From 3cc0b8490d36ce06b19d19b354905f645bbea361 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Fri, 23 Dec 2022 13:14:24 +0000
Subject: [PATCH] more

---
 Makefile       |  4 ++--
 cpdfcommand.ml |  2 +-
 cpdfjpeg.ml    | 39 +++++++++++++++++++++++++++++++++++++++
 cpdfjpeg.mli   |  2 ++
 4 files changed, 44 insertions(+), 3 deletions(-)
 create mode 100644 cpdfjpeg.ml
 create mode 100644 cpdfjpeg.mli

diff --git a/Makefile b/Makefile
index 8dde5b1..ac7ae08 100644
--- a/Makefile
+++ b/Makefile
@@ -6,8 +6,8 @@ DOC = cpdfunicodedata cpdferror cpdfdebug cpdfjson cpdfstrftime cpdfcoord \
       cpdfbookmarks cpdfpage cpdftruetype cpdfremovetext cpdfextracttext \
       cpdfembed cpdfaddtext cpdfimage cpdffont cpdftype cpdfpad cpdfocg \
       cpdfsqueeze cpdfdraft cpdfspot cpdfpagelabels cpdfcreate cpdfannot \
-      cpdfxobject cpdfimpose cpdftweak cpdftexttopdf cpdftoc cpdfdraw \
-      cpdfcommand
+      cpdfxobject cpdfimpose cpdftweak cpdftexttopdf cpdftoc cpdfjpeg \
+      cpdfdraw cpdfcommand
 
 MODS = $(NONDOC) $(DOC)
 
diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index 0020b3e..179b353 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -1932,7 +1932,7 @@ let addjpeg n =
   in
     try
       let data = Pdfio.bytes_of_string (contents_of_file filename) in
-      let w, h = Pdfjpeg.jpeg_dimensions data in
+      let w, h = Cpdfjpeg.jpeg_dimensions data in
       let d = 
         ["/Length", Pdf.Integer (Pdfio.bytes_size data);
          "/Filter", Pdf.Name "/DCTDecode";
diff --git a/cpdfjpeg.ml b/cpdfjpeg.ml
new file mode 100644
index 0000000..4b3e6cf
--- /dev/null
+++ b/cpdfjpeg.ml
@@ -0,0 +1,39 @@
+open Pdfutil
+open Pdfio
+
+(* Return the width and height of a JPEG image, per Michael Petrov's C version. *)
+exception Answer of int * int
+
+let jpeg_dimensions bs =
+  try
+    let get = bget bs in
+    let i = ref 0 in
+    if get !i = 0xFF && get (!i + 1) = 0xD8 && get (!i + 2) = 0xFF && get (!i + 3) = 0xE0 then
+      begin
+        i += 4;
+        if
+             get (!i + 2) = int_of_char 'J' && get (!i + 3) = int_of_char 'F'
+          && get (!i + 4) = int_of_char 'I' && get (!i + 5) = int_of_char 'F'
+          && get (!i + 6) = 0
+        then
+          let block_length = ref (get !i * 256 + get (!i + 1)) in
+            while !i < bytes_size bs do
+              i := !i + !block_length;
+              if !i > bytes_size bs then raise (Pdf.PDFError "jpeg_dimensions: too short") else
+              if get !i <> 0xFF then raise (Pdf.PDFError "jpeg_dimensions: not a valid block") else
+              if get (!i + 1) = 0xC0 then
+                raise (Answer (get (!i + 7) * 256 + get (!i + 8), (get (!i + 5) * 256 + get (!i + 6))))
+              else
+                begin
+                  i += 2;
+                  block_length := get !i * 256 + get (!i + 1)
+                end
+            done
+        else
+          raise (Pdf.PDFError "jpeg_dimensions: Not a valid JFIF string")
+      end
+    else
+      raise (Pdf.PDFError "jpeg_dimensions: Not a valid SOI header");
+    assert false
+ with
+   Answer (w, h) -> (w, h)
diff --git a/cpdfjpeg.mli b/cpdfjpeg.mli
new file mode 100644
index 0000000..ed45ae4
--- /dev/null
+++ b/cpdfjpeg.mli
@@ -0,0 +1,2 @@
+(** Return the dimensions of a JPEG *)
+val jpeg_dimensions : Pdfio.bytes -> int * int