From 5852c184a8eb1a88c8dedd0f794809323015ccd3 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Tue, 11 Jul 2023 16:53:21 +0100
Subject: [PATCH] Rehabilitating toc

---
 cpdftoc.ml      | 29 +++++++++++++++++++++++------
 cpdftruetype.ml |  4 ++--
 2 files changed, 25 insertions(+), 8 deletions(-)

diff --git a/cpdftoc.ml b/cpdftoc.ml
index c89b40b..2d447ba 100644
--- a/cpdftoc.ml
+++ b/cpdftoc.ml
@@ -14,16 +14,33 @@ let rec real_newline = function
   | x::r -> x::real_newline r
   | [] -> []
 
+let rec collate_runs cfn a = function
+  | [] -> rev (map rev a)
+  | (charcode, fontnum, font) as h::t ->
+      match a with
+      | [] -> collate_runs fontnum [[h]] t
+      | this::rest ->
+          if fontnum = cfn
+            then collate_runs cfn ((h::this)::rest) t
+            else collate_runs fontnum ([h]::this::rest) t
+
+let collate_runs = function
+  | [] -> []
+  | (_, fontnum, _)::_ as l -> collate_runs fontnum [] l
+
 (* Run of Font / Text elements from a fontpack and UTF8 text *)
 let of_utf8 fontpack fontsize t =
-  (* Use the fontpack to find which font is required, then return runs
-  of [Font (font, size); Text t] in a list *)
   let codepoints = Pdftext.codepoints_of_utf8 t in
-  (* 1. Text to (charcode, font, fontnum) *)
   let fonted = option_map (Cpdfembed.get_char fontpack) codepoints in
-  (* 2. FIXME Collate them *)
-  (* 3. Produce the runs, converting to char *)
-  flatten (map (function (charcode, fontnum, font) -> [Cpdftype.Font (font, fontsize); Cpdftype.Text [char_of_int charcode]]) fonted)
+  let collated = collate_runs fonted in
+    flatten
+      (map
+        (function
+         | [] -> []
+         | (_, _, font) as h::t ->
+             let charcodes = map (fun (c, _, _) -> char_of_int c) (h::t) in
+               [Cpdftype.Font (font, fontsize); Cpdftype.Text charcodes])
+        collated)
 
 (* Cpdftype codepoints from a font and PDFDocEndoding string *)
 let of_pdfdocencoding fontpack fontsize t =
diff --git a/cpdftruetype.ml b/cpdftruetype.ml
index 27e5482..91710af 100644
--- a/cpdftruetype.ml
+++ b/cpdftruetype.ml
@@ -289,7 +289,7 @@ let write_glyf_table subset cmap bs mk_b glyfoffset loca =
          with
            Not_found -> ())
       subset;
-  let locnums = expand_composites mk_b loca glyfoffset (sort compare (map fst (list_of_hashtbl locnums))) in
+  let locnums = (*expand_composites mk_b loca glyfoffset*) (sort compare (map fst (list_of_hashtbl locnums))) in
   (*if !dbg then*)
     (Printf.printf "We want glyfs for locations: ";
      iter (Printf.printf "%i ") locnums; Printf.printf "\n");
@@ -680,7 +680,7 @@ let parse ~subset data encoding =
                 write_font "one.ttf" one.subset_fontfile;
                 (*Printf.printf "\nHigher subset:\n";
                 debug_t (hd twos);*)
-                write_font "two.ttf" (hd twos).subset_fontfile;
+                if twos <> [] then write_font "two.ttf" (hd twos).subset_fontfile;
                 one::twos
 
 let parse ~subset data encoding =