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 2023-05-04 15:01:12 +01:00
parent 7351e73a3f
commit 04b395a526
1 changed files with 43 additions and 45 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

88
cpdfdraw.ml
View File

@ -55,43 +55,39 @@ type drawops =
| EndURL | EndURL
(* Per page resources *) (* Per page resources *)
let images = null_hash () type res =
let gss = null_hash () {images : (string, (string * int)) Hashtbl.t; (* (name, (pdf name, objnum)) *)
let current_url = ref None extgstates : (string, Pdf.pdfobject) Hashtbl.t; (* pdf name, pdf object *)
let fonts = null_hash () fonts : (string, int) Hashtbl.t; (* (pdf name, objnum)) *)
let form_xobjects = null_hash () form_xobjects : (string, int) Hashtbl.t; (* (pdf name, objnum)) *)
mutable time : Cpdfstrftime.t;
mutable current_url : string option;
mutable current_font : Pdftext.font;
mutable num : int}
let current_font = let res =
ref (Pdftext.StandardFont (Pdftext.TimesRoman, Pdftext.WinAnsiEncoding)) {images = null_hash ();
extgstates = null_hash ();
fonts = null_hash ();
form_xobjects = null_hash ();
time = Cpdfstrftime.dummy;
current_url = None;
current_font = Pdftext.StandardFont (Pdftext.TimesRoman, Pdftext.WinAnsiEncoding);
num = 0}
(* Fresh XObject names. If we are stamping over another page, manage clashes later. *) let fresh_name s =
let fresh_xobj_name () = "/Img0" res.num <- res.num + 1;
s ^ string_of_int res.num
let gsnum = ref ~-1 (* At end of page, we keep things for which we have indirects - we may use them on another page. *)
let fresh_gs_name () =
gsnum += 1;
"/gs" ^ string_of_int !gsnum
let fontnum = ref 0
let fresh_font_name pdf f =
fontnum += 1;
let n = "/F" ^ string_of_int !fontnum in
Hashtbl.add fonts n (Pdf.Indirect (Pdftext.write_font pdf f));
n
(* FIXME This will remove fonts, images etc, for moving on to the next page *)
let reset_state () = let reset_state () =
() Hashtbl.clear res.extgstates
let time = ref Cpdfstrftime.dummy
let process_specials pdf endpage filename bates batespad num page s = let process_specials pdf endpage filename bates batespad num page s =
let pairs = let pairs =
Cpdfaddtext.replace_pairs pdf endpage None filename bates batespad num page Cpdfaddtext.replace_pairs pdf endpage None filename bates batespad num page
in in
Cpdfaddtext.process_text !time s pairs Cpdfaddtext.process_text res.time s pairs
let rec ops_of_drawop pdf endpage filename bates batespad num page = function let rec ops_of_drawop pdf endpage filename bates batespad num page = function
| Push -> [Pdfops.Op_q] | Push -> [Pdfops.Op_q]
@ -130,39 +126,41 @@ let rec ops_of_drawop pdf endpage filename bates batespad num page = function
| SetDashPattern (x, y) -> [Pdfops.Op_d (x, y)] | SetDashPattern (x, y) -> [Pdfops.Op_d (x, y)]
| FormXObject (a, b, c, d, n, ops) -> create_form_xobject a b c d pdf endpage filename bates batespad num page n ops; [] | FormXObject (a, b, c, d, n, ops) -> create_form_xobject a b c d pdf endpage filename bates batespad num page n ops; []
| Use n -> [Pdfops.Op_Do n] | Use n -> [Pdfops.Op_Do n]
| Image s -> [Pdfops.Op_Do (try fst (Hashtbl.find images s) with _ -> Cpdferror.error ("Image not found: " ^ s))] | Image s -> [Pdfops.Op_Do (try fst (Hashtbl.find res.images s) with _ -> Cpdferror.error ("Image not found: " ^ s))]
| ImageXObject (s, obj) -> | ImageXObject (s, obj) ->
Hashtbl.add images s (fresh_xobj_name (), Pdf.addobj pdf obj); Hashtbl.add res.images s (fresh_name "/XObj", Pdf.addobj pdf obj);
[] []
| NewPage -> Pdfe.log ("NewPage remaining in graphic stream"); assert false | NewPage -> Pdfe.log ("NewPage remaining in graphic stream"); assert false
| Opacity v -> | Opacity v ->
let n = fresh_gs_name () in let n = fresh_name "/gs" in
Hashtbl.add gss n (Pdf.Dictionary [("/ca", Pdf.Real v)]); Hashtbl.add res.extgstates n (Pdf.Dictionary [("/ca", Pdf.Real v)]);
[Pdfops.Op_gs n] [Pdfops.Op_gs n]
| SOpacity v -> | SOpacity v ->
