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

more

This commit is contained in:
John Whitington 2021-12-21 15:25:59 +00:00
parent c711e3aa77
commit 9bdeccb343
8 changed files with 320 additions and 333 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Makefile
View File

@ -3,7 +3,7 @@ MODS = cpdfyojson cpdfxmlm \
cpdfunicodedata cpdferror cpdfdebug cpdfjson cpdfstrftime cpdfcoord \
cpdfattach cpdfpagespec cpdfposition cpdfpresent cpdfmetadata \
cpdfbookmarks cpdfpage cpdfaddtext cpdf cpdfimage cpdffont cpdftype \
cpdftexttopdf cpdftoc cpdfpad cpdfocg cpdfsqueeze cpdfspot \
cpdftexttopdf cpdftoc cpdfpad cpdfocg cpdfsqueeze cpdfdraft cpdfspot \
cpdfpagelabels cpdfcreate cpdfannot cpdfcommand
SOURCES = $(foreach x,$(MODS),$(x).ml $(x).mli) cpdfcommandrun.ml

318
cpdf.ml
View File

@ -1421,167 +1421,6 @@ let thinlines range width pdf =
in
Cpdfpage.process_pages (ppstub thinpage) pdf range
(* \section{Making draft documents} *)
(* Predicate on an xobject: true if an image xobject. *)
let isimage pdf (_, xobj) =
match Pdf.lookup_direct pdf "/Subtype" xobj with
| Some (Pdf.Name "/Image") -> true
| _ -> false
(* Given a set of resources for a page, and the name of a resource, determine if
that name refers to an image xobject. *)
let xobject_isimage pdf resources name =
match resources with
| Pdf.Dictionary _ ->
begin match Pdf.lookup_direct pdf "/XObject" resources with
| Some xobjects ->
isimage pdf ("", Pdf.lookup_fail "xobject not there" pdf name xobjects)
| _ -> false
end
| _ -> failwith "bad resources"
(* The subsitute for an image. *)
let substitute boxes =
if boxes then
rev
[Pdfops.Op_q;
Pdfops.Op_w 0.;
Pdfops.Op_G 0.;
Pdfops.Op_re (0., 0., 1., 1.);
Pdfops.Op_m (0., 0.);
Pdfops.Op_l (1., 1.);
Pdfops.Op_m (0., 1.);
Pdfops.Op_l (1., 0.);
Pdfops.Op_S;
Pdfops.Op_Q]
else
[]
(* Remove references to images from a graphics stream. *)
let rec remove_images_stream onlyremove boxes pdf resources prev = function
| [] -> rev prev
| (Pdfops.Op_Do name) as h::t ->
if xobject_isimage pdf resources name && (match onlyremove with None -> true | Some x -> x = name)
then remove_images_stream onlyremove boxes pdf resources (substitute boxes @ prev) t
else remove_images_stream onlyremove boxes pdf resources (h::prev) t
| Pdfops.InlineImage _ as h::t ->
if onlyremove <> None
then remove_images_stream onlyremove boxes pdf resources (h::prev) t
else remove_images_stream onlyremove boxes pdf resources (substitute boxes @ prev) t
| h::t ->
remove_images_stream onlyremove boxes pdf resources (h::prev) t
let rec process_form_xobject onlyremove boxes pdf form =
let form = Pdf.direct pdf form in
let page =
{Pdfpage.content = [form];
Pdfpage.mediabox = Pdf.Null;
Pdfpage.resources =
begin match Pdf.lookup_direct pdf "/Resources" form with
| Some r -> r
| None -> Pdf.Dictionary []
end;
Pdfpage.rotate = Pdfpage.Rotate0;
Pdfpage.rest = Pdf.Dictionary []}
in
let page', pdf =
remove_images_page onlyremove boxes pdf page
in
let form' =
match form with
| Pdf.Stream {contents = (dict, _)} ->
begin match
Pdfops.stream_of_ops
(Pdfops.parse_operators pdf (Pdf.Dictionary []) page'.Pdfpage.content)
with
| Pdf.Stream {contents = (_, Pdf.Got data)} ->
let dict' =
Pdf.add_dict_entry dict "/Length" (Pdf.Integer (bytes_size data))
in
Pdf.Stream {contents = (dict', Pdf.Got data)}
| _ -> assert false
end
| _ -> raise (Pdf.PDFError "not a stream")
in
form', pdf
(* Remove images from a page. *)
and remove_images_page onlyremove boxes pdf page =
let isform pdf xobj =
match Pdf.lookup_direct pdf "/Subtype" xobj with Some (Pdf.Name "/Form") -> true | _ -> false
in
let isimage pdf xobj =
