issue_owner_repo listlengths 2 2 | issue_body stringlengths 0 262k ⌀ | issue_title stringlengths 1 1.02k | issue_comments_url stringlengths 53 116 | issue_comments_count int64 0 2.49k | issue_created_at stringdate 1999-03-17 02:06:42 2025-06-23 11:41:49 | issue_updated_at stringdate 2000-02-10 06:43:57 2025-06-23 11:43:00 | issue_html_url stringlengths 34 97 | issue_github_id int64 132 3.17B | issue_number int64 1 215k |
|---|---|---|---|---|---|---|---|---|---|
[
"libjxl",
"libjxl"
] | I was asked to create this issue from the discord. libjxl produces substantially larger files than cjxl in lossless mode due to the usage of 14-bit XYB, which in addition to being larger is also not truly lossless.
Code used below:
```CPP
m_opt = JxlEncoderOptionsCreate(m_enc, nullptr);
if (quality < 0) quality = 0;
JXLEE(JxlEncoderOptionsSetDistance(m_opt, quality));
JXLEE(JxlEncoderOptionsSetLossless(m_opt, quality == 0 ? JXL_TRUE : JXL_FALSE));
JXLEE(JxlEncoderOptionsSetEffort(m_opt, 3));
static constexpr JxlPixelFormat PFMT {
.num_channels = 3,
.data_type = JXL_TYPE_UINT8,
.endianness = JXL_NATIVE_ENDIAN,
.align = 0
};
m_info.xsize = width;
m_info.ysize = height;
m_info.bits_per_sample = 8;
m_info.num_color_channels = 3;
m_info.intensity_target = 255;
m_info.orientation = JXL_ORIENT_IDENTITY;
JXLEE(JxlEncoderSetBasicInfo(m_enc, &m_info));
JxlColorEncoding color_profile;
JxlColorEncodingSetToSRGB(&color_profile, JXL_FALSE);
JXLEE(JxlEncoderSetColorEncoding(m_enc, &color_profile));
JXLEE(JxlEncoderAddImageFrame(m_opt, &PFMT, buf, width * height * 3));
JxlEncoderCloseInput(m_enc);
```
(`quality` is set to 0 in this circumstance) | 14-bit XYB encoding incorrectly used when JxlEncoderOptionsSetLossless is set to JXL_TRUE | https://api.github.com/repos/libjxl/libjxl/issues/257/comments | 5 | 2021-06-30T20:26:25Z | 2022-03-25T16:22:35Z | https://github.com/libjxl/libjxl/issues/257 | 934,099,251 | 257 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
```
1576 - DecodeTest.PixelTestOpaqueSrgbLossyNoise (Failed)
1637 - DecodeTest/DecodeTestParam.PixelTest/301x33RGBtoRGBu8#GetParam()=301x33RGBtoRGBu8 (Failed)
1661 - DecodeTest/DecodeTestParam.PixelTest/301x33RGBtoRGBAu8#GetParam()=301x33RGBtoRGBAu8 (Failed)
1957 - RoundtripTest.Uint8FrameRoundtripTest (Failed)
1960 - RoundtripTest.TestICCProfile (Failed)
```
**To Reproduce**
Steps to reproduce the behavior:
/ci.sh asan
| Tests fail on ASAN build | https://api.github.com/repos/libjxl/libjxl/issues/254/comments | 1 | 2021-06-30T16:42:55Z | 2021-07-01T07:39:13Z | https://github.com/libjxl/libjxl/issues/254 | 933,900,822 | 254 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Encoding `011.png` with modular mode crashes, lossy and lossless.
It also happens when encoding `.ppm` and `.pgx` files created from the `.png` file.
"Works" after transcoding to `.pfm`, but this changes the image and the resulting `.jxl` is larger than the original `.png`.
**To Reproduce**
```
~ $ cjxl -m 011.png 011.jxl
JPEG XL encoder v0.3.7 [AVX2]
build/libjxl-git/src/libjxl/lib/extras/codec_png.cc:497: PNG: no color_space/icc_pathname given, assuming sRGB
Read 822x1168 image, 38.3 MP/s
Encoding [Modular, lossless, squirrel], 8 threads.
build/libjxl-git/src/libjxl/lib/jxl/enc_ans.cc:220: JXL_DASSERT: n <= 255
fish: Job 1, 'cjxl -m 011.png 011.jxl' terminated by signal SIGILL (Illegal instruction)
```
**Expected behavior**
It should be possible to encode this image in modular mode.
**Environment**
- OS: Arch Linux
- Compiler version: clang 12.0.0
- CPU type: ryzen 2700x x86_64
- cjxl/djxl version string: 8193d7b370c36f3df0528f1b22c481c182627aa5
**Additional context**
Compiled with https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=libjxl-git and
```
CFLAGS="-march=native -O3 -pipe -fstack-protector-strong --param=ssp-buffer-size=4 -fno-plt -pthread -Wno-error -w"
CXXFLAGS="$CFLAGS"
```
https://user-images.githubusercontent.com/7374061/123955891-707f7100-d9aa-11eb-9ff4-d8af1667466f.png
| Modular mode crashes with SIGILL; JXL_DASSERT: n <= 255 | https://api.github.com/repos/libjxl/libjxl/issues/251/comments | 3 | 2021-06-30T11:58:30Z | 2022-03-29T18:42:04Z | https://github.com/libjxl/libjxl/issues/251 | 933,629,377 | 251 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
When decoding 003.jxl, which was losslessly transcoded from a jpg, the resulting png file is only 1953 B small and all black, even though djxl didn't throw any errors.
Decoding to the original jpg file is still possible, but only because `--strip` was not used, otherwise it would be impossible to decode this file at all.
**To Reproduce**
(unpack the attached zip file to get 003.jxl)
```
~ $ djxl 003.jxl 003.png
JPEG XL decoder v0.3.7 [AVX2]
Read 274618 compressed bytes.
Decoded to pixels.
752 x 1080, 14.64 MP/s [14.64, 14.64], 1 reps, 8 threads.
Allocations: 476 (max bytes in use: 3.297596E+07)
~ $ djxl 003.jxl 003.jpg
JPEG XL decoder v0.3.7 [AVX2]
Read 274618 compressed bytes.
Reconstructed to JPEG.
752 x 1080, 16.33 MP/s [16.33, 16.33], 6.64 MB/s [6.64, 6.64], 1 reps, 8 threads.
Allocations: 429 (max bytes in use: 3.297534E+07)
~ $ du -b 003.*
330011 003.jpg
274618 003.jxl
1953 003.png
```
**Expected behavior**
It should be possible to decode this file to png.
