Report natural language

Changes

@@ -8,6 +8,7 @@ o Verify compliance to PDF/UA via the Matterhorn protocol
 o Extract, edit and reapply document structure tree
 o Split structure tree when splitting PDF to save size
 o Combine structure trees when stamping PDFs
+o Report natural language on -info
 
 2.7 (February 2024)
 

cpdfmetadata.ml

@@ -442,6 +442,11 @@ let determine_subformats pdf =
           end;
           !formats
 
+let language pdf =
+  match Pdf.lookup_chain pdf pdf.Pdf.trailerdict ["/Root"; "/Lang"] with
+  | Some (Pdf.String x) -> Some x
+  | _ -> None
+
 let output_xmp_info ?(json=ref [("none", `Null)]) encoding pdf =
   let notjson = !json = [("none", `Null)] in
   let print_out tree title namespace name =
@@ -459,6 +464,9 @@ let output_xmp_info ?(json=ref [("none", `Null)]) encoding pdf =
     if notjson
       then Printf.printf "Subformats: %s\n" (combine_with_commas (determine_subformats pdf))
       else json =| ("Subformats", `List (map (fun x -> `String x) (determine_subformats pdf)));
+    if notjson
+      then Printf.printf "Language: %s\n" (match language pdf with None -> "" | Some x -> "\"" ^ x ^ "\"")
+      else json =| ("Language", match language pdf with None -> `Null | Some x -> `String x);
     match get_metadata pdf with
       None -> ()
     | Some metadata ->

cpdfmetadata.mli

@@ -87,6 +87,8 @@ val get_viewer_pref_item : string -> Pdf.t -> string
 
 val determine_subformats : Pdf.t -> string list
 
+val language : Pdf.t -> string option
+
 val adobe : string
 val xmp : string
 val dc : string

cpdfua.ml

@@ -164,7 +164,8 @@ let matterhorn_10_001 pdf =
   unimpl ()
 
 (* Natural language for text in page content cannot be determined. *)
-let matterhorn_11_001 pdf = todo ()
+let matterhorn_11_001 pdf =
+  unimpl ()
 
 (* Natural language for text in Alt, ActualText and E attributes cannot be
    determined. *)