From d96b6aabbb26861e449a722970aeaa6eceb17839 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Mon, 18 Dec 2023 20:50:52 +0000
Subject: [PATCH] Detection for ICCBased colour spaces

---
 cpdfimage.ml | 39 +++++++++++++++++++++------------------
 1 file changed, 21 insertions(+), 18 deletions(-)

diff --git a/cpdfimage.ml b/cpdfimage.ml
index 1f3576c..cd12ed3 100644
--- a/cpdfimage.ml
+++ b/cpdfimage.ml
@@ -500,13 +500,22 @@ let process pdf ~q ~qlossless ~path_to_convert =
               Sys.remove out2
         | Some (Pdf.Name "/Image"), _ ->
             (* 0. Test if this is one we can do - for now just Colourspace=RGB, BPC=8 *)
-            begin match Pdf.lookup_direct pdf "/ColorSpace" dict, Pdf.lookup_direct pdf "/BitsPerComponent" dict with
-            | Some (Pdf.Name "/DeviceRGB"), Some (Pdf.Integer 8) ->
+            let bpc = Pdf.lookup_direct pdf "/BitsPerComponent" dict in
+            let is_rgb =
+              match Pdf.lookup_direct pdf "/ColorSpace" dict with
+              | Some (Pdf.Name "/DeviceRGB") -> true
+              | Some (Pdf.Array [Pdf.Name "/ICCBased"; stream]) ->
+                  begin match Pdf.lookup_direct pdf "/N" stream with
+                  | Some (Pdf.Integer 3) -> true
+                  | _ -> false
+                  end
+              | _ -> false
+            in
+            begin match is_rgb, bpc with
+            | true, Some (Pdf.Integer 8) ->
                 let size = match Pdf.lookup_direct pdf "/Length" dict with Some (Pdf.Integer i) -> i | _ -> 0 in
-                (* 1. Decompress it - check we succeeded, bail if not *)
                 Pdfcodec.decode_pdfstream_until_unknown pdf s;
                 begin match Pdf.lookup_direct pdf "/Filter" (fst !reference) with Some _ -> () | None ->
-                  (* 1. Output to pnm *)
                   let out = Filename.temp_file "cpdf" "convertin" ^ ".pnm" in
                   let out2 = Filename.temp_file "cpdf" "convertout" ^ ".jpg" in
                   let fh = open_out_bin out in
@@ -515,7 +524,6 @@ let process pdf ~q ~qlossless ~path_to_convert =
                   let data = match s with Pdf.Stream {contents = _, Pdf.Got d} -> d | _ -> assert false in
                   pnm_to_channel_24 fh w h data;
                   close_out fh;
-                  (* 2. Convert to JPEG with convert *)
                   let retcode =
                     let command = 
                       (Filename.quote_command path_to_convert
@@ -524,35 +532,30 @@ let process pdf ~q ~qlossless ~path_to_convert =
                       (*Printf.printf "%S\n" command;*)
                       Sys.command command
                   in
-                  (* 3. Check smaller, Read file, and build new dictionary - removing ColorSpace, BitsPerComponent replacing Filter *)
                   if retcode = 0 then
                     begin
                       let result = open_in_bin out2 in
                       let newsize = in_channel_length result in
-                        Printf.printf "Lossless to JPEG %i -> %i\n" size newsize;
                       if newsize < size then
+                        Printf.printf "Lossless to JPEG %i -> %i\n" size newsize;
                         reference :=
-                          (Pdf.remove_dict_entry
-                            (Pdf.remove_dict_entry
-                              (Pdf.add_dict_entry
-                                (Pdf.add_dict_entry dict "/Length" (Pdf.Integer newsize))
-                                "/Filter"
-                                (Pdf.Name "/DCTDecode"))
-                             "/ColorSpace")
-                           "/BitsPerComponent"),
+                          (Pdf.add_dict_entry
+                            (Pdf.add_dict_entry dict "/Length" (Pdf.Integer newsize))
+                             "/Filter"
+                             (Pdf.Name "/DCTDecode")),
                           Pdf.Got (Pdfio.bytes_of_input_channel result)
                     end;
-                  (* 4. Clean up. *)
                   Sys.remove out;
                   Sys.remove out2
                 end
             | colspace, bpc ->
-              (*let colspace, bpc, filter = 
+              let colspace = Pdf.lookup_direct pdf "/ColorSpace" dict in
+              let colspace, bpc, filter = 
                 (match colspace with None -> "none" | Some x -> Pdfwrite.string_of_pdf x),
                 (match bpc with None -> "none" | Some x -> Pdfwrite.string_of_pdf x),
                 (match Pdf.lookup_direct pdf "/Filter" dict with None -> "none" | Some x -> Pdfwrite.string_of_pdf x)
               in
-                print_string (Printf.sprintf "%s (%s) [%s]\n" colspace bpc filter);*)
+                print_string (Printf.sprintf "%s (%s) [%s]\n" colspace bpc filter);
                 () (* an image we cannot or do not handle *)
             end
         | _ -> () (* not an image *)