let n = fresh_gs_name () in let n = fresh_name "/gs" in
Hashtbl.add gss n (Pdf.Dictionary [("/CA", Pdf.Real v)]); Hashtbl.add res.extgstates n (Pdf.Dictionary [("/CA", Pdf.Real v)]);
[Pdfops.Op_gs n] [Pdfops.Op_gs n]
| URL s -> | URL s ->
current_url := Some s; res.current_url <- Some s;
[] []
| EndURL -> | EndURL ->
current_url := None; res.current_url <- None;
[] []
| Font (s, f) -> | Font (s, f) ->
let n = fresh_font_name pdf (Pdftext.StandardFont (s, Pdftext.WinAnsiEncoding)) in let o = Pdftext.write_font pdf (Pdftext.StandardFont (s, Pdftext.WinAnsiEncoding)) in
let n = fresh_name "/F" in
Hashtbl.add res.fonts n o;
[Pdfops.Op_Tf (n, f)] [Pdfops.Op_Tf (n, f)]
| BT -> [Pdfops.Op_BT] | BT -> [Pdfops.Op_BT]
| ET -> [Pdfops.Op_ET] | ET -> [Pdfops.Op_ET]
| Text s -> | Text s ->
let charcodes = let charcodes =
implode (map char_of_int (option_map (Pdftext.charcode_extractor_of_font_real !current_font) (Pdftext.codepoints_of_utf8 s))) implode (map char_of_int (option_map (Pdftext.charcode_extractor_of_font_real res.current_font) (Pdftext.codepoints_of_utf8 s)))
in in
[Pdfops.Op_Tj charcodes] [Pdfops.Op_Tj charcodes]
| SpecialText s -> | SpecialText s ->
let s = process_specials pdf endpage filename bates batespad num page s in let s = process_specials pdf endpage filename bates batespad num page s in
let charcodes = let charcodes =
implode (map char_of_int (option_map (Pdftext.charcode_extractor_of_font_real !current_font) (Pdftext.codepoints_of_utf8 s))) implode (map char_of_int (option_map (Pdftext.charcode_extractor_of_font_real res.current_font) (Pdftext.codepoints_of_utf8 s)))
in in
[Pdfops.Op_Tj charcodes] [Pdfops.Op_Tj charcodes]
| Leading f -> [Pdfops.Op_TL f] | Leading f -> [Pdfops.Op_TL f]
@ -190,7 +188,7 @@ and create_form_xobject a b c d pdf endpage filename bates batespad num page n o
], ],
Pdf.Got data)} Pdf.Got data)}
in in
Hashtbl.add form_xobjects n (Pdf.addobj pdf obj) Hashtbl.add res.form_xobjects n (Pdf.addobj pdf obj)
(* Draw all the accumulated operators. *) (* Draw all the accumulated operators. *)
let draw_single ~filename ~bates ~batespad fast range pdf drawops = let draw_single ~filename ~bates ~batespad fast range pdf drawops =
@ -209,11 +207,11 @@ let draw_single ~filename ~bates ~batespad fast range pdf drawops =
(ilist 1 endpage) (ilist 1 endpage)
ss; ss;
let pdf = !pdf in let pdf = !pdf in
let images = list_of_hashtbl images in let images = list_of_hashtbl res.images in
let image_resources = map (fun (_, (n, o)) -> (n, Pdf.Indirect o)) images in let image_resources = map (fun (_, (n, o)) -> (n, Pdf.Indirect o)) images in
let gss_resources = list_of_hashtbl gss in let gss_resources = list_of_hashtbl res.extgstates in
let font_resources = list_of_hashtbl fonts in let font_resources = map (fun (n, o) -> (n, Pdf.Indirect o)) (list_of_hashtbl res.fonts) in
let form_resources = map (fun (n, o) -> (n, Pdf.Indirect o)) (list_of_hashtbl form_xobjects) in let form_resources = map (fun (n, o) -> (n, Pdf.Indirect o)) (list_of_hashtbl res.form_xobjects) in
match images, gss_resources, font_resources, form_resources with [], [], [], [] -> pdf | _ -> match images, gss_resources, font_resources, form_resources with [], [], [], [] -> pdf | _ ->
let pages = Pdfpage.pages_of_pagetree pdf in let pages = Pdfpage.pages_of_pagetree pdf in
let pages = let pages =
@ -252,7 +250,7 @@ let draw_single ~filename ~bates ~batespad fast range pdf drawops =
Pdfpage.change_pages true pdf pages Pdfpage.change_pages true pdf pages
let draw ~filename ~bates ~batespad fast range pdf drawops = let draw ~filename ~bates ~batespad fast range pdf drawops =
time := Cpdfstrftime.current_time (); res.time <- Cpdfstrftime.current_time ();
let pdf = ref pdf in let pdf = ref pdf in
let range = ref range in let range = ref range in
let chunks = ref (split_around (eq NewPage) drawops) in let chunks = ref (split_around (eq NewPage) drawops) in