**Environment**
- OS: Arch Linux
- Compiler version: clang 12.0.0
- CPU type: ryzen 2700x x86_64
- cjxl/djxl version string: 8193d7b370c36f3df0528f1b22c481c182627aa5
**Additional context**
Compiled with https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=libjxl-git and
```
CFLAGS="-march=native -O3 -pipe -fstack-protector-strong --param=ssp-buffer-size=4 -fno-plt -pthread -Wno-error -w"
CXXFLAGS="$CFLAGS"
```
003.jxl inside a zip file, otherwise github wouldn't accept it.
[003.zip](https://github.com/libjxl/libjxl/files/6736434/003.zip)
| Decoding to png results in a small black image. | https://api.github.com/repos/libjxl/libjxl/issues/248/comments | 3 | 2021-06-29T21:12:23Z | 2021-07-05T15:36:08Z | https://github.com/libjxl/libjxl/issues/248 | 933,107,249 | 248 |
[
"libjxl",
"libjxl"
] | Hello, I got an error when I tried to decode jxl file after trunction it to 1KB. I have attached below PNG image that I encoded. Looks like this issue happens with PNG images that have transparency.
Error message:
```
JPEG XL decoder v0.3.7 [AVX2,SSE4,Scalar]
Read 1024 compressed bytes.
Failed to decompress to pixels.
```
My steps:
1. cjxl <input_png> <output_jxl> -p --distance=15
2. <truncate output jxl file by std::filesystem::file_size in C++>
3. djxl <input_jxl> <output_png> --allow_partial_files --allow_more_progressive_steps
Environment:
- OS: Windows 10
- Compiler version: Clang x64, Visual Studio 2019
- CPU type: x86_64
- cjxl/djxl version string: decoder v0.3.7 [AVX2,SSE4,Scalar], encoder v0.3.7 [AVX2,SSE4,Scalar]
Image:
| Fail decoding of truncated JXL file | https://api.github.com/repos/libjxl/libjxl/issues/245/comments | 8 | 2021-06-29T14:37:13Z | 2021-11-19T15:56:54Z | https://github.com/libjxl/libjxl/issues/245 | 932,766,253 | 245 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
`-e 3` produces a smaller file than `-e 9` when losslessly compressing the attached jpg.
**To Reproduce**
```
cjxl -e 3 a.jpg a-3.jxl
cjxl -e 9 a.jpg a-9.jxl
du -b *
JPEG XL encoder v0.3.7 [AVX2]
Read 1200x1000 image, 112.8 MP/s
Encoding [Container | JPEG, lossless transcode, falcon | JPEG reconstruction data], 4 threads.
Compressed to 127967 bytes (0.853 bpp).
1200 x 1000, 30.81 MP/s [30.81, 30.81], 1 reps, 4 threads.
Including container: 128487 bytes (0.857 bpp).
JPEG XL encoder v0.3.7 [AVX2]
Read 1200x1000 image, 104.5 MP/s
Encoding [Container | JPEG, lossless transcode, tortoise | JPEG reconstruction data], 4 threads.
Compressed to 128733 bytes (0.858 bpp).
1200 x 1000, 0.42 MP/s [0.42, 0.42], 1 reps, 4 threads.
Including container: 129253 bytes (0.862 bpp).
128487 a-3.jxl
129253 a-9.jxl
161639 a.jpg
```
**Expected behavior**
`-e 9` should produce the smallest files out of all the presets.
**Environment**
- OS: Arch Linux
- Compiler version: clang 12.0.0
- CPU type: ryzen 3500U x86_64
- cjxl/djxl version string: f8790509f3588413413682b611cf11f18f4498b1
**Additional context**
Compiled with https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=libjxl-git and
```
CFLAGS="-march=native -O3 -pipe -fstack-protector-strong --param=ssp-buffer-size=4 -fno-plt -pthread -Wno-error -w"
CXXFLAGS="$CFLAGS"
```
https://user-images.githubusercontent.com/7374061/123633993-2cf5fd00-d81a-11eb-8be1-150ac3fbdd95.jpg
| Lossless JPEG -e 3 smaller than -e 9 | https://api.github.com/repos/libjxl/libjxl/issues/235/comments | 4 | 2021-06-28T12:20:23Z | 2021-11-18T09:41:58Z | https://github.com/libjxl/libjxl/issues/235 | 931,512,361 | 235 |
[
"libjxl",
"libjxl"
] | Arch Linux
GCC 11.1.0
JPEG XL encoder v0.3.7 [AVX2,SSE4,Scalar]
I tried to load an assortment of JXL images and there seems to be a bug specifically with JXLs that were transcoded from JPEG, it only seems to happen to the GIMP plugin specifically.
The issue does not occur with lossless JXL, or JXLs encoded with a -q < 100, only JPEG -> JXL
I get a GIMP error: "Opening '...' failed: JPEG XL image plug-in could not open image"
and then the following output in the console:
```
/home/.cache/yay/libjxl-git/src/libjxl/lib/jxl/image_metadata.cc:354: JXL_FAILURE: invalid min 0.001234 vs max 0.001001
/home/.cache/yay/libjxl-git/src/libjxl/lib/jxl/fields.cc:697: JXL_RETURN_IF_ERROR code=1: visitor.Visit(fields, PrintVisitors() ? "-- Read\n" : "")
/home/.cache/yay/libjxl-git/src/libjxl/lib/jxl/dec_file.cc:33: JXL_RETURN_IF_ERROR code=1: ReadImageMetadata(reader, &io->metadata.m)
/home/.cache/yay/libjxl-git/src/libjxl/lib/jxl/dec_file.cc:108: JXL_RETURN_IF_ERROR code=1: DecodeHeaders(&reader, io)
/home/.cache/yay/libjxl-git/src/libjxl/plugins/gimp/file-jxl-load.cc:95: JXL_RETURN_IF_ERROR code=1: DecodeFile(dparams, compressed, &io, &pool)
``` | GIMP plugin fails to open JXLs which were transcoded from JPEG | https://api.github.com/repos/libjxl/libjxl/issues/229/comments | 5 | 2021-06-26T00:30:34Z | 2021-08-25T08:55:46Z | https://github.com/libjxl/libjxl/issues/229 | 930,575,447 | 229 |
[
"libjxl",
"libjxl"
] | Hello honored devs,
cjxl keeps the alpha channel even when it is fully opaque. That's not a huge deal regarding file size but it negatively affects RAM consumption and especially decoding performance.
The encoder should automatically drop such unnecessary channels and maybe have a new parameter `--always-keep-alpha` if someone really wants to do that.
Thanks! | Ability to drop fully opaque alpha channel | https://api.github.com/repos/libjxl/libjxl/issues/219/comments | 3 | 2021-06-24T12:36:34Z | 2022-03-29T10:37:48Z | https://github.com/libjxl/libjxl/issues/219 | 929,184,869 | 219 |
[
"libjxl",
"libjxl"
] | **Is your feature request related to a problem? Please describe.**
JPEG XL seems to have several features that allows it to be a PSD replacement, but all the necessary metadata is currently encapsulated in the internal API. The only thing that is possible with the current API is to convert a PSD to JXL (the inverse conversion is not implemented). In any other cases, only the final blended image is accessible.