match Pdf.lookup_direct pdf "/Subtype" xobj with Some (Pdf.Name "/Image") -> true | _ -> false
in
(* Remove image xobjects and look into form ones *)
let form_xobjects, image_xobjects =
match Pdf.lookup_direct pdf "/XObject" page.Pdfpage.resources with
| Some (Pdf.Dictionary elts) ->
keep (function (_, p) -> isform pdf p) elts,
keep (function (_, p) -> isimage pdf p) elts
| _ -> [], []
in
let resources', pdf =
let names, pointers = split form_xobjects in
let form_xobjects', pdf =
let pdf = ref pdf
in let outputs = ref [] in
iter
(fun p ->
let p', pdf' = process_form_xobject onlyremove boxes !pdf p in
pdf := pdf';
outputs =| p')
pointers;
rev !outputs, !pdf
in
let nums = ref [] in
iter
(fun xobj ->
let objnum = Pdf.addobj pdf xobj in
nums =| objnum)
form_xobjects';
let image_xobjects' =
match onlyremove with
None -> []
| Some n -> option_map (function (n', _) as xobj -> if n = n' then None else Some xobj) image_xobjects
in
let newdict =
Pdf.Dictionary (image_xobjects' @ combine names (map (fun x -> Pdf.Indirect x) (rev !nums)))
in
Pdf.add_dict_entry page.Pdfpage.resources "/XObject" newdict, pdf
in
let content' =
remove_images_stream onlyremove boxes pdf page.Pdfpage.resources []
(Pdfops.parse_operators pdf page.Pdfpage.resources page.Pdfpage.content)
in
{page with
Pdfpage.content =
(let stream = Pdfops.stream_of_ops content' in
Pdfcodec.encode_pdfstream pdf Pdfcodec.Flate stream;
[stream]);
Pdfpage.resources = resources'}, pdf
(* Remove images from all pages in a document. *)
let draft onlyremove boxes range pdf =
let pages = Pdfpage.pages_of_pagetree pdf in
let pagenums = indx pages in
let pdf = ref pdf
in let pages' = ref [] in
iter2
(fun p pagenum ->
let p', pdf' =
if mem pagenum range
then remove_images_page onlyremove boxes !pdf p
else p, !pdf
in
pdf := pdf';
pages' =| p')
pages
pagenums;
Pdfpage.change_pages true !pdf (rev !pages')
(* Parse the new content to make sure syntactically ok, append
* as required. Rewrite the content *)
let append_page_content_page fast s before pdf n page =
@ -1776,153 +1615,6 @@ let remove_clipping pdf range =
in
Cpdfpage.process_pages (ppstub remove_clipping_page) pdf range
(* Image resolution *)
type xobj =
| Image of int * int (* width, height *)
| Form of Pdftransform.transform_matrix * Pdf.pdfobject * Pdf.pdfobject (* Will add actual data later. *)
let image_results = ref []
let add_image_result i =
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
let pageops = Pdfops.parse_operators pdf page.Pdfpage.resources page.Pdfpage.content
and transform = ref [ref Pdftransform.i_matrix] in
iter
(function
| Pdfops.Op_cm matrix ->
begin match !transform with
| [] -> raise (Failure "no transform")
| _ -> (hd !transform) := Pdftransform.matrix_compose !(hd !transform) matrix
end
| Pdfops.Op_Do xobject ->
let trans (x, y) =
match !transform with
| [] -> raise (Failure "no transform")
| _ -> Pdftransform.transform_matrix !(hd !transform) (x, y)
in
let o = trans (0., 0.)
and x = trans (1., 0.)
and y = trans (0., 1.)
in
(*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
| [] -> assert false
| (_, a, _) as h::_ when a = k -> h
| _::t -> lookup_image k t
in
begin match lookup_image xobject images with
| (pagenum, name, Form (xobj_matrix, content, resources)) ->
let content =
(* Add in matrix etc. *)
let total_matrix = Pdftransform.matrix_compose xobj_matrix !(hd !transform) in
let ops =
Pdfops.Op_cm total_matrix::
Pdfops.parse_operators pdf resources [content]
in
Pdfops.stream_of_ops ops
in
let page =
{Pdfpage.content = [content];
Pdfpage.mediabox = Pdfpage.rectangle_of_paper Pdfpaper.a4;
