From e928818ccb15d3bdbbd394478482246bc44e2451 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Mon, 4 Dec 2023 16:32:12 +0000
Subject: [PATCH] Scaffolding for JBIG2Globals

---
 cpdfcommand.ml     | 16 +++++++++++++++-
 cpdfdrawcontrol.ml |  4 ++--
 cpdfimage.ml       |  8 ++++----
 cpdfimage.mli      |  8 ++++----
 4 files changed, 25 insertions(+), 11 deletions(-)

diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index 5e23542..0ef5c9c 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -1787,9 +1787,11 @@ let set_input_image f s =
   with
     Sys_error _ -> error "Image file not found"
 
+let jbig2_global = ref None
+
 let set_input_png = set_input_image Cpdfimage.obj_of_png_data
 let set_input_jpeg = set_input_image Cpdfimage.obj_of_jpeg_data
-let set_input_jbig2 = set_input_image (Cpdfimage.obj_of_jbig2_data ~global:1)
+let set_input_jbig2 = set_input_image (Cpdfimage.obj_of_jbig2_data ?global:!jbig2_global)
 
 let embed_font_inner font =
   match font with
@@ -1896,6 +1898,12 @@ let setlistimagesjson () =
   setop ListImages ();
   args.format_json <- true
 
+let set_jbig2_global f =
+  jbig2_global := Some (Pdfio.bytes_of_string (contents_of_file f))
+
+let clear_jbig2_global () =
+  jbig2_global := None
+
 (* 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
@@ -1919,6 +1927,12 @@ and specs =
    ("-jbig2",
        Arg.String set_input_jbig2,
        " Load from a JBIG2 fragment, converting to PDF");
+   ("-jbig2-global",
+       Arg.String set_jbig2_global,
+       " Load a JBIG2 global stream");
+   ("-jbig2-global-clear",
+       Arg.Unit clear_jbig2_global,
+       " Forget any JBIG2 global stream");
    ("-idir",
        Arg.String set_input_dir,
        " Add a directory of files");
diff --git a/cpdfdrawcontrol.ml b/cpdfdrawcontrol.ml
index 71012df..e13c965 100644
--- a/cpdfdrawcontrol.ml
+++ b/cpdfdrawcontrol.ml
@@ -245,7 +245,7 @@ let addjpeg n =
   in
     try
       let data = Pdfio.bytes_of_string (contents_of_file filename) in
-        addop (Cpdfdraw.ImageXObject (name, Cpdfimage.obj_of_jpeg_data data))
+        addop (Cpdfdraw.ImageXObject (name, fst (Cpdfimage.obj_of_jpeg_data data)))
     with
       _ -> error "addjpeg: could not load JPEG"
 
@@ -256,7 +256,7 @@ let addpng n =
     | _ -> error "addpng: bad file specification"
   in
     let data = Pdfio.bytes_of_string (contents_of_file filename) in
-      addop (Cpdfdraw.ImageXObject (name, Cpdfimage.obj_of_png_data data))
+      addop (Cpdfdraw.ImageXObject (name, fst (Cpdfimage.obj_of_png_data data)))
 
 let addimage s =
   addop (Cpdfdraw.Image s)
diff --git a/cpdfimage.ml b/cpdfimage.ml
index 0e05505..251516c 100644
--- a/cpdfimage.ml
+++ b/cpdfimage.ml
@@ -388,7 +388,7 @@ let obj_of_jpeg_data data =
      "/Width", Pdf.Integer w;
      "/Height", Pdf.Integer h]
   in
-    Pdf.Stream {contents = (Pdf.Dictionary d, Pdf.Got data)}
+    Pdf.Stream {contents = (Pdf.Dictionary d, Pdf.Got data)}, []
 
 let obj_of_png_data data =
   let png = Cpdfpng.read_png (Pdfio.input_of_bytes data) in
@@ -406,7 +406,7 @@ let obj_of_png_data data =
      "/Width", Pdf.Integer png.width;
      "/Height", Pdf.Integer png.height]
   in
-    Pdf.Stream {contents = (Pdf.Dictionary d, Pdf.Got png.idat)}
+    Pdf.Stream {contents = (Pdf.Dictionary d, Pdf.Got png.idat)}, []
 
 let jbig2_dimensions data =
   (bget data 11 * 256 * 256 * 256 + bget data 12 * 256 * 256 + bget data 13 * 256 + bget data 14,
@@ -424,12 +424,12 @@ let obj_of_jbig2_data ?global:int data =
        ("/Width", Pdf.Integer w);
        ("/Height", Pdf.Integer h)]
   in
-    Pdf.Stream {contents = (Pdf.Dictionary d, Pdf.Got data)}
+    Pdf.Stream {contents = (Pdf.Dictionary d, Pdf.Got data)}, []
 
 let image_of_input fobj i =
   let pdf = Pdf.empty () in
   let data = Pdfio.bytes_of_input i 0 i.Pdfio.in_channel_length in
-  let obj = fobj data in
+  let obj, extras = fobj data in
   let w = match Pdf.lookup_direct pdf "/Width" obj with Some x -> Pdf.getnum pdf x | _ -> assert false in
   let h = match Pdf.lookup_direct pdf "/Height" obj with Some x -> Pdf.getnum pdf x | _ -> assert false in
   let page =
diff --git a/cpdfimage.mli b/cpdfimage.mli
index ef2c838..4c42d42 100644
--- a/cpdfimage.mli
+++ b/cpdfimage.mli
@@ -12,7 +12,7 @@ val image_resolution : Pdf.t -> int list -> float -> (int * string * int * int *
 val images : Pdf.t -> int list -> Cpdfyojson.Safe.t
 
 (**/**)
-val image_of_input : (Pdfio.bytes -> Pdf.pdfobject) -> Pdfio.input -> Pdf.t
-val obj_of_jpeg_data : Pdfio.bytes -> Pdf.pdfobject
-val obj_of_png_data : Pdfio.bytes -> Pdf.pdfobject
-val obj_of_jbig2_data : ?global:int -> Pdfio.bytes -> Pdf.pdfobject
+val image_of_input : (Pdfio.bytes -> Pdf.pdfobject * (int * Pdf.pdfobject) list) -> Pdfio.input -> Pdf.t
+val obj_of_jpeg_data : Pdfio.bytes -> Pdf.pdfobject * (int * Pdf.pdfobject) list
+val obj_of_png_data : Pdfio.bytes -> Pdf.pdfobject * (int * Pdf.pdfobject) list
+val obj_of_jbig2_data : ?global:Pdfio.bytes -> Pdfio.bytes -> Pdf.pdfobject * (int * Pdf.pdfobject) list