The feature request aims to make it possible to manipulate layered JXL images in image editor applications. To do that, the API needs to support reading and writing unblended images with their crop coordinates.
**Describe the solution you'd like**
The encoder can already take the input frame by frame. Therefore this could be implemented by allowing additional metadata (crop coordinates, blend mode) when calling `JxlEncoderAddImageFrame`.
The decoder can handle the output frame by frame as well, but the API is currently intended for full-sized frames. I propose to add an decoder option that, when enabled, outputs unblended, cropped frames. The current `JXL_DEC_FRAME` and `JXL_DEC_NEED_IMAGE_OUT_BUFFER` events could be reused to inform the application to resize the buffer when necessary.
**Describe alternatives you've considered**
TBD. The proposal is subject to change.
**Unresolved questions**
- Should we expose internal frames? For PSD-like usecases it's likely not used, but it might be useful for parsing animation files.
- Should we allow the application to have control over reference frames? This could be used control the "preferred final image", where some layers are hidden by default. It could also be used to create images where there's one "base" layer and multiple switchable "patch" layers, and the switchable variants would only depend on the base frame instead of sequentially depending on previous frames. | Expose APIs for crop coordinates/unblended frames | https://api.github.com/repos/libjxl/libjxl/issues/217/comments | 2 | 2021-06-24T06:59:28Z | 2022-03-29T10:33:31Z | https://github.com/libjxl/libjxl/issues/217 | 928,911,482 | 217 |
[
"libjxl",
"libjxl"
] | I packaged `libjxl` for nixpkgs; the `x86_64-linux` build works fine, but the `aarch64-linux` fails because apparently `skcms` triggers a GCC 9.3.0 `internal compiler error`:
https://github.com/NixOS/nixpkgs/pull/103160#issuecomment-866388610
Since this is just a free-time experiment for me and I don't have time to chase this up to skia or GCC, I'm reporting it here in case you are interested in following up on it. | build failure on aarch64-linux due to `skcms` triggering GCC internal compiler error | https://api.github.com/repos/libjxl/libjxl/issues/213/comments | 1 | 2021-06-22T22:57:51Z | 2022-03-29T10:31:42Z | https://github.com/libjxl/libjxl/issues/213 | 927,694,502 | 213 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
as in title when trying to run cjxl or djxl a error shows up "cjxl: error while loading shared libraries: libOpenEXR-3_0.so.27: cannot open shared object file: No such file or directory". seeing as this was a problem with openexr i have tried both the repository version of 3.0.4 and the git version from the aur. but this still happens with either version.
**To Reproduce**
Steps to reproduce the behavior:
1. try to run cjxl or djxl with either openexr3.0.4 or -git with any options
**Expected behavior**
cjxl or djxl should put a list of options
**Screenshots**
If applicable, add screenshots or example input/output images to help explain your problem.
**Environment**
- OS: Manjaro
- Compiler version: clang 12.0.0
- CPU type: x86_64
- cjxl/djxl version string: unable to retrieve as error stops program from running but from install is 3.7.r80-git
**Additional context**
Add any other context about the problem here.
| when using cjxl or djxl error:"cjxl: error while loading shared libraries: libOpenEXR-3_0.so.27: cannot open shared object file: No such file or directory" happens | https://api.github.com/repos/libjxl/libjxl/issues/204/comments | 8 | 2021-06-18T08:42:17Z | 2021-11-22T06:24:00Z | https://github.com/libjxl/libjxl/issues/204 | 924,682,850 | 204 |
[
"libjxl",
"libjxl"
] | **Is your feature request related to a problem? Please describe.**
It would be good to evaluate butteraugli on the CLIC-2021 perceptual quality task. This should provide additional information to he community with respect to its performance characteristics when compared to other potentially usable perceptual quality metrics that JPEG XL could optimize for.
**Describe the solution you'd like**
Please use the test data from the CLIC 2021 perceptual challenge to generate a CSV file with the decisions (see link below for the exact instructions):
https://github.com/fab-jul/clic2021-devkit/blob/main/README.md#perceptual-challenge
Please email them to me to get the final results / ranks. We'll publish these at: http://compression.cc/leaderboard/perceptual/test/
...and of course update this bug tracker.
| Evaluate butteraugli on the CLIC-2021 perceptual quality task | https://api.github.com/repos/libjxl/libjxl/issues/202/comments | 6 | 2021-06-17T20:54:26Z | 2022-03-29T10:30:30Z | https://github.com/libjxl/libjxl/issues/202 | 924,325,924 | 202 |
[
"libjxl",
"libjxl"
] | cjxl.exe image_21447_24bit.png image_21447_24bit.jxl -v -m -q 100 -s 5 -C 1 --num_threads=4
```
JPEG XL encoder v0.3.7 0.3.7-13649d2b [Scalar]
codec_png.cc:496: PNG: no color_space/icc_pathname given, assuming sRGB
Read 1563x1558 image, 20.7 MP/s
Encoding [Modular, lossless, hare], 4 threads.
transform.cc:70: JXL_FAILURE: Invalid channel range
```
https://github.com/libjxl/libjxl/commit/30dae3b5b0cb70134dd08964a6bb1f247f2ac412
| JXL_FAILURE: Invalid channel range | https://api.github.com/repos/libjxl/libjxl/issues/184/comments | 3 | 2021-06-16T10:32:08Z | 2021-06-17T10:17:08Z | https://github.com/libjxl/libjxl/issues/184 | 922,434,216 | 184 |
[
"libjxl",
"libjxl"
] | I followed the build for windows documentation
a lot of errors seems to be in test_util-int.h how can I skip the test build to see if it's work?
or how can I build it for windows? | Build for windows (advanced) doesn't build | https://api.github.com/repos/libjxl/libjxl/issues/180/comments | 15 | 2021-06-16T04:59:16Z | 2022-03-29T10:29:49Z | https://github.com/libjxl/libjxl/issues/180 | 922,095,994 | 180 |
[
"libjxl",
"libjxl"
] | Hello I built cjxl with the command
BUILD_TARGET=x86_64-w64-mingw32 SKIP_TEST=1 CC=clang-7 CXX=clang++-7 ./ci.sh release on docker,
build worked and I got the cjxl.exe but when I run it with a command line nothing is happening and it returns immediately without doing any thing or sending errors.
What dependencies should I install on windows to run the exe?
How can I modify the build command line so that all the dependencies are imbedded in the exe?
Thank you
| docker build for windows, can't run exe | https://api.github.com/repos/libjxl/libjxl/issues/179/comments | 8 | 2021-06-16T04:48:53Z | 2021-06-16T20:53:56Z | https://github.com/libjxl/libjxl/issues/179 | 922,088,884 | 179 |
[
"libjxl",
"libjxl"
] | ```
JPEG XL decoder v0.3.7 [AVX2,SSE4,Scalar]
/C/msys64/home/eustas/clients/libjxl/tools/cpu/cpu.cc:415: JXL_FAILURE: Unable to detect processor topology
Failed to choose default num_threads; you can avoid this error by specifying a --num_threads N argument.