Pdfpage.resources = resources;
Pdfpage.rotate = Pdfpage.Rotate0;
Pdfpage.rest = Pdf.Dictionary []}
in
let newpdf = Pdfpage.change_pages false pdf [page] in
image_resolution newpdf [pagenum] dpi
| (pagenum, name, Image (w, h)) ->
let lx = Pdfunits.convert 0. Pdfunits.PdfPoint Pdfunits.Inch (distance_between o x)
and ly = Pdfunits.convert 0. Pdfunits.PdfPoint Pdfunits.Inch (distance_between o y) in
let wdpi = float w /. lx
and hdpi = float h /. ly in
add_image_result (pagenum, xobject, w, h, wdpi, hdpi)
(*Printf.printf "%i, %s, %i, %i, %f, %f\n" pagenum xobject w h wdpi hdpi*)
(*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*)
end
| Pdfops.Op_q ->
begin match !transform with
| [] -> raise (Failure "Unbalanced q/Q ops")
| h::t ->
let h' = ref Pdftransform.i_matrix in
h' := !h;
transform := h'::h::t
end
| Pdfops.Op_Q ->
begin match !transform with
| [] -> raise (Failure "Unbalanced q/Q ops")
| _ -> transform := tl !transform
end
| _ -> ())
pageops
with
e -> Printf.printf "Error %s\n" (Printexc.to_string e); flprint "\n"
and image_resolution pdf range dpi =
let images = ref [] in
Cpdfpage.iter_pages
(fun pagenum page ->
(* 1. Get all image names and their native resolutions from resources as string * int * int *)
match Pdf.lookup_direct pdf "/XObject" page.Pdfpage.resources with
| Some (Pdf.Dictionary xobjects) ->
iter
(function (name, xobject) ->
match Pdf.lookup_direct pdf "/Subtype" xobject with
| Some (Pdf.Name "/Image") ->
let width =
match Pdf.lookup_direct pdf "/Width" xobject with
| Some x -> Pdf.getnum x
| None -> 1.
and height =
match Pdf.lookup_direct pdf "/Height" xobject with
| Some x -> Pdf.getnum x
| None -> 1.
in
images := (pagenum, name, Image (int_of_float width, int_of_float height))::!images
| Some (Pdf.Name "/Form") ->
let resources =
match Pdf.lookup_direct pdf "/Resources" xobject with
| None -> page.Pdfpage.resources (* Inherit from page or form above. *)
| Some r -> r
and contents =
xobject
and matrix =
match Pdf.lookup_direct pdf "/Matrix" xobject with
| Some (Pdf.Array [a; b; c; d; e; f]) ->
{Pdftransform.a = Pdf.getnum a; Pdftransform.b = Pdf.getnum b; Pdftransform.c = Pdf.getnum c;
Pdftransform.d = Pdf.getnum d; Pdftransform.e = Pdf.getnum e; Pdftransform.f = Pdf.getnum f}
| _ -> Pdftransform.i_matrix
in
images := (pagenum, name, Form (matrix, contents, resources))::!images
| _ -> ()
)
xobjects
| _ -> ())
pdf
range;
(* Now, split into differing pages, and call [image_resolution_page] on each one *)
let pagesplits =
map
(function (a, _, _)::_ as ls -> (a, ls) | _ -> assert false)
(collate (fun (a, _, _) (b, _, _) -> compare a b) (rev !images))
and pages =
Pdfpage.pages_of_pagetree pdf
in
iter
(function (pagenum, images) ->
let page = select pagenum pages in
image_resolution_page pdf page pagenum dpi images)
pagesplits
let image_resolution pdf range dpi =
image_results := [];
image_resolution pdf range dpi;
rev !image_results
(* copy the contents of the box f to the box t. If mediabox_if_missing is set,
the contents of the mediabox will be used if the from fox is not available. If
mediabox_is_missing is false, the page is unaltered. *)
@ -1964,13 +1656,3 @@ let remove_unused_resources_page pdf n page =
let remove_unused_resources pdf =
Cpdfpage.process_pages (ppstub (remove_unused_resources_page pdf)) pdf (ilist 1 (Pdfpage.endpage pdf))
let create_pdf pages pagesize =
let page =
{(Pdfpage.blankpage pagesize) with
Pdfpage.content = [Pdfops.stream_of_ops []];
Pdfpage.resources = Pdf.Dictionary []}
in
let pdf, pageroot = Pdfpage.add_pagetree (many page pages) (Pdf.empty ()) in
Pdfpage.add_root pageroot [] pdf

