From 84014abf5cea21bf097dac47e557e876ce430647 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Fri, 17 Jan 2025 11:30:01 +0800
Subject: [PATCH] Work on 28_005

---
 cpdfimage.ml |  2 +-
 cpdfua.ml    | 37 ++++++++++++++++++++++++++++++++++++-
 2 files changed, 37 insertions(+), 2 deletions(-)

diff --git a/cpdfimage.ml b/cpdfimage.ml
index eef8200..3c78bf5 100644
--- a/cpdfimage.ml
+++ b/cpdfimage.ml
@@ -845,7 +845,7 @@ let recompress_1bpp_ccitt_lossless ~pixel_threshold ~length_threshold pdf s dict
   let old = !reference in
   let restore () = reference := old in
   let w = match Pdf.lookup_direct pdf "/Width" dict with Some (Pdf.Integer i) -> i | _ -> error "bad width" in
-  let h = match Pdf.lookup_direct pdf "/Height" dict with Some (Pdf.Integer i) -> i | _ -> error "bad height" in
+  (*let h = match Pdf.lookup_direct pdf "/Height" dict with Some (Pdf.Integer i) -> i | _ -> error "bad height" in*)
   (*if w * h < pixel_threshold then (if !debug_image_processing then Printf.printf "pixel threshold not met\n%!") else*)
   let size = match Pdf.lookup_direct pdf "/Length" dict with Some (Pdf.Integer i) -> i | _ -> 0 in
   (*if size < length_threshold then (if !debug_image_processing then Printf.printf "length threshold not met\n%!") else*)
diff --git a/cpdfua.ml b/cpdfua.ml
index f6bb5a3..8410b27 100644
--- a/cpdfua.ml
+++ b/cpdfua.ml
@@ -702,8 +702,43 @@ let matterhorn_28_004 _ _ pdf =
 (* A form field does not have a TU entry and does not have an alternative
    description (in the form of an Alt entry in the enclosing structure
    element). *)
+let get_field_object_numbers pdf =
+  let rec get_field_object_numbers_inner obj =
+    match obj with
+    | Pdf.Indirect i ->
+        (* Is this referenced item a field (rather than an annotation alone?).
+           If so, count it, and recurse on an /Kids. *)
+        begin match Pdf.lookup_direct pdf "/T" (Pdf.Indirect i) with
+        | Some _ ->
+            begin match Pdf.lookup_direct pdf "/Kids" (Pdf.Indirect i) with
+            | None -> [i]
+            | Some kids -> i::get_field_object_numbers_inner kids
+            end
+        | None -> []
+        end
+    | _ ->
+       Pdfe.log "get_field_object_numbers_inner: non-indirect Kid";
+       []
+  in
+    match Pdf.lookup_chain pdf pdf.Pdf.trailerdict ["/Root"; "/AcroForm"; "/Fields"] with
+    | Some (Pdf.Array toplevelfields) ->
+        flatten (map get_field_object_numbers_inner toplevelfields) 
+    | _ ->
+        []
+
 let matterhorn_28_005 _ _ pdf =
-  unimpl ()
+  let missing_tu =
+    option_map
+      (function objnum ->
+        match Pdf.lookup_direct pdf "/T" (Pdf.Indirect objnum) with
+        | Some _ -> None
+        | None -> Some objnum)
+      (get_field_object_numbers pdf)
+  in
+   if missing_tu = [] then () else
+     (* Check for alts in enclosing. We look for /StructParent (from merged annoation, not field) *)
+     let parent_tree = read_parent_tree pdf in
+     ()
 
 (* An annotation with subtype undefined in ISO 32000 does not meet 7.18.1. *)
 let matterhorn_28_006 _ _ pdf =