``` | MSYS2 build fails to detect architecture (and number of threads) | https://api.github.com/repos/libjxl/libjxl/issues/168/comments | 3 | 2021-06-14T14:32:51Z | 2021-11-08T20:44:10Z | https://github.com/libjxl/libjxl/issues/168 | 920,476,110 | 168 |
[
"libjxl",
"libjxl"
] | Hello,
djxl crashes on some (not all) JXL files during decoding.
```
$ djxl jpegxl-logo.jxl output.png --num_threads=1
Segmentation fault
```
Test file: [jpegxl-logo.jxl](https://github.com/novomesk/qt-jpegxl-image-plugin/blob/main/testfiles/jpegxl-logo.jxl)
```
(gdb) run
Starting program: C:\msys64\mingw64\bin\djxl.exe jpegxl-logo.jxl output.png "--num_threads=1"
[New Thread 10912.0x3630]
[New Thread 10912.0x32d0]
[New Thread 10912.0x1a7c]
[New Thread 10912.0xc94]
Thread 5 received signal SIGSEGV, Segmentation fault.
[Switching to Thread 10912.0xc94]
0x00007ff64ad95967 in jxl::N_AVX2::ComputePixelChannel<hwy::N_AVX2::Simd<float, 8ull> >(hwy::N_AVX2::Simd<float, 8ull>, float, float const*, float const*, float const*, decltype (Zero((hwy::N_AVX2::Simd<float, 8ull>)()))*, decltype (Zero((hwy::N_AVX2::Simd<float, 8ull>)()))*, decltype (Zero((hwy::N_AVX2::Simd<float, 8ull>)()))*, unsigned long long) (x=8, gap=0xfddbdfb7e0, sm=0xfddbdfb840,
mc=0xfddbdfb8a0, row_bottom=0x2414425fb80, row=0x2414425f980,
row_top=0x2414425f780, dc_factor=0.000206910816, d=...)
at C:/msys64/home/daniel/libjxl/lib/jxl/compressed_dc.cc:96
96 *gap = MaxWorkaround(*gap, Abs((*mc - *sm) / dc_quant));
(gdb) bt
#0 0x00007ff64ad95967 in jxl::N_AVX2::ComputePixelChannel<hwy::N_AVX2::Simd<float, 8ull> >(hwy::N_AVX2::Simd<float, 8ull>, float, float const*, float const*, float const*, decltype (Zero((hwy::N_AVX2::Simd<float, 8ull>)()))*, decltype (Zero((hwy::N_AVX2::Simd<float, 8ull>)()))*, decltype (Zero((hwy::N_AVX2::Simd<float, 8ull>)()))*, unsigned long long) (x=8, gap=0xfddbdfb7e0,
sm=0xfddbdfb840, mc=0xfddbdfb8a0, row_bottom=0x2414425fb80,
row=0x2414425f980, row_top=0x2414425f780, dc_factor=0.000206910816, d=...)
at C:/msys64/home/daniel/libjxl/lib/jxl/compressed_dc.cc:96
#1 jxl::N_AVX2::ComputePixel<hwy::N_AVX2::Simd<float, 8ull> > (x=8,
out_rows=0xfddbdfd310, rows_bottom=0xfddbdfd330, rows=0xfddbdfd350,
rows_top=0xfddbdfd370, dc_factors=0xfddb5fd400)
at C:/msys64/home/daniel/libjxl/lib/jxl/compressed_dc.cc:114
#2 operator() (__closure=0xfddb5fc200, y=1)
at C:/msys64/home/daniel/libjxl/lib/jxl/compressed_dc.cc:188
#3 0x00007ff64adb4090 in jxl::ThreadPool::RunCallState<jxl::Status(long long unsigned int), jxl::N_AVX2::AdaptiveDCSmoothing(float const*, jxl::Image3F*, jxl::ThreadPool*)::<lambda(int, int)> >::CallDataFunc(void *, uint32_t, size_t) (
jpegxl_opaque=0xfddb5fc0b0, value=1, thread_id=0)
at C:/msys64/home/daniel/libjxl/lib/jxl/base/data_parallel.h:88
#4 0x00007ff64ae06bb2 in jpegxl::ThreadParallelRunner::RunRange (
self=0xfddb5ff930, command=4294967439, thread=0)
at C:/msys64/home/daniel/libjxl/lib/threads/thread_parallel_runner_internal.cc:137
#5 0x00007ff64ae06cb8 in jpegxl::ThreadParallelRunner::ThreadFunc (
self=0xfddb5ff930, thread=0)
at C:/msys64/home/daniel/libjxl/lib/threads/thread_parallel_runner_internal.cc:167
#6 0x00007ff64b4977aa in std::__invoke_impl<void, void (*)(jpegxl::ThreadParallelRunner*, int), jpegxl::ThreadParallelRunner*, unsigned int> (
__f=@0x241441e31d8: 0x7ff64ae06bc0 <jpegxl::ThreadParallelRunner::ThreadFunc(jpegxl::ThreadParallelRunner*, int)>)
at C:/msys64/mingw64/include/c++/10.3.0/bits/invoke.h:60
#7 0x00007ff64b4ae2c0 in std::__invoke<void (*)(jpegxl::ThreadParallelRunner*, int), jpegxl::ThreadParallelRunner*, unsigned int> (
__fn=@0x241441e31d8: 0x7ff64ae06bc0 <jpegxl::ThreadParallelRunner::ThreadFunc(jpegxl::ThreadParallelRunner*, int)>)
at C:/msys64/mingw64/include/c++/10.3.0/bits/invoke.h:95
#8 0x00007ff64b45bbc0 in std::thread::_Invoker<std::tuple<void (*)(jpegxl::ThreadParallelRunner*, int), jpegxl::ThreadParallelRunner*, unsigned int> >::_M_invoke<0ull, 1ull, 2ull> (this=0x241441e31c8)
at C:/msys64/mingw64/include/c++/10.3.0/thread:264
#9 0x00007ff64b45bbe7 in std::thread::_Invoker<std::tuple<void (*)(jpegxl::ThreadParallelRunner*, int), jpegxl::ThreadParallelRunner*, unsigned int> >::operator() (this=0x241441e31c8) at C:/msys64/mingw64/include/c++/10.3.0/thread:271
#10 0x00007ff64b45b94c in std::thread::_State_impl<std::thread::_Invoker<std::tuple<void (*)(jpegxl::ThreadParallelRunner*, int), jpegxl::ThreadParallelRunner*, unsigned int> > >::_M_run (this=0x241441e31c0)
at C:/msys64/mingw64/include/c++/10.3.0/thread:215
#11 0x00007ff81eed06e1 in ?? () from C:\msys64\mingw64\bin\libstdc++-6.dll
#12 0x00007ff821d54f33 in ?? () from C:\msys64\mingw64\bin\libwinpthread-1.dll
#13 0x00007ff8340daf5a in msvcrt!_beginthreadex ()
from C:\WINDOWS\System32\msvcrt.dll
#14 0x00007ff8340db02c in msvcrt!_endthreadex ()
from C:\WINDOWS\System32\msvcrt.dll
#15 0x00007ff833827034 in KERNEL32!BaseThreadInitThunk ()
from C:\WINDOWS\System32\kernel32.dll
