From fa831170c3b60b15048c6219ddc6b7bbd7362512 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Mon, 23 Sep 2024 15:37:27 +0100
Subject: [PATCH] Redacting with process_struct_trees

---
 cpdfcommand.ml   |  2 +-
 cpdfpage.ml      | 29 ++++++++++++++++-------------
 cpdfpage.mli     |  2 +-
 cpdfpagespec.mli |  2 ++
 4 files changed, 20 insertions(+), 15 deletions(-)

diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index a726490..07be95e 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -4650,7 +4650,7 @@ let go () =
   | Some Redact ->
       let pdf = get_single_pdf args.op false in
       let range = parse_pagespec_allow_empty pdf (get_pagespec ()) in
-        write_pdf false (Cpdfpage.redact pdf range)
+        write_pdf false (Cpdfpage.redact ~process_struct_tree:args.process_struct_trees pdf range)
 
 (* Advise the user if a combination of command line flags makes little sense,
 or error out if it make no sense at all. *)
diff --git a/cpdfpage.ml b/cpdfpage.ml
index 4aae5e6..c53f86a 100644
--- a/cpdfpage.ml
+++ b/cpdfpage.ml
@@ -996,17 +996,20 @@ let copy_box f t mediabox_if_missing pdf range =
     pdf
     range
 
-(* TODO: Trim structure tree *)
-let redact pdf range =
-  process_pages
-    (Pdfpage.ppstub
-       (fun pnum page ->
-          if mem pnum range then
-            {page with
-               Pdfpage.content = [];
-               Pdfpage.resources = Pdf.Dictionary [];
-               Pdfpage.rest = Pdf.remove_dict_entry page.Pdfpage.rest "/Annots"}
-          else
-            page))
+let redact ~process_struct_tree pdf range =
+  let pdf =
+    process_pages
+      (Pdfpage.ppstub
+         (fun pnum page ->
+            if mem pnum range then
+              {page with
+                 Pdfpage.content = [];
+                 Pdfpage.resources = Pdf.Dictionary [];
+                 Pdfpage.rest = Pdf.remove_dict_entry page.Pdfpage.rest "/Annots"}
+            else
+              page))
+      pdf
+      range
+  in
+    if process_struct_tree then Pdfst.trim_structure_tree pdf (Cpdfpagespec.invert_range (Pdfpage.endpage pdf) range);
     pdf
-    range
diff --git a/cpdfpage.mli b/cpdfpage.mli
index 243ec16..02c4cf1 100644
--- a/cpdfpage.mli
+++ b/cpdfpage.mli
@@ -131,4 +131,4 @@ val alluprightonly : int list -> Pdf.t -> bool
 (** When a page is transformed, its patterns must be too. *)
 val change_pattern_matrices_page : Pdf.t -> Pdftransform.transform_matrix -> Pdfpage.t -> Pdfpage.t
 
-val redact : Pdf.t -> int list -> Pdf.t
+val redact : process_struct_tree:bool -> Pdf.t -> int list -> Pdf.t
diff --git a/cpdfpagespec.mli b/cpdfpagespec.mli
index c3d92fb..ab25f93 100644
--- a/cpdfpagespec.mli
+++ b/cpdfpagespec.mli
@@ -32,3 +32,5 @@ val validate_pagespec : string -> bool
 (** Return a string for the given range. Knows how to identify all, odd, even,
 x-y ranges etc. *)
 val string_of_pagespec : Pdf.t -> int list -> string
+
+val invert_range : int -> int list -> int list