Can now filter page contents; very basic

3 files changed, 35 insertions, 3 deletions

diff --git a/src/qpdf/object.cpp b/src/qpdf/object.cpp
index 9963ff2..a644f03 100644
--- a/src/qpdf/object.cpp
+++ b/src/qpdf/object.cpp
@@ -825,7 +825,21 @@ void init_object(py::module& m)
     tokenfilter
         .def(py::init<>())
         .def("handle_token", &QPDFObjectHandle::TokenFilter::handleToken)
-        .def("handle_eof", &QPDFObjectHandle::TokenFilter::handleEOF);
+        .def("handle_eof", &QPDFObjectHandle::TokenFilter::handleEOF)
+        .def("write_token", &TokenFilterPublicist::writeToken)
+        ;
+
+    py::class_<QPDFTokenizer::Token> token(m, "Token");
+    token
+        .def_property_readonly("type_", &QPDFTokenizer::Token::getType)
+        .def_property_readonly("value", &QPDFTokenizer::Token::getValue)
+        .def_property_readonly("raw_value",
+            [](const QPDFTokenizer::Token& t) -> py::bytes {
+                return t.getRawValue();
+            }
+        )
+        .def_property_readonly("error_msg", &QPDFTokenizer::Token::getErrorMessage)
+        .def("__eq__", &QPDFTokenizer::Token::operator==);
 
     m.def("_encode",
         [](py::handle handle) {
diff --git a/src/qpdf/object_parsers.h b/src/qpdf/object_parsers.h
index df99b78..5b9675e 100644
--- a/src/qpdf/object_parsers.h
+++ b/src/qpdf/object_parsers.h
@@ -12,6 +12,7 @@
 #include <pybind11/stl.h>
 
 #include "pikepdf.h"
+#include <qpdf/QPDFTokenizer.hh>
 
 
 class PyParserCallbacks : public QPDFObjectHandle::ParserCallbacks {
@@ -151,7 +152,7 @@ public:
     }
 
     void handleEOF() override {
-        PYBIND11_OVERLOAD_PURE_NAME(
+        PYBIND11_OVERLOAD_NAME(
             void,
             QPDFObjectHandle::TokenFilter,
             "handle_eof", /* Python name */
@@ -159,3 +160,9 @@ public:
         );
     }
 };
+
+
+class TokenFilterPublicist : public QPDFObjectHandle::TokenFilter {
+public:
+    using QPDFObjectHandle::TokenFilter::writeToken;
+};
diff --git a/src/qpdf/page.cpp b/src/qpdf/page.cpp
index 4ea3b36..9d1ed90 100644
--- a/src/qpdf/page.cpp
+++ b/src/qpdf/page.cpp
@@ -12,9 +12,11 @@
 #include <cctype>
 
 #include "pikepdf.h"
+#include "object_parsers.h"
 
 #include <qpdf/QPDFPageObjectHelper.hh>
 #include <qpdf/Pipeline.hh>
+#include <qpdf/Pl_Buffer.hh>
 
 void init_page(py::module& m)
 {
@@ -32,6 +34,15 @@ void init_page(py::module& m)
         .def("_parse_page_contents", &QPDFPageObjectHelper::parsePageContents)
         .def("remove_unreferenced_resources", &QPDFPageObjectHelper::removeUnreferencedResources)
         .def("as_form_xobject", &QPDFPageObjectHelper::getFormXObjectForPage)
-        .def("_filter_page_contents", &QPDFPageObjectHelper::filterPageContents)
+        .def("_filter_page_contents",
+            [](QPDFPageObjectHelper &poh, QPDFObjectHandle::TokenFilter &tf) {
+                Pl_Buffer pl_buffer("filter_page");
+                poh.filterPageContents(&tf, &pl_buffer);
+                PointerHolder<Buffer> phbuf(pl_buffer.getBuffer());
+                poh.getObjectHandle().getKey("/Contents").replaceStreamData(
+                    phbuf, QPDFObjectHandle::newNull(), QPDFObjectHandle::newNull()
+                );
+            }
+        )
         ;
 }