From b8aaf29420a409063c9548166a445d98b75eb082 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Fri, 22 Dec 2023 16:45:53 +0000
Subject: [PATCH] Scaffolding for 1bpp reprocessing

---
 cpdfcommand.ml | 15 ++++++++++++---
 cpdfimage.ml   | 14 ++++++++++++--
 cpdfimage.mli  |  2 +-
 3 files changed, 25 insertions(+), 6 deletions(-)

diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index ce7efc3..cd5e0bd 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -523,7 +523,8 @@ type args =
    mutable idir_only_pdfs : bool;
    mutable no_warn_rotate : bool;
    mutable jpegquality : int;
-   mutable jpegqualitylossless : int}
+   mutable jpegqualitylossless : int;
+   mutable onebppmethod : string}
 
 let args =
   {op = None;
@@ -647,7 +648,8 @@ let args =
    idir_only_pdfs = false;
    no_warn_rotate = false;
    jpegquality = 100;
-   jpegqualitylossless = 100}
+   jpegqualitylossless = 100;
+   onebppmethod = ""}
 
 let reset_arguments () =
   args.op <- None;
@@ -754,6 +756,7 @@ let reset_arguments () =
   args.idir_only_pdfs <- false;
   args.jpegquality <- 100;
   args.jpegqualitylossless <- 100;
+  args.onebppmethod <- "";
   (* Do not reset original_filename or cpdflin or was_encrypted or
    was_decrypted_with_owner or recrypt or producer or creator or path_to_* or
    gs_malformed or gs_quiet or no-warn-rotate, since we want these to work
@@ -1927,6 +1930,9 @@ let setjpegquality q =
 let setjpegqualitylossless q =
   args.jpegqualitylossless <- q
 
+let set1bppmethod m =
+  args.onebppmethod <- m
+
 (* Parse a control file, make an argv, and then make Arg parse it. *)
 let rec make_control_argv_and_parse filename =
   control_args := !control_args @ parse_control_file filename
@@ -2710,6 +2716,9 @@ and specs =
    ("-lossless-to-jpeg",
      Arg.Int setjpegqualitylossless,
      " Set JPEG quality for existing lossless images");
+   ("-1bpp-method",
+     Arg.String set1bppmethod,
+     " Set 1bpp compression method for existing images");
    ("-squeeze",
      Arg.Unit setsqueeze,
      " Squeeze");
@@ -4447,7 +4456,7 @@ 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 ~path_to_convert:args.path_to_convert;
+        Cpdfimage.process pdf ~q:args.jpegquality ~qlossless:args.jpegqualitylossless ~onebppmethod:args.onebppmethod ~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 fc42c66..be110fe 100644
--- a/cpdfimage.ml
+++ b/cpdfimage.ml
@@ -590,10 +590,16 @@ 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_ccitt pdf s dict reference =
+  ()
+
+let recompress_1bpp_jbig2_lossless pdf s dict reference =
+  ()
+
 (* 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 CCITT; anything to JBIG2 lossless (no globals yet) *)
-let process ?q ?qlossless ?jbig2 pdf ~path_to_convert =
+let process ?q ?qlossless ?onebppmethod pdf ~path_to_convert =
   let process_obj _ s =
     match s with
     | Pdf.Stream ({contents = dict, _} as reference) ->
@@ -610,7 +616,11 @@ let process ?q ?qlossless ?jbig2 pdf ~path_to_convert =
             end
         | Some (Pdf.Name "/Image"), _, Some (Pdf.Integer 1), _
         | Some (Pdf.Name "/Image"), _, _, Some (Pdf.Boolean true) ->
-            Printf.printf "1bpp\n"
+           begin match onebppmethod with
+           | Some "CCITT" -> recompress_1bpp_ccitt pdf s dict reference
+           | Some "JBIG2Lossless" -> recompress_1bpp_jbig2_lossless pdf s dict reference
+           | _ -> ()
+           end
         | Some (Pdf.Name "/Image"), _, _, _ ->
             begin match qlossless with
             | Some qlossless -> lossless_to_jpeg pdf ~qlossless ~path_to_convert s dict reference
diff --git a/cpdfimage.mli b/cpdfimage.mli
index f13707b..5542d32 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 -> ?jbig2:int -> Pdf.t -> path_to_convert:string -> unit
+val process : ?q:int -> ?qlossless:int -> ?onebppmethod:string -> Pdf.t -> path_to_convert:string -> unit
 
 (**/**)
 val image_of_input : (unit -> Pdfio.bytes -> Pdf.pdfobject * (int * Pdf.pdfobject) list) -> Pdfio.input -> Pdf.t