Pythonize.cxx

Go to the documentation of this file.

// Bindings
#include "cpyrt.h"
 
using namespace cppjit;
#include "CPPFunction.h"
#include "CPPInstance.h"
#include "CPPOverload.h"
#include "Compatibility.h"
#include "Converters.h"
#include "CustomPyTypes.h"
#include "LowLevelViews.h"
#include "ProxyWrappers.h"
#include "PyCallable.h"
#include "PyStrings.h"
#include "Pythonize.h"
#include "TypeManip.h"
#include "Utility.h"
#include "cppjit_interop.h"
 
// Standard
#include <algorithm>
#include <complex>
#include <deque>
#include <set>
#include <sstream>
#include <stdexcept>
#include <string>
#include <utility>
 
//- data and local helpers ---------------------------------------------------
namespace cppjit::cpyrt {
extern PyObject* gThisModule;
std::unordered_map<std::string, std::vector<PyObject*>>& pythonizations();
} // namespace cppjit::cpyrt
 
namespace {
 
// for convenience
using namespace cppjit::cpyrt;
 
//-----------------------------------------------------------------------------
bool HasAttrDirect(PyObject* pyclass, PyObject* pyname,
                   bool mustBecpyrt = false) {
  // prevents calls to Py_TYPE(pyclass)->tp_getattr, which is unnecessary for
  // our purposes here and could tickle problems w/ spurious lookups into ROOT
  // meta
  PyObject* dct = PyObject_GetAttr(pyclass, PyStrings::gDict);
  if (dct) {
    PyObject* attr = PyObject_GetItem(dct, pyname);
    Py_DECREF(dct);
    if (attr) {
      bool ret = !mustBecpyrt || CPPOverload_Check(attr);
      Py_DECREF(attr);
      return ret;
    }
  }
  PyErr_Clear();
  return false;
}
 
PyObject* GetAttrDirect(PyObject* pyclass, PyObject* pyname) {
  // get an attribute without causing getattr lookups
  PyObject* dct = PyObject_GetAttr(pyclass, PyStrings::gDict);
  if (dct) {
    PyObject* attr = PyObject_GetItem(dct, pyname);
    Py_DECREF(dct);
    return attr;
  }
  return nullptr;
}
 
//-----------------------------------------------------------------------------
inline bool IsTemplatedSTLClass(const std::string& name,
                                const std::string& klass) {
  // Scan the name of the class and determine whether it is a template
  // instantiation.
  auto pos = name.find(klass);
  return pos == 5 && name.rfind("std::", 0, 5) == 0 &&
         name.find("::", name.rfind(">")) == std::string::npos;
}
 
// to prevent compiler warnings about const char* -> char*
inline PyObject* CallPyObjMethod(PyObject* obj, const char* meth) {
  // Helper; call method with signature: obj->meth().
  Py_INCREF(obj);
  PyObject* result =
      PyObject_CallMethod(obj, const_cast<char*>(meth), const_cast<char*>(""));
  Py_DECREF(obj);
  return result;
}
 
//-----------------------------------------------------------------------------
inline PyObject* CallPyObjMethod(PyObject* obj, const char* meth,
                                 PyObject* arg1) {
  // Helper; call method with signature: obj->meth(arg1).
  Py_INCREF(obj);
  PyObject* result = PyObject_CallMethod(obj, const_cast<char*>(meth),
                                         const_cast<char*>("O"), arg1);
  Py_DECREF(obj);
  return result;
}
 
//-----------------------------------------------------------------------------
PyObject* PyStyleIndex(PyObject* self, PyObject* index) {
  // Helper; converts python index into straight C index.
  Py_ssize_t idx = PyLong_AsSsize_t(index);
  if (idx == (Py_ssize_t)-1 && PyErr_Occurred())
    return nullptr;
 
  Py_ssize_t size = PySequence_Size(self);
  if (idx >= size || (idx < 0 && idx < -size)) {
    PyErr_SetString(PyExc_IndexError, "index out of range");
    return nullptr;
  }
 
  PyObject* pyindex = nullptr;
  if (idx >= 0) {
    Py_INCREF(index);
    pyindex = index;
  } else
    pyindex = PyLong_FromSsize_t(size + idx);
 
  return pyindex;
}
 
//-----------------------------------------------------------------------------
inline bool AdjustSlice(const Py_ssize_t nlen, Py_ssize_t& start,
                        Py_ssize_t& stop, Py_ssize_t& step) {
  // Helper; modify slice range to match the container.
  if ((step > 0 && stop <= start) || (step < 0 && start <= stop))
    return false;
 
  if (start < 0)
    start = 0;
  if (start >= nlen)
    start = nlen - 1;
  if (step >= nlen)
    step = nlen;
 
  stop = step > 0 ? std::min(nlen, stop) : (stop >= 0 ? stop : -1);
  return true;
}
 
//-----------------------------------------------------------------------------
inline PyObject* CallSelfIndex(CPPInstance* self, PyObject* idx,
                               PyObject* pymeth) {
  // Helper; call method with signature: meth(pyindex).
  Py_INCREF((PyObject*)self);
  PyObject* pyindex = PyStyleIndex((PyObject*)self, idx);
  if (!pyindex) {
    Py_DECREF((PyObject*)self);
    return nullptr;
  }
 
  PyObject* result =
      PyObject_CallMethodOneArg((PyObject*)self, pymeth, pyindex);
  Py_DECREF(pyindex);
  Py_DECREF((PyObject*)self);
  return result;
}
 
//- "smart pointer" behavior ---------------------------------------------------
static PyObject* smart_follow(PyObject* self, PyObject* name,
                              PyObject* method) {
  if (!PyUnicode_Check(name))
    PyErr_SetString(PyExc_TypeError,
                    "getattr(): attribute name must be string");
 
  PyObject* pyptr = PyObject_CallMethodNoArgs(self, method);
  if (!pyptr)
    return nullptr;
 
  // prevent a potential infinite loop
  if (Py_TYPE(pyptr) == Py_TYPE(self)) {
    PyObject* val1 = PyObject_Str((PyObject*)Py_TYPE(self));
    PyObject* val2 = PyObject_Str(name);
    PyErr_Format(PyExc_AttributeError, "%s object has no attribute \'%s\'",
                 PyUnicode_AsUTF8(val1), PyUnicode_AsUTF8(val2));
    Py_DECREF(val2);
    Py_DECREF(val1);
 
    Py_DECREF(pyptr);
    return nullptr;
  }
 
  PyObject* result = PyObject_GetAttr(pyptr, name);
  Py_DECREF(pyptr);
  return result;
}
 
PyObject* DeRefGetAttr(PyObject* self, PyObject* name) {
  // Follow operator*() if present (available in python as __deref__), so that
  // smart pointers behave as expected.
  if (name == PyStrings::gTypeCode || name == PyStrings::gCTypesType) {
    // TODO: these calls come from TemplateProxy and are unlikely to be needed
    // in practice, whereas as-is, they can accidentally dereference the result
    // of end() on some STL containers. Obviously, this is a dumb hack that
    // should be resolved more fundamentally.
    PyErr_SetString(PyExc_AttributeError, PyUnicode_AsUTF8(name));
    return nullptr;
  }
 
  return smart_follow(self, name, PyStrings::gDeref);
}
 
//-----------------------------------------------------------------------------
PyObject* ValueGetAttr(PyObject* self, PyObject* name) {
  // std::optional and std::expected require has_value() before operator*().
  PyObject* has_value = PyObject_CallMethodNoArgs(self, PyStrings::gHasValue);
  if (!has_value)
    return nullptr;
 
  const int contains_value = PyObject_IsTrue(has_value);
  Py_DECREF(has_value);
  if (contains_value < 0)
    return nullptr;
  if (!contains_value) {
    PyErr_SetObject(PyExc_AttributeError, name);
    return nullptr;
  }
 
  return DeRefGetAttr(self, name);
}
 
//-----------------------------------------------------------------------------
PyObject* FollowGetAttr(PyObject* self, PyObject* name) {
  // Follow operator->() if present (available in python as __follow__), so that
  // smart pointers behave as expected.
  return smart_follow(self, name, PyStrings::gFollow);
}
 
//- pointer checking bool converter -------------------------------------------
PyObject* NullCheckBool(PyObject* self) {
  if (!CPPInstance_Check(self)) {
    PyErr_SetString(PyExc_TypeError, "C++ object proxy expected");
    return nullptr;
  }
 
  if (!((CPPInstance*)self)->GetObject())
    Py_RETURN_FALSE;
 
  return PyObject_CallMethodNoArgs(self, PyStrings::gCppBool);
}
 
//- vector behavior as primitives ----------------------------------------------
 
// TODO: can probably use the below getters in the InitializerListConverter
struct ItemGetter {
  ItemGetter(PyObject* pyobj) : fPyObject(pyobj) { Py_INCREF(fPyObject); }
  virtual ~ItemGetter() { Py_DECREF(fPyObject); }
  virtual Py_ssize_t size() = 0;
  virtual PyObject* get() = 0;
  PyObject* fPyObject;
};
 
struct CountedItemGetter : public ItemGetter {
  CountedItemGetter(PyObject* pyobj) : ItemGetter(pyobj), fCur(0) {}
  Py_ssize_t fCur;
};
 
struct TupleItemGetter : public CountedItemGetter {
  using CountedItemGetter::CountedItemGetter;
  Py_ssize_t size() override { return PyTuple_GET_SIZE(fPyObject); }
  PyObject* get() override {
    if (fCur < PyTuple_GET_SIZE(fPyObject)) {
      PyObject* item = PyTuple_GET_ITEM(fPyObject, fCur++);
      Py_INCREF(item);
      return item;
    }
    PyErr_SetString(PyExc_StopIteration, "end of tuple");
    return nullptr;
  }
};
 
struct ListItemGetter : public CountedItemGetter {
  using CountedItemGetter::CountedItemGetter;
  Py_ssize_t size() override { return PyList_GET_SIZE(fPyObject); }
  PyObject* get() override {
    if (fCur < PyList_GET_SIZE(fPyObject)) {
      PyObject* item = PyList_GET_ITEM(fPyObject, fCur++);
      Py_INCREF(item);
      return item;
    }
    PyErr_SetString(PyExc_StopIteration, "end of list");
    return nullptr;
  }
};
 
struct SequenceItemGetter : public CountedItemGetter {
  using CountedItemGetter::CountedItemGetter;
  Py_ssize_t size() override {
    Py_ssize_t sz = PySequence_Size(fPyObject);
    if (sz < 0) {
      PyErr_Clear();
      return PyObject_LengthHint(fPyObject, 8);
    }
    return sz;
  }
  PyObject* get() override { return PySequence_GetItem(fPyObject, fCur++); }
};
 
struct IterItemGetter : public ItemGetter {
  using ItemGetter::ItemGetter;
  Py_ssize_t size() override { return PyObject_LengthHint(fPyObject, 8); }
  PyObject* get() override {
    return (*(Py_TYPE(fPyObject)->tp_iternext))(fPyObject);
  }
};
 
static ItemGetter* GetGetter(PyObject* args) {
  // Create an ItemGetter to loop over the iterable argument, if any.
  ItemGetter* getter = nullptr;
 
  if (PyTuple_GET_SIZE(args) == 1) {
    PyObject* fi = PyTuple_GET_ITEM(args, 0);
    if (PyUnicode_Check(fi) || PyBytes_Check(fi))
      return nullptr; // do not accept string to fill std::vector<char>
 