#16 0x00007ff834362651 in ntdll!RtlUserThreadStart ()
from C:\WINDOWS\SYSTEM32\ntdll.dll
#17 0x0000000000000000 in ?? ()
```
configuration:
`cmake -G "MSYS Makefiles" -DCMAKE_INSTALL_PREFIX=/mingw64 -DCMAKE_BUILD_TYPE=Debug -DJPEGXL_ENABLE_PLUGINS=OFF -DBUILD_TESTING=OFF -DJPEGXL_WARNINGS_AS_ERRORS=OFF -DJPEGXL_ENABLE_SJPEG=OFF -DJPEGXL_ENABLE_BENCHMARK=OFF -DJPEGXL_ENABLE_EXAMPLES=OFF -DJPEGXL_ENABLE_MANPAGES=OFF -DJPEGXL_FORCE_SYSTEM_BROTLI=ON ..`
I observed that when I added `-DHWY_COMPILE_ONLY_SCALAR` to CXX_FLAGS, the djxl worked correctly. | crash on Windows/MSYS2 | https://api.github.com/repos/libjxl/libjxl/issues/165/comments | 4 | 2021-06-14T11:44:01Z | 2021-11-12T13:57:53Z | https://github.com/libjxl/libjxl/issues/165 | 920,329,106 | 165 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
fuzzer_corpus fails on an assert. This failure was introduced by https://github.com/libjxl/libjxl/pull/25
**To Reproduce**
Steps to reproduce the behavior:
```bash
mkdir corpusdir
build/tools/fuzzer_corpus -j 0 corpusdir
```
Result:
```
Generating ImageSpec<size=8x8 * chan=3 depth=8 alpha=0 (premult=0) x frames=1 seed=407987, speed=1, butteraugli=1.5, modular_mode=0, lossy_palette=0, noise=0, preview=0, fuzzer_friendly=0, is_reconstructible_jpeg=1, orientation=4> as a6abe1aab9970bb0771cf78b376749e2
../lib/jxl/enc_frame.cc:1066: JXL_ASSERT: metadata->m.xyb_encoded == (cparams.color_transform == ColorTransform::kXYB)
Illegal instruction
```
**Environment**
ci.sh release build; Linux build. | fuzzer_corpus hits an assert | https://api.github.com/repos/libjxl/libjxl/issues/150/comments | 2 | 2021-06-10T13:32:26Z | 2021-06-10T16:24:45Z | https://github.com/libjxl/libjxl/issues/150 | 917,351,879 | 150 |
[
"libjxl",
"libjxl"
] | Could you please consider adding a new release tag? I would like to package libjxl for Void Linux (draft PR at https://github.com/void-linux/void-packages/pull/31397), which requires a release version. However, the last release (0.3.7) is outdated (old license, does not work with qt-jpegxl-image-plugin). | Request for new release tag | https://api.github.com/repos/libjxl/libjxl/issues/144/comments | 4 | 2021-06-10T09:28:05Z | 2021-08-06T19:52:12Z | https://github.com/libjxl/libjxl/issues/144 | 917,124,689 | 144 |
[
"libjxl",
"libjxl"
] | Sorry to be the bearer of bad news but the libvips fuzzers appear to have something against libjxl today :(
This image causes an integer overflow when decoding using the latest code on the `main` branch.
[clusterfuzz-testcase-minimized-6288583458684928.txt](https://github.com/libjxl/libjxl/files/6626055/clusterfuzz-testcase-minimized-6288583458684928.txt)
```
/src/libjxl/lib/jxl/modular/encoding/encoding.cc:228:24: runtime error: signed integer overflow: 1107427842 + 1073742852 cannot be represented in type 'int'
#0 0x10eedda in jxl::DecodeModularChannelMAANS(jxl::BitReader*, jxl::ANSSymbolReader*, std::__1::vector<unsigned char, std::__1::allocator<unsigned char> > const&, std::__1::vector<jxl::PropertyDecisionNode, std::__1::allocator<jxl::PropertyDecisionNode> > const&, jxl::weighted::Header const&, int, unsigned long, jxl::Image*) libjxl/lib/jxl/modular/encoding/encoding.cc:228:24
#1 0x10f179c in jxl::ModularDecode(jxl::BitReader*, jxl::Image&, jxl::GroupHeader&, unsigned long, jxl::ModularOptions*, std::__1::vector<jxl::PropertyDecisionNode, std::__1::allocator<jxl::PropertyDecisionNode> > const*, jxl::ANSCode const*, std::__1::vector<unsigned char, std::__1::allocator<unsigned char> > const*, bool) libjxl/lib/jxl/modular/encoding/encoding.cc:487:5
#2 0x10f1cc8 in jxl::ModularGenericDecompress(jxl::BitReader*, jxl::Image&, jxl::GroupHeader*, unsigned long, jxl::ModularOptions*, int, std::__1::vector<jxl::PropertyDecisionNode, std::__1::allocator<jxl::PropertyDecisionNode> > const*, jxl::ANSCode const*, std::__1::vector<unsigned char, std::__1::allocator<unsigned char> > const*, bool) libjxl/lib/jxl/modular/encoding/encoding.cc:517:21
#3 0x1456f7b in jxl::ModularFrameDecoder::DecodeGlobalInfo(jxl::BitReader*, jxl::FrameHeader const&, bool) libjxl/lib/jxl/dec_modular.cc:211:23
#4 0x13a4938 in jxl::FrameDecoder::ProcessDCGlobal(jxl::BitReader*) libjxl/lib/jxl/dec_frame.cc:381:46
#5 0x13a1191 in jxl::FrameDecoder::ProcessSections(jxl::FrameDecoder::SectionInfo const*, unsigned long,
```
Perhaps further clamping might be required here:
https://github.com/libjxl/libjxl/blob/87ebbe9d5cd7581afbcce650e6879bccf80e3beb/lib/jxl/modular/encoding/encoding.cc#L225-L228 | Possible integer overflow in DecodeModularChannelMAANS | https://api.github.com/repos/libjxl/libjxl/issues/140/comments | 1 | 2021-06-09T18:36:09Z | 2021-06-10T07:09:47Z | https://github.com/libjxl/libjxl/issues/140 | 916,536,180 | 140 |
[
"libjxl",
"libjxl"
] | Hello, the following file, discovered via fuzz testing, causes a divide by zero when decoded using the latest code on the `main` branch.
