From be610916d421c654289dd056670fd224a687b006 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Wed, 19 Oct 2022 14:36:26 +0100
Subject: [PATCH] more

---
 cpdfcommand.ml   |  4 ++--
 cpdfembed.ml     |  2 +-
 cpdftexttopdf.ml | 30 ++++++++++++++++++++----------
 3 files changed, 23 insertions(+), 13 deletions(-)

diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index 4c28f80..603c4af 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -3068,7 +3068,7 @@ let embed_font () =
               Pdfio.bytes_of_string (contents_of_file (Filename.concat dirname filename)),
               Filename.remove_extension filename
             in
-            Cpdfembed.EmbedInfo {fontfile; fontname; fontencoding = args.fontencoding}
+            Cpdfembed.EmbedInfo {fontfile; fontname; encoding = args.fontencoding}
           with
             e -> error (Printf.sprintf "Can't load font for embedding: %s\n" (Printexc.to_string e))
           end
@@ -3078,7 +3078,7 @@ let embed_font () =
     | OtherFont f ->
         ExistingNamedFont f
     | FontToEmbed fontfile ->
-        EmbedInfo {fontfile; fontname = args.fontname; fontencoding = args.fontencoding}
+        EmbedInfo {fontfile; fontname = args.fontname; encoding = args.fontencoding}
 
 (* Main function *)
 let go () =
diff --git a/cpdfembed.ml b/cpdfembed.ml
index 6643012..d483fed 100644
--- a/cpdfembed.ml
+++ b/cpdfembed.ml
@@ -1,7 +1,7 @@
 (* Embed a font *)
 open Pdfutil
 
-type t = Pdftext.font list * (int, int * int) Hashtbl.t
+type t = Pdftext.font list * (int, int * int) Hashtbl.t (* Table returns font number and charcode for given unicode codepoint *)
 
 type cpdffont =
   PreMadeFontPack of t
diff --git a/cpdftexttopdf.ml b/cpdftexttopdf.ml
index e7d1894..99014c0 100644
--- a/cpdftexttopdf.ml
+++ b/cpdftexttopdf.ml
@@ -1,14 +1,25 @@
 open Pdfutil
 
-let rec of_utf8_with_newlines used charcode_extractor t =
+(* Return set of unicode characters in this text *)
+let used_characters t =
+  let codepoints = Pdftext.codepoints_of_utf8 t in
+    setify codepoints
+
+(* Just first font, expand later. Move into cpdfembed? *)
+let get_char (fonts, table) u =
+  match Hashtbl.find table u with
+  | (n, charcode) -> Some charcode
+  | exception Not_found -> None
+
+let rec of_utf8_with_newlines fontpack t =
   let items = ref [] in
   let buf = ref [] in
   let codepoints = Pdftext.codepoints_of_utf8 t in
   let charcodes_of_codepoints cs =
     option_map
       (fun u ->
-         match charcode_extractor u with
-         | Some c -> Hashtbl.replace used c (); Some (char_of_int c)
+         match get_char fontpack u with
+         | Some c -> Some (char_of_int c)
          | None -> Printf.printf "No glyph for unicode U+%04X in this font\n" u; None)
       cs
   in
@@ -30,21 +41,20 @@ let rec of_utf8_with_newlines used charcode_extractor t =
 
 let typeset ~papersize ~font ~fontsize text =
   let pdf = Pdf.empty () in
-  let font =
+  let codepoints = used_characters (Pdfio.string_of_bytes text) in
+  let font, fontpack =
     match font with
-    | Cpdfembed.PreMadeFontPack t -> hd (fst t)
+    | Cpdfembed.PreMadeFontPack t -> (hd (fst t), t)
     | Cpdfembed.EmbedInfo {fontfile; fontname; encoding} ->
-        hd (fst (Cpdfembed.embed_truetype pdf ~fontfile ~fontname ~codepoints ~encoding))
+        let embedded = Cpdfembed.embed_truetype pdf ~fontfile ~fontname ~codepoints ~encoding in
+         (hd (fst embedded), embedded)
     | Cpdfembed.ExistingNamedFont _ -> raise (Pdf.PDFError "Can't use existing named font for text-to-PDF")
   in
-  let charcode_extractor = Pdftext.charcode_extractor_of_font_real font in
+  let instrs = of_utf8_with_newlines fontpack (Pdfio.string_of_bytes text) in
   let margin =
     Pdfunits.convert
       72. (Pdfpaper.unit papersize) (Pdfunits.PdfPoint) (Pdfpaper.width papersize) /. 15.
   in
-  let used = null_hash () in
-  let instrs = of_utf8_with_newlines used charcode_extractor (Pdfio.string_of_bytes text) in
-  let codepoints = map fst (list_of_hashtbl used) in
   let pages =
     Cpdftype.typeset
       margin margin margin margin papersize pdf