    // TODO: this only tests for new-style buffers, which is too strict, but a
    // generic check for Py_TYPE(fi)->tp_as_buffer is too loose (note that the
    // main use case is numpy, which offers the new interface)
    if (PyObject_CheckBuffer(fi))
      return nullptr;
 
    if (PyTuple_CheckExact(fi))
      getter = new TupleItemGetter(fi);
    else if (PyList_CheckExact(fi))
      getter = new ListItemGetter(fi);
    else if (PySequence_Check(fi))
      getter = new SequenceItemGetter(fi);
    else {
      PyObject* iter = PyObject_GetIter(fi);
      if (iter) {
        getter = new IterItemGetter{iter};
        Py_DECREF(iter);
      } else
        PyErr_Clear();
    }
  }
 
  return getter;
}
 
namespace {
 
void compileSpanHelpers() {
  static bool compiled = false;
 
  if (compiled)
    return;
 
  compiled = true;
 
  auto code = R"(
namespace __cppjit_internal {
 
template <class T>
struct ptr_iterator {
   T *cur;
   T *end;
 
   ptr_iterator(T *c, T *e) : cur(c), end(e) {}
 
   T &operator*() const { return *cur; }
   ptr_iterator &operator++()
   {
      ++cur;
      return *this;
   }
   bool operator==(const ptr_iterator &other) const { return cur == other.cur; }
   bool operator!=(const ptr_iterator &other) const { return !(*this == other); }
};
 
template <class T>
ptr_iterator<T> make_iter(T *begin, T *end)
{
   return {begin, end};
}
 
} // namespace __cppjit_internal
 
// Note: for const span<T>, T is const-qualified here
template <class T>
auto __cppjit_internal_begin(T &s) noexcept
{
   return __cppjit_internal::make_iter(s.data(), s.data() + s.size());
}
 
// Note: for const span<T>, T is const-qualified here
template <class T>
auto __cppjit_internal_end(T &s) noexcept
{
   // end iterator = begin iterator with cur == end
   return __cppjit_internal::make_iter(s.data() + s.size(), s.data() + s.size());
}
    )";
  interop::Compile(code, /*silent*/ true);
}
 
PyObject* spanBegin() {
  static PyObject* pyFunc = nullptr;
  if (!pyFunc) {
    compileSpanHelpers();
    PyObject* py_ns = cpyrt::GetScopeProxy(interop::GetGlobalScope());
    pyFunc = PyObject_GetAttrString(py_ns, "__cppjit_internal_begin");
    if (!pyFunc) {
      PyErr_Format(PyExc_RuntimeError,
                   "cppyy internal error: failed to locate helper "
                   "'__cppjit_internal_begin' for std::span pythonization");
    }
  }
  return pyFunc;
}
 
PyObject* spanEnd() {
  static PyObject* pyFunc = nullptr;
  if (!pyFunc) {
    compileSpanHelpers();
    PyObject* py_ns = cpyrt::GetScopeProxy(interop::GetGlobalScope());
    pyFunc = PyObject_GetAttrString(py_ns, "__cppjit_internal_end");
    if (!pyFunc) {
      PyErr_Format(PyExc_RuntimeError,
                   "cppyy internal error: failed to locate helper "
                   "'__cppjit_internal_end' for std::span pythonization");
    }
  }
  return pyFunc;
}
 
} // namespace
 
static PyObject* SpanBegin(PyObject* self, PyObject*) {
  auto begin = spanBegin();
  if (!begin)
    return nullptr;
  return PyObject_CallOneArg(begin, self);
}
 
static PyObject* SpanEnd(PyObject* self, PyObject*) {
  auto end = spanEnd();
  if (!end)
    return nullptr;
  return PyObject_CallOneArg(end, self);
}
 
static bool FillVector(PyObject* vecin, PyObject* args, ItemGetter* getter) {
  Py_ssize_t sz = getter->size();
  if (sz < 0)
    return false;
 
  // reserve memory as applicable
  if (0 < sz) {
    PyObject* res =
        PyObject_CallMethod(vecin, (char*)"reserve", (char*)"n", sz);
    Py_DECREF(res);
  } else // i.e. sz == 0, so empty container: done
    return true;
 
  bool fill_ok = true;
 
  // two main options: a list of lists (or tuples), or a list of objects; the
  // former are emplace_back'ed, the latter push_back'ed
  PyObject* fi = PySequence_GetItem(PyTuple_GET_ITEM(args, 0), 0);
  if (!fi)
    PyErr_Clear();
  if (fi && (PyTuple_CheckExact(fi) || PyList_CheckExact(fi))) {
    // use emplace_back to construct the vector entries one by one
    PyObject* eb_call = PyObject_GetAttrString(vecin, (char*)"emplace_back");
    PyObject* vtype =
        GetAttrDirect((PyObject*)Py_TYPE(vecin), PyStrings::gValueTypePtr);
    bool value_is_vector = false;
    if (vtype && PyLong_Check(vtype)) {
      // if the value_type is a vector, then allow for initialization from
      // sequences
      if (interop::GetTypeAsString(PyLong_AsVoidPtr(vtype))
              .rfind("std::vector", 0) != std::string::npos)
        value_is_vector = true;
    } else
      PyErr_Clear();
    Py_XDECREF(vtype);
 
    if (eb_call) {
      PyObject* eb_args;
      for (int i = 0; /* until break */; ++i) {
        PyObject* item = getter->get();
        if (item) {
          if (value_is_vector && PySequence_Check(item)) {
            eb_args = PyTuple_New(1);
            PyTuple_SET_ITEM(eb_args, 0, item);
          } else if (PyTuple_CheckExact(item)) {
            eb_args = item;
          } else if (PyList_CheckExact(item)) {
            Py_ssize_t isz = PyList_GET_SIZE(item);
            eb_args = PyTuple_New(isz);
            for (Py_ssize_t j = 0; j < isz; ++j) {
              PyObject* iarg = PyList_GET_ITEM(item, j);
              Py_INCREF(iarg);
              PyTuple_SET_ITEM(eb_args, j, iarg);
            }
            Py_DECREF(item);
          } else {
            Py_DECREF(item);
            PyErr_Format(PyExc_TypeError, "argument %d is not a tuple or list",
                         i);
            fill_ok = false;
            break;
          }
          PyObject* ebres = PyObject_CallObject(eb_call, eb_args);
          Py_DECREF(eb_args);
          if (!ebres) {
            fill_ok = false;
            break;
          }
          Py_DECREF(ebres);
        } else {
          if (PyErr_Occurred()) {
            if (!(PyErr_ExceptionMatches(PyExc_IndexError) ||
                  PyErr_ExceptionMatches(PyExc_StopIteration)))
              fill_ok = false;
            else {
              PyErr_Clear();
            }
          }
          break;
        }
      }
      Py_DECREF(eb_call);
    }
  } else {
    // use push_back to add the vector entries one by one
    PyObject* pb_call = PyObject_GetAttrString(vecin, (char*)"push_back");
    if (pb_call) {
      for (;;) {
        PyObject* item = getter->get();
        if (item) {
          PyObject* pbres =
              PyObject_CallFunctionObjArgs(pb_call, item, nullptr);
          Py_DECREF(item);
          if (!pbres) {
            fill_ok = false;
            break;
          }
          Py_DECREF(pbres);
        } else {
          if (PyErr_Occurred()) {
            if (!(PyErr_ExceptionMatches(PyExc_IndexError) ||
                  PyErr_ExceptionMatches(PyExc_StopIteration)))
              fill_ok = false;
            else {
              PyErr_Clear();
            }
          }
          break;
        }
      }
      Py_DECREF(pb_call);
    }
  }
  Py_XDECREF(fi);
 
  return fill_ok;
}
 
PyObject* VectorIAdd(PyObject* self, PyObject* args, PyObject* /* kwds */) {
  // Implement fast __iadd__ on std::vector (generic __iadd__ is in Python)
  ItemGetter* getter = GetGetter(args);
 
  if (getter) {
    bool fill_ok = FillVector(self, args, getter);
    delete getter;
 
    if (!fill_ok)
      return nullptr;
 
    Py_INCREF(self);
    return self;
  }
 
  // if no getter, it could still be b/c we have a buffer (e.g. numpy); looping
  // over a buffer here is slow, so use insert() instead
  if (PyTuple_GET_SIZE(args) == 1) {
    PyObject* fi = PyTuple_GET_ITEM(args, 0);
    if (PyObject_CheckBuffer(fi) &&
        !(PyUnicode_Check(fi) || PyBytes_Check(fi))) {
      PyObject* vend = PyObject_CallMethodNoArgs(self, PyStrings::gEnd);
      if (vend) {
        // when __iadd__ is overriden, the operation does not end with
        // calling the __iadd__ method, but also assigns the result to the
        // lhs of the iadd. For example, performing vec += arr, Python
        // first calls our override, and then does vec = vec.iadd(arr).
        PyObject* it = PyObject_CallMethodObjArgs(self, PyStrings::gInsert,
                                                  vend, fi, nullptr);
        Py_DECREF(vend);
 
        if (!it)
          return nullptr;
 
        Py_DECREF(it);
        // Assign the result of the __iadd__ override to the std::vector
        Py_INCREF(self);
        return self;
      }
    }
  }
 
  if (!PyErr_Occurred())
    PyErr_SetString(PyExc_TypeError, "argument is not iterable");
  return nullptr; // error already set
}
 
PyObject* VectorInit(PyObject* self, PyObject* args, PyObject* /* kwds */) {
  // Specialized vector constructor to allow construction from containers;
  // allowing such construction from initializer_list instead would possible,
  // but can be error-prone. This use case is common enough for std::vector to
  // implement it directly, except for arrays (which can be passed wholesale)
  // and strings (which won't convert properly as they'll be seen as buffers)
 
  ItemGetter* getter = GetGetter(args);
 
  if (getter) {
    // construct an empty vector, then back-fill it
    PyObject* result = PyObject_CallMethodNoArgs(self, PyStrings::gRealInit);
    if (!result) {
      delete getter;
      return nullptr;
    }
 
    bool fill_ok = FillVector(self, args, getter);
    delete getter;
 
    if (!fill_ok) {
      Py_DECREF(result);
      return nullptr;
    }
 
    return result;
  }
 
  // The given argument wasn't iterable: simply forward to regular constructor
  PyObject* realInit = PyObject_GetAttr(self, PyStrings::gRealInit);
  if (realInit) {
    PyObject* result = PyObject_Call(realInit, args, nullptr);
    Py_DECREF(realInit);
    return result;
  }
 
  return nullptr;
}
 
//---------------------------------------------------------------------------
PyObject* VectorData(PyObject* self, PyObject*) {
  PyObject* pydata = CallPyObjMethod(self, "__real_data");
  if (!LowLevelView_Check(pydata) && !CPPInstance_Check(pydata))
    return pydata;
 
  PyObject* pylen = PyObject_CallMethodNoArgs(self, PyStrings::gSize);
  if (!pylen) {
    PyErr_Clear();
    return pydata;
  }
 
  long clen = PyLong_AsLong(pylen);
  Py_DECREF(pylen);
 
  if (CPPInstance_Check(pydata)) {
    ((CPPInstance*)pydata)->CastToArray(clen);
    return pydata;
  }
 
  ((LowLevelView*)pydata)->resize((size_t)clen);
  return pydata;
}
 
//---------------------------------------------------------------------------
PyObject* VectorArray(PyObject* self, PyObject* args, PyObject* kwargs) {
  PyObject* pydata = VectorData(self, nullptr);
  PyObject* arrcall = PyObject_GetAttr(pydata, PyStrings::gArray);
  PyObject* newarr = PyObject_Call(arrcall, args, kwargs);
  Py_DECREF(arrcall);
  Py_DECREF(pydata);
  return newarr;
}
 
//-----------------------------------------------------------------------------
static PyObject* vector_iter(PyObject* v) {
  vectoriterobject* vi = PyObject_GC_New(vectoriterobject, &VectorIter_Type);
  if (!vi)
    return nullptr;
 
  vi->ii_container = v;
 
  // tell the iterator code to set a life line if this container is a temporary
  // (checked before taking the iterator's own reference)
  vi->vi_flags = vectoriterobject::kDefault;
#if PY_VERSION_HEX >= 0x030e0000
  if (PyUnstable_Object_IsUniqueReferencedTemporary(v) ||
      (((CPPInstance*)v)->fFlags & CPPInstance::kIsValue))
#else
  if (Py_REFCNT(v) <= 1 || (((CPPInstance*)v)->fFlags & CPPInstance::kIsValue))
#endif
    vi->vi_flags = vectoriterobject::kNeedLifeLine;
 
