Thanks. I’ve added a bug report regarding this issue:
1 Like
Hi @LeWhoo,
This is something I have seen in the past, in principle there is support for proper, direct conversion of numpy arrays into std::vector only for flat arrays. Unfortunately, the reproducer you wrote does not work even in upstream cppyy:
$: conda activate cppyy-bare
$: conda list
# packages in environment at /home/vpadulan/programs/mambaforge/envs/cppyy-bare:
#
# Name Version Build Channel
_libgcc_mutex 0.1 conda_forge conda-forge
_openmp_mutex 4.5 2_gnu conda-forge
binutils 2.39 hdd6e379_1 conda-forge
binutils_impl_linux-64 2.39 he00db2b_1 conda-forge
binutils_linux-64 2.39 h5fc0e48_13 conda-forge
bzip2 1.0.8 h7f98852_4 conda-forge
c-compiler 1.5.2 h0b41bf4_0 conda-forge
ca-certificates 2023.5.7 hbcca054_0 conda-forge
cppyy 2.4.2 py310h2efb557_1 conda-forge
cppyy-backend 1.14.10 py310hdf3cbec_0 conda-forge
cppyy-cling 6.27.1 py310h07cc95c_0 conda-forge
cpycppyy 1.12.12 py310hdf3cbec_0 conda-forge
cxx-compiler 1.5.2 hf52228f_0 conda-forge
gcc 11.3.0 h02d0930_13 conda-forge
gcc_impl_linux-64 11.3.0 hab1b70f_19 conda-forge
gcc_linux-64 11.3.0 he6f903b_13 conda-forge
gxx 11.3.0 h02d0930_13 conda-forge
gxx_impl_linux-64 11.3.0 hab1b70f_19 conda-forge
gxx_linux-64 11.3.0 hc203a17_13 conda-forge
kernel-headers_linux-64 2.6.32 he073ed8_15 conda-forge
ld_impl_linux-64 2.39 hcc3a1bd_1 conda-forge
libffi 3.4.2 h7f98852_5 conda-forge
libgcc-devel_linux-64 11.3.0 h210ce93_19 conda-forge
libgcc-ng 12.2.0 h65d4601_19 conda-forge
libgomp 12.2.0 h65d4601_19 conda-forge
libllvm9 9.0.1 default_hc23dcda_7 conda-forge
libnsl 2.0.0 h7f98852_0 conda-forge
libsanitizer 11.3.0 h239ccf8_19 conda-forge
libsqlite 3.41.2 h2797004_1 conda-forge
libstdcxx-devel_linux-64 11.3.0 h210ce93_19 conda-forge
libstdcxx-ng 12.2.0 h46fd767_19 conda-forge
libuuid 2.38.1 h0b41bf4_0 conda-forge
libzlib 1.2.13 h166bdaf_4 conda-forge
ncurses 6.3 h27087fc_1 conda-forge
numpy 1.24.3 py310ha4c1d20_0 conda-forge
openssl 3.1.0 hd590300_3 conda-forge
pip 23.1.2 pyhd8ed1ab_0 conda-forge
python 3.10.10 he550d4f_0_cpython conda-forge
python_abi 3.10 3_cp310 conda-forge
readline 8.2 h8228510_1 conda-forge
setuptools 67.7.2 pyhd8ed1ab_0 conda-forge
sysroot_linux-64 2.12 he073ed8_15 conda-forge
tk 8.6.12 h27826a3_0 conda-forge
tzdata 2023c h71feb2d_0 conda-forge
wheel 0.40.0 pyhd8ed1ab_0 conda-forge
xz 5.2.6 h166bdaf_0 conda-forge
# repro.py
import cppyy
import numpy as np
b = np.arange(2*2, dtype=np.uint32).reshape(2,2)
a = cppyy.gbl.std.vector["std::vector<unsigned int>"](b)
$: python repro.py
Traceback (most recent call last):
File "/home/vpadulan/projects/rootcode/github-issues/github-12718/repro_cppyy.py", line 4, in <module>
a = cppyy.gbl.std.vector["std::vector<unsigned int>"](b)
TypeError: Template method resolution failed:
none of the 9 overloaded methods succeeded. Full details:
vector<std::vector<unsigned int> >::vector<std::vector<unsigned int> >(std::initializer_list<std::vector<std::vector<unsigned int> >::value_type> __l, const std::vector<std::vector<unsigned int> >::allocator_type& __a = std::vector<std::vector<unsigned int, std::allocator<unsigned int> >, std::allocator<std::vector<unsigned int, std::allocator<unsigned int> > > >::allocator_type()) =>
TypeError: could not convert argument 1 (buffer itemsize (4) does not match expected size (24))
