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

-process-images-info finished

This commit is contained in:
John Whitington 2023-12-31 11:13:58 +00:00
parent 96a9142542
commit 8859f17584
1 changed files with 67 additions and 50 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

117
cpdfimage.ml
View File

@ -4,6 +4,9 @@ open Cpdferror
let debug_image_processing = ref false let debug_image_processing = ref false
let remove x =
try Sys.remove x with _ -> ()
let pnm_white ch = output_char ch ' ' let pnm_white ch = output_char ch ' '
let pnm_newline ch = output_char ch '\n' let pnm_newline ch = output_char ch '\n'
let pnm_output_string = Stdlib.output_string let pnm_output_string = Stdlib.output_string
@ -97,20 +100,20 @@ let write_image ~raw ?path_to_p2p ?path_to_im pdf resources name image =
begin match begin match
Sys.command (Filename.quote_command path_to_im [pnm; png]) Sys.command (Filename.quote_command path_to_im [pnm; png])
with with
0 -> Sys.remove pnm 0 -> remove pnm
| _ -> | _ ->
Pdfe.log "Call to imagemagick failed: did you specify -p2p or -im correctly?\n"; Pdfe.log "Call to imagemagick failed: did you specify -p2p or -im correctly?\n";
Sys.remove pnm remove pnm
end end
end end
| Some path_to_p2p -> | Some path_to_p2p ->
begin match begin match
Sys.command (Filename.quote_command path_to_p2p ~stdout:png ["-gamma"; "0.45"; "-quiet"; pnm]) Sys.command (Filename.quote_command path_to_p2p ~stdout:png ["-gamma"; "0.45"; "-quiet"; pnm])
with with
| 0 -> Sys.remove pnm | 0 -> remove pnm
| _ -> | _ ->
Pdfe.log "Call to pnmtopng failed: did you specify -p2p correctly?\n"; Pdfe.log "Call to pnmtopng failed: did you specify -p2p correctly?\n";
Sys.remove pnm remove pnm
end end
end end
| _ -> | _ ->
@ -507,18 +510,28 @@ let jpeg_to_jpeg pdf ~pixel_threshold ~length_threshold ~percentage_threshold ~q
let command = let command =
(Filename.quote_command path_to_convert [out; "-quality"; string_of_int q ^ "%"; out2]) (Filename.quote_command path_to_convert [out; "-quality"; string_of_int q ^ "%"; out2])
in in
(*Printf.printf "%S\n" command;*) Sys.command command (*0Printf.printf "%S\n" command;*) Sys.command command
in in
if retcode = 0 then if retcode = 0 then
begin begin
let result = open_in_bin out2 in let result = open_in_bin out2 in
let newsize = in_channel_length result in let newsize = in_channel_length result in
if newsize < size then if newsize < size then
if !debug_image_processing then Printf.printf "JPEG to JPEG %i -> %i (%i%%)\n%!" size newsize (int_of_float (float newsize /. float size *. 100.)); begin
reference := Pdf.add_dict_entry dict "/Length" (Pdf.Integer newsize), Pdf.Got (Pdfio.bytes_of_input_channel result) if !debug_image_processing then Printf.printf "JPEG to JPEG %i -> %i (%i%%)\n%!" size newsize (int_of_float (float newsize /. float size *. 100.));
reference := Pdf.add_dict_entry dict "/Length" (Pdf.Integer newsize), Pdf.Got (Pdfio.bytes_of_input_channel result)
end
else
begin
if !debug_image_processing then Printf.printf "no size reduction\n%!"
end
end
else
begin
Printf.printf "external process failed\n%!"
end; end;
Sys.remove out; remove out;
Sys.remove out2 remove out2
let suitable_num pdf dict = let suitable_num pdf dict =
match Pdf.lookup_direct pdf "/ColorSpace" dict with match Pdf.lookup_direct pdf "/ColorSpace" dict with
@ -566,7 +579,7 @@ let lossless_out pdf ~pixel_threshold ~length_threshold extension s dict referen
in in
print_string (Pdfwrite.string_of_pdf dict); print_string (Pdfwrite.string_of_pdf dict);
print_string (Printf.sprintf "%s (%s) [%s]\n" colspace bpc filter);*) print_string (Printf.sprintf "%s (%s) [%s]\n" colspace bpc filter);*)
if !debug_image_processing then Printf.printf "not suitable\n%!"; if !debug_image_processing then Printf.printf "colourspace not suitable\n%!";
None (* an image we cannot or do not handle *) None (* an image we cannot or do not handle *)
let lossless_to_jpeg pdf ~pixel_threshold ~length_threshold ~percentage_threshold ~qlossless ~path_to_convert s dict reference = let lossless_to_jpeg pdf ~pixel_threshold ~length_threshold ~percentage_threshold ~qlossless ~path_to_convert s dict reference =
@ -601,8 +614,8 @@ let lossless_to_jpeg pdf ~pixel_threshold ~length_threshold ~percentage_threshol
end; end;
close_in result close_in result
end; end;
Sys.remove out; remove out;
Sys.remove out2 remove out2