  Py_INCREF(v);
 
  PyObject* pyvalue_type =
      PyObject_GetAttr((PyObject*)Py_TYPE(v), PyStrings::gValueTypePtr);
  if (pyvalue_type) {
    PyObject* pyvalue_size =
        GetAttrDirect((PyObject*)Py_TYPE(v), PyStrings::gValueSize);
    if (pyvalue_size) {
      vi->vi_stride = PyLong_AsLong(pyvalue_size);
      Py_DECREF(pyvalue_size);
    } else {
      PyErr_Clear();
      vi->vi_stride = 0;
    }
 
    if (PyLong_Check(pyvalue_type)) {
      interop::TCppType_t value_type = PyLong_AsVoidPtr(pyvalue_type);
      value_type = interop::ResolveType(value_type);
      vi->vi_klass = interop::GetScopeFromType(value_type);
      if (!vi->vi_klass) {
        // look for a special case of pointer to a class type (which is a
        // builtin, but it is more useful to treat it polymorphically by
        // allowing auto-downcasts)
        const std::string& clean_type = TypeManip::clean_type(
            interop::GetTypeAsString(value_type), false, false);
        interop::TCppScope_t c = interop::GetScope(clean_type);
        if (c &&
            TypeManip::compound(interop::GetTypeAsString(value_type)) == "*") {
          vi->vi_klass = c;
          vi->vi_flags = vectoriterobject::kIsPolymorphic;
        }
      }
      if (interop::IsPointerType(value_type))
        vi->vi_flags = vectoriterobject::kIsPolymorphic;
      if (vi->vi_klass) {
        vi->vi_converter = nullptr;
        if (!vi->vi_flags) {
          value_type = interop::ResolveType(value_type);
          if (interop::GetTypeAsString(value_type).back() !=
              '*') // meaning, object stored by-value
            vi->vi_flags = vectoriterobject::kNeedLifeLine;
        }
      } else
        vi->vi_converter = CreateConverter(value_type);
      if (!vi->vi_stride)
        vi->vi_stride = interop::SizeOfType(value_type);
 
    } else if (CPPScope_Check(pyvalue_type)) {
      vi->vi_klass = ((CPPClass*)pyvalue_type)->fCppType;
      vi->vi_converter = nullptr;
      if (!vi->vi_stride)
        vi->vi_stride = interop::SizeOf(vi->vi_klass);
      if (!vi->vi_flags)
        vi->vi_flags = vectoriterobject::kNeedLifeLine;
    }
 
    PyObject* pydata = CallPyObjMethod(v, "__real_data");
    if (!pydata || Utility::GetBuffer(pydata, '*', 1, vi->vi_data, false) == 0)
      vi->vi_data = CPPInstance_Check(pydata)
                        ? ((CPPInstance*)pydata)->GetObjectRaw()
                        : nullptr;
    Py_XDECREF(pydata);
 
  } else {
    PyErr_Clear();
    vi->vi_data = nullptr;
    vi->vi_stride = 0;
    vi->vi_converter = nullptr;
    vi->vi_klass = nullptr;
    vi->vi_flags = 0;
  }
 
  Py_XDECREF(pyvalue_type);
 
  vi->ii_pos = 0;
  vi->ii_len = PySequence_Size(v);
 
  PyObject_GC_Track(vi);
  return (PyObject*)vi;
}
 
PyObject* VectorGetItem(CPPInstance* self, PySliceObject* index) {
  // Implement python's __getitem__ for std::vector<>s.
  if (PySlice_Check(index)) {
    if (!self->GetObject()) {
      PyErr_SetString(PyExc_TypeError, "unsubscriptable object");
      return nullptr;
    }
 
    PyObject* pyclass = (PyObject*)Py_TYPE((PyObject*)self);
    PyObject* nseq = PyObject_CallObject(pyclass, nullptr);
 
    Py_ssize_t start, stop, step;
    PySlice_GetIndices((PyObject*)index, PyObject_Length((PyObject*)self),
                       &start, &stop, &step);
 
    const Py_ssize_t nlen = PySequence_Size((PyObject*)self);
    if (!AdjustSlice(nlen, start, stop, step))
      return nseq;
 
    const Py_ssize_t sign = step < 0 ? -1 : 1;
    for (Py_ssize_t i = start; i * sign < stop * sign; i += step) {
      PyObject* pyidx = PyLong_FromSsize_t(i);
      PyObject* item = PyObject_CallMethodOneArg((PyObject*)self,
                                                 PyStrings::gGetNoCheck, pyidx);
      CallPyObjMethod(nseq, "push_back", item);
      Py_DECREF(item);
      Py_DECREF(pyidx);
    }
 
    return nseq;
  }
 
  return CallSelfIndex(self, (PyObject*)index, PyStrings::gGetNoCheck);
}
 
static interop::TCppScope_t sVectorBoolTypeID;
 
PyObject* VectorBoolGetItem(CPPInstance* self, PyObject* idx) {
  // std::vector<bool> is a special-case in C++, and its return type depends on
  // the compiler: treat it special here as well
  if (!CPPInstance_Check(self) || self->ObjectIsA() != sVectorBoolTypeID) {
    PyErr_Format(PyExc_TypeError,
                 "require object of type std::vector<bool>, but %s given",
                 interop::GetScopedFinalName(self->ObjectIsA()).c_str());
    return nullptr;
  }
 
  if (!self->GetObject()) {
    PyErr_SetString(PyExc_TypeError, "unsubscriptable object");
    return nullptr;
  }
 
  if (PySlice_Check(idx)) {
    PyObject* pyclass = (PyObject*)Py_TYPE((PyObject*)self);
    PyObject* nseq = PyObject_CallObject(pyclass, nullptr);
 
    Py_ssize_t start, stop, step;
    PySlice_GetIndices((PyObject*)idx, PyObject_Length((PyObject*)self), &start,
                       &stop, &step);
    const Py_ssize_t nlen = PySequence_Size((PyObject*)self);
    if (!AdjustSlice(nlen, start, stop, step))
      return nseq;
 
    const Py_ssize_t sign = step < 0 ? -1 : 1;
    for (Py_ssize_t i = start; i * sign < stop * sign; i += step) {
      PyObject* pyidx = PyLong_FromSsize_t(i);
      PyObject* item = PyObject_CallMethodOneArg((PyObject*)self,
                                                 PyStrings::gGetItem, pyidx);
      CallPyObjMethod(nseq, "push_back", item);
      Py_DECREF(item);
      Py_DECREF(pyidx);
    }
 
    return nseq;
  }
 
  PyObject* pyindex = PyStyleIndex((PyObject*)self, idx);
  if (!pyindex)
    return nullptr;
 
  int index = (int)PyLong_AsLong(pyindex);
  Py_DECREF(pyindex);
 
  // get hold of the actual std::vector<bool> (no cast, as vector is never a
  // base)
  std::vector<bool>* vb = (std::vector<bool>*)self->GetObject();
 
  // finally, return the value
  if (bool((*vb)[index]))
    Py_RETURN_TRUE;
  Py_RETURN_FALSE;
}
 
PyObject* VectorBoolSetItem(CPPInstance* self, PyObject* args) {
  // std::vector<bool> is a special-case in C++, and its return type depends on
  // the compiler: treat it special here as well
  if (!CPPInstance_Check(self) || self->ObjectIsA() != sVectorBoolTypeID) {
    PyErr_Format(PyExc_TypeError,
                 "require object of type std::vector<bool>, but %s given",
                 interop::GetScopedFinalName(self->ObjectIsA()).c_str());
    return nullptr;
  }
 
  if (!self->GetObject()) {
    PyErr_SetString(PyExc_TypeError, "unsubscriptable object");
    return nullptr;
  }
 
  int bval = 0;
  PyObject* idx = nullptr;
  if (!PyArg_ParseTuple(args, const_cast<char*>("Oi:__setitem__"), &idx, &bval))
    return nullptr;
 
  PyObject* pyindex = PyStyleIndex((PyObject*)self, idx);
  if (!pyindex)
    return nullptr;
 
  int index = (int)PyLong_AsLong(pyindex);
  Py_DECREF(pyindex);
 
  // get hold of the actual std::vector<bool> (no cast, as vector is never a
  // base)
  std::vector<bool>* vb = (std::vector<bool>*)self->GetObject();
 
  // finally, set the value
  (*vb)[index] = (bool)bval;
 
  Py_RETURN_NONE;
}
 
//- array behavior as primitives ----------------------------------------------
PyObject* ArrayInit(PyObject* self, PyObject* args, PyObject* /* kwds */) {
  // std::array is normally only constructed using aggregate initialization,
  // which is a concept that does not exist in python, so use this custom
  // constructor to to fill the array using setitem
 
  if (args && PyTuple_GET_SIZE(args) == 1 &&
      PySequence_Check(PyTuple_GET_ITEM(args, 0))) {
    // construct the empty array, then fill it
    PyObject* result = PyObject_CallMethodNoArgs(self, PyStrings::gRealInit);
    if (!result)
      return nullptr;
 
    PyObject* items = PyTuple_GET_ITEM(args, 0);
    Py_ssize_t fillsz = PySequence_Size(items);
    if (PySequence_Size(self) != fillsz) {
      PyErr_Format(PyExc_ValueError,
                   "received sequence of size %zd where %zd expected", fillsz,
                   PySequence_Size(self));
      Py_DECREF(result);
      return nullptr;
    }
 
    PyObject* si_call = PyObject_GetAttr(self, PyStrings::gSetItem);
    for (Py_ssize_t i = 0; i < fillsz; ++i) {
      PyObject* item = PySequence_GetItem(items, i);
      PyObject* index = PyLong_FromSsize_t(i);
      PyObject* sires =
          PyObject_CallFunctionObjArgs(si_call, index, item, nullptr);
      Py_DECREF(index);
      Py_DECREF(item);
      if (!sires) {
        Py_DECREF(si_call);
        Py_DECREF(result);
        return nullptr;
      } else
        Py_DECREF(sires);
    }
    Py_DECREF(si_call);
 
    return result;
  } else
    PyErr_Clear();
 
  // The given argument wasn't iterable: simply forward to regular constructor
  PyObject* realInit = PyObject_GetAttr(self, PyStrings::gRealInit);
  if (realInit) {
    PyObject* result = PyObject_Call(realInit, args, nullptr);
    Py_DECREF(realInit);
    return result;
  }
 
  return nullptr;
}
 
//- map behavior as primitives ------------------------------------------------
static PyObject* MapFromPairs(PyObject* self, PyObject* pairs) {
  // construct an empty map, then fill it with the key, value pairs
  PyObject* result = PyObject_CallMethodNoArgs(self, PyStrings::gRealInit);
  if (!result)
    return nullptr;
 
  PyObject* si_call = PyObject_GetAttr(self, PyStrings::gSetItem);
  for (Py_ssize_t i = 0; i < PySequence_Size(pairs); ++i) {
    PyObject* pair = PySequence_GetItem(pairs, i);
    PyObject* sires = nullptr;
    if (pair && PySequence_Check(pair) && PySequence_Size(pair) == 2) {
      PyObject* key = PySequence_GetItem(pair, 0);
      PyObject* value = PySequence_GetItem(pair, 1);
      sires = PyObject_CallFunctionObjArgs(si_call, key, value, nullptr);
      Py_DECREF(value);
      Py_DECREF(key);
    }
    Py_DECREF(pair);
    if (!sires) {
      Py_DECREF(si_call);
      Py_DECREF(result);
      if (!PyErr_Occurred())
        PyErr_SetString(
            PyExc_TypeError,
            "Failed to fill map (argument not a dict or sequence of pairs)");
      return nullptr;
    } else
      Py_DECREF(sires);
  }
  Py_DECREF(si_call);
 
  return result;
}
 
PyObject* MapInit(PyObject* self, PyObject* args, PyObject* /* kwds */) {
  // Specialized map constructor to allow construction from mapping containers
  // and from tuples of pairs ("initializer_list style").
 
