From 59178af85c1ed995dd07c644f2a5fe2da4ed9a1c Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Mon, 3 Jun 2024 16:29:51 +0100
Subject: [PATCH] First plausible struct tree extraction

---
 cpdfua.ml | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)

diff --git a/cpdfua.ml b/cpdfua.ml
index ad571d9..f5c571d 100644
--- a/cpdfua.ml
+++ b/cpdfua.ml
@@ -331,4 +331,20 @@ let mark pdf =
            pdf.Pdf.root <- pdf3.Pdf.root
    | None -> assert false
 
-let extract_struct_tree pdf = `String ""
+let extract_struct_tree pdf =
+  match Pdf.lookup_obj pdf pdf.Pdf.root with
+  | Pdf.Dictionary d ->
+      begin match lookup "/StructTreeRoot" d with
+      | None -> `List []
+      | Some x ->
+          let objs = Pdf.objects_referenced ["/Pg"] [] pdf x in
+            `List
+               (map
+                  (fun objnum ->
+                     let jsonobj =
+                       Cpdfjson.json_of_object ~utf8:true ~no_stream_data:false ~parse_content:false pdf (function _ -> ()) (Pdf.lookup_obj pdf objnum)
+                     in
+                       `Tuple [`Int objnum; jsonobj])
+                  objs)
+      end
+  | _ -> error "extract_struct_tree: no root"