fenix/cpdf-source
1
0
Fork 0
mirror of https://github.com/johnwhitington/cpdf-source.git synced 2025-06-05 22:09:39 +02:00
Code Issues Packages Projects Releases Wiki Activity

Work on lossless->JPEG

Browse Source
This commit is contained in:
John Whitington
2023-12-17 12:58:32 +00:00
parent adeb7c6c33
commit 81c719aae3
1 changed files with 46 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files

79
cpdfimage.ml
View File

@ -457,41 +457,54 @@ let image_of_input fobj i =
let pdf, pageroot = Pdfpage.add_pagetree [page] pdf in let pdf, pageroot = Pdfpage.add_pagetree [page] pdf in
Pdfpage.add_root pageroot [] pdf Pdfpage.add_root pageroot [] pdf
(* For each image xobject, process it through convert to reduce JPEG quality if we can. *) (* FIXME Make sure this process is ok for masks too - do we get them, is it allowed etc. *)
(* FIXME Only do if quality < 100 *)
(* For each image xobject, process it through convert to reduce size. *)
let process pdf ~q ~qlossless ~path_to_convert = let process pdf ~q ~qlossless ~path_to_convert =
let process_obj _ s = let process_obj _ s =
match s with match s with
| Pdf.Stream ({contents = dict, _} as reference) -> | Pdf.Stream ({contents = dict, _} as reference) ->
begin match Pdf.lookup_direct pdf "/Subtype" dict, Pdf.lookup_direct pdf "/Filter" dict with begin match Pdf.lookup_direct pdf "/Subtype" dict, Pdf.lookup_direct pdf "/Filter" dict with
| Some (Pdf.Name "/Image"), Some (Pdf.Name "/DCTDecode" | Pdf.Array [Pdf.Name "/DCTDecode"]) -> | Some (Pdf.Name "/Image"), Some (Pdf.Name "/DCTDecode" | Pdf.Array [Pdf.Name "/DCTDecode"]) ->
Pdf.getstream s; Pdf.getstream s;
let out = Filename.temp_file "cpdf" "convertin" ^ ".jpg" in let out = Filename.temp_file "cpdf" "convertin" ^ ".jpg" in
let out2 = Filename.temp_file "cpdf" "convertout" ^ ".jpg" in let out2 = Filename.temp_file "cpdf" "convertout" ^ ".jpg" in
let fh = open_out_bin out in let fh = open_out_bin out in
let size = let size =
begin match s with Pdf.Stream {contents = _, Pdf.Got d} -> Pdfio.bytes_to_output_channel fh d; bytes_size d | _ -> 0 end begin match s with Pdf.Stream {contents = _, Pdf.Got d} -> Pdfio.bytes_to_output_channel fh d; bytes_size d | _ -> 0 end
in
close_out fh;
let retcode =
let command =
(Filename.quote_command path_to_convert
[out; "-quality"; string_of_int q ^ "%"; out2])
in in
(*Printf.printf "%S\n" command;*) close_out fh;
Sys.command command let retcode =
in let command =
if retcode = 0 then (Filename.quote_command path_to_convert
begin [out; "-quality"; string_of_int q ^ "%"; out2])
let result = open_in_bin out2 in in
let newsize = in_channel_length result in (*Printf.printf "%S\n" command;*)
if newsize < size then Sys.command command
(*Printf.printf "%i -> %i\n" size newsize;*) in
reference := Pdf.add_dict_entry dict "/Length" (Pdf.Integer newsize), Pdf.Got (Pdfio.bytes_of_input_channel result) if retcode = 0 then
end; begin
Sys.remove out; let result = open_in_bin out2 in
Sys.remove out2 let newsize = in_channel_length result in
| _ -> () if newsize < size then
end (*Printf.printf "%i -> %i\n" size newsize;*)
| _ -> () reference := Pdf.add_dict_entry dict "/Length" (Pdf.Integer newsize), Pdf.Got (Pdfio.bytes_of_input_channel result)
end;
Sys.remove out;
Sys.remove out2
| Some (Pdf.Name "/Image"), _ ->
(* 0. Test if this is one we can do - for now just Colourspace=RGB, BPC=8 *)
begin match Pdf.lookup_direct pdf "/ColorSpace" dict, Pdf.lookup_direct pdf "/BitsPerComponent" dict with
| Some (Pdf.Name "/DeviceRGB"), Some (Pdf.Integer 8) ->
Printf.printf "Found a lossless(rgb, 8) image to JPEGify\n"
(* 1. Decompress it - check we succeeded, bail if not *)
(* 1. Output to pnm *)
(* 2. Convert to JPEG with convert *)
(* 3. Check smaller, Read file, and build new dictionary - removing ColorSpace, BitsPerComponent replacing Filter *)
| _ -> Printf.printf "I"
end
| _ -> () (* not an image *)
end
| _ -> () (* not a stream *)
in in
Pdf.objiter process_obj pdf Pdf.objiter process_obj pdf