  // PyMapping_Check is not very discriminatory, as it basically only checks for
  // the existence of  __getitem__, hence the most common cases of tuple and
  // list are dropped straight-of-the-bat (the PyMapping_Items call will fail on
  // them).
  if (PyTuple_GET_SIZE(args) == 1 &&
      PyMapping_Check(PyTuple_GET_ITEM(args, 0)) &&
      !(PyTuple_Check(PyTuple_GET_ITEM(args, 0)) ||
        PyList_Check(PyTuple_GET_ITEM(args, 0)))) {
    PyObject* assoc = PyTuple_GET_ITEM(args, 0);
    // in p3, PyMapping_Items isn't a macro, but a function that short-circuits
    // dict
    PyObject* items = PyMapping_Items(assoc);
    if (items && PySequence_Check(items)) {
      PyObject* result = MapFromPairs(self, items);
      Py_DECREF(items);
      return result;
    }
 
    Py_XDECREF(items);
    PyErr_Clear();
 
    // okay to fall through as long as 'self' has not been created (is done in
    // MapFromPairs)
  }
 
  // tuple of pairs case (some mapping types are sequences)
  if (PyTuple_GET_SIZE(args) == 1 &&
      PySequence_Check(PyTuple_GET_ITEM(args, 0)))
    return MapFromPairs(self, PyTuple_GET_ITEM(args, 0));
 
  // The given argument wasn't a mapping or tuple of pairs: forward to regular
  // constructor
  PyObject* realInit = PyObject_GetAttr(self, PyStrings::gRealInit);
  if (realInit) {
    PyObject* result = PyObject_Call(realInit, args, nullptr);
    Py_DECREF(realInit);
    return result;
  }
 
  return nullptr;
}
 
PyObject* STLContainsWithFind(PyObject* self, PyObject* obj) {
  // Implement python's __contains__ for std::map/std::set
  PyObject* result = nullptr;
 
  PyObject* iter = CallPyObjMethod(self, "find", obj);
  if (CPPInstance_Check(iter)) {
    PyObject* end = PyObject_CallMethodNoArgs(self, PyStrings::gEnd);
    if (CPPInstance_Check(end)) {
      if (!PyObject_RichCompareBool(iter, end, Py_EQ)) {
        Py_INCREF(Py_True);
        result = Py_True;
      }
    }
    Py_XDECREF(end);
  }
  Py_XDECREF(iter);
 
  if (!result) {
    PyErr_Clear(); // e.g. wrong argument type, which should always lead to
                   // False
    Py_INCREF(Py_False);
    result = Py_False;
  }
 
  return result;
}
 
//- set behavior as primitives ------------------------------------------------
PyObject* SetInit(PyObject* self, PyObject* args, PyObject* /* kwds */) {
  // Specialized set constructor to allow construction from Python sets.
  if (PyTuple_GET_SIZE(args) == 1 && PySet_Check(PyTuple_GET_ITEM(args, 0))) {
    PyObject* pyset = PyTuple_GET_ITEM(args, 0);
 
    // construct an empty set, then fill it
    PyObject* result = PyObject_CallMethodNoArgs(self, PyStrings::gRealInit);
    if (!result)
      return nullptr;
 
    PyObject* iter = PyObject_GetIter(pyset);
    if (iter) {
      PyObject* ins_call = PyObject_GetAttrString(self, (char*)"insert");
 
      IterItemGetter getter{iter};
      Py_DECREF(iter);
 
      PyObject* item = getter.get();
      while (item) {
        PyObject* insres =
            PyObject_CallFunctionObjArgs(ins_call, item, nullptr);
        Py_DECREF(item);
        if (!insres) {
          Py_DECREF(ins_call);
          Py_DECREF(result);
          return nullptr;
        } else
          Py_DECREF(insres);
        item = getter.get();
      }
      Py_DECREF(ins_call);
    }
 
    return result;
  }
 
  // The given argument wasn't iterable: simply forward to regular constructor
  PyObject* realInit = PyObject_GetAttr(self, PyStrings::gRealInit);
  if (realInit) {
    PyObject* result = PyObject_Call(realInit, args, nullptr);
    Py_DECREF(realInit);
    return result;
  }
 
  return nullptr;
}
 
//- STL container iterator support --------------------------------------------
static const ptrdiff_t PS_END_ADDR = 7;   // non-aligned address, so no clash
static const ptrdiff_t PS_FLAG_ADDR = 11; // id.
static const ptrdiff_t PS_COLL_ADDR = 13; // id.
 
PyObject* STLIterNext(PyObject* self); // defined below; used by STLSequenceIter
 
PyObject* LLSequenceIter(PyObject* self) {
  // Implement python's __iter__ for low level views used through STL-type
  // begin()/end()
  PyObject* iter = PyObject_CallMethodNoArgs(self, PyStrings::gBegin);
 
  if (LowLevelView_Check(iter)) {
    // builtin pointer iteration: can only succeed if a size is available
    Py_ssize_t sz = PySequence_Size(self);
    if (sz == -1) {
      Py_DECREF(iter);
      return nullptr;
    }
    PyObject* lliter = Py_TYPE(iter)->tp_iter(iter);
    ((indexiterobject*)lliter)->ii_len = sz;
    Py_DECREF(iter);
    return lliter;
  }
 
  if (iter) {
    Py_DECREF(iter);
    PyErr_SetString(PyExc_TypeError,
                    "unrecognized iterator type for low level views");
  }
 
  return nullptr;
}
 
PyObject* STLSequenceIter(PyObject* self) {
  // Implement python's __iter__ for std::iterator<>s
  PyObject* iter = PyObject_CallMethodNoArgs(self, PyStrings::gBegin);
  if (iter) {
    PyObject* end = PyObject_CallMethodNoArgs(self, PyStrings::gEnd);
    if (end) {
      if (CPPInstance_Check(iter)) {
        // Guarantee the returned iterator implements Python's iterator
        // protocol. The deferred, name-based __next__ install can miss when an
        // iterator's canonical return-type spelling and its scoped-final-name
        // diverge (e.g libc++'s std::__1::__wrap_iter on macOS) leaving the
        // type without
        // __next__, so iter() rejects it. begin()/end() resolving here proves
        // this is an STL forward iterator, so install the protocol now if it is
        // still absent
        PyTypeObject* itype = Py_TYPE(iter);
        if (!PyIter_Check(iter)) { // no tp_iternext, or the
                                   // _PyObject_NextNotImplemented sentinel
          itype->tp_iternext = (iternextfunc)STLIterNext;
          Utility::AddToClass((PyObject*)itype, "__next__",
                              (PyCFunction)STLIterNext, METH_NOARGS);
          if (!itype->tp_iter) {
            itype->tp_iter = (getiterfunc)PyObject_SelfIter;
            Utility::AddToClass((PyObject*)itype, "__iter__",
                                (PyCFunction)PyObject_SelfIter, METH_NOARGS);
          }
          PyType_Modified(itype);
        }
        // use the data member cache to store extra state on the iterator
        // object, without it being visible on the Python side
        auto& dmc = ((CPPInstance*)iter)->GetDatamemberCache();
        dmc.push_back(std::make_pair(PS_END_ADDR, end));
 
        // set a flag, indicating first iteration (reset in __next__)
        Py_INCREF(Py_False);
        dmc.push_back(std::make_pair(PS_FLAG_ADDR, Py_False));
 
        // make sure the iterated over collection remains alive for the duration
        Py_INCREF(self);
        dmc.push_back(std::make_pair(PS_COLL_ADDR, self));
      } else {
        // could store "end" on the object's dictionary anyway, but if end()
        // returns a user-customized object, then its __next__ is probably
        // custom, too
        Py_DECREF(end);
      }
    }
  }
  return iter;
}
 
//- generic iterator support over a sequence with operator[] and size ---------
//-----------------------------------------------------------------------------
static PyObject* index_iter(PyObject* c) {
  indexiterobject* ii = PyObject_GC_New(indexiterobject, &IndexIter_Type);
  if (!ii)
    return nullptr;
 
  Py_INCREF(c);
  ii->ii_container = c;
  ii->ii_pos = 0;
  ii->ii_len = PySequence_Size(c);
 
  PyObject_GC_Track(ii);
  return (PyObject*)ii;
}
 
//- safe indexing for STL-like vector w/o iterator dictionaries ---------------
/* replaced by indexiterobject iteration, but may still have some future use ...
PyObject* CheckedGetItem(PyObject* self, PyObject* obj)
{
// Implement a generic python __getitem__ for STL-like classes that are missing
the
// reflection info for their iterators. This is then used for iteration by means
of
// consecutive indices, it such index is of integer type.
    Py_ssize_t size = PySequence_Size(self);
    Py_ssize_t idx  = PyLong_AsSsize_t(obj);
    if ((size == (Py_ssize_t)-1 || idx == (Py_ssize_t)-1) && PyErr_Occurred()) {
    // argument conversion problem: let method itself resolve anew and report
        PyErr_Clear();
        return PyObject_CallMethodOneArg(self, PyStrings::gGetNoCheck, obj);
    }
 
    bool inbounds = false;
    if (idx < 0) idx += size;
    if (0 <= idx && 0 <= size && idx < size)
        inbounds = true;
 
    if (inbounds)
        return PyObject_CallMethodOneArg(self, PyStrings::gGetNoCheck, obj);
    else
        PyErr_SetString( PyExc_IndexError, "index out of range" );
 
    return nullptr;
}*/
 
//- pair as sequence to allow tuple unpacking --------------------------------
PyObject* PairUnpack(PyObject* self, PyObject* pyindex) {
  // For std::map<> iteration, unpack std::pair<>s into tuples for the loop.
  long idx = PyLong_AsLong(pyindex);
  if (idx == -1 && PyErr_Occurred())
    return nullptr;
 
  if (!CPPInstance_Check(self) || !((CPPInstance*)self)->GetObject()) {
    PyErr_SetString(PyExc_TypeError, "unsubscriptable object");
    return nullptr;
  }
 
  if ((int)idx == 0)
    return PyObject_GetAttr(self, PyStrings::gFirst);
  else if ((int)idx == 1)
    return PyObject_GetAttr(self, PyStrings::gSecond);
 
  // still here? Trigger stop iteration
  PyErr_SetString(PyExc_IndexError, "out of bounds");
  return nullptr;
}
 
//- simplistic len() functions -----------------------------------------------
PyObject* ReturnTwo(CPPInstance*, PyObject*) { return PyLong_FromLong(2); }
 
//- shared/unique_ptr behavior -----------------------------------------------
PyObject* SmartPtrInit(PyObject* self, PyObject* args, PyObject* /* kwds */) {
  // since the shared/unique pointer will take ownership, we need to relinquish
  // it
  PyObject* realInit = PyObject_GetAttr(self, PyStrings::gRealInit);
  if (realInit) {
    PyObject* result = PyObject_Call(realInit, args, nullptr);
    Py_DECREF(realInit);
    if (result && PyTuple_GET_SIZE(args) == 1 &&
        CPPInstance_Check(PyTuple_GET_ITEM(args, 0))) {
      CPPInstance* cppinst = (CPPInstance*)PyTuple_GET_ITEM(args, 0);
      if (!(cppinst->fFlags & CPPInstance::kIsSmartPtr))
        cppinst->CppOwns();
    }
    return result;
  }
  return nullptr;
}
 
