From 122285db59335aadb6b8d918c1e71f0a08621987 Mon Sep 17 00:00:00 2001
From: John Whitington <john@coherentgraphics.co.uk>
Date: Tue, 31 Oct 2023 16:51:57 +0000
Subject: [PATCH] Implement listing of page info in JSON

---
 cpdfcommand.ml |  2 +-
 cpdfpage.ml    | 50 ++++++++++++++++++++++++++++++++++++--------------
 cpdfpage.mli   |  2 +-
 3 files changed, 38 insertions(+), 16 deletions(-)

diff --git a/cpdfcommand.ml b/cpdfcommand.ml
index 1be56d6..ee8eaf2 100644
--- a/cpdfcommand.ml
+++ b/cpdfcommand.ml
@@ -3489,7 +3489,7 @@ let go () =
       | (_, pagespec, _, _, _, _)::_, _ ->
           let pdf = get_single_pdf args.op true in
             let range = parse_pagespec_allow_empty pdf pagespec in
-              Cpdfpage.output_page_info pdf range
+              Cpdfpage.output_page_info ~json:args.format_json pdf range
       | _ -> error "list-bookmarks: bad command line"
       end
   | Some Metadata ->
diff --git a/cpdfpage.ml b/cpdfpage.ml
index e30a026..672b8c1 100644
--- a/cpdfpage.ml
+++ b/cpdfpage.ml
@@ -154,7 +154,7 @@ let change_pattern_matrices_page pdf tr page =
     {page with Pdfpage.resources = change_pattern_matrices_resources pdf tr page.Pdfpage.resources used}
 
 (* Output information for each page *)
-let output_page_info pdf range =
+let output_page_info ?(json=false) pdf range =
   let pages = Pdfpage.pages_of_pagetree pdf
   and labels = Pdfpagelabels.read pdf in
     let getbox page box =
@@ -173,19 +173,41 @@ let output_page_info pdf range =
     and rotation page =
       Pdfpage.int_of_rotation page.Pdfpage.rotate
     in
-      iter
-        (fun pnum ->
-           let page = select pnum pages in
-             Printf.printf "Page %i:\n" pnum;
-             Printf.printf "Label: %s\n"
-               (try Pdfpagelabels.pagelabeltext_of_pagenumber pnum labels with Not_found -> "");
-             Printf.printf "MediaBox: %s\n" (getbox page "/MediaBox");
-             Printf.printf "CropBox: %s\n" (getbox page "/CropBox");
-             Printf.printf "BleedBox: %s\n" (getbox page "/BleedBox");
-             Printf.printf "TrimBox: %s\n" (getbox page "/TrimBox");
-             Printf.printf "ArtBox: %s\n" (getbox page "/ArtBox");
-             Printf.printf "Rotation: %i\n" (rotation page))
-        range
+      let json_entry_of_pnum pnum =
+        let getbox_json page box =
+          match getbox page box with
+          | "" -> `Null
+          | s ->
+            let a, b, c, d = Cpdfcoord.parse_rectangle (Pdf.empty ()) s in
+              `List [`Float a; `Float b; `Float c; `Float d]
+        in
+        let page = select pnum pages in
+          `Assoc
+            [("Page", `Int pnum);
+             ("Label", (`String (try Pdfpagelabels.pagelabeltext_of_pagenumber pnum labels with Not_found -> "")));
+             ("MediaBox", getbox_json page "/MediaBox");
+             ("CropBox", getbox_json page "/CropBox");
+             ("BleedBox", getbox_json page "/BleedBox");
+             ("TrimBox", getbox_json page "/TrimBox");
+             ("ArtBox", getbox_json page "/ArtBox");
+             ("Rotation", `Int (rotation page))]
+      in
+        if json then
+          flprint (Cpdfyojson.Safe.pretty_to_string (`List (map json_entry_of_pnum range)))
+        else
+          iter
+            (fun pnum ->
+               let page = select pnum pages in
+                 Printf.printf "Page %i:\n" pnum;
+                 Printf.printf "Label: %s\n"
+                   (try Pdfpagelabels.pagelabeltext_of_pagenumber pnum labels with Not_found -> "");
+                 Printf.printf "MediaBox: %s\n" (getbox page "/MediaBox");
+                 Printf.printf "CropBox: %s\n" (getbox page "/CropBox");
+                 Printf.printf "BleedBox: %s\n" (getbox page "/BleedBox");
+                 Printf.printf "TrimBox: %s\n" (getbox page "/TrimBox");
+                 Printf.printf "ArtBox: %s\n" (getbox page "/ArtBox");
+                 Printf.printf "Rotation: %i\n" (rotation page))
+            range
 
 let process_pages f pdf range =
   let pages = Pdfpage.pages_of_pagetree pdf in
diff --git a/cpdfpage.mli b/cpdfpage.mli
index e890f0f..b445fae 100644
--- a/cpdfpage.mli
+++ b/cpdfpage.mli
@@ -1,7 +1,7 @@
 (** Working with pages *)
 
 (** Print page info (Mediabox etc) to standard output. *)
-val output_page_info : Pdf.t -> int list -> unit
+val output_page_info : ?json:bool -> Pdf.t -> int list -> unit
 
 (** Given a function from page number and page to page, a document, and a list
 of page numbers to apply it to, apply the function to all those pages. *)