let lossless_resample pdf ~pixel_threshold ~length_threshold ~percentage_threshold~interpolate ~path_to_convert s dict reference = let lossless_resample pdf ~pixel_threshold ~length_threshold ~percentage_threshold~interpolate ~path_to_convert s dict reference =
match lossless_out pdf ~pixel_threshold ~length_threshold ".png" s dict reference with None -> () | Some (out, out2, size, components, w, h) -> match lossless_out pdf ~pixel_threshold ~length_threshold ".png" s dict reference with None -> () | Some (out, out2, size, components, w, h) ->
@ -636,50 +649,54 @@ let lossless_resample pdf ~pixel_threshold ~length_threshold ~percentage_thresho
end; end;
close_in result close_in result
end; end;
Sys.remove out; remove out;
Sys.remove out2 remove out2
let recompress_1bpp_jbig2_lossless ~pixel_threshold ~length_threshold ~path_to_jbig2enc pdf s dict reference = let recompress_1bpp_jbig2_lossless ~pixel_threshold ~length_threshold ~path_to_jbig2enc pdf s dict reference =
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
if w * h < pixel_threshold then (if !debug_image_processing then Printf.printf "pixel threshold not met\n%!") else (* (but also, jbig2enc fails on tiny images) *) if w * h < pixel_threshold then (if !debug_image_processing then Printf.printf "pixel threshold not met\n%!") else (* (but also, jbig2enc fails on tiny images) *)
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
if size < length_threshold then if !debug_image_processing then Printf.printf "length threshold not met\n%!" else if size < length_threshold then (if !debug_image_processing then Printf.printf "length threshold not met\n%!") else
Pdfcodec.decode_pdfstream_until_unknown pdf s; begin
match Pdf.lookup_direct pdf "/Filter" (fst !reference) with Some _ -> if !debug_image_processing then Printf.printf "could not decode - skipping\n%!" | None -> Pdfcodec.decode_pdfstream_until_unknown pdf s;
let out = Filename.temp_file "cpdf" "convertin" ^ ".pnm" in match Pdf.lookup_direct pdf "/Filter" (fst !reference) with
let out2 = Filename.temp_file "cpdf" "convertout" ^ ".jbig2" in | Some x -> if !debug_image_processing then Printf.printf "could not decode - skipping %s length %i\n%!" (Pdfwrite.string_of_pdf x) size
let fh = open_out_bin out in | None ->
let data = match s with Pdf.Stream {contents = _, Pdf.Got d} -> d | _ -> assert false in let out = Filename.temp_file "cpdf" "convertin" ^ ".pnm" in
pnm_to_channel_1_inverted fh w h data; let out2 = Filename.temp_file "cpdf" "convertout" ^ ".jbig2" in
close_out fh; let fh = open_out_bin out in
let retcode = let data = match s with Pdf.Stream {contents = _, Pdf.Got d} -> d | _ -> assert false in
let command = Filename.quote_command ~stdout:out2 path_to_jbig2enc ["-p"; out] in pnm_to_channel_1_inverted fh w h data;
(*Printf.printf "%S\n" command;*) Sys.command command close_out fh;
in let retcode =
if retcode = 0 then let command = Filename.quote_command ~stdout:out2 path_to_jbig2enc ["-p"; out] in
begin (*Printf.printf "%S\n" command;*) Sys.command command
let result = open_in_bin out2 in in
let newsize = in_channel_length result in if retcode = 0 then
if newsize < size then begin
begin let result = open_in_bin out2 in
if !debug_image_processing then Printf.printf "1bpp to JBIG2 %i -> %i (%i%%)\n%!" size newsize (int_of_float (float newsize /. float size *. 100.)); let newsize = in_channel_length result in
reference := if newsize < size then
(Pdf.remove_dict_entry begin
(Pdf.add_dict_entry if !debug_image_processing then Printf.printf "1bpp to JBIG2 %i -> %i (%i%%)\n%!" size newsize (int_of_float (float newsize /. float size *. 100.));
(Pdf.add_dict_entry dict "/Length" (Pdf.Integer newsize)) reference :=
"/Filter" (Pdf.remove_dict_entry
(Pdf.Name "/JBIG2Decode")) "/DecodeParms"), (Pdf.add_dict_entry
Pdf.Got (Pdfio.bytes_of_input_channel result) (Pdf.add_dict_entry dict "/Length" (Pdf.Integer newsize))
end "/Filter"
else (Pdf.Name "/JBIG2Decode")) "/DecodeParms"),
begin Pdf.Got (Pdfio.bytes_of_input_channel result)
if !debug_image_processing then Printf.printf "no size reduction\n%!" end
end; else
close_in result begin
end; if !debug_image_processing then Printf.printf "no size reduction\n%!"
Sys.remove out; end;
Sys.remove out2 close_in result
end;
remove out;
remove out2
end
(* JPEG to JPEG: RGB and CMYK JPEGS *) (* JPEG to JPEG: RGB and CMYK JPEGS *)
(* Lossless to JPEG: 8bpp Grey, 8bpp RGB, 8bpp CMYK including separation and ICCBased colourspaces *) (* Lossless to JPEG: 8bpp Grey, 8bpp RGB, 8bpp CMYK including separation and ICCBased colourspaces *)