10
cpdf.mli
View File

@ -113,9 +113,6 @@ val blacklines : Cpdfaddtext.color -> int list -> Pdf.t -> Pdf.t
(** Make all fills on certain pages black. *)
val blackfills : Cpdfaddtext.color -> int list -> Pdf.t -> Pdf.t
(** Remove images from a PDF, optionally adding crossed boxes. *)
val draft : string option -> bool -> int list -> Pdf.t -> Pdf.t
val remove_all_text : int list -> Pdf.t -> Pdf.t
(**/**)
@ -134,13 +131,6 @@ val print_dict_entry : Pdf.t -> string -> unit
val remove_clipping : Pdf.t -> int list -> Pdf.t
val image_resolution : Pdf.t -> int list -> float -> (int * string * int * int * float * float) list
val copy_box : string -> string -> bool -> Pdf.t -> int list -> Pdf.t
val remove_unused_resources : Pdf.t -> Pdf.t
val create_pdf : int -> Pdfpaper.t -> Pdf.t

6
cpdfcommand.ml
View File

@ -3649,7 +3649,7 @@ let go () =
| Some Draft ->
let pdf = get_single_pdf args.op false in
let range = parse_pagespec_allow_empty pdf (get_pagespec ()) in
write_pdf false (Cpdf.draft args.removeonly args.boxes range pdf)
write_pdf false (Cpdfdraft.draft args.removeonly args.boxes range pdf)
| Some (AddText text) ->
let pdf = get_single_pdf args.op false in
let range = parse_pagespec_allow_empty pdf (get_pagespec ()) in
@ -3778,7 +3778,7 @@ let go () =
| Some (ImageResolution f) ->
let pdf = get_single_pdf args.op true in
let range = parse_pagespec_allow_empty pdf (get_pagespec ()) in
let images = Cpdf.image_resolution pdf range f in
let images = Cpdfimage.image_resolution pdf range f in
iter
(function (pagenum, xobject, w, h, wdpi, hdpi) ->
if wdpi < f || hdpi < f then
@ -3854,7 +3854,7 @@ let go () =
let pdf = get_single_pdf args.op false in
write_pdf false (Cpdfbookmarks.bookmarks_open_to_level n pdf)
| Some CreatePDF ->
let pdf = Cpdf.create_pdf args.createpdf_pages args.createpdf_pagesize in
let pdf = Cpdfcreate.blank_document_paper args.createpdf_pagesize args.createpdf_pages in
write_pdf false pdf
| Some RemoveAllText ->
let pdf = get_single_pdf args.op false in

