From d4f0bd202d0434a5c25e3c2a4b3e86f6b0731107 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Fri, 1 Oct 2021 15:21:03 +0100
Subject: [PATCH] First successful JSON round trip

---
 cpdfJSON.ml | 79 +++++++++++++++++++++++++++++++++++++++++++----------
 1 file changed, 65 insertions(+), 14 deletions(-)

diff --git a/cpdfJSON.ml b/cpdfJSON.ml
index f092e6f..ea4c41a 100644
--- a/cpdfJSON.ml
+++ b/cpdfJSON.ml
@@ -1,3 +1,5 @@
+(*FIXME failwiths -> Pdf.PDFError or similar *)
+(*FIXME flprintfs to eprintf *)
 open Pdfutil
 
 module J = Tjjson
@@ -9,13 +11,26 @@ let soi = string_of_int
 let string_of_float _ = failwith "use sof"
 let string_of_int _ = failwith "use soi"
 
+let rec object_of_json = function
+  | J.Null -> P.Null
+  | J.Bool b -> P.Boolean b
+  | J.Number n -> Pdf.Indirect (int_of_string n)
+  | J.String s -> P.String s
+  | J.Array objs -> P.Array (map object_of_json objs)
+  | J.Object ["I", J.Number i] -> P.Integer (int_of_string i)
+  | J.Object ["F", J.Number f] -> P.Real (float_of_string f)
+  | J.Object ["N", J.String n] -> P.Name n
+  | J.Object ["S", J.Array [dict; J.String data]] ->
+      P.Stream (ref (object_of_json dict, P.Got (Pdfio.bytes_of_string data)))
+  | J.Object elts -> P.Dictionary (map (fun (n, o) -> (n, object_of_json o)) elts)
+
 let rec json_of_object pdf fcs no_stream_data = function
-  | P.Null -> J.String "null"
+  | P.Null -> J.Null
   | P.Boolean b -> J.Bool b
-  | P.Integer i -> J.Number (soi i)
-  | P.Real r -> J.Number (sof r)
+  | P.Integer i -> J.Object [("I", J.Number (soi i))]
+  | P.Real r -> J.Object [("F", J.Number (sof r))]
   | P.String s -> J.String s
-  | P.Name n -> J.String n
+  | P.Name n -> J.Object [("N", J.String n)]
   | P.Array objs -> J.Array (map (json_of_object pdf fcs no_stream_data) objs)
   | P.Dictionary elts ->
       iter
@@ -41,8 +56,8 @@ let rec json_of_object pdf fcs no_stream_data = function
               match stream with P.Got b -> Pdfio.string_of_bytes b | P.ToGet _ -> "failure: toget"
        end
       in
-        json_of_object pdf fcs no_stream_data (P.Array [P.Dictionary dict; P.String str])
-  | P.Stream _ -> J.String "error: stream with not-a-dictioary"
+        json_of_object pdf fcs no_stream_data (P.Dictionary [("S", P.Array [P.Dictionary dict; P.String str])])
+  | P.Stream _ -> J.String "error: stream with not-a-dictionary"
   | P.Indirect i ->
       begin match P.lookup_obj pdf i with
       | P.Stream {contents = (P.Dictionary dict as d, _)} ->
@@ -197,9 +212,12 @@ let json_of_pdf parse_content no_stream_data 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);
+      (Pdf.Dictionary [("/CPDFJSONformatversion", Pdf.Integer 2);
                        ("/CPDFJSONcontentparsed", Pdf.Boolean parse_content);
-                       ("/CPDFJSONstreamdataincluded", Pdf.Boolean (not no_stream_data))]))
+                       ("/CPDFJSONstreamdataincluded", Pdf.Boolean (not no_stream_data));
+                       ("/CPDFJSONmajorpdfversion", Pdf.Integer pdf.Pdf.major);
+                       ("/CPDFJSONminorpdfversion", Pdf.Integer pdf.Pdf.minor);
+                      ]))
   in
   let content_streams = ref [] in
   let fcs n = content_streams := n::!content_streams in
@@ -239,22 +257,55 @@ let json_of_pdf parse_content no_stream_data pdf =
           pairs_parsed)
 
 let pdf_of_json json =
-  let major = 0 in
-  let minor = 1 in
-  let root = 2 in
-  let objmap = P.pdfobjmap_empty () in 
+  (*flprint (J.show json);
+  flprint "\n";*)
+  let objs = match json with J.Array objs -> objs | _ -> failwith "bad json top level" in
+  let params = ref Pdf.Null in
+  let trailerdict = ref Pdf.Null in
+    let objects =
+      option_map
+        (function
+         | J.Array [J.Number n; o] ->
+             let objnum = int_of_string n in
+               begin match objnum with
+               | -1 -> params := object_of_json o; None 
+               | 0 -> trailerdict := object_of_json o; None
+               | n when n < 0 -> None
+               | n ->  Some (n, object_of_json o)
+               end
+         | _ -> failwith "json bad obj")
+        objs
+    in
+      (*List.iter (fun (i, o) -> flprint (soi i); flprint "\n"; flprint (Pdfwrite.string_of_pdf o); flprint "\n") objects;*)
+  let major =
+    match Pdf.lookup_direct (Pdf.empty ()) "/CPDFJSONmajorpdfversion" !params with 
+      Some (Pdf.Integer i) -> i | _ -> failwith "bad major version"
+  in
+  let minor =
+    match Pdf.lookup_direct (Pdf.empty ()) "/CPDFJSONminorpdfversion" !params with 
+      Some (Pdf.Integer i) -> i | _ -> failwith "bad minor version"
+  in
+  (*flprint (Pdfwrite.string_of_pdf !trailerdict);*)
+  let root =
+    match !trailerdict with Pdf.Dictionary d ->
+      begin match lookup "/Root" d with
+        Some (Pdf.Indirect i) -> i | _ -> failwith "bad root"
+      end
+    | _ -> failwith "bad root 2"
+  in
+  let objmap = P.pdfobjmap_empty () in
+  List.iter (fun (k, v) -> Hashtbl.add objmap k (ref (P.Parsed v), 0)) objects;
   let objects =
     {P.maxobjnum = 0;
      P.parse = None;
      P.pdfobjects = objmap;
      P.object_stream_ids = Hashtbl.create 0}
   in
-  let trailerdict = P.Null in
     {P.major;
      P.minor;
      P.root;
      P.objects;
-     P.trailerdict;
+     P.trailerdict = !trailerdict;
      P.was_linearized = false;
      P.saved_encryption = None}