fenix/cpdf-source
1
0
Fork 0
mirror of https://github.com/johnwhitington/cpdf-source.git synced 2025-06-05 22:09:39 +02:00
Code Issues Packages Projects Releases Wiki Activity

Second stab at naked_ops

Browse Source
This commit is contained in:
John Whitington
2025-01-29 16:53:34 +08:00
parent f5b3a1df04
commit 25a999eba3
1 changed files with 38 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
cpdfua.ml
View File

@ -153,8 +153,9 @@ let string_of_st st =
Cpdfyojson.Safe.pretty_to_string (convert st) Cpdfyojson.Safe.pretty_to_string (convert st)
(* Return a list of ops for all pages and form xobjects in a document. *) (* Return a list of ops for all pages and form xobjects in a document. *)
let all_ops pdf = let all_ops ?(xobjects=true) pdf =
let form_xobject_ops = let form_xobject_ops =
if xobjects then
let objnums = let objnums =
Pdf.objselect Pdf.objselect
(function Pdf.Stream s when Pdf.lookup_direct pdf "/Subtype" (Pdf.Stream s) = Some (Pdf.Name "/Form") -> true | _ -> false) (function Pdf.Stream s when Pdf.lookup_direct pdf "/Subtype" (Pdf.Stream s) = Some (Pdf.Name "/Form") -> true | _ -> false)
@ -166,6 +167,8 @@ let all_ops pdf =
let resources = match Pdf.lookup_direct pdf "/Resources" stream with Some d -> d | None -> Pdf.Dictionary [] in let resources = match Pdf.lookup_direct pdf "/Resources" stream with Some d -> d | None -> Pdf.Dictionary [] in
Pdfops.parse_operators pdf resources [stream]) Pdfops.parse_operators pdf resources [stream])
objnums objnums
else
[]
in in
let page_ops = let page_ops =
map map
@ -213,15 +216,34 @@ let matterhorn_01_004 _ _ pdf =
(* Which operations are real? *) (* Which operations are real? *)
let op_is_real = function let op_is_real = function
| Pdfops.( Op_m _ | Op_l _ | Op_c _ | Op_v _ | Op_y _ | Op_h | Op_re _ | Op_S | Op_s | Op_f | Op_F | Op_f' | Pdfops.( Op_m _ | Op_l _ | Op_c _ | Op_v _ | Op_y _ | Op_h | Op_re _ | Op_S | Op_s | Op_f | Op_F | Op_f'
| Op_B | Op_B' | Op_b | Op_b' | Op_n | Op_W | Op_W' | Op_BT | Op_ET | Op_Tj _ | Op_TJ _ | Op_' _ | Op_B | Op_B' | Op_b | Op_b' | Op_n | Op_W | Op_W' | Op_Tj _ | Op_TJ _ | Op_' _
| Op_'' _ | Op_sh _ | InlineImage _ | Op_Do _) -> true | Op_'' _ | Op_sh _ | InlineImage _ | Op_Do _) -> true
| _ -> false | _ -> false
(* Look at a list of ops and return operators neither marked as neither artifect nor content *) (* Look at a list of ops and return operators neither marked as neither
let naked_ops ops = [] artifect nor content. Assumes artifact/content markers not nested - so each
op is either unmarked, marked as artifact or marked as content. But, of
course, other marked content may be present. If it is, it must be properly
nested. *)
type mc = Artifact | Content | Other
let rec naked_ops acc stack = function
| [] -> rev acc
| Pdfops.Op_BDC (_, Pdf.Dictionary d)::t when lookup "/MCID" d <> None -> naked_ops acc (Content::stack) t
| Pdfops.Op_BDC _::t -> naked_ops acc (Other::stack) t
| Pdfops.Op_BMC "/Artifact"::t -> naked_ops acc (Artifact::stack) t
| Pdfops.Op_EMC::t ->
if stack = [] then (Printf.printf "Empty stack!\n"; merror ()) else naked_ops acc (tl stack) t
| h::t ->
if List.exists (function Artifact | Content -> true | _ -> false) stack
then naked_ops acc stack t
else naked_ops (h::acc) stack t
let print_ops =
iter (fun op -> Printf.printf "%s\n" (Pdfops.string_of_op op))
let matterhorn_01_005 _ _ pdf = let matterhorn_01_005 _ _ pdf =
iter (fun ops -> if List.exists op_is_real (naked_ops ops) then merror ()) (all_ops pdf) iter (fun ops -> if List.exists op_is_real (let n = naked_ops [] [] ops in print_ops n; n) then merror ()) (all_ops ~xobjects:false pdf)
(* Suspects entry has a value of true. *) (* Suspects entry has a value of true. *)
let matterhorn_01_007 _ _ pdf = let matterhorn_01_007 _ _ pdf =