//- string behavior as primitives --------------------------------------------
// TODO: this is wrong, b/c it doesn't order
static int PyObject_Compare(PyObject* one, PyObject* other) {
  return !PyObject_RichCompareBool(one, other, Py_EQ);
}
static inline PyObject* cpyrt_PyString_FromCppString(std::string_view s,
                                                     bool native = true) {
  if (native)
    return PyBytes_FromStringAndSize(s.data(), s.size());
  return PyUnicode_FromStringAndSize(s.data(), s.size());
}
 
static inline PyObject* cpyrt_PyString_FromCppString(std::wstring_view s,
                                                     bool native = true) {
  PyObject* pyobj = PyUnicode_FromWideChar(s.data(), s.size());
  if (pyobj && native) {
    PyObject* pybytes = PyUnicode_AsEncodedString(pyobj, "UTF-8", "strict");
    Py_DECREF(pyobj);
    pyobj = pybytes;
  }
  return pyobj;
}
 
#define CPPJIT_IMPL_STRING_PYTHONIZATION(type, name)                           \
  static inline PyObject* name##StringGetData(PyObject* self,                  \
                                              bool native = true) {            \
    if (cpyrt::CPPInstance_Check(self)) {                                      \
      type* obj = ((type*)((CPPInstance*)self)->GetObject());                  \
      if (obj)                                                                 \
        return cpyrt_PyString_FromCppString(*obj, native);                     \
    }                                                                          \
    PyErr_Format(PyExc_TypeError, "object mismatch (%s expected)", #type);     \
    return nullptr;                                                            \
  }                                                                            \
                                                                               \
  PyObject* name##StringStr(PyObject* self) {                                  \
    PyObject* pyobj = name##StringGetData(self, false);                        \
    if (!pyobj) {                                                              \
      /* do a native conversion to make printing possible (debatable) */       \
      PyErr_Clear();                                                           \
      PyObject* pybytes = name##StringGetData(self, true);                     \
      if (pybytes) { /* should not fail */                                     \
        pyobj = PyObject_Str(pybytes);                                         \
        Py_DECREF(pybytes);                                                    \
      }                                                                        \
    }                                                                          \
    return pyobj;                                                              \
  }                                                                            \
                                                                               \
  PyObject* name##StringBytes(PyObject* self) {                                \
    return name##StringGetData(self, true);                                    \
  }                                                                            \
                                                                               \
  PyObject* name##StringRepr(PyObject* self) {                                 \
    PyObject* data = name##StringGetData(self, true);                          \
    if (data) {                                                                \
      PyObject* repr = PyObject_Repr(data);                                    \
      Py_DECREF(data);                                                         \
      return repr;                                                             \
    }                                                                          \
    return nullptr;                                                            \
  }                                                                            \
                                                                               \
  PyObject* name##StringIsEqual(PyObject* self, PyObject* obj) {               \
    PyObject* data = name##StringGetData(self, PyBytes_Check(obj));            \
    if (data) {                                                                \
      PyObject* result = PyObject_RichCompare(data, obj, Py_EQ);               \
      Py_DECREF(data);                                                         \
      return result;                                                           \
    }                                                                          \
    return nullptr;                                                            \
  }                                                                            \
                                                                               \
  PyObject* name##StringIsNotEqual(PyObject* self, PyObject* obj) {            \
    PyObject* data = name##StringGetData(self, PyBytes_Check(obj));            \
    if (data) {                                                                \
      PyObject* result = PyObject_RichCompare(data, obj, Py_NE);               \
      Py_DECREF(data);                                                         \
      return result;                                                           \
    }                                                                          \
    return nullptr;                                                            \
  }
 
// Only define STLStringCompare:
#define CPPJIT_IMPL_STRING_PYTHONIZATION_CMP(type, name)                       \
  CPPJIT_IMPL_STRING_PYTHONIZATION(type, name)                                 \
  PyObject* name##StringCompare(PyObject* self, PyObject* obj) {               \
    PyObject* data = name##StringGetData(self, PyBytes_Check(obj));            \
    int result = 0;                                                            \
    if (data) {                                                                \
      result = PyObject_Compare(data, obj);                                    \
      Py_DECREF(data);                                                         \
    }                                                                          \
    if (PyErr_Occurred())                                                      \
      return nullptr;                                                          \
    return PyLong_FromLong(result);                                            \
  }
 
CPPJIT_IMPL_STRING_PYTHONIZATION_CMP(std::string, STL)
CPPJIT_IMPL_STRING_PYTHONIZATION_CMP(std::wstring, STLW)
CPPJIT_IMPL_STRING_PYTHONIZATION_CMP(std::string_view, STLView)
 
static inline std::string* GetSTLString(CPPInstance* self) {
  if (!CPPInstance_Check(self)) {
    PyErr_SetString(PyExc_TypeError, "std::string object expected");
    return nullptr;
  }
 
  std::string* obj = (std::string*)self->GetObject();
  if (!obj)
    PyErr_SetString(PyExc_ReferenceError, "attempt to access a null-pointer");
 
  return obj;
}
 
PyObject* STLStringDecode(CPPInstance* self, PyObject* args, PyObject* kwds) {
  std::string* obj = GetSTLString(self);
  if (!obj)
    return nullptr;
 
  char* keywords[] = {(char*)"encoding", (char*)"errors", (char*)nullptr};
  const char* encoding = nullptr;
  const char* errors = nullptr;
  if (!PyArg_ParseTupleAndKeywords(args, kwds, const_cast<char*>("s|s"),
                                   keywords, &encoding, &errors))
    return nullptr;
 
  return PyUnicode_Decode(obj->data(), obj->size(), encoding, errors);
}
 
PyObject* STLStringContains(CPPInstance* self, PyObject* pyobj) {
  std::string* obj = GetSTLString(self);
  if (!obj)
    return nullptr;
 
  const char* needle = PyUnicode_AsUTF8(pyobj);
  if (!needle)
    return nullptr;
 
  if (obj->find(needle) != std::string::npos) {
    Py_RETURN_TRUE;
  }
 
  Py_RETURN_FALSE;
}
 
PyObject* STLStringReplace(CPPInstance* self, PyObject* args,
                           PyObject* /*kwds*/) {
  std::string* obj = GetSTLString(self);
  if (!obj)
    return nullptr;
 
  // both str and std::string have a method "replace", but the Python version
  // only accepts strings and takes no keyword arguments, whereas the C++
  // version has no overload that takes a string
 
  if (2 <= PyTuple_GET_SIZE(args) &&
      PyUnicode_Check(PyTuple_GET_ITEM(args, 0))) {
    PyObject* pystr = PyUnicode_FromStringAndSize(obj->data(), obj->size());
    PyObject* meth = PyObject_GetAttrString(pystr, (char*)"replace");
    Py_DECREF(pystr);
    PyObject* result = PyObject_CallObject(meth, args);
    Py_DECREF(meth);
    return result;
  }
 
  PyObject* cppreplace =
      PyObject_GetAttrString((PyObject*)self, (char*)"__cpp_replace");
  if (cppreplace) {
    PyObject* result = PyObject_Call(cppreplace, args, nullptr);
    Py_DECREF(cppreplace);
    return result;
  }
 
  PyErr_SetString(PyExc_AttributeError,
                  "\'std::string\' object has no attribute \'replace\'");
  return nullptr;
}
 
#define CPYRT_STRING_FINDMETHOD(name, cppname, pyname)                         \
  PyObject* STLString##name(CPPInstance* self, PyObject* args,                 \
                            PyObject* /*kwds*/) {                              \
    std::string* obj = GetSTLString(self);                                     \
    if (!obj)                                                                  \
      return nullptr;                                                          \
                                                                               \
    PyObject* cppmeth =                                                        \
        PyObject_GetAttrString((PyObject*)self, (char*)#cppname);              \
    if (cppmeth) {                                                             \
      PyObject* result = PyObject_Call(cppmeth, args, nullptr);                \
      Py_DECREF(cppmeth);                                                      \
      if (result) {                                                            \
        if (PyLongOrInt_AsULong64(result) ==                                   \
            (PY_ULONG_LONG)std::string::npos) {                                \
          Py_DECREF(result);                                                   \
          return PyLong_FromLong(-1);                                          \
        }                                                                      \
        return result;                                                         \
      }                                                                        \
      PyErr_Clear();                                                           \
    }                                                                          \
                                                                               \
    PyObject* pystr = PyUnicode_FromStringAndSize(obj->data(), obj->size());   \
    PyObject* pymeth = PyObject_GetAttrString(pystr, (char*)#pyname);          \
    Py_DECREF(pystr);                                                          \
    PyObject* result = PyObject_CallObject(pymeth, args);                      \
    Py_DECREF(pymeth);                                                         \
    return result;                                                             \
  }
 
// both str and std::string have method "find" and "rfin"; try the C++ version
// first and fall back on the Python one in case of failure
// clang-format off
CPYRT_STRING_FINDMETHOD( Find, __cpp_find,  find)
CPYRT_STRING_FINDMETHOD(RFind, __cpp_rfind, rfind)
// clang-format on
 
PyObject* STLStringGetAttr(CPPInstance* self, PyObject* attr_name) {
  std::string* obj = GetSTLString(self);
  if (!obj)
    return nullptr;
 
  PyObject* pystr = PyUnicode_FromStringAndSize(obj->data(), obj->size());
  PyObject* attr = PyObject_GetAttr(pystr, attr_name);
  Py_DECREF(pystr);
  return attr;
}
 
Py_hash_t STLStringHash(PyObject* self) {
  // std::string objects hash to the same values as Python strings to allow
  // matches in dictionaries etc.
  PyObject* data = STLStringGetData(self, false);
  Py_hash_t h = PyUnicode_Type.tp_hash(data);
  Py_DECREF(data);
  return h;
}
 
//- string_view behavior as primitive ----------------------------------------
PyObject* StringViewInit(PyObject* self, PyObject* args, PyObject* /* kwds */) {
  // if constructed from a Python unicode object, the constructor will convert
  // it to a temporary byte string, which is likely to go out of scope too soon;
  // so buffer it as needed
  PyObject* realInit = PyObject_GetAttr(self, PyStrings::gRealInit);
  if (realInit) {
    PyObject *strbuf = nullptr, *newArgs = nullptr;
    if (PyTuple_GET_SIZE(args) == 1) {
      PyObject* arg0 = PyTuple_GET_ITEM(args, 0);
      if (PyUnicode_Check(arg0)) {
        // convert to the expected bytes array to control the temporary
        strbuf = PyUnicode_AsEncodedString(arg0, "UTF-8", "strict");
        newArgs = PyTuple_New(1);
        Py_INCREF(strbuf);
        PyTuple_SET_ITEM(newArgs, 0, strbuf);
      } else if (PyBytes_Check(arg0)) {
        // tie the life time of the provided string to the string_view
        Py_INCREF(arg0);
        strbuf = arg0;
      }
    }
 
    PyObject* result =
        PyObject_Call(realInit, newArgs ? newArgs : args, nullptr);
 
    Py_XDECREF(newArgs);
    Py_DECREF(realInit);
 
    // if construction was successful and a string buffer was used, add a
    // life line to it from the string_view bound object
    if (result && self && strbuf)
      PyObject_SetAttr(self, PyStrings::gLifeLine, strbuf);
    Py_XDECREF(strbuf);
 
    return result;
  }
  return nullptr;
}
 
//- STL iterator behavior ----------------------------------------------------
PyObject* STLIterNext(PyObject* self) {
  // Python iterator protocol __next__ for STL forward iterators.
  bool mustIncrement = true;
  PyObject* last = nullptr;
  if (CPPInstance_Check(self)) {
    auto& dmc = ((CPPInstance*)self)->GetDatamemberCache();
    for (auto& p : dmc) {
      if (p.first == PS_END_ADDR) {
        last = p.second;
        Py_INCREF(last);
      } else if (p.first == PS_FLAG_ADDR) {
        mustIncrement = p.second == Py_True;
        if (!mustIncrement) {
          Py_DECREF(p.second);
          Py_INCREF(Py_True);
          p.second = Py_True;
        }
      }
    }
  }
 