163
cpdfdraft.ml Normal file
View File

@ -0,0 +1,163 @@
open Pdfutil
open Pdfio
(* \section{Making draft documents} *)
(* Predicate on an xobject: true if an image xobject. *)
let isimage pdf (_, xobj) =
match Pdf.lookup_direct pdf "/Subtype" xobj with
| Some (Pdf.Name "/Image") -> true
| _ -> false
(* Given a set of resources for a page, and the name of a resource, determine if
that name refers to an image xobject. *)
let xobject_isimage pdf resources name =
match resources with
| Pdf.Dictionary _ ->
begin match Pdf.lookup_direct pdf "/XObject" resources with
| Some xobjects ->
isimage pdf ("", Pdf.lookup_fail "xobject not there" pdf name xobjects)
| _ -> false
end
| _ -> failwith "bad resources"
(* The subsitute for an image. *)
let substitute boxes =
if boxes then
rev
[Pdfops.Op_q;
Pdfops.Op_w 0.;
Pdfops.Op_G 0.;
Pdfops.Op_re (0., 0., 1., 1.);
Pdfops.Op_m (0., 0.);
Pdfops.Op_l (1., 1.);
Pdfops.Op_m (0., 1.);
Pdfops.Op_l (1., 0.);
Pdfops.Op_S;
Pdfops.Op_Q]
else
[]
(* Remove references to images from a graphics stream. *)
let rec remove_images_stream onlyremove boxes pdf resources prev = function
| [] -> rev prev
| (Pdfops.Op_Do name) as h::t ->
if xobject_isimage pdf resources name && (match onlyremove with None -> true | Some x -> x = name)
then remove_images_stream onlyremove boxes pdf resources (substitute boxes @ prev) t
else remove_images_stream onlyremove boxes pdf resources (h::prev) t
| Pdfops.InlineImage _ as h::t ->
if onlyremove <> None
then remove_images_stream onlyremove boxes pdf resources (h::prev) t
else remove_images_stream onlyremove boxes pdf resources (substitute boxes @ prev) t
| h::t ->
remove_images_stream onlyremove boxes pdf resources (h::prev) t
let rec process_form_xobject onlyremove boxes pdf form =
let form = Pdf.direct pdf form in
let page =
{Pdfpage.content = [form];
Pdfpage.mediabox = Pdf.Null;
Pdfpage.resources =
begin match Pdf.lookup_direct pdf "/Resources" form with
| Some r -> r
| None -> Pdf.Dictionary []
end;
Pdfpage.rotate = Pdfpage.Rotate0;
Pdfpage.rest = Pdf.Dictionary []}
in
let page', pdf =
remove_images_page onlyremove boxes pdf page
in
let form' =
match form with
| Pdf.Stream {contents = (dict, _)} ->
begin match
Pdfops.stream_of_ops
(Pdfops.parse_operators pdf (Pdf.Dictionary []) page'.Pdfpage.content)
with
| Pdf.Stream {contents = (_, Pdf.Got data)} ->
let dict' =
Pdf.add_dict_entry dict "/Length" (Pdf.Integer (bytes_size data))
in
Pdf.Stream {contents = (dict', Pdf.Got data)}
| _ -> assert false
end
| _ -> raise (Pdf.PDFError "not a stream")
in
form', pdf
(* Remove images from a page. *)
and remove_images_page onlyremove boxes pdf page =
let isform pdf xobj =
match Pdf.lookup_direct pdf "/Subtype" xobj with Some (Pdf.Name "/Form") -> true | _ -> false
in
let isimage pdf xobj =
match Pdf.lookup_direct pdf "/Subtype" xobj with Some (Pdf.Name "/Image") -> true | _ -> false
in
(* Remove image xobjects and look into form ones *)
let form_xobjects, image_xobjects =
match Pdf.lookup_direct pdf "/XObject" page.Pdfpage.resources with
| Some (Pdf.Dictionary elts) ->
keep (function (_, p) -> isform pdf p) elts,
keep (function (_, p) -> isimage pdf p) elts
| _ -> [], []
in
let resources', pdf =
let names, pointers = split form_xobjects in
let form_xobjects', pdf =
let pdf = ref pdf
in let outputs = ref [] in
iter
(fun p ->
let p', pdf' = process_form_xobject onlyremove boxes !pdf p in
pdf := pdf';
outputs =| p')
pointers;
rev !outputs, !pdf
in
let nums = ref [] in
iter
(fun xobj ->
let objnum = Pdf.addobj pdf xobj in
nums =| objnum)
form_xobjects';
let image_xobjects' =
match onlyremove with
None -> []
| Some n -> option_map (function (n', _) as xobj -> if n = n' then None else Some xobj) image_xobjects
in
let newdict =
Pdf.Dictionary (image_xobjects' @ combine names (map (fun x -> Pdf.Indirect x) (rev !nums)))
in
Pdf.add_dict_entry page.Pdfpage.resources "/XObject" newdict, pdf
in
let content' =
remove_images_stream onlyremove boxes pdf page.Pdfpage.resources []
(Pdfops.parse_operators pdf page.Pdfpage.resources page.Pdfpage.content)
in
{page with
Pdfpage.content =
(let stream = Pdfops.stream_of_ops content' in
Pdfcodec.encode_pdfstream pdf Pdfcodec.Flate stream;
[stream]);
Pdfpage.resources = resources'}, pdf
(* Remove images from all pages in a document. *)
let draft onlyremove boxes range pdf =
let pages = Pdfpage.pages_of_pagetree pdf in
let pagenums = indx pages in
let pdf = ref pdf
in let pages' = ref [] in
iter2
(fun p pagenum ->
let p', pdf' =
if mem pagenum range
then remove_images_page onlyremove boxes !pdf p
else p, !pdf
in
pdf := pdf';
pages' =| p')
pages
pagenums;
Pdfpage.change_pages true !pdf (rev !pages')