[clusterfuzz-testcase-minimized-jpegsave_buffer_fuzzer-5146219264475136.txt](https://github.com/libjxl/libjxl/files/6622107/clusterfuzz-testcase-minimized-jpegsave_buffer_fuzzer-5146219264475136.txt)
```
/src/libjxl/lib/jxl/splines.cc:105:39: runtime error: division by zero
#0 0x14cb3e8 in jxl::N_AVX2::(anonymous namespace)::DrawGaussian(jxl::Image3<float>*, jxl::Rect const&, jxl::Rect const&, jxl::Spline::Point const&, float, float const*, float, std::__1::vector<int, std::__1::allocator<int> >&, std::__1::vector<int, std::__1::allocator<int> >&, std::__1::vector<float, std::__1::allocator<float> >&) libjxl/lib/jxl/splines.cc:105:39
#1 0x14c7cbf in jxl::N_AVX2::(anonymous namespace)::DrawFromPoints(jxl::Image3<float>*, jxl::Rect const&, jxl::Rect const&, jxl::Spline const&, bool, std::__1::vector<std::__1::pair<jxl::Spline::Point, float>, std::__1::allocator<std::__1::pair<jxl::Spline::Point, float> > > const&, float) libjxl/lib/jxl/splines.cc:162:5
#2 0x14c2de5 in jxl::Status jxl::Splines::Apply<true>(jxl::Image3<float>*, jxl::Rect const&, jxl::Rect const&, jxl::ColorCorrelationMap const&) const libjxl/lib/jxl/splines.cc:507:5
#3 0x1479729 in jxl::FinalizeImageRect(jxl::Image3<float>*, jxl::Rect const&, std::__1::vector<std::__1::pair<jxl::Plane<float>*, jxl::Rect>, std::__1::allocator<std::__1::pair<jxl::Plane<float>*, jxl::Rect> > > const&, jxl::PassesDecoderState*, unsigned long, jxl::ImageBundle*, jxl::Rect const&) libjxl/lib/jxl/dec_reconstruct.cc:900:5
#4 0x1481683 in operator() libjxl/lib/jxl/dec_reconstruct.cc:1160:12
```
The point of failure is:
https://github.com/libjxl/libjxl/blob/6946efdf32adfde7cb7d715ae06912da5521dac7/lib/jxl/splines.cc#L105
which appears to be due to `sigma` being calculated as zero here:
https://github.com/libjxl/libjxl/blob/6946efdf32adfde7cb7d715ae06912da5521dac7/lib/jxl/splines.cc#L160-L161
I'm happy to help fix this, but am unsure what the right approach is here. Perhaps we should avoid the `DrawGaussian` call when `sigma` is zero, or maybe this is a `JXL_FAILURE` condition? | Possible divide by zero in DrawGaussian function of splines.cc | https://api.github.com/repos/libjxl/libjxl/issues/129/comments | 11 | 2021-06-09T08:23:49Z | 2021-08-20T18:39:56Z | https://github.com/libjxl/libjxl/issues/129 | 915,932,026 | 129 |
[
"libjxl",
"libjxl"
] | **Is your feature request related to a problem? Please describe.**
The current settings are not optimal
**Describe the solution you'd like**
Add `-I 0 -P 0 --palette=0` by default when `--lossy-palette` is enabled and if they are not specified
| Change the default settings for --lossy-palette | https://api.github.com/repos/libjxl/libjxl/issues/119/comments | 2 | 2021-06-07T22:57:41Z | 2025-04-27T03:11:52Z | https://github.com/libjxl/libjxl/issues/119 | 914,021,898 | 119 |
[
"libjxl",
"libjxl"
] | Dear all,
Can anyone clarify me which class code should I change in order to dump/write to file the predicted image by JPEG-XL in modular mode for lossless compression?
Kind regards,
F | Dump image prediction | https://api.github.com/repos/libjxl/libjxl/issues/116/comments | 7 | 2021-06-07T14:55:19Z | 2021-06-09T12:22:35Z | https://github.com/libjxl/libjxl/issues/116 | 913,642,229 | 116 |
[
"libjxl",
"libjxl"
] | Hello,
Can someone help building cjxl and djxl for mac os? (I managed to build a debian version with the advanced guide for debian)
may be a documentation page to build for different platform like macos, ios, and android could be done at some point?
Thank you
| How to build for mac os | https://api.github.com/repos/libjxl/libjxl/issues/115/comments | 9 | 2021-06-07T05:57:22Z | 2022-03-29T10:27:45Z | https://github.com/libjxl/libjxl/issues/115 | 913,129,113 | 115 |
[
"libjxl",
"libjxl"
] | It would be great if the plugin showed a dialog with encode options before actual saving.
The options could be: Distance, Quality, Effort, Progressive (which corresponds with `cjxl -h`).
A brief explanation of the options could be there as well.
(The plugin can't easily decide itself whether it should use lossy/lossless in the same way as cjxl because the input are pixels, not image formats.) | GIMP plugin: Add encode options dialog | https://api.github.com/repos/libjxl/libjxl/issues/100/comments | 0 | 2021-06-04T06:00:07Z | 2021-08-25T08:54:14Z | https://github.com/libjxl/libjxl/issues/100 | 911,149,090 | 100 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Using container when losslessly transcoding JPEG to JPEG XL makes filesize reported by cjxl smaller than the one reported by the filesystem. It also doesn't change in cjxl whether you use a container or not, contrary to the reality reported by the filesystem.
**To Reproduce**
(on Linux and PowerShell)
1. `cjxl file.jpg file.jxl --strip; ls -l file.jxl`
Without container file sizes reported both by `cjxl` and `ls -l` are the same...
2. `cjxl file.jpg file.jxl; ls -l file.jxl`
...but they no longer match if you decide to use a container!
Note how the `cjxl`-reported file size is the same as in 1.
**Expected behavior**
Filesize should be reported properly by cjxl.
**Environment**
- cjxl/djxl version string: dfc730a8dd7d94f0ce9ec32573e7c9c7e178fb15 | File size reported by cjxl doesn't take the container cost into account | https://api.github.com/repos/libjxl/libjxl/issues/99/comments | 0 | 2021-06-04T05:28:20Z | 2021-06-14T15:38:21Z | https://github.com/libjxl/libjxl/issues/99 | 911,132,343 | 99 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
When `-Wp,-D_GLIBCXX_ASSERTIONS` is added to CXXFLAGS, many tests fail when their processes unexpectedly abort. In my tests exactly 79 tests fail, a list of these compiled by another user is here: https://aur.archlinux.org/pkgbase/libjxl/#comment-811172
**To Reproduce**
With the latest release, add `-Wp,-D_GLIBCXX_ASSERTIONS` to CXXFLAGS. Try to build and run tests with cmake, e.g.
cmake --build . -- -j8
cmake --build . -- test
**Expected behavior**
Tests complete successfully.
