From 686318718df00cc58cc4bdf88d0f4814608a78c4 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Thu, 3 Oct 2024 16:18:09 +0100
Subject: [PATCH] More structure tree work

---
 cpdftexttopdf.ml | 40 ++++++++++++++++++++++++++++++++--------
 cpdftype.ml      |  5 +++--
 2 files changed, 35 insertions(+), 10 deletions(-)

diff --git a/cpdftexttopdf.ml b/cpdftexttopdf.ml
index 9922761..e6789bd 100644
--- a/cpdftexttopdf.ml
+++ b/cpdftexttopdf.ml
@@ -94,6 +94,30 @@ let typeset ~process_struct_tree ?subformat ?title ~papersize ~font ~fontsize te
       let firstfont = hd (keep (function Cpdftype.Font _ -> true | _ -> false) tagged) in 
         [firstfont; Cpdftype.BeginDocument] @ tagged
   in
+  let pages, tags = Cpdftype.typeset ~process_struct_tree margin margin margin margin papersize pdf instrs in
+  iter (fun x -> Printf.printf "PAGE\n"; iter (fun (_, i) -> Printf.printf "Tag number %i\n" i) x) tags;
+  (* We make (tag number, page number, mcid) triples *)
+  let tagtriples =
+    flatten
+      (map2
+       (fun pn tags ->
+          map2 (fun (_, tagnum) mcid -> (tagnum, pn, mcid)) tags (indx0 tags))
+       (indx0 tags)
+       tags)
+  in
+    Printf.printf "(tag number, page number, mcid) triples:\n";
+    iter (fun (tagnum, pn, mcid) -> Printf.printf "%i, %i, %i\n" tagnum pn mcid) tagtriples;
+  (* Now work out the nodes and what each /K and /Pg in them is *)
+  let pages_and_mcids =
+    []
+  in
+    Printf.printf "Pages and their MCIDs\n";
+    iter
+      (fun (page, mcids) ->
+         Printf.printf "Page %i\n";
+         iter (Printf.printf "%i ") mcids;
+         Printf.printf "\n")
+      pages_and_mcids;
     if subformat = Some Cpdfua.PDFUA2 then
       begin
         let str = Pdf.addobj pdf Pdf.Null in
@@ -118,11 +142,11 @@ let typeset ~process_struct_tree ?subformat ?title ~papersize ~font ~fontsize te
         Pdf.addobj_given_num pdf (str, Pdf.Dictionary [("/Type", Pdf.Name "/StructTreeRoot"); ("/K", Pdf.Array [Pdf.Indirect p]); ("/ParentTree", Pdf.Indirect parent_tree)]);
         Pdf.replace_chain pdf ["/Root"] ("/StructTreeRoot", (Pdf.Indirect str))
       end;
-  let pages, tags = Cpdftype.typeset ~process_struct_tree margin margin margin margin papersize pdf instrs in
-  let pages =
-    map
-      (fun p -> if process_struct_tree then {p with Pdfpage.rest = Pdf.add_dict_entry p.Pdfpage.rest "/StructParents" (Pdf.Integer 1)} else p)
-      pages
-  in
-    let pdf, pageroot = Pdfpage.add_pagetree pages pdf in
-      Pdfpage.add_root pageroot [] pdf
+    let pages =
+      map2
+        (fun pn p -> if process_struct_tree then {p with Pdfpage.rest = Pdf.add_dict_entry p.Pdfpage.rest "/StructParents" (Pdf.Integer pn)} else p)
+        (indx0 pages)
+        pages
+    in
+      let pdf, pageroot = Pdfpage.add_pagetree pages pdf in
+        Pdfpage.add_root pageroot [] pdf
diff --git a/cpdftype.ml b/cpdftype.ml
index 4fc061a..552ff11 100644
--- a/cpdftype.ml
+++ b/cpdftype.ml
@@ -246,7 +246,7 @@ let typeset ~process_struct_tree lmargin rmargin tmargin bmargin papersize pdf i
   Hashtbl.clear width_table_cache;
   let debug = false in
   if debug then (print_endline "***input:\n\n"; print_endline (to_string i));
-  let i = number_tags 1 i in
+  let i = number_tags 0 i in
   let i = layout lmargin rmargin papersize i in
   if debug then (print_endline "***after layout:\n\n"; print_endline (to_string i));
   let i = paginate tmargin bmargin papersize i in
@@ -275,7 +275,8 @@ let typeset ~process_struct_tree lmargin rmargin tmargin bmargin papersize pdf i
        Pdfpage.rest = make_annotations pdf !thispageannotations}
     in
       pages := page::!pages;
-      tagsout := rev !tags::!tagsout
+      tagsout := rev !tags::!tagsout;
+      tags := []
   in
   let rec typeset_element = function
     | Text cps ->