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

Subformat identification for -info done

This commit is contained in:
John Whitington 2023-11-01 20:23:31 +00:00
parent 29ac4c8a64
commit 536110d7fe
1 changed files with 54 additions and 37 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

71
cpdfmetadata.ml
View File

@ -286,7 +286,7 @@ let pdfe = "http://www.aiim.org/pdfe/ns/id/"
let pdfuaid = "http://www.aiim.org/pdfua/ns/id/" let pdfuaid = "http://www.aiim.org/pdfua/ns/id/"
let pdfvtid = "http://www.npes.org/pdfvt/ns/id/" let pdfvtid = "http://www.npes.org/pdfvt/ns/id/"
let combine_with_spaces strs = let combine_with_commas strs =
String.trim String.trim
(fold_left (fun x y -> x ^ (if x <> "" then ", " else "") ^ y) "" strs) (fold_left (fun x y -> x ^ (if x <> "" then ", " else "") ^ y) "" strs)
@ -296,7 +296,7 @@ let collect_list_items = function
E (((n, n'), _), elts) when E (((n, n'), _), elts) when
n = rdf && (n' = "Alt" || n' = "Seq" || n' = "Bag") n = rdf && (n' = "Alt" || n' = "Seq" || n' = "Bag")
-> ->
combine_with_spaces combine_with_commas
(option_map (option_map
(function (function
E (((n, n'), _), [D d]) when n = rdf && n' = "li" -> E (((n, n'), _), [D d]) when n = rdf && n' = "li" ->
@ -321,43 +321,60 @@ let rec get_data_for namespace name = function
x :: _ -> Some x x :: _ -> Some x
| _ -> None | _ -> None
(* -PDF/A: <pdfaid:part>2</pdfaid:part> <pdfaid:conformance>B</pdfaid:conformance> (* PDF/A: <pdfaid:part>2</pdfaid:part> <pdfaid:conformance>B</pdfaid:conformance>
PDF/E: <pdfe:ISO_PDFEVersion>PDF/E-1</pdfe:ISO_PDFEVersion> <pdfxid:GTS_PDFXVersion>PDF/X-4</pdfxid:GTS_PDFXVersion> PDF/E: <pdfe:ISO_PDFEVersion>PDF/E-1</pdfe:ISO_PDFEVersion>
PDF/VT: <pdfxid:GTS_PDFXVersion>PDF/X-4</pdfxid:GTS_PDFXVersion> <pdfvtid:GTS_PDFVTVersion>PDF/VT-1</pdfvtid:GTS_PDFVTVersion> PDF/VT: <pdfvtid:GTS_PDFVTVersion>PDF/VT-1</pdfvtid:GTS_PDFVTVersion>
-PDF/UA: <pdfuaid:part>1</pdfuaid:part> PDF/UA: <pdfuaid:part>1</pdfuaid:part>
PDF/X: <pdfxid:GTS_PDFXVersion>PDF/X-4</pdfxid:GTS_PDFXVersion> (Fallback DID /GTS_PDFXVersion(PDF/X-1:2001)) *) PDF/X: <pdfxid:GTS_PDFXVersion>PDF/X-4</pdfxid:GTS_PDFXVersion> (Fallback DID /GTS_PDFXVersion) *)
(* FIXME Can it have multiple subformats? *)
let determine_subformat pdf = let determine_subformat pdf =
let formats = ref [] in
let fallback_pdfx () =
match Pdf.lookup_direct pdf "/Info" pdf.Pdf.trailerdict with
| Some d ->
begin match Pdf.lookup_direct pdf "/GTS_PDFXVersion" d with
| Some (Pdf.String s) -> formats =| s
| _ -> ()
end
| None -> ()
in
match get_metadata pdf with match get_metadata pdf with
| None -> "PDF" | None ->
fallback_pdfx ();
combine_with_commas !formats
| Some metadata -> | Some metadata ->
try
let _, tree = xmltree_of_bytes metadata in let _, tree = xmltree_of_bytes metadata in
flprint (string_of_xmltree tree);
(* PDF/E *) (* PDF/E *)
match get_data_for pdfe "ISO_PDFEVersion" tree with begin match get_data_for pdfe "ISO_PDFEVersion" tree with
| Some s -> s | Some s -> formats =| s
| None -> | None -> ()
end;
(* PDF/UA *) (* PDF/UA *)
match get_data_for pdfuaid "part" tree with begin match get_data_for pdfuaid "part" tree with
| Some s -> "PDF/UA-" ^ s | Some s -> formats =| "PDF/UA-" ^ s
| None -> | None -> ()
end;
(* PDF/A *) (* PDF/A *)
match get_data_for pdfaid "part" tree with begin match get_data_for pdfaid "part" tree with
| Some part -> | Some part ->
let conformance = let conformance =
match get_data_for pdfaid "conformance" tree with match get_data_for pdfaid "conformance" tree with
| Some s -> String.lowercase_ascii s | Some s -> String.lowercase_ascii s
| None -> "" | None -> ""
in in
"PDF/A-" ^ part ^ conformance formats =| "PDF/A-" ^ part ^ conformance
| None -> | None -> ()
"PDFn" end;
with (* PDF/X *)
_ -> begin match get_data_for pdfxid "GTS_PDFXVersion" tree with
(* Fallback DID /GTS... *) | Some s -> formats =| s
"qPDF" | None -> fallback_pdfx ()
end;
(* PDF/VT *)
begin match get_data_for pdfvtid "GTS_PDFVTVersion" tree with
| Some s -> formats =| s
| None -> ()
end;
combine_with_commas !formats
let output_xmp_info ?(json=ref [("none", `Null)]) encoding pdf = let output_xmp_info ?(json=ref [("none", `Null)]) encoding pdf =
let notjson = !json = [("none", `Null)] in let notjson = !json = [("none", `Null)] in
@ -373,7 +390,7 @@ let output_xmp_info ?(json=ref [("none", `Null)]) encoding pdf =
else else
json =| (title, `String data) json =| (title, `String data)
in in
Printf.printf "Subformat: %s\n" (determine_subformat pdf); Printf.printf "Subformats: %s\n" (determine_subformat pdf);
match get_metadata pdf with match get_metadata pdf with
None -> () None -> ()
| Some metadata -> | Some metadata ->