2
cpdfdraft.mli Normal file
View File

@ -0,0 +1,2 @@
(** Remove images from a PDF, optionally adding crossed boxes. *)
val draft : string option -> bool -> int list -> Pdf.t -> Pdf.t

148
cpdfimage.ml
View File

@ -125,3 +125,151 @@ let extract_images path_to_p2p path_to_im encoding dedup dedup_per_page pdf rang
iter (extract_images_form_xobject path_to_p2p path_to_im encoding dedup dedup_per_page pdf serial stem pnum) forms)
pages
(indx pages)
(* Image resolution *)
type xobj =
| Image of int * int (* width, height *)
| Form of Pdftransform.transform_matrix * Pdf.pdfobject * Pdf.pdfobject (* Will add actual data later. *)
let image_results = ref []
let add_image_result i =
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
let pageops = Pdfops.parse_operators pdf page.Pdfpage.resources page.Pdfpage.content
and transform = ref [ref Pdftransform.i_matrix] in
iter
(function
| Pdfops.Op_cm matrix ->
begin match !transform with
| [] -> raise (Failure "no transform")
| _ -> (hd !transform) := Pdftransform.matrix_compose !(hd !transform) matrix
end
| Pdfops.Op_Do xobject ->
let trans (x, y) =
match !transform with
| [] -> raise (Failure "no transform")
| _ -> Pdftransform.transform_matrix !(hd !transform) (x, y)
in
let o = trans (0., 0.)
and x = trans (1., 0.)
and y = trans (0., 1.)
in
(*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
| [] -> assert false
| (_, a, _) as h::_ when a = k -> h
| _::t -> lookup_image k t
in
begin match lookup_image xobject images with
| (pagenum, name, Form (xobj_matrix, content, resources)) ->
let content =
(* Add in matrix etc. *)
let total_matrix = Pdftransform.matrix_compose xobj_matrix !(hd !transform) in
let ops =
Pdfops.Op_cm total_matrix::
Pdfops.parse_operators pdf resources [content]
in
Pdfops.stream_of_ops ops
in
let page =
{Pdfpage.content = [content];
Pdfpage.mediabox = Pdfpage.rectangle_of_paper Pdfpaper.a4;
Pdfpage.resources = resources;
Pdfpage.rotate = Pdfpage.Rotate0;
Pdfpage.rest = Pdf.Dictionary []}
in
let newpdf = Pdfpage.change_pages false pdf [page] in
image_resolution newpdf [pagenum] dpi
| (pagenum, name, Image (w, h)) ->
let lx = Pdfunits.convert 0. Pdfunits.PdfPoint Pdfunits.Inch (distance_between o x)
and ly = Pdfunits.convert 0. Pdfunits.PdfPoint Pdfunits.Inch (distance_between o y) in
let wdpi = float w /. lx
and hdpi = float h /. ly in
add_image_result (pagenum, xobject, w, h, wdpi, hdpi)
(*Printf.printf "%i, %s, %i, %i, %f, %f\n" pagenum xobject w h wdpi hdpi*)
(*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*)