**Environment**
- OS: Arch Linux
- Compiler version: clang 11.0.1
- CPU type: x86_64
- cjxl/djxl version string: [v0.3.7 | SIMD supported: SSE4,Scalar]
**Additional context**
Arch Linux recently added this option to its default CXXFLAGS for building packages. Users trying to build libjxl as a package on Arch Linux are likely to run into this problem.
Here is sample output from running one of the tests directly:
"/home/adam/Downloads/jxl/libjxl/src/build/lib/tests/butteraugli_test" "--gtest_filter=ButteraugliTest.Distmap" "--gtest_also_run_disabled_tests"
Running main() from /build/gtest/src/googletest-release-1.10.0/googletest/src/gtest_main.cc
Note: Google Test filter = ButteraugliTest.Distmap
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from ButteraugliTest
[ RUN ] ButteraugliTest.Distmap
/usr/bin/../lib64/gcc/x86_64-pc-linux-gnu/11.1.0/../../../../include/c++/11.1.0/bits/atomic_base.h:268: void std::atomic_flag::clear(std::memory_order): Assertion '__b != memory_order_acq_rel' failed.
zsh: abort (core dumped) "/home/adam/Downloads/jxl/libjxl/src/build/lib/tests/butteraugli_test"
Possibly related bug: https://github.com/libjxl/libjxl/issues/64 (this user also had this compiler option enabled and failed to build libjxl) | Many tests fail when built with -Wp,-D_GLIBCXX_ASSERTIONS | https://api.github.com/repos/libjxl/libjxl/issues/98/comments | 2 | 2021-06-03T19:57:03Z | 2021-06-04T03:27:50Z | https://github.com/libjxl/libjxl/issues/98 | 910,809,777 | 98 |
[
"libjxl",
"libjxl"
] | It would be good to add a document that keeps track of software that has jxl support. Such a list serves several purposes:
- thank/acknowledge other projects for integrating jxl support
- point end-users to software that can read/write jxl
- keep track of the adoption status of jxl
- in case of a (security) bug, it's easier to see who might be affected and check if they are updated (in case they use static linking)
Here is a first attempt at making such a list. Please add missing software in the comments! When we have a more or less complete list, I'll make a pull request to add the list as a markdown document.
## Browsers
- Chromium: behind a flag since version 91, tracking bug: https://bugs.chromium.org/p/chromium/issues/detail?id=1178058
- Firefox: behind a flag since version 90, tracking bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1539075
- Safari: not supported, tracking bug: https://bugs.webkit.org/show_bug.cgi?id=208235
- Edge: behind a flag since version 91, start with `.\msedge.exe --enable-features=JXL`
## Image libraries
- ImageMagick: supported since 7.0.10-54 (https://imagemagick.org/)
- libvips: supported since 8.11 (https://libvips.github.io/libvips/)
- Imlib2: https://github.com/alistair7/imlib2-jxl
## OS-level support / UI frameworks / file browser plugins
- Qt / KDE: plugin available: https://github.com/novomesk/qt-jpegxl-image-plugin
- GDK-pixbuf: plugin available in libjxl repo
- gThumb: https://ubuntuhandbook.org/index.php/2021/04/gthumb-3-11-3-adds-jpeg-xl-support/
- MacOS viewer/QuickLook plugin: https://github.com/yllan/JXLook
- Windows Imaging Component: https://github.com/mirillis/jpegxl-wic
- Windows thumbnail handler: https://github.com/saschanaz/jxl-winthumb
- OpenMandriva Lx (since 4.3 RC)
## Image editors
- GIMP: plugin available in libjxl repo, no official support, tracking bug: https://gitlab.gnome.org/GNOME/gimp/-/issues/4681
- Photoshop: no plugin available yet, no official support yet
## Image viewers
- XnView: https://www.xnview.com/en/
- ImageGlass: https://imageglass.org/
- Any viewer based on Qt, KDE, GDK-pixbuf, ImageMagick, libvips or imlib2 (see above)
- Qt viewers: gwenview, digiKam, KolourPaint, KPhotoAlbum, LXImage-Qt, qimgv, qView, nomacs, VookiImageViewer
## Online tools
- Squoosh: https://squoosh.app/
- Cloudinary: https://cloudinary.com/blog/cloudinary_supports_jpeg_xl
- MConverter: https://mconverter.eu/ | Add list of applications/projects using libjxl | https://api.github.com/repos/libjxl/libjxl/issues/96/comments | 6 | 2021-06-03T13:58:55Z | 2021-11-21T14:28:55Z | https://github.com/libjxl/libjxl/issues/96 | 910,522,993 | 96 |
[
"libjxl",
"libjxl"
] | **Is your feature request related to a problem? Please describe.**
Attempting to losslessly transcode a jpg to a progressively encoded jxl fails.
When you add `-p` to a lossless transcode, you get the following message:
`Error: progressive lossless JPEG transcode is not yet implemented.`
**Describe the solution you'd like**
A progressively transcoded jxl should be produced.
| Progressive Transcoding | https://api.github.com/repos/libjxl/libjxl/issues/92/comments | 1 | 2021-06-03T00:17:31Z | 2022-03-29T10:26:41Z | https://github.com/libjxl/libjxl/issues/92 | 909,978,904 | 92 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Attempting to subsample colors causes encoding to fail.
**To Reproduce**
Add `--resampling=2` (or any other valid value other than 1).
An error occurs: `Failed to compress to VarDCT.`
**Expected behavior**
Produce an image.
**Environment**
- OS: Windows 10
- Compiler version: various
- CPU type: x86_64
- cjxl/djxl version string: [v0.3.7 | SIMD supported: AVX, Scalar]
| Subsampling Fails | https://api.github.com/repos/libjxl/libjxl/issues/91/comments | 16 | 2021-06-03T00:07:59Z | 2021-07-09T14:53:08Z | https://github.com/libjxl/libjxl/issues/91 | 909,974,861 | 91 |
[
"libjxl",
"libjxl"
] | In the case of premultiplied (associated) alpha, `djxl` and other applications ignore that image header field, and treat everything as if it is the usual non-premultiplied (unassociated) alpha like in PNG.
It would make sense to do the same thing here as with the Orientation: by default, the decoder should return pixels in a default way (orientation corrected, unassociated alpha), so applications that want to interpret the `alpha_associated` info themselves and get the raw data would have to explicitly ask the decoder to not do the normalization.
| Premultiplied alpha not rendered correctly | https://api.github.com/repos/libjxl/libjxl/issues/81/comments | 0 | 2021-06-02T07:37:58Z | 2021-06-03T12:59:33Z | https://github.com/libjxl/libjxl/issues/81 | 909,212,792 | 81 |
[
"libjxl",
"libjxl"
] | The bitstream can do premultiplied alpha, but currently cjxl has no way to select whether to do that or not.
