From ece1e9f280c075faadf88846bf7dbe599274125f Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Tue, 12 Oct 2021 15:35:08 +0100
Subject: [PATCH] more

---
 cpdfcommand.ml | 15 +++++++++------
 cpdfjson.ml    | 21 ++++++++++++++-------
 cpdfjson.mli   |  2 +-
 3 files changed, 24 insertions(+), 14 deletions(-)

diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index 05e3471..a22287c 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -1,3 +1,4 @@
+(* FIXME add -output-json-precombine-contents *)
 (* cpdf command line tools *)
 let demo = false
 let noncomp = false
@@ -3406,17 +3407,19 @@ let write_json output pdf =
   | Stdout ->
       Cpdfjson.to_output
         (Pdfio.output_of_channel stdout)
-        args.jsonparsecontentstreams
-        args.jsonnostreamdata
-        args.jsondecompressstreams
+        ~parse_content:args.jsonparsecontentstreams
+        ~no_stream_data:args.jsonnostreamdata
+        ~decompress_streams:args.jsondecompressstreams
+        ~precombine_page_content:false (* FIXME add arg *)
         pdf
   | File filename ->
       let f = open_out filename in
         Cpdfjson.to_output
           (Pdfio.output_of_channel f)
-          args.jsonparsecontentstreams
-          args.jsonnostreamdata
-          args.jsondecompressstreams
+          ~parse_content:args.jsonparsecontentstreams
+          ~no_stream_data:args.jsonnostreamdata
+          ~decompress_streams:args.jsondecompressstreams
+          ~precombine_page_content:false (* FIXME add arg *)
           pdf;
         close_out f
 
diff --git a/cpdfjson.ml b/cpdfjson.ml
index a343048..7589e98 100644
--- a/cpdfjson.ml
+++ b/cpdfjson.ml
@@ -333,9 +333,10 @@ let parse_content_stream pdf resources bs =
     `List (map (json_of_op pdf false) ops)
 
 (* We need to make sure each page only has one page content stream. Otherwise,
-   if not split on op boundaries, each one would fail to parse on its own. *)
-(* Future improvement. Don't blow up shared content streams. *)
-let precombine_page_content pdf =
+   if not split on op boundaries, each one would fail to parse on its own. The
+   caller should really only do this on otherwise-failing files, since it could
+   blow up any shared content streams *)
+let do_precombine_page_content pdf =
   let pages' =
     map
       (fun page ->
@@ -351,8 +352,14 @@ let precombine_page_content pdf =
   in
     Pdfpage.change_pages true pdf pages'
 
-let json_of_pdf parse_content no_stream_data decompress_streams pdf =
-  let pdf = if parse_content then precombine_page_content pdf else pdf in
+let json_of_pdf
+  ~parse_content
+  ~no_stream_data
+  ~decompress_streams
+  ~precombine_page_content
+  pdf
+=
+  let pdf = if parse_content && precombine_page_content then do_precombine_page_content pdf else pdf in
   if decompress_streams then
     Pdf.objiter (fun _ obj -> Pdfcodec.decode_pdfstream_until_unknown pdf obj) pdf;
   Pdf.remove_unreferenced pdf;
@@ -405,8 +412,8 @@ let json_of_pdf parse_content no_stream_data decompress_streams pdf =
           pairs_parsed)
 
 (* FIXME Proper streaming to output *)
-let to_output o parse_content no_stream_data decompress_streams pdf =
-  let json = json_of_pdf parse_content no_stream_data decompress_streams pdf in
+let to_output o ~parse_content ~no_stream_data ~decompress_streams ~precombine_page_content pdf =
+  let json = json_of_pdf ~parse_content ~no_stream_data ~decompress_streams ~precombine_page_content pdf in
     o.Pdfio.output_string (J.pretty_to_string json)
 
 (* FIXME Proper streaming to output / from input, rather than making a big string first. *)
diff --git a/cpdfjson.mli b/cpdfjson.mli
index 039b312..1e7b837 100644
--- a/cpdfjson.mli
+++ b/cpdfjson.mli
@@ -1,2 +1,2 @@
-val to_output : Pdfio.output -> bool -> bool -> bool -> Pdf.t -> unit
+val to_output : Pdfio.output -> parse_content:bool -> no_stream_data:bool -> decompress_streams:bool -> precombine_page_content:bool -> Pdf.t -> unit
 val of_input : Pdfio.input -> Pdf.t