fenix/cpdf-source
1
0
Fork 0
mirror of https://github.com/johnwhitington/cpdf-source.git synced 2024-12-27 09:13:04 +01:00
Code Issues Packages Projects Releases Wiki Activity

Remove squeeze debug commands

Browse Source
This commit is contained in:
John Whitington 2024-12-05 13:42:11 +00:00
parent 8279a06ec0
commit 05a7d1a9d7
1 changed files with 12 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
cpdfsqueeze.ml
View File

@ -49,7 +49,10 @@ let pdfobjeq pdf x y =
begin match y with Pdf.Stream _ -> Pdf.getstream y | _ -> () end; begin match y with Pdf.Stream _ -> Pdf.getstream y | _ -> () end;
compare x y compare x y
let memory () = Printf.printf "%i bytes in use\n%!" (Gc.(quick_stat ()).heap_words * 4)
let really_squeeze pdf = let really_squeeze pdf =
(*Printf.printf "Beginning of really_squeeze: %!"; memory ();*)
let objs = ref [] in let objs = ref [] in
Pdf.objiter (fun objnum _ -> objs := objnum :: !objs) pdf; Pdf.objiter (fun objnum _ -> objs := objnum :: !objs) pdf;
let toprocess = let toprocess =
@ -57,6 +60,7 @@ let really_squeeze pdf =
(fun x -> length x > 1) (fun x -> length x > 1)
(collate (pdfobjeq pdf) (sort (pdfobjeq pdf) !objs)) (collate (pdfobjeq pdf) (sort (pdfobjeq pdf) !objs))
in in
(*Printf.printf "Stage 1 done%!\n"; memory ();*)
(* Remove any pools of objects which are page objects, since Adobe Reader (* Remove any pools of objects which are page objects, since Adobe Reader
* gets confused when there are duplicate page objects. *) * gets confused when there are duplicate page objects. *)
let toprocess = let toprocess =
@ -69,6 +73,7 @@ let really_squeeze pdf =
| _ -> Some l) | _ -> Some l)
toprocess toprocess
in in
(*Printf.printf "Stage 2 done%!\n"; memory ();*)
let pdfr = ref pdf in let pdfr = ref pdf in
let changetable = Hashtbl.create 100 in let changetable = Hashtbl.create 100 in
iter iter
@ -79,9 +84,15 @@ let really_squeeze pdf =
Pdf.renumber is run twice. This is bizarre, since Pdf.renumber is Pdf.renumber is run twice. This is bizarre, since Pdf.renumber is
an old, well-understood function in use for years -- what is an old, well-understood function in use for years -- what is
going on? Furthermore, if we run it 3 times, it gets bigger again! *) going on? Furthermore, if we run it 3 times, it gets bigger again! *)
(*Printf.printf "Stage 3 done\n%!"; memory ();*)
pdfr := Pdf.renumber changetable !pdfr; pdfr := Pdf.renumber changetable !pdfr;
(*Printf.printf "Stage 4 done\n%!"; memory ();*)
pdfr := Pdf.renumber changetable !pdfr; pdfr := Pdf.renumber changetable !pdfr;
(*Printf.printf "Stage 5 done\n%!"; memory ();*)
Pdf.remove_unreferenced !pdfr; Pdf.remove_unreferenced !pdfr;
(*Printf.printf "Stage 6 done\n%!"; memory ();*)
(*Gc.compact ();*)
(*Printf.printf "Compacted:\n%!"; memory ();*)
pdf.Pdf.root <- !pdfr.Pdf.root; pdf.Pdf.root <- !pdfr.Pdf.root;
pdf.Pdf.objects <- !pdfr.Pdf.objects; pdf.Pdf.objects <- !pdfr.Pdf.objects;
pdf.Pdf.trailerdict <- !pdfr.Pdf.trailerdict pdf.Pdf.trailerdict <- !pdfr.Pdf.trailerdict
@ -96,6 +107,7 @@ this object will fail, and we bail out. *)
(* FIXME: XObjects inside xobjects? *) (* FIXME: XObjects inside xobjects? *)
let xobjects_done = ref [] let xobjects_done = ref []
let squeeze_form_xobject pdf objnum = let squeeze_form_xobject pdf objnum =
if mem objnum !xobjects_done then () else if mem objnum !xobjects_done then () else
begin begin