From a9204a03652046dd6c8579257673893e1ae183d3 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Thu, 13 Jul 2023 14:44:57 +0100
Subject: [PATCH] Move collate into cpdfembed, patch up -draw

---
 cpdfdraw.ml   | 28 ++++++++++++++++------------
 cpdfembed.ml  | 15 +++++++++++++++
 cpdfembed.mli |  2 ++
 cpdftoc.ml    | 16 +---------------
 4 files changed, 34 insertions(+), 27 deletions(-)

diff --git a/cpdfdraw.ml b/cpdfdraw.ml
index 3942267..80afcc6 100644
--- a/cpdfdraw.ml
+++ b/cpdfdraw.ml
@@ -136,17 +136,21 @@ let process_specials pdf endpage filename bates batespad num page s =
   in
     Cpdfaddtext.process_text (res ()).time s pairs
 
-(* FIXME: implement for other kinds of font *)
 let runs_of_utf8 s =
-  match (res ()).current_fontpack with
-  | ((f::_, _) as fontpack) ->
-      let codepoints = Pdftext.codepoints_of_utf8 s in
-      let charcodes = option_map (Cpdfembed.get_char fontpack) codepoints in
-      let fontname =
-        fst (Hashtbl.find (res ()).fonts f)
-      in
-        [Pdfops.Op_Tf (fontname, (res ()).font_size); Pdfops.Op_Tj (implode (map (fun (c, _, _) -> char_of_int c) charcodes))]
-  | _ -> failwith "charcodes_of_utf8: unknown font"
+  let fontpack = (res ()).current_fontpack in
+  let codepoints = Pdftext.codepoints_of_utf8 s in
+  let triples = option_map (Cpdfembed.get_char fontpack) codepoints in
+  let collated = Cpdfembed.collate_runs triples in
+    flatten
+     (map
+       (fun l ->
+         if l = [] then [] else
+           let f = match l with (_, _, f)::_ -> f | _ -> assert false in
+           let fontname = fst (Hashtbl.find (res ()).fonts f) in
+           let charcodes = map (fun (c, _, _) -> char_of_int c) l in
+             [Pdfops.Op_Tf (fontname, (res ()).font_size);
+              Pdfops.Op_Tj (implode charcodes)])
+      collated)
 
 let extgstate kind v =
   try Hashtbl.find (res ()).extgstates (kind, v) with
@@ -239,9 +243,9 @@ let rec ops_of_drawop pdf endpage filename bates batespad num page = function
         match cpdffont with
         | PreMadeFontPack fp -> fp
         | EmbedInfo {fontfile; fontname; encoding} ->
-            Cpdfembed.embed_truetype pdf ~fontfile ~fontname ~codepoints:[int_of_char 'a'] ~encoding
+            Cpdfembed.embed_truetype pdf ~fontfile ~fontname ~codepoints:[int_of_char 'H'] ~encoding
         | ExistingNamedFont ->
-            error "-draw does not support using an exsiting named font"
+            error "-draw does not support using an existing named font"
       in
       let ns =
         map
diff --git a/cpdfembed.ml b/cpdfembed.ml
index 0840ca6..7f3d54b 100644
--- a/cpdfembed.ml
+++ b/cpdfembed.ml
@@ -92,3 +92,18 @@ let embed_truetype pdf ~fontfile ~fontname ~codepoints ~encoding =
   let fs = Cpdftruetype.parse ~subset:codepoints fontfile encoding in
   let subsets_and_their_fonts = map (make_single_font ~fontname ~encoding pdf) fs in
     (map snd subsets_and_their_fonts, make_fontpack_hashtable subsets_and_their_fonts)
+
+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
+
diff --git a/cpdfembed.mli b/cpdfembed.mli
index 4592a51..7cdadd7 100644
--- a/cpdfembed.mli
+++ b/cpdfembed.mli
@@ -16,3 +16,5 @@ val get_char : t -> int -> (int * int * Pdftext.font) option
 val embed_truetype :
   Pdf.t -> fontfile:Pdfio.bytes -> fontname:string -> codepoints:int list ->
   encoding:Pdftext.encoding -> t
+
+val collate_runs : ('a * 'b * 'c) list -> ('a * 'b * 'c) list list
diff --git a/cpdftoc.ml b/cpdftoc.ml
index efd326a..e4bb34a 100644
--- a/cpdftoc.ml
+++ b/cpdftoc.ml
@@ -14,20 +14,6 @@ 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
-
 let rec width_of_runs runs =
   match runs with
   | Cpdftype.Font (f, fontsize)::Cpdftype.Text t::more ->
@@ -39,7 +25,7 @@ let rec width_of_runs runs =
 let of_utf8 fontpack fontsize t =
   let codepoints = Pdftext.codepoints_of_utf8 t in
   let fonted = option_map (Cpdfembed.get_char fontpack) codepoints in
-  let collated = collate_runs fonted in
+  let collated = Cpdfembed.collate_runs fonted in
     flatten
       (map
         (function