From ce2ff087b82bc787f6e37e85d2a4c2a06cbc676c Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Wed, 29 Dec 2021 15:58:03 +0000
Subject: [PATCH] more

---
 cpdfcommand.ml    | 2 +-
 cpdfjson.ml       | 5 +++--
 cpdftexttopdf.ml  | 9 ++++++---
 cpdftexttopdf.mli | 2 +-
 4 files changed, 11 insertions(+), 7 deletions(-)

diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index 06b7d1d..639ecac 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -3917,7 +3917,7 @@ let go () =
       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
+      let pdf = Cpdftexttopdf.typeset ~papersize:args.createpdf_pagesize ~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/cpdfjson.ml b/cpdfjson.ml
index 8b85e3d..3162c32 100644
--- a/cpdfjson.ml
+++ b/cpdfjson.ml
@@ -465,8 +465,9 @@ let rec ppstring_single_object pdf = function
   | x -> x
 
 let preprocess_strings pdf =
-    Pdf.objselfmap (ppstring_single_object pdf) pdf;
-    pdf.Pdf.trailerdict <- ppstring_single_object pdf pdf.Pdf.trailerdict
+    Pdf.objselfmap (ppstring_single_object pdf) pdf
+    (* Skip the trailer dictionary since may mess up /ID if it happens to begin with UTF16BE BOM *)
+    (*pdf.Pdf.trailerdict <- ppstring_single_object pdf pdf.Pdf.trailerdict*)
 
 let json_of_pdf
   ~parse_content ~no_stream_data ~decompress_streams
diff --git a/cpdftexttopdf.ml b/cpdftexttopdf.ml
index d7d0618..0d19755 100644
--- a/cpdftexttopdf.ml
+++ b/cpdftexttopdf.ml
@@ -28,13 +28,16 @@ let rec of_utf8_with_newlines t =
       if c <> "" then items := Text (explode c)::!items;
     rev !items
 
-(* FIXME margins, hyphenation of too-long words, efficiency *)
-let typeset ~font ~fontsize text =
+let typeset ~papersize ~font ~fontsize text =
   let pdf = Pdf.empty () in
+  let margin =
+    Pdfunits.convert 72. (Pdfpaper.unit papersize) (Pdfunits.PdfPoint) (Pdfpaper.width papersize) /. 15.
+  in
   let f = (Pdftext.StandardFont (font, Pdftext.WinAnsiEncoding), fontsize) in
   let pages =
     Cpdftype.typeset
-      20. 20. 20. 20. Pdfpaper.a4 pdf ([Cpdftype.Font f; Cpdftype.BeginDocument] @ of_utf8_with_newlines (Pdfio.string_of_bytes text))
+      margin margin margin margin papersize pdf
+      ([Cpdftype.Font f; Cpdftype.BeginDocument] @ 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
index 5deb16a..346b05e 100644
--- a/cpdftexttopdf.mli
+++ b/cpdftexttopdf.mli
@@ -1 +1 @@
-val typeset : font:Pdftext.standard_font -> fontsize:float -> Pdfio.bytes -> Pdf.t
+val typeset : papersize:Pdfpaper.t -> font:Pdftext.standard_font -> fontsize:float -> Pdfio.bytes -> Pdf.t