executable");
+   ("-jbig2enc",
+     Arg.String setjbig2encpath,
+     " Path to jbig2enc executable");
    ("-jpeg-to-jpeg",
      Arg.Int setjpegquality,
      " Set JPEG quality for existing JPEGs");
@@ -4456,7 +4464,7 @@ let go () =
         write_pdf false (Cpdfchop.chop ~x ~y ~columns:args.impose_columns ~btt:args.impose_btt ~rtl:args.impose_rtl pdf range)
   | Some ProcessImages ->
       let pdf = get_single_pdf args.op false in
-        Cpdfimage.process pdf ~q:args.jpegquality ~qlossless:args.jpegqualitylossless ~onebppmethod:args.onebppmethod ~path_to_convert:args.path_to_convert;
+        Cpdfimage.process pdf ~q:args.jpegquality ~qlossless:args.jpegqualitylossless ~onebppmethod:args.onebppmethod ~path_to_jbig2enc:args.path_to_jbig2enc ~path_to_convert:args.path_to_convert;
         write_pdf false pdf
 
 (* Advise the user if a combination of command line flags makes little sense,
diff --git a/cpdfimage.ml b/cpdfimage.ml
index 653ee4d..45a6d94 100644
--- a/cpdfimage.ml
+++ b/cpdfimage.ml
@@ -492,6 +492,7 @@ let image_of_input fobj i =
 (* FIXME if lossy only 5% smaller, ignore? Set this parameter... *)
 (* FIXME error handling for Sys.remove, others *)
 (* FIXME Use raw format for all, and make it fast *)
+(* FIXME combine path_to_im / path_to_convert *)
 let jpeg_to_jpeg pdf ~q ~path_to_convert s dict reference =
   Pdf.getstream s;
   let out = Filename.temp_file "cpdf" "convertin" ^ ".jpg" in
@@ -590,13 +591,18 @@ let lossless_to_jpeg pdf ~qlossless ~path_to_convert s dict reference =
       print_string (Printf.sprintf "%s (%s) [%s]\n" colspace bpc filter);
       () (* an image we cannot or do not handle *)
 
-let recompress_1bpp_jbig2_lossless pdf s dict reference =
+let recompress_1bpp_jbig2_lossless ~path_to_jbig2enc pdf s dict reference =
+  (* 1. Decode it *)
+  (* 2. Check we are ok *)
+  (* 3. Call jbig2enc *)
+  (* 4. Read in result *)
+  (* 5. Set data and dictionary *)
   ()
 
 (* JPEG to JPEG: RGB and CMYK JPEGS *)
 (* Lossless to JPEG: 8bpp Grey, 8bpp RGB, 8bpp CMYK including separation add ICCBased colourspaces *)
 (* 1 bit: anything to JBIG2 lossless (no globals) *)
-let process ?q ?qlossless ?onebppmethod pdf ~path_to_convert =
+let process ?q ?qlossless ?onebppmethod pdf ~path_to_jbig2enc ~path_to_convert =
   let process_obj _ s =
     match s with
     | Pdf.Stream ({contents = dict, _} as reference) ->
@@ -614,7 +620,7 @@ let process ?q ?qlossless ?onebppmethod pdf ~path_to_convert =
         | Some (Pdf.Name "/Image"), _, Some (Pdf.Integer 1), _
         | Some (Pdf.Name "/Image"), _, _, Some (Pdf.Boolean true) ->
            begin match onebppmethod with
-           | Some "JBIG2" -> recompress_1bpp_jbig2_lossless pdf s dict reference
+           | Some "JBIG2" -> recompress_1bpp_jbig2_lossless ~path_to_jbig2enc pdf s dict reference
            | _ -> ()
            end
         | Some (Pdf.Name "/Image"), _, _, _ ->
diff --git a/cpdfimage.mli b/cpdfimage.mli
index 5542d32..7cc6c7a 100644
--- a/cpdfimage.mli
+++ b/cpdfimage.mli
@@ -11,7 +11,7 @@ val image_resolution : Pdf.t -> int list -> float -> (int * string * int * int *
 (** List images in JSON format *)
 val images : Pdf.t -> int list -> Cpdfyojson.Safe.t
 
-val process : ?q:int -> ?qlossless:int -> ?onebppmethod:string -> Pdf.t -> path_to_convert:string -> unit
+val process : ?q:int -> ?qlossless:int -> ?onebppmethod:string -> Pdf.t -> path_to_jbig2enc:string -> path_to_convert:string -> unit
 
 (**/**)
 val image_of_input : (unit -> Pdfio.bytes -> Pdf.pdfobject * (int * Pdf.pdfobject) list) -> Pdfio.input -> Pdf.t