end
| Pdfops.Op_q ->
begin match !transform with
| [] -> raise (Failure "Unbalanced q/Q ops")
| h::t ->
let h' = ref Pdftransform.i_matrix in
h' := !h;
transform := h'::h::t
end
| Pdfops.Op_Q ->
begin match !transform with
| [] -> raise (Failure "Unbalanced q/Q ops")
| _ -> transform := tl !transform
end
| _ -> ())
pageops
with
e -> Printf.printf "Error %s\n" (Printexc.to_string e); flprint "\n"
and image_resolution pdf range dpi =
let images = ref [] in
Cpdfpage.iter_pages
(fun pagenum page ->
(* 1. Get all image names and their native resolutions from resources as string * int * int *)
match Pdf.lookup_direct pdf "/XObject" page.Pdfpage.resources with
| Some (Pdf.Dictionary xobjects) ->
iter
(function (name, xobject) ->
match Pdf.lookup_direct pdf "/Subtype" xobject with
| Some (Pdf.Name "/Image") ->
let width =
match Pdf.lookup_direct pdf "/Width" xobject with
| Some x -> Pdf.getnum x
| None -> 1.
and height =
match Pdf.lookup_direct pdf "/Height" xobject with
| Some x -> Pdf.getnum x
| None -> 1.
in
images := (pagenum, name, Image (int_of_float width, int_of_float height))::!images
| Some (Pdf.Name "/Form") ->
let resources =
match Pdf.lookup_direct pdf "/Resources" xobject with
| None -> page.Pdfpage.resources (* Inherit from page or form above. *)
| Some r -> r
and contents =
xobject
and matrix =
match Pdf.lookup_direct pdf "/Matrix" xobject with
| Some (Pdf.Array [a; b; c; d; e; f]) ->
{Pdftransform.a = Pdf.getnum a; Pdftransform.b = Pdf.getnum b; Pdftransform.c = Pdf.getnum c;
Pdftransform.d = Pdf.getnum d; Pdftransform.e = Pdf.getnum e; Pdftransform.f = Pdf.getnum f}
| _ -> Pdftransform.i_matrix
in
images := (pagenum, name, Form (matrix, contents, resources))::!images
| _ -> ()
)
xobjects
| _ -> ())
pdf
range;
(* Now, split into differing pages, and call [image_resolution_page] on each one *)
let pagesplits =
map
(function (a, _, _)::_ as ls -> (a, ls) | _ -> assert false)
(collate (fun (a, _, _) (b, _, _) -> compare a b) (rev !images))
and pages =
Pdfpage.pages_of_pagetree pdf
in
iter
(function (pagenum, images) ->
let page = select pagenum pages in
image_resolution_page pdf page pagenum dpi images)
pagesplits
let image_resolution pdf range dpi =
image_results := [];
image_resolution pdf range dpi;
rev !image_results

2
cpdfimage.mli
View File

@ -1,3 +1,5 @@
val extract_images : string ->
string ->
Cpdfmetadata.encoding -> bool -> bool -> Pdf.t -> int list -> string -> unit
val image_resolution : Pdf.t -> int list -> float -> (int * string * int * int * float * float) list