From b874221aca2525d09fdfa6976a08055398c5efd4 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Fri, 1 Oct 2021 12:16:55 +0100
Subject: [PATCH] -j support

---
 Makefile       |  2 +-
 cpdfJSON.ml    | 14 +++++++++++---
 cpdfJSON.mli   |  1 +
 cpdfcommand.ml | 10 ++++++++++
 4 files changed, 23 insertions(+), 4 deletions(-)

diff --git a/Makefile b/Makefile
index 57c0f88..c54c421 100644
--- a/Makefile
+++ b/Makefile
@@ -18,7 +18,7 @@ OCAMLLDFLAGS = -g
 all : native-code native-code-library byte-code-library top htdoc
 
 clean ::
-	rm -rf doc foo foo2 out.pdf out2.pdf *.cmt *.cmti
+	rm -rf doc foo foo2 out.pdf out2.pdf *.cmt *.cmti *.json
 
 LIBINSTALL_FILES = cpdf.a cpdf.cma cpdf.cmxa \
 $(foreach x,$(MODS),$x.mli) $(foreach x,$(MODS),$x.cmi) \
diff --git a/cpdfJSON.ml b/cpdfJSON.ml
index 2caf6fb..0ea16d8 100644
--- a/cpdfJSON.ml
+++ b/cpdfJSON.ml
@@ -195,6 +195,12 @@ let json_of_pdf parse_content no_stream_data pdf =
   let pdf = if parse_content then precombine_page_content pdf else pdf in
   Pdf.remove_unreferenced pdf;
   let trailerdict = (0, json_of_object pdf (fun x -> ()) no_stream_data pdf.P.trailerdict) in
+  let parameters =
+    (-1, json_of_object pdf (fun x -> ()) false
+      (Pdf.Dictionary [("/CPDFJSONformatversion", Pdf.Integer 1);
+                       ("/CPDFJSONcontentparsed", Pdf.Boolean parse_content);
+                       ("/CPDFJSONstreamdataincluded", Pdf.Boolean (not no_stream_data))]))
+  in
   let content_streams = ref [] in
   let fcs n = content_streams := n::!content_streams in
   let pairs =
@@ -203,7 +209,7 @@ let json_of_pdf parse_content no_stream_data pdf =
         (fun i pdfobj ->
           ps := (i, json_of_object pdf fcs no_stream_data pdfobj)::!ps)
         pdf;
-      trailerdict::!ps
+      parameters::trailerdict::!ps
   in
     if parse_content then
       iter (fun n -> Pdfcodec.decode_pdfstream_until_unknown pdf (P.lookup_obj pdf n)) !content_streams;
@@ -233,9 +239,11 @@ let json_of_pdf parse_content no_stream_data pdf =
           pairs_parsed)
 
 (* FIXME Proper streaming to output, rather than making a big string first. *)
-let to_output output parse_content no_stream_data pdf =
+let to_output o parse_content no_stream_data pdf =
   let b = Buffer.create 256 in
   let formatter = Format.formatter_of_buffer b in
     J.format formatter (json_of_pdf parse_content no_stream_data pdf);
     Format.pp_print_flush formatter ();
-    output.Pdfio.output_string (Buffer.contents b)
+    o.Pdfio.output_string (Buffer.contents b)
+
+let of_input i = Pdf.empty ()
diff --git a/cpdfJSON.mli b/cpdfJSON.mli
index 97eafa6..8a42e23 100644
--- a/cpdfJSON.mli
+++ b/cpdfJSON.mli
@@ -1 +1,2 @@
 val to_output : Pdfio.output -> bool -> bool -> Pdf.t -> unit
+val of_input : Pdfio.input -> Pdf.t
diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index f4aa021..3d70e8f 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -1168,6 +1168,13 @@ let set_input s =
   args.original_filename <- s;
   args.inputs <- (InFile s, "all", "", "", ref false, None)::args.inputs
 
+let set_json_input s =
+  args.original_filename <- s;
+  let fh = open_in_bin s in
+  let pdf = CpdfJSON.of_input (Pdfio.input_of_channel fh) in
+    close_in fh;
+    args.inputs <- (AlreadyInMemory pdf, "all", "", "", ref false, None)::args.inputs
+
 let set_input_dir s =
   let names = sort compare (leafnames_of_dir s) in
     args.inputs <-
@@ -2197,6 +2204,9 @@ and specs =
    ("-output-json-no-stream-data",
      Arg.Unit setjsonnostreamdata,
      " Skip stream data for brevity");
+   ("-j",
+     Arg.String set_json_input,
+     "Load a PDF JSON file");
    ("-ocg-list",
      Arg.Unit (setop OCGList),
      " List optional content groups");