From 8c2440fb7a183ce3a6fe858ba5ff5485f648704b Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Mon, 16 Sep 2024 16:37:12 +0100
Subject: [PATCH] Added manual artifact control + turn off auto

---
 cpdfcommand.ml      |  3 +++
 cpdfdraw.ml         | 31 +++++++++----------------------
 cpdfdraw.mli        |  4 ++++
 cpdfdrawcontrol.ml  |  4 ++++
 cpdfdrawcontrol.mli |  3 +++
 5 files changed, 23 insertions(+), 22 deletions(-)

diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index 801f323..34db148 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -2810,6 +2810,9 @@ let specs =
    ("-end-stag", Arg.Unit Cpdfdrawcontrol.endstag, " End structure branch");
    ("-auto-tags", Arg.Unit (fun _ -> Cpdfdrawcontrol.autotags true), " Auto-tag paragraphs and figures");
    ("-no-auto-tags", Arg.Unit (fun _ -> Cpdfdrawcontrol.autotags false), " Don't auto-tag paragraphs and figures");
+   ("-artifact", Arg.Unit (fun _ -> Cpdfdrawcontrol.artifact ()), " Begin an artifact");
+   ("-end-artifact", Arg.Unit (fun _ -> Cpdfdrawcontrol.endartifact ()), "End an artifact");
+   ("-no-auto-artifacts", Arg.Unit (fun _ -> Cpdfdrawcontrol.autoartifacts false), " Don't mark untagged content as artifacts");
    ("-rect", Arg.String Cpdfdrawcontrol.addrect, " Draw rectangle");
    ("-to", Arg.String Cpdfdrawcontrol.addto, " Move to");
    ("-line", Arg.String Cpdfdrawcontrol.addline, " Add line to");
diff --git a/cpdfdraw.ml b/cpdfdraw.ml
index d018b4b..87380c8 100644
--- a/cpdfdraw.ml
+++ b/cpdfdraw.ml
@@ -1,6 +1,8 @@
 open Pdfutil
 open Cpdferror
 
+let do_add_artifacts = ref false
+
 type colspec =
    NoCol
  | RGB of float * float * float
@@ -56,6 +58,8 @@ type drawops =
   | Rise of float
   | STag of string
   | EndSTag
+  | BeginArtifact
+  | EndArtifact
 
 (*let rec string_of_drawop = function
   | Qq o -> "Qq (" ^ string_of_drawops o ^ ")"
@@ -418,6 +422,10 @@ let rec ops_of_drawop struct_tree dryrun pdf endpage filename bates batespad num
   | EndSTag ->
       structdata =| StDataEndTree;
       []
+  | BeginArtifact ->
+      [Pdfops.Op_BMC "/Artifact"]
+  | EndArtifact ->
+      [Pdfops.Op_EMC]
 
 and ops_of_drawops struct_tree dryrun pdf endpage filename bates batespad num page drawops =
   flatten (map (ops_of_drawop struct_tree dryrun pdf endpage filename bates batespad num page) drawops)
@@ -562,28 +570,6 @@ type st =
  | StItem of {kind : string; pageobjnum : int option; alt : string option; children : st list}
 
 (* Build a tree from the MCIDs and structure tree instructions gathered *)
-(*let make_structure_tree pdf items =
-  (* Make map of page numbers to pageobjnums, and create a reference to keep track. *)
-  let pagenum = ref 0 in
-  let items_out = ref [] in
-  let pageobjnums =
-    let objnums = Pdf.page_reference_numbers pdf in
-      combine (indx objnums) objnums
-  in
-  (* Process the items, making the st list tree data structure *)
-  let process = function
-    | StDataMCID (n, mcid, alt) ->
-        items_out =| StItem {kind = n; alt; pageobjnum = lookup !pagenum pageobjnums; children = [StMCID mcid]}
-    | StDataPage n ->
-        pagenum := n
-    | StDataBeginTree s ->
-        ()
-    | StDataEndTree ->
-        ()
-  in
-    iter process items;
-    !items_out*)
-
 let rec make_structure_tree pageobjnums pdf pagenum = function
   | [] -> []
   | StDataMCID (n, mcid, alt)::t ->
@@ -659,6 +645,7 @@ let write_structure_tree pdf st =
 
 let draw ~struct_tree ~fast ~underneath ~filename ~bates ~batespad range pdf drawops =
   (*Printf.printf "%s\n" (string_of_drawops drawops);*)
+  do_add_artifacts := struct_tree;
   resstack := [empty_res ()];
   Hashtbl.clear !fontpacks;
   (res ()).time <- Cpdfstrftime.current_time ();
diff --git a/cpdfdraw.mli b/cpdfdraw.mli
index 0208a62..759a72a 100644
--- a/cpdfdraw.mli
+++ b/cpdfdraw.mli
@@ -55,6 +55,10 @@ type drawops =
   | Rise of float
   | STag of string
   | EndSTag
+  | BeginArtifact
+  | EndArtifact
+
+val do_add_artifacts : bool ref
 
 (** Calling [draw fast underneath filename bates batespad range pdf drawops] draws on
     top of all the pages in the range. *)
diff --git a/cpdfdrawcontrol.ml b/cpdfdrawcontrol.ml
index b3ea128..806e2e5 100644
--- a/cpdfdrawcontrol.ml
+++ b/cpdfdrawcontrol.ml
@@ -102,6 +102,10 @@ let endtag () = ()
 let endstag () =
   addop Cpdfdraw.EndSTag
 let autotags b = ()
+let autoartifacts b =
+  Cpdfdraw.do_add_artifacts := b
+let artifact () = ()
+let endartifact () = ()
 
 let addrect s =
   let x, y, w, h = Cpdfcoord.parse_rectangle (Pdf.empty ()) s in
diff --git a/cpdfdrawcontrol.mli b/cpdfdrawcontrol.mli
index 789fba7..dac3f48 100644
--- a/cpdfdrawcontrol.mli
+++ b/cpdfdrawcontrol.mli
@@ -19,6 +19,9 @@ val addstag : string -> unit
 val endtag : unit -> unit
 val endstag : unit -> unit
 val autotags : bool -> unit
+val artifact : unit -> unit
+val endartifact : unit -> unit
+val autoartifacts : bool -> unit
 val addrect : string -> unit
 val addto : string -> unit
 val addline : string -> unit