From 9516784ece47fbcb719d875f76bc6b611a540843 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Mon, 6 Dec 2021 15:55:46 -0800
Subject: [PATCH] more

---
 Makefile          |  2 +-
 cpdfcommand.ml    | 46 ++++------------------------------------------
 cpdftexttopdf.ml  | 39 +++++++++++++++++++++++++++++++++++++++
 cpdftexttopdf.mli |  1 +
 cpdftoc.ml        |  1 -
 5 files changed, 45 insertions(+), 44 deletions(-)
 create mode 100644 cpdftexttopdf.ml
 create mode 100644 cpdftexttopdf.mli

diff --git a/Makefile b/Makefile
index b340916..f9cd91c 100644
--- a/Makefile
+++ b/Makefile
@@ -1,7 +1,7 @@
 # Build the cpdf command line tools and top level
 MODS = cpdfyojson cpdfxmlm \
        cpdfunicodedata cpdferror cpdfjson cpdfstrftime cpdfcoord cpdfattach \
-       cpdfpagespec cpdfposition cpdf cpdffont cpdftype cpdftoc cpdfcommand
+       cpdfpagespec cpdfposition cpdf cpdffont cpdftype cpdftexttopdf cpdftoc cpdfcommand
 
 SOURCES = $(foreach x,$(MODS),$(x).ml $(x).mli) cpdfcommandrun.ml
 
diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index d77a831..a40aeb8 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -2918,47 +2918,6 @@ let collate (names, pdfs, ranges) =
     done;
     split3 (rev !nis)
 
-let of_utf8 (f, fontsize) t =
-     Pdftext.codepoints_of_utf8 t
-  |> option_map (Pdftext.charcode_extractor_of_font_real f)
-  |> map char_of_int
-  |> implode
-
-let of_pdfdocencoding (f, fontsize) t =
-  of_utf8 (f, fontsize) (Pdftext.utf8_of_pdfdocstring t)
-
-let rec of_utf8_with_newlines t =
-  let items = ref [] in
-  let buf = Buffer.create 256 in
-    String.iter
-      (function
-       | '\n' ->
-           let c = Buffer.contents buf in
-             if c <> "" then items := Cpdftype.Text (explode c)::!items;
-             items := Cpdftype.NewLine::!items;
-             Buffer.clear buf
-       | x ->
-           Buffer.add_char buf x)
-      t;
-    (* Do last one *)
-    let c = Buffer.contents buf in
-      if c <> "" then items := Text (explode c)::!items;
-    rev !items
-
-(* FIXME margins, hyphenation of too-long words, efficiency *)
-let typeset text =
-  let pdf = Pdf.empty () in
-  let f = 
-    (begin match args.font with StandardFont sf -> Pdftext.StandardFont (sf, Pdftext.WinAnsiEncoding) | _ -> failwith "typeset bad font" end,
-     args.fontsize)
-  in
-  let pages =
-    Cpdftype.typeset
-      20. 20. 20. 20. Pdfpaper.a4 pdf ([Cpdftype.Font f] @ of_utf8_with_newlines (string_of_bytes text))
-  in
-    let pdf, pageroot = Pdfpage.add_pagetree pages pdf in
-      Pdfpage.add_root pageroot [] pdf
-
 (* Main function *)
 let go () =
   match args.op with
@@ -3888,7 +3847,10 @@ let go () =
         write_pdf false pdf
   | Some (Typeset filename) ->
       let text = Pdfio.bytes_of_input_channel (open_in filename) in
-      let pdf = typeset text in
+      let font =
+        match args.font with StandardFont f -> f | _ -> error "text to PDF: not a standard font"
+      in
+      let pdf = Cpdftexttopdf.typeset ~font ~fontsize:args.fontsize text in
         write_pdf false pdf
 
 (* Advise the user if a combination of command line flags makes little sense,
diff --git a/cpdftexttopdf.ml b/cpdftexttopdf.ml
new file mode 100644
index 0000000..020795b
--- /dev/null
+++ b/cpdftexttopdf.ml
@@ -0,0 +1,39 @@
+open Pdfutil
+
+let of_utf8 (f, fontsize) t =
+     Pdftext.codepoints_of_utf8 t
+  |> option_map (Pdftext.charcode_extractor_of_font_real f)
+  |> map char_of_int
+  |> implode
+
+let of_pdfdocencoding (f, fontsize) t =
+  of_utf8 (f, fontsize) (Pdftext.utf8_of_pdfdocstring t)
+
+let rec of_utf8_with_newlines t =
+  let items = ref [] in
+  let buf = Buffer.create 256 in
+    String.iter
+      (function
+       | '\n' ->
+           let c = Buffer.contents buf in
+             if c <> "" then items := Cpdftype.Text (explode c)::!items;
+             items := Cpdftype.NewLine::!items;
+             Buffer.clear buf
+       | x ->
+           Buffer.add_char buf x)
+      t;
+    (* Do last one *)
+    let c = Buffer.contents buf in
+      if c <> "" then items := Text (explode c)::!items;
+    rev !items
+
+(* FIXME margins, hyphenation of too-long words, efficiency *)
+let typeset ~font ~fontsize text =
+  let pdf = Pdf.empty () in
+  let f = (Pdftext.StandardFont (font, Pdftext.WinAnsiEncoding), fontsize) in
+  let pages =
+    Cpdftype.typeset
+      20. 20. 20. 20. Pdfpaper.a4 pdf ([Cpdftype.Font f] @ of_utf8_with_newlines (Pdfio.string_of_bytes text))
+  in
+    let pdf, pageroot = Pdfpage.add_pagetree pages pdf in
+      Pdfpage.add_root pageroot [] pdf
diff --git a/cpdftexttopdf.mli b/cpdftexttopdf.mli
new file mode 100644
index 0000000..5deb16a
--- /dev/null
+++ b/cpdftexttopdf.mli
@@ -0,0 +1 @@
+val typeset : font:Pdftext.standard_font -> fontsize:float -> Pdfio.bytes -> Pdf.t
diff --git a/cpdftoc.ml b/cpdftoc.ml
index bfcc6db..0fa4f94 100644
--- a/cpdftoc.ml
+++ b/cpdftoc.ml
@@ -93,4 +93,3 @@ let typeset_table_of_contents ~font ~fontsize ~title pdf =
   let labels' = label::map (fun l -> {l with Pdfpagelabels.startpage = l.Pdfpagelabels.startpage + toc_pages_len}) labels in
     Pdfpagelabels.write pdf labels';
     pdf
-