From 4cb6fca6af2fdaf2f7b07e99a2af0d65cff60e44 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Thu, 3 Oct 2024 12:32:26 +0100
Subject: [PATCH] Return tags from typesetter

---
 cpdftexttopdf.ml |  2 +-
 cpdftoc.ml       |  2 +-
 cpdftype.ml      | 11 ++++++++---
 cpdftype.mli     |  2 +-
 4 files changed, 11 insertions(+), 6 deletions(-)

diff --git a/cpdftexttopdf.ml b/cpdftexttopdf.ml
index dc0567d..90cc306 100644
--- a/cpdftexttopdf.ml
+++ b/cpdftexttopdf.ml
@@ -92,6 +92,6 @@ let typeset ~process_struct_tree ?subformat ?title ~papersize ~font ~fontsize te
       let firstfont = hd (keep (function Cpdftype.Font _ -> true | _ -> false) tagged) in 
         [firstfont; Cpdftype.BeginDocument] @ tagged
   in
-  let pages = Cpdftype.typeset ~process_struct_tree margin margin margin margin papersize pdf instrs in
+  let pages, tags = Cpdftype.typeset ~process_struct_tree margin margin margin margin papersize pdf instrs in
     let pdf, pageroot = Pdfpage.add_pagetree pages pdf in
       Pdfpage.add_root pageroot [] pdf
diff --git a/cpdftoc.ml b/cpdftoc.ml
index 81d818f..43c8319 100644
--- a/cpdftoc.ml
+++ b/cpdftoc.ml
@@ -156,7 +156,7 @@ let typeset_table_of_contents ~font ~fontsize ~title ~bookmark pdf =
             Cpdftype.NewLine])
       (Pdfmarks.read_bookmarks pdf)
   in
-  let toc_pages =
+  let toc_pages, _ =
     let title =
       let glue = Cpdftype.VGlue (fontsize *. 2.) in
         if title = "" then [] else
diff --git a/cpdftype.ml b/cpdftype.ml
index 2a331e1..4fc061a 100644
--- a/cpdftype.ml
+++ b/cpdftype.ml
@@ -263,6 +263,8 @@ let typeset ~process_struct_tree lmargin rmargin tmargin bmargin papersize pdf i
   let thispageannotations = ref [] in
   let thisdestrectangles = ref [] in
   let pages = ref [] in
+  let tags = ref [] in
+  let tagsout = ref [] in
   let write_page () =
     let ops = if process_struct_tree then add_artifacts (rev !ops) else rev !ops in
     let page =
@@ -272,7 +274,8 @@ let typeset ~process_struct_tree lmargin rmargin tmargin bmargin papersize pdf i
        Pdfpage.rotate = Pdfpage.Rotate0;
        Pdfpage.rest = make_annotations pdf !thispageannotations}
     in
-      pages := page :: !pages
+      pages := page::!pages;
+      tagsout := rev !tags::!tagsout
   in
   let rec typeset_element = function
     | Text cps ->
@@ -340,9 +343,11 @@ let typeset ~process_struct_tree lmargin rmargin tmargin bmargin papersize pdf i
             thispageannotations := map annot !thisdestrectangles @ !thispageannotations;
         s.dest <- None;
         thisdestrectangles := []
-   | Tag (s, _) -> ops := Pdfops.Op_BDC ("/" ^ s, Pdf.Dictionary [("/MCID", Pdf.Integer (mcid ()))])::!ops
+   | Tag (s, i) ->
+       tags := (s, i)::!tags;
+       ops := Pdfops.Op_BDC ("/" ^ s, Pdf.Dictionary [("/MCID", Pdf.Integer (mcid ()))])::!ops
    | EndTag -> ops := Pdfops.Op_EMC::!ops
   in
     iter typeset_element i;
     write_page ();
-    rev !pages
+    (rev !pages, rev !tagsout)
diff --git a/cpdftype.mli b/cpdftype.mli
index a0cd39c..8946bf6 100644
--- a/cpdftype.mli
+++ b/cpdftype.mli
@@ -24,6 +24,6 @@ val font_widths : string -> Pdftext.font -> float -> float array
 val width_of_string : float array -> char list -> float
 
 (** [typeset lmargin rmargin tmargin bmargin papersize pdf contents] builds a list of pages of typset content. *)
-val typeset : process_struct_tree:bool -> float -> float -> float -> float -> Pdfpaper.t -> Pdf.t -> t -> Pdfpage.t list
+val typeset : process_struct_tree:bool -> float -> float -> float -> float -> Pdfpaper.t -> Pdf.t -> t -> Pdfpage.t list * (string * int) list list
 
 val add_artifacts : Pdfops.t list -> Pdfops.t list