  PyObject* next = nullptr;
  if (last) {
    // handle special case of empty container (i.e. self is end)
    if (!PyObject_RichCompareBool(last, self, Py_EQ)) {
      bool iter_valid = true;
      if (mustIncrement) {
        // prefer preinc, but allow post-inc; in both cases, it is "self" that
        // has the updated state to dereference
        PyObject* iter = PyObject_CallMethodNoArgs(self, PyStrings::gPreInc);
        if (!iter) {
          PyErr_Clear();
          static PyObject* dummy = PyLong_FromLong(1l);
          iter = PyObject_CallMethodOneArg(self, PyStrings::gPostInc, dummy);
        }
        iter_valid = iter && PyObject_RichCompareBool(last, self, Py_NE);
        Py_XDECREF(iter);
      }
 
      if (iter_valid) {
        next = PyObject_CallMethodNoArgs(self, PyStrings::gDeref);
        if (!next)
          PyErr_Clear();
      }
    }
    Py_DECREF(last);
  }
 
  if (!next)
    PyErr_SetString(PyExc_StopIteration, "");
  return next;
}
 
//- STL complex<T> behavior --------------------------------------------------
#define COMPLEX_METH_GETSET(name, cppname)                                     \
  static PyObject* name##ComplexGet(PyObject* self, void*) {                   \
    return PyObject_CallMethodNoArgs(self, cppname);                           \
  }                                                                            \
  static int name##ComplexSet(PyObject* self, PyObject* value, void*) {        \
    PyObject* result = PyObject_CallMethodOneArg(self, cppname, value);        \
    if (result) {                                                              \
      Py_DECREF(result);                                                       \
      return 0;                                                                \
    }                                                                          \
    return -1;                                                                 \
  }                                                                            \
  PyGetSetDef name##Complex{(char*)#name, (getter)name##ComplexGet,            \
                            (setter)name##ComplexSet, nullptr, nullptr};
 
COMPLEX_METH_GETSET(real, PyStrings::gCppReal)
COMPLEX_METH_GETSET(imag, PyStrings::gCppImag)
 
static PyObject* ComplexComplex(PyObject* self) {
  PyObject* real = PyObject_CallMethodNoArgs(self, PyStrings::gCppReal);
  if (!real)
    return nullptr;
  double r = PyFloat_AsDouble(real);
  Py_DECREF(real);
  if (r == -1. && PyErr_Occurred())
    return nullptr;
 
  PyObject* imag = PyObject_CallMethodNoArgs(self, PyStrings::gCppImag);
  if (!imag)
    return nullptr;
  double i = PyFloat_AsDouble(imag);
  Py_DECREF(imag);
  if (i == -1. && PyErr_Occurred())
    return nullptr;
 
  return PyComplex_FromDoubles(r, i);
}
 
static PyObject* ComplexRepr(PyObject* self) {
  PyObject* real = PyObject_CallMethodNoArgs(self, PyStrings::gCppReal);
  if (!real)
    return nullptr;
  double r = PyFloat_AsDouble(real);
  Py_DECREF(real);
  if (r == -1. && PyErr_Occurred())
    return nullptr;
 
  PyObject* imag = PyObject_CallMethodNoArgs(self, PyStrings::gCppImag);
  if (!imag)
    return nullptr;
  double i = PyFloat_AsDouble(imag);
  Py_DECREF(imag);
  if (i == -1. && PyErr_Occurred())
    return nullptr;
 
  std::ostringstream s;
  s << '(' << r << '+' << i << "j)";
  return PyUnicode_FromString(s.str().c_str());
}
 
static PyObject* ComplexDRealGet(CPPInstance* self, void*) {
  return PyFloat_FromDouble(((std::complex<double>*)self->GetObject())->real());
}
 
static int ComplexDRealSet(CPPInstance* self, PyObject* value, void*) {
  double d = PyFloat_AsDouble(value);
  if (d == -1.0 && PyErr_Occurred())
    return -1;
  ((std::complex<double>*)self->GetObject())->real(d);
  return 0;
}
 
PyGetSetDef ComplexDReal{(char*)"real", (getter)ComplexDRealGet,
                         (setter)ComplexDRealSet, nullptr, nullptr};
 
static PyObject* ComplexDImagGet(CPPInstance* self, void*) {
  return PyFloat_FromDouble(((std::complex<double>*)self->GetObject())->imag());
}
 
static int ComplexDImagSet(CPPInstance* self, PyObject* value, void*) {
  double d = PyFloat_AsDouble(value);
  if (d == -1.0 && PyErr_Occurred())
    return -1;
  ((std::complex<double>*)self->GetObject())->imag(d);
  return 0;
}
 
PyGetSetDef ComplexDImag{(char*)"imag", (getter)ComplexDImagGet,
                         (setter)ComplexDImagSet, nullptr, nullptr};
 
static PyObject* ComplexDComplex(CPPInstance* self) {
  double r = ((std::complex<double>*)self->GetObject())->real();
  double i = ((std::complex<double>*)self->GetObject())->imag();
  return PyComplex_FromDoubles(r, i);
}
 
} // unnamed namespace
 
//- public functions ---------------------------------------------------------
namespace cppjit::cpyrt {
std::set<std::string> gIteratorTypes;
}
 
static inline bool run_pythonizors(PyObject* pyclass, PyObject* pyname,
                                   const std::vector<PyObject*>& v) {
  PyObject* args = PyTuple_New(2);
  Py_INCREF(pyclass);
  PyTuple_SET_ITEM(args, 0, pyclass);
  Py_INCREF(pyname);
  PyTuple_SET_ITEM(args, 1, pyname);
 
  bool pstatus = true;
  for (auto pythonizor : v) {
    PyObject* result = PyObject_CallObject(pythonizor, args);
    if (!result) {
      pstatus = false; // TODO: detail the error handling
      break;
    }
    Py_DECREF(result);
  }
  Py_DECREF(args);
 
  return pstatus;
}
 
bool cpyrt::Pythonize(PyObject* pyclass, interop::TCppScope_t scope) {
  const std::string& name = interop::GetScopedFinalName(scope);
 
  // Add pre-defined pythonizations (for STL and ROOT) to classes based on their
  // signature and/or class name.
  if (!pyclass)
    return false;
 
  CPPScope* klass = (CPPScope*)pyclass;
 
  //- method name based pythonization ------------------------------------------
 
  // for smart pointer style classes that are otherwise not known as such; would
  // prefer operator-> as that returns a pointer (which is simpler since it
  // never has to deal with ref-assignment), but operator* plays better with STL
  // iters and algorithms
  // optional and expected use operator* to access a contained value, rather
  // than to provide pointer-like access to an object.
  const bool has_value_semantics = IsTemplatedSTLClass(name, "optional") ||
                                   IsTemplatedSTLClass(name, "expected");
  if (HasAttrDirect(pyclass, PyStrings::gDeref) &&
      !interop::IsSmartPtr(klass->fCppType)) {
    if (has_value_semantics)
      Utility::AddToClass(pyclass, "__getattr__", (PyCFunction)ValueGetAttr,
                          METH_O);
    else
      Utility::AddToClass(pyclass, "__getattr__", (PyCFunction)DeRefGetAttr,
                          METH_O);
  } else if (HasAttrDirect(pyclass, PyStrings::gFollow) &&
             !interop::IsSmartPtr(klass->fCppType))
    Utility::AddToClass(pyclass, "__getattr__", (PyCFunction)FollowGetAttr,
                        METH_O);
 
  // for pre-check of nullptr for boolean types
  if (HasAttrDirect(pyclass, PyStrings::gCppBool)) {
    const char* pybool_name = "__bool__";
    Utility::AddToClass(pyclass, pybool_name, (PyCFunction)NullCheckBool,
                        METH_NOARGS);
  }
 
  // for STL containers, and user classes modeled after them
  // the attribute must be a cpyrt overload, otherwise the check gives false
  // positives in the case where the class has a non-function attribute that is
  // called "size".
  {
    // only pythonize as __len__ if the class actually models a container,
    // i.e. it is iterable (begin/end) or indexable (operator[]), and size()
    // returns an integral value; e.g. a size() returning a custom dimension
    // object must not become __len__ (len() would fail on a valid object).
    // The size method and the container interface may come from a base class
    // (checked with PyObject_HasAttr), as long as at least one of them is
    // freshly declared here (else the base's __len__ is inherited via the
    // normal MRO and there is nothing to do).
    bool sizeDirect =
        HasAttrDirect(pyclass, PyStrings::gSize, /*mustBecpyrt=*/true);
    bool ifaceDirect = (HasAttrDirect(pyclass, PyStrings::gBegin) &&
                        HasAttrDirect(pyclass, PyStrings::gEnd)) ||
                       HasAttrDirect(pyclass, PyStrings::gGetItem);
    PyObject* size_attr = (sizeDirect || ifaceDirect)
                              ? PyObject_GetAttr(pyclass, PyStrings::gSize)
                              : nullptr;
    if (!size_attr)
      PyErr_Clear();
    else if (!CPPOverload_Check(size_attr)) {
      // non-function attribute called "size": not pythonizable
      Py_DECREF(size_attr);
      size_attr = nullptr;
    }
    bool isContainer = size_attr &&
                       PyObject_HasAttr(pyclass, PyStrings::gBegin) &&
                       PyObject_HasAttr(pyclass, PyStrings::gEnd);
    if (size_attr && !isContainer) {
      // indexable via a real C++ operator[]? (every class "has" __getitem__
      // through the CPPInstance defaults, so require an actual overload)
      PyObject* gi = PyObject_GetAttr(pyclass, PyStrings::gGetItem);
      if (gi) {
        isContainer = CPPOverload_Check(gi);
        Py_DECREF(gi);
      } else
        PyErr_Clear();
    }
    if (isContainer) {
      auto isIntegral = [](const std::string& tn) {
        static const char* const integrals[] = {
            "char",           "signed char", "unsigned char",     "short",
            "unsigned short", "int",         "unsigned int",      "long",
            "unsigned long",  "long long",   "unsigned long long"};
        for (const char* i : integrals)
          if (tn == i)
            return true;
        return false;
      };
      // find the scope (this class or a base) that declares size()
      std::vector<interop::TCppMethod_t> sizes;
      std::deque<interop::TCppScope_t> todo{klass->fCppType};
      while (sizes.empty() && !todo.empty()) {
        interop::TCppScope_t s = todo.front();
        todo.pop_front();
        sizes = interop::GetMethodsFromName(s, "size");
        for (interop::TCppIndex_t i = 0; i < interop::GetNumBases(s); ++i)
          todo.push_back(interop::GetBaseScope(s, i));
      }
      for (interop::TCppMethod_t m : sizes) {
        if (isIntegral(interop::GetMethodReturnTypeAsString(m))) {
          Utility::AddToClass(pyclass, "__len__", "size");
          break;
        }
      }
    }
    Py_XDECREF(size_attr);
  }
 
  if (HasAttrDirect(pyclass, PyStrings::gContains)) {
    Utility::AddToClass(pyclass, "__contains__", "contains");
  }
 
  if (!IsTemplatedSTLClass(name, "vector") && // vector is dealt with below
      !((PyTypeObject*)pyclass)->tp_iter) {
    if (HasAttrDirect(pyclass, PyStrings::gBegin) &&
        HasAttrDirect(pyclass, PyStrings::gEnd)) {
      // obtain the name of the return type
      const auto& methods =
          interop::GetMethodsFromName(klass->fCppType, "begin");
      if (!methods.empty()) {
        // check return type; if not explicitly an iterator, add it to the
        // "known" return types to add the "next" method on use
        interop::TCppMethod_t meth = methods[0];
        const std::string& resname = interop::GetMethodReturnTypeAsString(meth);
        bool isIterator = gIteratorTypes.find(resname) != gIteratorTypes.end();
        if (!isIterator && interop::GetScope(resname)) {
          if (resname.find("iterator") == std::string::npos)
            gIteratorTypes.insert(resname);
          isIterator = true;
        }
 
        if (isIterator) {
          // install iterator protocol a la STL
          ((PyTypeObject*)pyclass)->tp_iter = (getiterfunc)STLSequenceIter;
          Utility::AddToClass(pyclass, "__iter__", (PyCFunction)STLSequenceIter,
                              METH_NOARGS);
        } else {
          // still okay if this is some pointer type of builtin persuasion
          // (general class won't work: the return type needs to understand the
          // iterator protocol)
          std::string resolved = interop::ResolveName(resname);
          if (resolved.back() == '*' &&
              interop::IsBuiltin(resolved.substr(0, resolved.size() - 1))) {
            ((PyTypeObject*)pyclass)->tp_iter = (getiterfunc)LLSequenceIter;
            Utility::AddToClass(pyclass, "__iter__",
                                (PyCFunction)LLSequenceIter, METH_NOARGS);
          }
        }
      }
    }
    if (!((PyTypeObject*)pyclass)->tp_iter && // no iterator resolved
        HasAttrDirect(pyclass, PyStrings::gGetItem) &&
        PyObject_HasAttr(pyclass, PyStrings::gLen)) {
      // Python will iterate over __getitem__ using integers, but C++ operator[]
      // will never raise a StopIteration. A checked getitem (raising IndexError
      // if beyond size()) works in some cases but would mess up if operator[]
      // is meant to implement an associative container. So, this has to be
      // implemented as an iterator protocol.
      ((PyTypeObject*)pyclass)->tp_iter = (getiterfunc)index_iter;
      Utility::AddToClass(pyclass, "__iter__", (PyCFunction)index_iter,
                          METH_NOARGS);
    }
  }
 
