fenix/cpdf-source
1
0
Fork 0
mirror of https://github.com/johnwhitington/cpdf-source.git synced 2025-03-02 10:38:03 +01:00
Code Issues Packages Projects Releases Wiki Activity

Plumbing for -proces-images -dpi-threshold

Browse Source
This commit is contained in:
John Whitington 2024-01-04 17:40:15 +00:00
parent 3a39b5497a
commit ecbccd4567
3 changed files with 15 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
cpdfcommand.ml
View File

@ -4321,13 +4321,13 @@ let go () =
if args.format_json then if args.format_json then
flprint flprint
(Cpdfyojson.Safe.pretty_to_string (Cpdfyojson.Safe.pretty_to_string
(`List (map (fun (pagenum, xobject, w, h, wdpi, hdpi) -> (`List (map (fun (pagenum, xobject, w, h, wdpi, hdpi, objnum) ->
`Assoc [("Page", `Int pagenum); ("XObject", `String xobject); ("W", `Int w); ("H", `Int h); ("Xdpi", `Float wdpi); ("Ydpi", `Float hdpi)]) images))) `Assoc [("Object", `Int objnum); ("Page", `Int pagenum); ("XObject", `String xobject); ("W", `Int w); ("H", `Int h); ("Xdpi", `Float wdpi); ("Ydpi", `Float hdpi)]) images)))
else else
iter iter
(function (pagenum, xobject, w, h, wdpi, hdpi) -> (function (pagenum, xobject, w, h, wdpi, hdpi, objnum) ->
if wdpi < f || hdpi < f then if wdpi < f || hdpi < f then
Printf.printf "%i, %s, %i, %i, %f, %f\n" pagenum xobject w h wdpi hdpi) Printf.printf "%i, %s, %i, %i, %f, %f, %i\n" pagenum xobject w h wdpi hdpi objnum)
images images
| Some ListImages -> | Some ListImages ->
let pdf = get_single_pdf args.op true in let pdf = get_single_pdf args.op true in

