WIP: PDF 1.5 cross-reference stream support

Perhaps only for Go.

1 files changed, 39 insertions, 1 deletions

diff --git a/pdf/pdf.go b/pdf/pdf.go
index 92d18e9..e9f42a7 100644
--- a/pdf/pdf.go
+++ b/pdf/pdf.go
@@ -671,11 +671,49 @@ func (u *Updater) parse(lex *Lexer, stack *[]Object) (Object, error) {
 	}
 }
 
+// TODO: ISO 32000-2:2020 7.5.8.2 Cross-reference stream dictionary
+//   - How to update this kind of file? Hybrid-reference file?
+//     ⋄ XRefStm could point to this old-new xref. But it's actually a /Prev.
+func (u *Updater) loadXrefStream(
+	lex *Lexer, loadedEntries map[uint]struct{}) error {
+	var stack []Object
+	for {
+		object, err := u.parse(lex, &stack)
+		if err != nil {
+			return fmt.Errorf("invalid xref table: %s", err)
+		} else if object.Kind == End {
+			return errors.New("invalid xref table")
+		}
+
+		// For the sake of simplicity, keep stacking until we find an object.
+		if object.Kind != Indirect {
+			stack = append(stack, object)
+			continue
+		}
+
+		object = object.Array[0]
+		if object.Kind != Stream {
+			return errors.New("invalid xref table")
+		}
+		if typ, ok := object.Dict["Type"]; !ok || typ.String != "XRef" {
+			return errors.New("invalid xref stream")
+		}
+		if filter, ok := object.Dict["Filter"]; !ok ||
+			filter.String != "FlateDecode" {
+			return errors.New("invalid xref stream")
+		}
+
+		return errors.New("unimplemented")
+	}
+}
+
+// TODO: Also return the trailer dictionary, but remove any XRef/stream fields,
+// perhaps as per Table 17 + /Filter, /Length.
 func (u *Updater) loadXref(lex *Lexer, loadedEntries map[uint]struct{}) error {
 	var throwawayStack []Object
 	if keyword, _ := u.parse(lex,
 		&throwawayStack); keyword.Kind != Keyword || keyword.String != "xref" {
-		return errors.New("invalid xref table")
+		return u.loadXrefStream(lex, loadedEntries)
 	}
 	for {
 		object, _ := u.parse(lex, &throwawayStack)