  // operator==/!= are used in op_richcompare of CPPInstance, which subsequently
  // allows comparisons to None; if no operator is available, a hook is
  // installed for lazy lookups in the global and/or class namespace
  if (HasAttrDirect(pyclass, PyStrings::gEq, true) &&
      interop::GetMethodsFromName(klass->fCppType, "__eq__").empty()) {
    PyObject* cppol = PyObject_GetAttr(pyclass, PyStrings::gEq);
    if (!klass->fOperators)
      klass->fOperators = new Utility::PyOperators();
    klass->fOperators->fEq = cppol;
    // re-insert the forwarding __eq__ from the CPPInstance in case there was a
    // Python-side override in the base class
    static PyObject* top_eq = nullptr;
    if (!top_eq) {
      PyObject* top_cls = PyObject_GetAttrString(gThisModule, "CPPInstance");
      top_eq = PyObject_GetAttr(top_cls, PyStrings::gEq);
      Py_DECREF(top_eq); // make it borrowed
      Py_DECREF(top_cls);
    }
    PyObject_SetAttr(pyclass, PyStrings::gEq, top_eq);
  }
 
  if (HasAttrDirect(pyclass, PyStrings::gNe, true) &&
      interop::GetMethodsFromName(klass->fCppType, "__ne__").empty()) {
    PyObject* cppol = PyObject_GetAttr(pyclass, PyStrings::gNe);
    if (!klass->fOperators)
      klass->fOperators = new Utility::PyOperators();
    klass->fOperators->fNe = cppol;
    // re-insert the forwarding __ne__ (same reason as above for __eq__)
    static PyObject* top_ne = nullptr;
    if (!top_ne) {
      PyObject* top_cls = PyObject_GetAttrString(gThisModule, "CPPInstance");
      top_ne = PyObject_GetAttr(top_cls, PyStrings::gNe);
      Py_DECREF(top_ne); // make it borrowed
      Py_DECREF(top_cls);
    }
    PyObject_SetAttr(pyclass, PyStrings::gNe, top_ne);
  }
 
  // The pseudo-constructor is not meant for standard-library aggregates;
  // detect those by walking the enclosing scopes instead of testing the
  // exposed name, which may or may not be qualified (e.g. std::tuple is
  // reported without its namespace here).
  bool stdclass = false;
  for (interop::TCppScope_t p =
           interop::GetParentScope(((CPPClass*)pyclass)->fCppType);
       p; p = interop::GetParentScope(p)) {
    if (interop::IsNamespace(p) && interop::GetName(p) == "std") {
      stdclass = true;
      break;
    }
  }
 
  if (interop::IsAggregate(((CPPClass*)pyclass)->fCppType) && !stdclass) {
    // create a pseudo-constructor to allow initializer-style object creation
    interop::TCppScope_t kls = ((CPPClass*)pyclass)->fCppType;
    std::vector<interop::TCppScope_t> datamems;
    interop::GetDatamembers(kls, datamems);
    if (!datamems.empty()) {
      std::string rname = name;
      TypeManip::cppscope_to_legalname(rname);
 
      std::ostringstream initdef;
      initdef << "namespace __cppjit_internal {\n"
              << "void init_" << rname << "(" << name << "** self";
      bool codegen_ok = true;
      std::vector<std::string> arg_types, arg_names, arg_defaults;
      const int ndata = datamems.size();
      arg_types.reserve(ndata);
      arg_names.reserve(ndata);
      arg_defaults.reserve(ndata);
      for (auto data : datamems) {
        if (interop::IsStaticDatamember(data) || !interop::IsPublicData(data))
          continue;
 
        interop::TCppType_t datammember_type = interop::GetDatamemberType(data);
        const std::string& res =
            interop::IsEnumType(datammember_type)
                ? interop::GetScopedFinalName(
                      interop::GetScopeFromType(datammember_type))
                : interop::GetTypeAsString(
                      interop::ResolveType(datammember_type));
        const std::string& cpd = TypeManip::compound(res);
        std::string res_clean = TypeManip::clean_type(res, false, true);
 
        if (res_clean == "internal_enum_type_t")
          res_clean = res; // restore (properly scoped name)
 
        if (res.rfind(']') == std::string::npos &&
            res.rfind(')') == std::string::npos) {
          if (!cpd.empty())
            arg_types.push_back(res_clean + cpd);
          else
            arg_types.push_back("const " + res_clean + "&");
          arg_names.push_back(interop::GetFinalName(data));
          if ((!cpd.empty() && cpd.back() == '*') ||
              interop::IsBuiltin(res_clean))
            arg_defaults.push_back("0");
          else {
            interop::TCppScope_t klsid = interop::GetScope(res_clean);
            if (interop::IsDefaultConstructable(klsid))
              arg_defaults.push_back(res_clean + "{}");
          }
        } else {
          codegen_ok =
              false; // TODO: how to support arrays, anonymous enums, etc?
          break;
        }
      }
 
      if (codegen_ok && !arg_types.empty()) {
        bool defaults_ok = arg_defaults.size() == arg_types.size();
        for (std::vector<std::string>::size_type i = 0; i < arg_types.size();
             ++i) {
          initdef << ", " << arg_types[i] << " " << arg_names[i];
          if (defaults_ok)
            initdef << " = " << arg_defaults[i];
        }
        initdef << ") {\n  *self = new " << name << "{";
        for (std::vector<std::string>::size_type i = 0; i < arg_names.size();
             ++i) {
          if (i != 0)
            initdef << ", ";
          initdef << arg_names[i];
        }
        initdef << "};\n} }";
 
        if (interop::Compile(initdef.str(), true /* silent */)) {
          interop::TCppScope_t cis = interop::GetScope("__cppjit_internal");
          const auto& methods =
              interop::GetMethodsFromName(cis, "init_" + rname);
          if (methods.size()) {
            if (!Utility::AddToClass(pyclass, "__init__",
                                     new CPPFunction(cis, methods[0])))
              PyErr_Clear();
          }
        }
      }
    }
  }
 
  //- class name based pythonization -------------------------------------------
 
  if (IsTemplatedSTLClass(name, "span")) {
    // libstdc++ (GCC >= 15) implements std::span::iterator using a private
    // nested tag type, which makes the iterator non-instantiable by
    // CallFunc-generated wrappers (the return type cannot be named without
    // violating access rules).
    //
    // To preserve correct Python iteration semantics, we replace begin()/end()
    // for std::span to return a custom pointer-based iterator instead.
    Utility::AddToClass(pyclass, "begin", (PyCFunction)SpanBegin, METH_NOARGS);
    Utility::AddToClass(pyclass, "end", (PyCFunction)SpanEnd, METH_NOARGS);
  }
 
  if (IsTemplatedSTLClass(name, "vector")) {
 
    // std::vector<bool> is a special case in C++
    if (!sVectorBoolTypeID)
      sVectorBoolTypeID = interop::GetScope("std::vector<bool>");
    if (klass->fCppType == sVectorBoolTypeID) {
      Utility::AddToClass(pyclass, "__getitem__",
                          (PyCFunction)VectorBoolGetItem, METH_O);
      Utility::AddToClass(pyclass, "__setitem__",
                          (PyCFunction)VectorBoolSetItem);
    } else {
      // constructor that takes python collections
      Utility::AddToClass(pyclass, "__real_init", "__init__");
      Utility::AddToClass(pyclass, "__init__", (PyCFunction)VectorInit,
                          METH_VARARGS | METH_KEYWORDS);
 
      // data with size
      if (!Utility::AddToClass(pyclass, "__real_data", "data"))
        PyErr_Clear(); // no 'data' method to alias
      Utility::AddToClass(pyclass, "data", (PyCFunction)VectorData);
 
      // checked getitem
      if (HasAttrDirect(pyclass, PyStrings::gLen)) {
        Utility::AddToClass(pyclass, "_getitem__unchecked", "__getitem__");
        Utility::AddToClass(pyclass, "__getitem__", (PyCFunction)VectorGetItem,
                            METH_O);
      }
 
      // vector-optimized iterator protocol
      ((PyTypeObject*)pyclass)->tp_iter = (getiterfunc)vector_iter;
 
      // optimized __iadd__
      Utility::AddToClass(pyclass, "__iadd__", (PyCFunction)VectorIAdd,
                          METH_VARARGS | METH_KEYWORDS);
 
      // helpers for iteration
      interop::TCppType_t value_type =
          interop::GetTypeFromScope(interop::GetNamed("value_type", scope));
      interop::TCppType_t vtype = interop::ResolveType(value_type);
 
      // numpy array conversion; only for vectors of non-class types:
      // data() on a vector of class instances hands back a proxy of the
      // first element, so forwarding __array__ to it would yield that
      // element instead of the full vector (numpy's generic sequence
      // protocol handles such vectors correctly on its own)
      if (vtype && !interop::GetScopeFromType(vtype))
        Utility::AddToClass(pyclass, "__array__", (PyCFunction)VectorArray,
                            METH_VARARGS | METH_KEYWORDS /* unused */);
 
      if (vtype) { // actually resolved?
        PyObject* pyvalue_type = PyLong_FromVoidPtr(vtype.data);
        PyObject_SetAttr(pyclass, PyStrings::gValueTypePtr, pyvalue_type);
        Py_DECREF(pyvalue_type);
        pyvalue_type =
            PyUnicode_FromString(interop::GetTypeAsString(vtype).c_str());
        PyObject_SetAttr(pyclass, PyStrings::gValueType, pyvalue_type);
        Py_DECREF(pyvalue_type);
      }
 
      size_t typesz = interop::SizeOfType(vtype);
      if (typesz) {
        PyObject* pyvalue_size = PyLong_FromSsize_t(typesz);
        PyObject_SetAttr(pyclass, PyStrings::gValueSize, pyvalue_size);
        Py_DECREF(pyvalue_size);
      }
    }
  }
 
  else if (IsTemplatedSTLClass(name, "array")) {
    // constructor that takes python associative collections
    Utility::AddToClass(pyclass, "__real_init", "__init__");
    Utility::AddToClass(pyclass, "__init__", (PyCFunction)ArrayInit,
                        METH_VARARGS | METH_KEYWORDS);
  }
 