23
cpdfimage.ml
View File

@ -187,11 +187,7 @@ type xobj =
let image_results = ref [] let image_results = ref []
let add_image_result i = let rec image_resolution_page pdf page pagenum dpi images =
image_results := i::!image_results
(* Given a page and a list of (pagenum, name, thing) *)
let rec image_resolution_page pdf page pagenum dpi (images : (int * string * xobj) list) =
try try
let pageops = Pdfops.parse_operators pdf page.Pdfpage.resources page.Pdfpage.content let pageops = Pdfops.parse_operators pdf page.Pdfpage.resources page.Pdfpage.content
and transform = ref [ref Pdftransform.i_matrix] in and transform = ref [ref Pdftransform.i_matrix] in
@ -215,11 +211,11 @@ let rec image_resolution_page pdf page pagenum dpi (images : (int * string * xob
(*i Printf.printf "o = %f, %f, x = %f, %f, y = %f, %f\n" (fst o) (snd o) (fst x) (snd x) (fst y) (snd y); i*) (*i Printf.printf "o = %f, %f, x = %f, %f, y = %f, %f\n" (fst o) (snd o) (fst x) (snd x) (fst y) (snd y); i*)
let rec lookup_image k = function let rec lookup_image k = function
| [] -> assert false | [] -> assert false
| (_, a, _) as h::_ when a = k -> h | (_, a, _, _) as h::_ when a = k -> h
| _::t -> lookup_image k t | _::t -> lookup_image k t
in in
begin match lookup_image xobject images with begin match lookup_image xobject images with
| (pagenum, name, Form (xobj_matrix, content, resources)) -> | (pagenum, name, Form (xobj_matrix, content, resources), objnum) ->
let content = let content =
(* Add in matrix etc. *) (* Add in matrix etc. *)
let total_matrix = Pdftransform.matrix_compose xobj_matrix !(hd !transform) in let total_matrix = Pdftransform.matrix_compose xobj_matrix !(hd !transform) in
@ -238,12 +234,12 @@ let rec image_resolution_page pdf page pagenum dpi (images : (int * string * xob
in in
let newpdf = Pdfpage.change_pages false pdf [page] in let newpdf = Pdfpage.change_pages false pdf [page] in
image_resolution newpdf [pagenum] dpi image_resolution newpdf [pagenum] dpi
| (pagenum, name, Image (w, h)) -> | (pagenum, name, Image (w, h), objnum) ->
let lx = Pdfunits.points (distance_between o x) Pdfunits.Inch in let lx = Pdfunits.points (distance_between o x) Pdfunits.Inch in
let ly = Pdfunits.points (distance_between o y) Pdfunits.Inch in let ly = Pdfunits.points (distance_between o y) Pdfunits.Inch in
let wdpi = float w /. lx let wdpi = float w /. lx
and hdpi = float h /. ly in and hdpi = float h /. ly in
add_image_result (pagenum, xobject, w, h, wdpi, hdpi) image_results := (pagenum, xobject, w, h, wdpi, hdpi, objnum)::!image_results
(*Printf.printf "%i, %s, %i, %i, %f, %f\n" pagenum xobject w h wdpi hdpi*) (*Printf.printf "%i, %s, %i, %i, %f, %f\n" pagenum xobject w h wdpi hdpi*)
(*i else (*i else
Printf.printf "S %i, %s, %i, %i, %f, %f\n" pagenum xobject (int_of_float w) (int_of_float h) wdpi hdpi i*) Printf.printf "S %i, %s, %i, %i, %f, %f\n" pagenum xobject (int_of_float w) (int_of_float h) wdpi hdpi i*)
@ -275,6 +271,7 @@ and image_resolution pdf range dpi =
| Some (Pdf.Dictionary xobjects) -> | Some (Pdf.Dictionary xobjects) ->
iter iter
(function (name, xobject) -> (function (name, xobject) ->
let objnum = match xobject with Pdf.Indirect i -> i | _ -> 0 in
match Pdf.lookup_direct pdf "/Subtype" xobject with match Pdf.lookup_direct pdf "/Subtype" xobject with
| Some (Pdf.Name "/Image") -> | Some (Pdf.Name "/Image") ->
let width = let width =
@ -286,7 +283,7 @@ and image_resolution pdf range dpi =
| Some x -> Pdf.getnum pdf x | Some x -> Pdf.getnum pdf x
| None -> 1. | None -> 1.
in in
images := (pagenum, name, Image (int_of_float width, int_of_float height))::!images images := (pagenum, name, Image (int_of_float width, int_of_float height), objnum)::!images
| Some (Pdf.Name "/Form") -> | Some (Pdf.Name "/Form") ->
let resources = let resources =
match Pdf.lookup_direct pdf "/Resources" xobject with match Pdf.lookup_direct pdf "/Resources" xobject with
@ -301,7 +298,7 @@ and image_resolution pdf range dpi =
Pdftransform.d = Pdf.getnum pdf d; Pdftransform.e = Pdf.getnum pdf e; Pdftransform.f = Pdf.getnum pdf f} Pdftransform.d = Pdf.getnum pdf d; Pdftransform.e = Pdf.getnum pdf e; Pdftransform.f = Pdf.getnum pdf f}
| _ -> Pdftransform.i_matrix | _ -> Pdftransform.i_matrix
in in
images := (pagenum, name, Form (matrix, contents, resources))::!images images := (pagenum, name, Form (matrix, contents, resources), objnum)::!images
| _ -> () | _ -> ()
) )
xobjects xobjects
@ -311,8 +308,8 @@ and image_resolution pdf range dpi =
(* Now, split into differing pages, and call [image_resolution_page] on each one *) (* Now, split into differing pages, and call [image_resolution_page] on each one *)
let pagesplits = let pagesplits =
map map
(function (a, _, _)::_ as ls -> (a, ls) | _ -> assert false) (function (a, _, _, _)::_ as ls -> (a, ls) | _ -> assert false)
(collate (fun (a, _, _) (b, _, _) -> compare a b) (rev !images)) (collate (fun (a, _, _, _) (b, _, _, _) -> compare a b) (rev !images))
and pages = and pages =
Pdfpage.pages_of_pagetree pdf Pdfpage.pages_of_pagetree pdf
in in

2
cpdfimage.mli
View File

@ -9,7 +9,7 @@ val extract_images :
Cpdfmetadata.encoding -> bool -> bool -> Pdf.t -> int list -> string -> unit Cpdfmetadata.encoding -> bool -> bool -> Pdf.t -> int list -> string -> unit
(** Report image resolutions. *) (** Report image resolutions. *)
val image_resolution : Pdf.t -> int list -> float -> (int * string * int * int * float * float) list val image_resolution : Pdf.t -> int list -> float -> (int * string * int * int * float * float * int) list
(** List images in JSON format *) (** List images in JSON format *)
val images : Pdf.t -> int list -> Cpdfyojson.Safe.t val images : Pdf.t -> int list -> Cpdfyojson.Safe.t