More updates to support ContentStreamInstruction/InlineImage

3 files changed, 36 insertions, 7 deletions

diff --git a/src/pikepdf/_qpdf.pyi b/src/pikepdf/_qpdf.pyi
index 5017134..b15f572 100644
--- a/src/pikepdf/_qpdf.pyi
+++ b/src/pikepdf/_qpdf.pyi
@@ -663,6 +663,24 @@ class NameTree(MutableMapping[Union[str, bytes], Object]):
     @property
     def obj(self) -> Object: ...
 
+class ContentStreamInstruction:
+    @property
+    def operands(self) -> _ObjectList: ...
+    @property
+    def operator(self) -> Operator: ...
+    def __getitem__(self, index: int) -> Union[_ObjectList, Operator]: ...
+    def __len__(self) -> int: ...
+
+class ContentStreamInlineImage:
+    @property
+    def operands(self) -> _ObjectList: ...
+    @property
+    def operator(self) -> Operator: ...
+    def __getitem__(self, index: int) -> Union[_ObjectList, Operator]: ...
+    def __len__(self) -> int: ...
+    @property
+    def iimage(self) -> PdfInlineImage: ...
+
 def _Null() -> Any: ...
 def _encode(handle: Any) -> Object: ...
 def _new_array(arg0: Iterable) -> Object: ...
diff --git a/src/pikepdf/models/__init__.py b/src/pikepdf/models/__init__.py
index 7aa709d..f0de95c 100644
--- a/src/pikepdf/models/__init__.py
+++ b/src/pikepdf/models/__init__.py
@@ -21,8 +21,16 @@ from .outlines import (
 )
 
 # Operands, Operator
-ContentStreamOperands = Collection[Union[Object, PdfInlineImage]]
-ContentStreamInstructions = Tuple[ContentStreamOperands, Operator]
+_OldContentStreamOperands = Collection[Union[Object, PdfInlineImage]]
+_OldContentStreamInstructions = Tuple[_OldContentStreamOperands, Operator]
+
+ContentStreamInstructions = Union[
+    _qpdf.ContentStreamInstruction, _qpdf.ContentStreamInlineImage
+]
+
+UnparseableContentStreamInstructions = Union[
+    ContentStreamInstructions, _OldContentStreamInstructions
+]
 
 
 class PdfParsingError(Exception):
@@ -111,7 +119,7 @@ def parse_content_stream(
 
 
 def unparse_content_stream(
-    instructions: Collection[ContentStreamInstructions],
+    instructions: Collection[UnparseableContentStreamInstructions],
 ) -> bytes:
     """
     Given a parsed list of instructions/operand-operators, convert to bytes suitable
diff --git a/src/qpdf/parsers.cpp b/src/qpdf/parsers.cpp
index dcc8bae..5efc760 100644
--- a/src/qpdf/parsers.cpp
+++ b/src/qpdf/parsers.cpp
@@ -152,6 +152,9 @@ py::bytes unparse_content_stream(py::iterable contentstream)
     ss.imbue(std::locale::classic());
 
     for (const auto &item : contentstream) {
+        // First iteration: print nothing
+        // All others: print "\n" to delimit previous
+        // Result is no leading or trailing delimiter
         ss << delim;
         delim = "\n";
 
@@ -169,12 +172,10 @@ py::bytes unparse_content_stream(py::iterable contentstream)
         } catch (py::cast_error &) {
         }
 
+        // Fallback: instruction is some combination of Python iterables.
+        // Destructure and convert to C++ types...
         auto operands_op = py::reinterpret_borrow<py::sequence>(item);
 
-        // First iteration: print nothing
-        // All others: print "\n" to delimit previous
-        // Result is no leading or trailing delimiter
-
         if (operands_op.size() != 2) {
             errmsg << "Wrong number of operands at content stream instruction " << n
                    << "; expected 2";
@@ -253,6 +254,7 @@ void init_parsers(py::module_ &m)
                 throw py::index_error(
                     std::string("Invalid index ") + std::to_string(index));
             })
+        .def("__len__", [](ContentStreamInstruction &csi) { return 2; })
         .def("__repr__", [](ContentStreamInstruction &csi) {
             return "pikepdf.ContentStreamInstruction()";
         });
@@ -273,6 +275,7 @@ void init_parsers(py::module_ &m)
                 throw py::index_error(
                     std::string("Invalid index ") + std::to_string(index));
             })
+        .def("__len__", [](ContentStreamInlineImage &csii) { return 2; })
         .def_property_readonly("iimage",
             [](ContentStreamInlineImage &csii) { return csii.get_inline_image(); })
         .def("__repr__", [](ContentStreamInlineImage &csii) {