vector<std::vector<unsigned int> >::vector<std::vector<unsigned int> >(std::vector<std::vector<unsigned int> >&& __rv, const std::vector<std::vector<unsigned int> >::allocator_type& __m) =>
TypeError: takes at least 2 arguments (1 given)
vector<std::vector<unsigned int> >::vector<std::vector<unsigned int> >(std::vector<std::vector<unsigned int> >&&) =>
TypeError: could not convert argument 1
vector<std::vector<unsigned int> >::vector<std::vector<unsigned int> >(const std::vector<std::vector<unsigned int> >& __x, const std::vector<std::vector<unsigned int> >::allocator_type& __a) =>
TypeError: takes at least 2 arguments (1 given)
vector<std::vector<unsigned int> >::vector<std::vector<unsigned int> >(const std::vector<std::vector<unsigned int> >::allocator_type& __a) =>
TypeError: could not convert argument 1
vector<std::vector<unsigned int> >::vector<std::vector<unsigned int> >(const std::vector<std::vector<unsigned int> >& __x) =>
TypeError: could not convert argument 1
vector<std::vector<unsigned int> >::vector<std::vector<unsigned int> >() =>
TypeError: takes at most 0 arguments (1 given)
vector<std::vector<unsigned int> >::vector<std::vector<unsigned int> >(std::vector<char>::size_type __n, const std::vector<std::vector<unsigned int> >::value_type& __value, const std::vector<std::vector<unsigned int> >::allocator_type& __a = std::vector<std::vector<unsigned int, std::allocator<unsigned int> >, std::allocator<std::vector<unsigned int, std::allocator<unsigned int> > > >::allocator_type()) =>
TypeError: takes at least 2 arguments (1 given)
vector<std::vector<unsigned int> >::vector<std::vector<unsigned int> >(std::vector<char>::size_type __n, const std::vector<std::vector<unsigned int> >::allocator_type& __a = std::vector<std::vector<unsigned int, std::allocator<unsigned int> >, std::allocator<std::vector<unsigned int, std::allocator<unsigned int> > > >::allocator_type()) =>
TypeError: could not convert argument 1 (an integer is required)
vector<std::vector<unsigned int> >::vector<std::vector<unsigned int> >() =>
TypeError: takes at most 0 arguments (1 given)
I am especially suspicious of
TypeError: could not convert argument 1 (buffer itemsize (4) does not match expected size (24))
I will try to address this but at this point there is no clear answer.
Cheers,
Vincenzo
Interesting, especially that @wlav above reported that it worked for him in cppyy. I hope it can be fixed. Meanwhile, in my class that basically makes std::vector pretend it is a list or a numpy array, I manually do something like:
if (isinstance(value, list) and self.basic_vec_type.split()[-1] == "float") or isinstance(value, np.ndarray):
if self.ndim == 1: value = array.array(cpp_to_array_typecodes[self.basic_vec_type], value)
if self.ndim == 2: value = [array.array(cpp_to_array_typecodes[self.basic_vec_type], el) for el in value]
if self.ndim == 3: value = [[array.array(cpp_to_array_typecodes[self.basic_vec_type], el1) for el1 in el] for el in value]
so manual looping with the use of the array.array. Ugly, but…
@LeWhoo … I did say that, and I mean that …
$ cat repro_cppyy.py
# repro.py
import cppyy
import numpy as np
b = np.arange(2*2, dtype=np.uint32).reshape(2,2)
a = cppyy.gbl.std.vector["std::vector<unsigned int>"](b)
print(b, a, sep='\n') # <- added
$ python repro_cppyy.py
[[0 1]
[2 3]]
{ { 0, 1 }, { 2, 3 } }
Runs fine on both Mac and Linux (I’m too lazy to check on Windows atm.).
1 Like
This topic was automatically closed after 10 days. New replies are no longer allowed.