From 707525d8da3305c507630ed13ef370669792e8fc Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Thu, 28 Dec 2023 15:48:30 +0000
Subject: [PATCH] Extend -list-images

---
 cpdfcommand.ml |  4 ++--
 cpdfimage.ml   | 29 ++++++++++++++++++++++-------
 cpdfimage.mli  |  6 +++---
 3 files changed, 27 insertions(+), 12 deletions(-)

diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index 5d994f4..ec2dd22 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -4306,9 +4306,9 @@ let go () =
           begin match json with
           | `List l ->
               iter
-                (function (`Assoc [(_, `Int i); (_, `List pages); (_, `String name); (_, `Int w); (_, `Int h); (_, `String cs)]) ->
+                (function (`Assoc [(_, `Int i); (_, `List pages); (_, `String name); (_, `Int w); (_, `Int h); (_, `Int size); (_, `Int bpc); (_, `String cs); (_, `String filter)]) ->
                    let pages = combine_with_spaces (map (function `Int i -> string_of_int i | _ -> "") pages) in
-                     flprint (Printf.sprintf "%i, %s, %s, %i, %i, %s\n" i pages name w h cs)
+                     flprint (Printf.sprintf "%i, %s, %s, %i, %i, %i, %i, %s, %s\n" i pages name w h size bpc cs filter)
                  | _ -> ())
                 l
           | _ -> ()
diff --git a/cpdfimage.ml b/cpdfimage.ml
index a8829e3..0f6372c 100644
--- a/cpdfimage.ml
+++ b/cpdfimage.ml
@@ -348,8 +348,8 @@ let images pdf range =
         begin match xobject with
         | Pdf.Indirect i ->
             begin match Hashtbl.find images i with
-            | (pagenums, n, w, h, cs) ->
-                Hashtbl.replace images i (pagenum::pagenums, n, w, h, cs)
+            | (pagenums, n, w, h, s, bpc, cs, f) ->
+                Hashtbl.replace images i (pagenum::pagenums, n, w, h, s, bpc, cs, f)
             | exception Not_found ->
                 let width =
                   match Pdf.lookup_direct pdf "/Width" xobject with
@@ -359,12 +359,24 @@ let images pdf range =
                   match Pdf.lookup_direct pdf "/Height" xobject with
                   | Some x -> Pdf.getnum pdf x
                   | None -> 1.
+                and size =
+                  match Pdf.lookup_direct pdf "/Length" xobject with
+                  | Some (Pdf.Integer x) -> x
+                  | _ -> 0
+                and bpc =
+                  match Pdf.lookup_direct pdf "/BitsPerComponent" xobject with
+                  | Some (Pdf.Integer x) -> x
+                  | _ -> 0
                 and colourspace =
                   match Pdf.lookup_direct pdf "/ColorSpace" xobject with
                   | Some x -> Some (Pdfspace.string_of_colourspace (Pdfspace.read_colourspace pdf resources x))
                   | None -> None
+                and filter =
+                  match Pdf.lookup_direct pdf "/Filter" xobject with
+                  | Some (Pdf.Array [x]) | Some x -> Some (Pdfwrite.string_of_pdf x)
+                  | None -> None
                 in
-                  Hashtbl.replace images i ([pagenum], name, int_of_float width, int_of_float height, colourspace)
+                  Hashtbl.replace images i ([pagenum], name, int_of_float width, int_of_float height, size, bpc, colourspace, filter)
             end
         | _ -> ()
         end
@@ -397,17 +409,20 @@ let images pdf range =
         pdf
         range;
         let images = list_of_hashtbl images in
-        let images = map (fun (i, (pnums, n, w, h, c)) -> (i, (setify (sort compare pnums), n, w, h, c))) images in
-        let images = sort (fun (_, (pnums, _, _, _, _)) (_, (pnums', _, _, _, _)) -> compare (hd pnums) (hd pnums')) images in
+        let images = map (fun (i, (pnums, n, w, h, s, bpc, c, filter)) -> (i, (setify (sort compare pnums), n, w, h, s, bpc, c, filter))) images in
+        let images = sort (fun (_, (pnums, _, _, _, _, _, _, _)) (_, (pnums', _, _, _, _, _, _, _)) -> compare (hd pnums) (hd pnums')) images in
          `List
            (map
-             (fun (i, (pnums, n, w, h, cs)) ->
+             (fun (i, (pnums, n, w, h, size, bpc, cs, filter)) ->
                `Assoc [("Object", `Int i);
                        ("Pages", `List (map (fun x -> `Int x) pnums));
                        ("Name", `String n);
                        ("Width", `Int w);
                        ("Height", `Int h);
-                       ("Colourspace", match cs with None -> `Null | Some s -> `String s)])
+                       ("Bytes", `Int size);
+                       ("BitsPerComponent", `Int bpc);
+                       ("Colourspace", match cs with None -> `Null | Some s -> `String s);
+                       ("Filter", match filter with None -> `Null | Some s -> `String s)])
              images)
 
 let obj_of_jpeg_data data =
diff --git a/cpdfimage.mli b/cpdfimage.mli
index 0adda57..df25a00 100644
--- a/cpdfimage.mli
+++ b/cpdfimage.mli
@@ -1,9 +1,9 @@
 (** Images *)
 
 (** Extract images. *)
-val extract_images : ?raw:bool -> ?path_to_p2p:string ->
-           ?path_to_im:string ->
-           Cpdfmetadata.encoding -> bool -> bool -> Pdf.t -> int list -> string -> unit
+val extract_images :
+  ?raw:bool -> ?path_to_p2p:string -> ?path_to_im:string ->
+  Cpdfmetadata.encoding -> bool -> bool -> Pdf.t -> int list -> string -> unit
 
 (** Report image resolutions. *)        
 val image_resolution : Pdf.t -> int list -> float -> (int * string * int * int * float * float) list