  else if (IsTemplatedSTLClass(name, "map") ||
           IsTemplatedSTLClass(name, "unordered_map")) {
    // constructor that takes python associative collections
    Utility::AddToClass(pyclass, "__real_init", "__init__");
    Utility::AddToClass(pyclass, "__init__", (PyCFunction)MapInit,
                        METH_VARARGS | METH_KEYWORDS);
    // From C++20, std::map/unordered_map have a native contains() that the
    // generic contains->__contains__ mapping above will pick up. Strong-types
    // reflection does not always expose it, so fall back to a find()-based
    // __contains__ when none was reflected (still O(log n)/O(1), never
    // iterating).
    if (!HasAttrDirect(pyclass, PyStrings::gContains))
      Utility::AddToClass(pyclass, "__contains__",
                          (PyCFunction)STLContainsWithFind, METH_O);
  }
 
  else if (IsTemplatedSTLClass(name, "set")) {
    // constructor that takes python associative collections
    Utility::AddToClass(pyclass, "__real_init", "__init__");
    Utility::AddToClass(pyclass, "__init__", (PyCFunction)SetInit,
                        METH_VARARGS | METH_KEYWORDS);
    // From C++20, std::set has a native contains() that the generic
    // contains->__contains__ mapping above will pick up. Strong-types
    // reflection does not always expose it, so fall back to a find()-based
    // __contains__ when none was reflected (still O(log n), never iterating).
    if (!HasAttrDirect(pyclass, PyStrings::gContains))
      Utility::AddToClass(pyclass, "__contains__",
                          (PyCFunction)STLContainsWithFind, METH_O);
  }
 
  else if (IsTemplatedSTLClass(name, "pair")) {
    Utility::AddToClass(pyclass, "__getitem__", (PyCFunction)PairUnpack,
                        METH_O);
    Utility::AddToClass(pyclass, "__len__", (PyCFunction)ReturnTwo,
                        METH_NOARGS);
  }
 
  if (IsTemplatedSTLClass(name, "shared_ptr") ||
      IsTemplatedSTLClass(name, "unique_ptr")) {
    Utility::AddToClass(pyclass, "__real_init", "__init__");
    Utility::AddToClass(pyclass, "__init__", (PyCFunction)SmartPtrInit,
                        METH_VARARGS | METH_KEYWORDS);
  }
 
  else if (!((PyTypeObject*)pyclass)->tp_iter &&
           (name.find("iterator") != std::string::npos ||
            gIteratorTypes.find(name) != gIteratorTypes.end())) {
    ((PyTypeObject*)pyclass)->tp_iternext = (iternextfunc)STLIterNext;
    Utility::AddToClass(pyclass, "__next__", (PyCFunction)STLIterNext,
                        METH_NOARGS);
    ((PyTypeObject*)pyclass)->tp_iter = (getiterfunc)PyObject_SelfIter;
    Utility::AddToClass(pyclass, "__iter__", (PyCFunction)PyObject_SelfIter,
                        METH_NOARGS);
  }
 
  else if (name == "std::basic_string<char>" ||
           name == "std::basic_string<char,std::char_traits<char>,std::"
                   "allocator<char> >" ||
           name == "std::basic_string<char, std::char_traits<char>, "
                   "std::allocator<char> >" ||
           name == "std::__1::basic_string<char>" || // libc++ inline namespace
           name == "std::string") { // typedef preserved by GetScopedFinalName
                                    // on libc++
    Utility::AddToClass(pyclass, "__repr__", (PyCFunction)STLStringRepr,
                        METH_NOARGS);
    Utility::AddToClass(pyclass, "__str__", (PyCFunction)STLStringStr,
                        METH_NOARGS);
    Utility::AddToClass(pyclass, "__bytes__", (PyCFunction)STLStringBytes,
                        METH_NOARGS);
    Utility::AddToClass(pyclass, "__cmp__", (PyCFunction)STLStringCompare,
                        METH_O);
    Utility::AddToClass(pyclass, "__eq__", (PyCFunction)STLStringIsEqual,
                        METH_O);
    Utility::AddToClass(pyclass, "__ne__", (PyCFunction)STLStringIsNotEqual,
                        METH_O);
    // Python's `in` operator needs __contains__ on the proxy regardless of the
    // C++ standard cpyrt is built with; it never dispatches to C++'s
    // std::string::contains (C++23+). The old `__cplusplus <= 202302L` guard
    // wrongly dropped it when built with -std=c++2c (__cplusplus == 202400L).
    Utility::AddToClass(pyclass, "__contains__", (PyCFunction)STLStringContains,
                        METH_O);
    Utility::AddToClass(pyclass, "decode", (PyCFunction)STLStringDecode,
                        METH_VARARGS | METH_KEYWORDS);
    Utility::AddToClass(pyclass, "__cpp_find", "find");
    Utility::AddToClass(pyclass, "find", (PyCFunction)STLStringFind,
                        METH_VARARGS | METH_KEYWORDS);
    Utility::AddToClass(pyclass, "__cpp_rfind", "rfind");
    Utility::AddToClass(pyclass, "rfind", (PyCFunction)STLStringRFind,
                        METH_VARARGS | METH_KEYWORDS);
    Utility::AddToClass(pyclass, "__cpp_replace", "replace");
    Utility::AddToClass(pyclass, "replace", (PyCFunction)STLStringReplace,
                        METH_VARARGS | METH_KEYWORDS);
    Utility::AddToClass(pyclass, "__getattr__", (PyCFunction)STLStringGetAttr,
                        METH_O);
 
    // to allow use of std::string in dictionaries and findable with str
    ((PyTypeObject*)pyclass)->tp_hash = (hashfunc)STLStringHash;
  }
 
  else if (name == "std::basic_string_view<char>" ||
           name == "std::basic_string_view<char,std::char_traits<char> >" ||
           name == "std::basic_string_view<char, std::char_traits<char> >" ||
           name ==
               "std::__1::basic_string_view<char>" || // libc++ inline namespace
           name == "std::string_view") {              // typedef preserved by
                                         // GetScopedFinalName on libc++
    Utility::AddToClass(pyclass, "__real_init", "__init__");
    Utility::AddToClass(pyclass, "__init__", (PyCFunction)StringViewInit,
                        METH_VARARGS | METH_KEYWORDS);
    Utility::AddToClass(pyclass, "__bytes__", (PyCFunction)STLViewStringBytes,
                        METH_NOARGS);
    Utility::AddToClass(pyclass, "__cmp__", (PyCFunction)STLViewStringCompare,
                        METH_O);
    Utility::AddToClass(pyclass, "__eq__", (PyCFunction)STLViewStringIsEqual,
                        METH_O);
    Utility::AddToClass(pyclass, "__ne__", (PyCFunction)STLViewStringIsNotEqual,
                        METH_O);
    Utility::AddToClass(pyclass, "__repr__", (PyCFunction)STLViewStringRepr,
                        METH_NOARGS);
    Utility::AddToClass(pyclass, "__str__", (PyCFunction)STLViewStringStr,
                        METH_NOARGS);
  }
 
  // The first condition was already present in upstream cpyrt. The others
  // are special to ROOT, because its reflection layer gives us the types
  // without the "std::" namespace. On some platforms, that applies only to the
  // template arguments, and on others also to the "basic_string".
  else if (name == "std::basic_string<wchar_t,std::char_traits<wchar_t>,std::"
                   "allocator<wchar_t> >" ||
           name == "std::basic_string<wchar_t, std::char_traits<wchar_t>, "
                   "std::allocator<wchar_t> >" ||
           name == "basic_string<wchar_t,char_traits<wchar_t>,allocator<wchar_"
                   "t> >" ||
           name == "std::basic_string<wchar_t,char_traits<wchar_t>,allocator<"
                   "wchar_t> >" ||
           name == "std::__1::basic_string<wchar_t,std::__1::char_traits<wchar_"
                   "t>,std::__1::allocator<wchar_t> >" ||
           name == "std::wstring") {
    Utility::AddToClass(pyclass, "__repr__", (PyCFunction)STLWStringRepr,
                        METH_NOARGS);
    Utility::AddToClass(pyclass, "__str__", (PyCFunction)STLWStringStr,
                        METH_NOARGS);
    Utility::AddToClass(pyclass, "__bytes__", (PyCFunction)STLWStringBytes,
                        METH_NOARGS);
    Utility::AddToClass(pyclass, "__cmp__", (PyCFunction)STLWStringCompare,
                        METH_O);
    Utility::AddToClass(pyclass, "__eq__", (PyCFunction)STLWStringIsEqual,
                        METH_O);
    Utility::AddToClass(pyclass, "__ne__", (PyCFunction)STLWStringIsNotEqual,
                        METH_O);
  }
 
  else if (name == "complex<double>" || name == "std::complex<double>") {
    Utility::AddToClass(pyclass, "__cpp_real", "real");
    PyObject_SetAttrString(
        pyclass, "real",
        PyDescr_NewGetSet((PyTypeObject*)pyclass, &ComplexDReal));
    Utility::AddToClass(pyclass, "__cpp_imag", "imag");
    PyObject_SetAttrString(
        pyclass, "imag",
        PyDescr_NewGetSet((PyTypeObject*)pyclass, &ComplexDImag));
    Utility::AddToClass(pyclass, "__complex__", (PyCFunction)ComplexDComplex,
                        METH_NOARGS);
    Utility::AddToClass(pyclass, "__repr__", (PyCFunction)ComplexRepr,
                        METH_NOARGS);
  }
 
  else if (IsTemplatedSTLClass(name, "complex")) {
    Utility::AddToClass(pyclass, "__cpp_real", "real");
    PyObject_SetAttrString(
        pyclass, "real",
        PyDescr_NewGetSet((PyTypeObject*)pyclass, &realComplex));
    Utility::AddToClass(pyclass, "__cpp_imag", "imag");
    PyObject_SetAttrString(
        pyclass, "imag",
        PyDescr_NewGetSet((PyTypeObject*)pyclass, &imagComplex));
    Utility::AddToClass(pyclass, "__complex__", (PyCFunction)ComplexComplex,
                        METH_NOARGS);
    Utility::AddToClass(pyclass, "__repr__", (PyCFunction)ComplexRepr,
                        METH_NOARGS);
  }
 
  // direct user access; there are two calls here:
  //   - explicit pythonization: won't fall through to the base classes and is
  //   preferred if present
  //   - normal pythonization: only called if explicit isn't present, falls
  //   through to base classes
  // The lookups live in Compatibility.h because they also accept the old
  // cppyy names of these hooks.
  bool bUserOk = true;
  PyObject* res = nullptr;
  PyObject* pyname = PyUnicode_FromString(name.c_str());
  if (PyObject* exPythonize = compat::ExplicitPythonizeHookName(pyclass)) {
    res = PyObject_CallMethodObjArgs(pyclass, exPythonize, pyclass, pyname,
                                     nullptr);
    bUserOk = (bool)res;
  } else if (PyObject* func = compat::GetPythonizeHook(pyclass)) {
    res = PyObject_CallFunctionObjArgs(func, pyclass, pyname, nullptr);
    Py_DECREF(func);
    bUserOk = (bool)res;
  }
  if (!bUserOk) {
    Py_DECREF(pyname);
    return false;
  } else {
    Py_XDECREF(res);
    // pyname handed to args tuple below
  }
 
  // call registered pythonizors, if any: first run the namespace-specific
  // pythonizors, then the global ones (the idea is to allow writing a
  // pythonizor that see all classes)
  bool pstatus = true;
  std::string outer_scope = TypeManip::extract_namespace(name);
  auto& pyzMap = pythonizations();
  if (!outer_scope.empty()) {
    auto p = pyzMap.find(outer_scope);
    if (p != pyzMap.end()) {
      PyObject* subname = PyUnicode_FromString(
          name.substr(outer_scope.size() + 2, std::string::npos).c_str());
      pstatus = run_pythonizors(pyclass, subname, p->second);
      Py_DECREF(subname);
    }
  }
 
  if (pstatus) {
    auto p = pyzMap.find("");
    if (p != pyzMap.end())
      pstatus = run_pythonizors(pyclass, pyname, p->second);
  }
 
  Py_DECREF(pyname);
 
  // phew! all done ...
  return pstatus;
}