fenix
/
cpdf-source
mirror of https://github.com/johnwhitington/cpdf-source.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Split up process_images

This commit is contained in:
John Whitington 2023-12-20 12:11:55 +00:00
parent 5631cfd724
commit f1b39a0c84
1 changed files with 103 additions and 98 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

47
cpdfimage.ml
View File

@ -492,14 +492,8 @@ let image_of_input fobj i =
(* FIXME What about predictors? Audit to see if files get smaller. *) (* FIXME What about predictors? Audit to see if files get smaller. *)
(* FIXME if lossy only 5% smaller, ignore? Set this parameter... *) (* FIXME if lossy only 5% smaller, ignore? Set this parameter... *)
(* FIXME error handling for Sys.remove, others *) (* FIXME error handling for Sys.remove, others *)
(* FIXME DeviceCYMK - use the convert CYMK samples format *) (* FIXME Use raw format for all, and make it fast *)
(* FIXME Test JPEG to JPEG on CYMK - is colourspace retained, or do we need to add -colorspace CMYK? *) let jpeg_to_jpeg pdf ~q ~path_to_convert s dict reference =
let process pdf ~q ~qlossless ~path_to_convert =
let process_obj _ s =
match s with
| Pdf.Stream ({contents = dict, _} as reference) ->
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"]) ->
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
@ -526,10 +520,8 @@ let process pdf ~q ~qlossless ~path_to_convert =
end; end;
Sys.remove out; Sys.remove out;
Sys.remove out2 Sys.remove out2
| Some (Pdf.Name "/Image"), _ ->
(* 0. Test if this is one we can do - for now just Colourspace=RGB, BPC=8 *) let suitable_num pdf dict =
let bpc = Pdf.lookup_direct pdf "/BitsPerComponent" dict in
let suitable_num =
match Pdf.lookup_direct pdf "/ColorSpace" dict with match Pdf.lookup_direct pdf "/ColorSpace" dict with
| Some (Pdf.Name "/DeviceRGB") -> 3 | Some (Pdf.Name "/DeviceRGB") -> 3
| Some (Pdf.Name "/DeviceGray") -> 1 | Some (Pdf.Name "/DeviceGray") -> 1
@ -542,27 +534,31 @@ let process pdf ~q ~qlossless ~path_to_convert =
| _ -> 0 | _ -> 0
end end
| _ -> 0 | _ -> 0
in
begin match suitable_num, bpc with let lossless_to_jpeg pdf ~qlossless ~path_to_convert s dict reference =
(* 0. Test if this is one we can do - for now just Colourspace=RGB, BPC=8 *)
let bpc = Pdf.lookup_direct pdf "/BitsPerComponent" dict in
let components = suitable_num pdf dict in
match components, bpc with
| (1 | 3 | 4), Some (Pdf.Integer 8) -> | (1 | 3 | 4), Some (Pdf.Integer 8) ->
let size = match Pdf.lookup_direct pdf "/Length" dict with Some (Pdf.Integer i) -> i | _ -> 0 in let size = match Pdf.lookup_direct pdf "/Length" dict with Some (Pdf.Integer i) -> i | _ -> 0 in
Pdfcodec.decode_pdfstream_until_unknown pdf s; Pdfcodec.decode_pdfstream_until_unknown pdf s;
begin match Pdf.lookup_direct pdf "/Filter" (fst !reference) with Some _ -> () | None -> begin match Pdf.lookup_direct pdf "/Filter" (fst !reference) with Some _ -> () | None ->
let w = match Pdf.lookup_direct pdf "/Width" dict with Some (Pdf.Integer i) -> i | _ -> error "bad width" in let w = match Pdf.lookup_direct pdf "/Width" dict with Some (Pdf.Integer i) -> i | _ -> error "bad width" in
let h = match Pdf.lookup_direct pdf "/Height" dict with Some (Pdf.Integer i) -> i | _ -> error "bad height" in let h = match Pdf.lookup_direct pdf "/Height" dict with Some (Pdf.Integer i) -> i | _ -> error "bad height" in
let out = Filename.temp_file "cpdf" "convertin" ^ (if suitable_num < 4 then ".pnm" else ".cmyk") in let out = Filename.temp_file "cpdf" "convertin" ^ (if suitable_num pdf dict < 4 then ".pnm" else ".cmyk") 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 data = match s with Pdf.Stream {contents = _, Pdf.Got d} -> d | _ -> assert false in let data = match s with Pdf.Stream {contents = _, Pdf.Got d} -> d | _ -> assert false in
(if suitable_num = 3 then pnm_to_channel_24 else (if components = 3 then pnm_to_channel_24 else
if suitable_num = 4 then cmyk_to_channel_32 else pnm_to_channel_8) fh w h data; if components = 4 then cmyk_to_channel_32 else pnm_to_channel_8) fh w h data;
close_out fh; close_out fh;
let retcode = let retcode =
let command = let command =
(Filename.quote_command path_to_convert (Filename.quote_command path_to_convert
((if suitable_num = 4 then ["-depth"; "8"; "-size"; string_of_int w ^ "x" ^ string_of_int h] else []) @ ((if components = 4 then ["-depth"; "8"; "-size"; string_of_int w ^ "x" ^ string_of_int h] else []) @
[out; "-quality"; string_of_int qlossless ^ "%"] @ [out; "-quality"; string_of_int qlossless ^ "%"] @
(if suitable_num = 1 then ["-colorspace"; "Gray"] else if suitable_num = 4 then ["-colorspace"; "CMYK"] else []) @ (if components = 1 then ["-colorspace"; "Gray"] else if components = 4 then ["-colorspace"; "CMYK"] else []) @
[out2])) [out2]))
in in
(*Printf.printf "%S\n" command;*) (*Printf.printf "%S\n" command;*)
@ -574,7 +570,7 @@ let process pdf ~q ~qlossless ~path_to_convert =
let newsize = in_channel_length result in let newsize = in_channel_length result in
if newsize < size then if newsize < size then
begin begin
Printf.printf "Lossless to JPEG %i -> %i (components %i) \n" size newsize suitable_num; Printf.printf "Lossless to JPEG %i -> %i (components %i) \n" size newsize components;
reference := reference :=
(Pdf.add_dict_entry (Pdf.add_dict_entry
(Pdf.add_dict_entry dict "/Length" (Pdf.Integer newsize)) (Pdf.add_dict_entry dict "/Length" (Pdf.Integer newsize))
@ -596,7 +592,16 @@ let process pdf ~q ~qlossless ~path_to_convert =
in in
print_string (Printf.sprintf "%s (%s) [%s]\n" colspace bpc filter); print_string (Printf.sprintf "%s (%s) [%s]\n" colspace bpc filter);
() (* an image we cannot or do not handle *) () (* an image we cannot or do not handle *)
end
let process pdf ~q ~qlossless ~path_to_convert =
let process_obj _ s =
match s with
| Pdf.Stream ({contents = dict, _} as reference) ->
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"]) ->
jpeg_to_jpeg pdf ~q ~path_to_convert s dict reference
| Some (Pdf.Name "/Image"), _ ->
lossless_to_jpeg pdf ~qlossless ~path_to_convert s dict reference
| _ -> () (* not an image *) | _ -> () (* not an image *)
end end
| _ -> () (* not a stream *) | _ -> () (* not a stream *)