From c8006e8cdb2473e937125ed02a3748fe24177323 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Sat, 21 Sep 2024 18:44:43 +0100
Subject: [PATCH] Very basic redaction

---
 Changes        |  1 +
 cpdfcommand.ml |  9 ++++++++-
 cpdfmanual.tex |  1 +
 cpdfpage.ml    | 15 +++++++++++++++
 cpdfpage.mli   |  2 ++
 5 files changed, 27 insertions(+), 1 deletion(-)

diff --git a/Changes b/Changes
index 5fac41b..0b33baa 100644
--- a/Changes
+++ b/Changes
@@ -16,6 +16,7 @@ New features:
 * -typeset can make PDF/UA documents
 * -jpeg, -png and friends can make PDF/UA documents
 o -stretch scales without preserving aspect ratio
+o -redact removes whole pages
 
 * = Supported by a grant from NLnet
 
diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index fbd2399..a726490 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -228,6 +228,7 @@ type op =
   | ExtractStructTree
   | ReplaceStructTree of string
   | SetLanguage of string
+  | Redact
 
 let string_of_op = function
   | PrintFontEncoding _ -> "PrintFontEncoding"
@@ -379,6 +380,7 @@ let string_of_op = function
   | ExtractStructTree -> "ExtractStructTree"
   | ReplaceStructTree _ -> "ReplaceStructTree"
   | SetLanguage _ -> "SetLanguage"
+  | Redact -> "Redact"
 
 (* Inputs: filename, pagespec. *)
 type input_kind = 
@@ -925,7 +927,7 @@ let banned banlist = function
   | Decrypt | Encrypt | CombinePages _ -> true (* Never allowed *)
   | AddBookmarks _ | PadBefore | PadAfter | PadEvery _ | PadMultiple _ | PadMultipleBefore _
   | Merge | Split | SplitOnBookmarks _ | SplitMax _ | Spray | RotateContents _ | Rotate _
-  | Rotateby _ | Upright | VFlip | HFlip | Impose _ | Chop _ | ChopHV _ ->
+  | Rotateby _ | Upright | VFlip | HFlip | Impose _ | Chop _ | ChopHV _ | Redact ->
       mem Pdfcrypt.NoAssemble banlist
   | TwoUp | TwoUpStack | RemoveBookmarks | AddRectangle | RemoveText|
     Draft | Shift | ShiftBoxes | Scale | ScaleToFit|Stretch|RemoveAttachedFiles|
@@ -2907,6 +2909,7 @@ let specs =
    ("-print-struct-tree", Arg.Unit (fun () -> setop PrintStructTree ()), " Print structure tree");
    ("-extract-struct-tree", Arg.Unit (fun () -> setop ExtractStructTree ()), " Extract structure tree in JSON format");
    ("-replace-struct-tree", Arg.String (fun s -> setop (ReplaceStructTree s) ()), " Replace structure tree from JSON");
+   ("-redact", Arg.Unit (fun () -> setop Redact ()), " Redact entire pages");
    (* These items are undocumented *)
    ("-debug", Arg.Unit setdebug, "");
    ("-debug-crypt", Arg.Unit (fun () -> args.debugcrypt <- true), "");
@@ -4644,6 +4647,10 @@ let go () =
       let pdf = get_single_pdf args.op false in
         Cpdfmetadata.set_language pdf s;
         write_pdf false pdf
+  | 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)
 
 (* 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/cpdfmanual.tex b/cpdfmanual.tex
index 74ab3eb..78b4c3e 100644
--- a/cpdfmanual.tex
+++ b/cpdfmanual.tex
@@ -1,4 +1,5 @@
 %Document -stretch
+%Document -redact
 \documentclass{book}
 % Edit here to produce cpdfmanual.pdf, cpdflibmanual.pdf, pycpdfmanual.pdf,
 % dotnetcpdflibmanual.pdf, jcpdflibmanual.pdf jscpdflibmanual.pdf etc.
diff --git a/cpdfpage.ml b/cpdfpage.ml
index 83e6451..4aae5e6 100644
--- a/cpdfpage.ml
+++ b/cpdfpage.ml
@@ -995,3 +995,18 @@ let copy_box f t mediabox_if_missing pdf range =
                else page))
     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))
+    pdf
+    range
diff --git a/cpdfpage.mli b/cpdfpage.mli
index e139882..243ec16 100644
--- a/cpdfpage.mli
+++ b/cpdfpage.mli
@@ -130,3 +130,5 @@ 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