Current behavior is that it does whatever the input format does, e.g. non-premultiplied alpha on PNG input and premultiplied alpha on EXR input.
It would be nice to have a flag to control this.
Also, in the non-premultiplied case, it would be good to have an option in the lossless mode to say "I don't care about the invisible pixels", so they can be made black or whatever else is good for compression (in lossy mode this is already done, but in lossless mode there is currently no way to do it, i.e. we now do what `cwebp -exact` does).
| Add option to cjxl do clear invisible pixels and/or do premultiplied alpha | https://api.github.com/repos/libjxl/libjxl/issues/76/comments | 2 | 2021-06-01T15:29:08Z | 2021-07-26T08:57:40Z | https://github.com/libjxl/libjxl/issues/76 | 908,416,264 | 76 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
cjxl allows command-line parameters for lossy Modular and lossy palette in the same command despite them being incompatible with each other by JPEG XL's design, thus resulting in buggy images.
**To Reproduce**
`cjxl smart_ptr.jpg smart_ptr.jxl -j -m -Q 50 --lossy-palette`
**Expected behavior**
The encoder either ignores one of the parameters or exits with a failure message.
**Screenshots**
**Environment**
- cjxl/djxl version: af6ece2e3b6bdece69cb4ce8f8cb7d630c5d72cb | Disable lossy palette when using lossy modular | https://api.github.com/repos/libjxl/libjxl/issues/75/comments | 0 | 2021-06-01T13:28:45Z | 2021-06-02T18:41:47Z | https://github.com/libjxl/libjxl/issues/75 | 908,294,850 | 75 |
[
"libjxl",
"libjxl"
] | From reading JxlDecoderSetPreviewOutBuffer() in decode.h it's not clear which color space the resulting buffer will have.
My first guess was that's the same as encode srgb for JXL_TYPE_UINT8 and linear-srgb for JXL_TYPE_FLOAT. But it seems to be JxlDecoderGetColorAsEncodedProfile()
=> Could please add some pointers for the user?
Moved this from gitlab https://gitlab.com/wg1/jpeg-xl/-/issues/243 as suggested.
If at all possible it would be nice to have a simple output-color-space option reducing all these possibilities of JxlDecoderGetColorAsEncodedProfile() return value to something easy-to-use. | Decode API: clarify color space of returned pixels | https://api.github.com/repos/libjxl/libjxl/issues/70/comments | 3 | 2021-06-01T07:21:46Z | 2021-11-20T11:34:35Z | https://github.com/libjxl/libjxl/issues/70 | 907,984,205 | 70 |
[
"libjxl",
"libjxl"
] | oss-fuzz might have found a nasty write-after-free bug in git master libjxl. This file:
http://www.rollthepotato.net/~john/.clusterfuzz-testcase-minimized-jpegsave_file_fuzzer-4933665846067200
Generate this asan error:
```
| ==291946==ERROR: AddressSanitizer: heap-use-after-free on address 0x622000015380 at pc 0x000000525dac bp 0x7fb1ddff69b0 sp 0x7fb1ddff6178
| WRITE of size 32 at 0x622000015380 thread T2
| SCARINESS: 55 (multi-byte-write-heap-use-after-free)
| #0 0x525dab in __asan_memset /src/llvm-project/compiler-rt/lib/asan/asan_interceptors_memintrinsics.cpp:26:3
| #1 0x2db1720 in Upsample<2, 2> libjxl/lib/jxl/dec_upsample.cc:106:3
| #2 0x2db1720 in jxl::N_AVX2::UpsampleRect(unsigned long, float const*, jxl::Plane<float> const&, jxl::Rect const&, jxl::Plane<float>*, jxl::Rect const&, long, unsigned long, float*, unsigned long) libjxl/lib/jxl/dec_upsample.cc:245:7
| #3 0x2ddfce0 in UpsampleRect libjxl/lib/jxl/dec_upsample.cc:336:3
| #4 0x2ddfce0 in jxl::Upsampler::UpsampleRect(jxl::Image3<float> const&, jxl::Rect const&, jxl::Image3<float>*, jxl::Rect const&, long, unsigned long, float*) const libjxl/lib/jxl/dec_upsample.cc:347:5
| #5 0x2d8ed4e in jxl::FinalizeImageRect(jxl::Image3<float>*, jxl::Rect const&, std::__1::vector<std::__1::pair<jxl::Plane<float>*, jxl::Rect>, std::__1::allocator<std::__1::pair<jxl::Plane<float>*, jxl::Rect> > > const&, jxl::PassesDecoderState*, unsigned long, jxl::ImageBundle*, jxl::Rect const&) libjxl/lib/jxl/dec_reconstruct.cc:822:24
| #6 0x2d9ad27 in operator() libjxl/lib/jxl/dec_reconstruct.cc:1048:12
| #7 0x2d9ad27 in jxl::ThreadPool::RunCallState<jxl::FinalizeFrameDecoding(jxl::ImageBundle*, jxl::PassesDecoderState*, jxl::ThreadPool*, bool, bool)::$_9, jxl::FinalizeFrameDecoding(jxl::ImageBundle*, jxl::PassesDecoderState*, jxl::ThreadPool*, bool, bool)::$_10>::CallDataFunc(void*, unsigned int, unsigned long) libjxl/lib/jxl/base/data_parallel.h:88:14
| #8 0x3177779 in RunRange libjxl/lib/threads/thread_parallel_runner_internal.cc:137:7
...
```
<details>
<summary>Full report</summary>
</details> | oss-fuzz reports a possible write-after-free in libjxl | https://api.github.com/repos/libjxl/libjxl/issues/66/comments | 3 | 2021-05-31T18:00:40Z | 2021-06-04T21:40:12Z | https://github.com/libjxl/libjxl/issues/66 | 907,641,254 | 66 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
The build fails on aarch64:
```
In file included from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:36:
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h: In lambda function:
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h:166:53: note: use '-flax-vector-conversions' to permit conversions between vectors with differing element types or numbers of subparts
166 | Vec128<uint8_t, 16>(vreinterpretq_s16_u8(exp16)))
| ~~~~~~~~~~~~~~~~~~~~^~~~~~~
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h:166:54: error: cannot convert 'int16x8_t' to 'uint8x16_t'
166 | Vec128<uint8_t, 16>(vreinterpretq_s16_u8(exp16)))
| ^~~~~
| |
| int16x8_t
In file included from /usr/include/hwy/ops/arm_neon-inl.h:18,
from /usr/include/hwy/highway.h:282,
from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:26:
/usr/lib/gcc/aarch64-redhat-linux/11/include/arm_neon.h:5281:34: note: initializing argument 1 of 'int16x8_t vreinterpretq_s16_u8(uint8x16_t)'
5281 | vreinterpretq_s16_u8 (uint8x16_t __a)
| ~~~~~~~~~~~^~~
In file included from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:36:
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h:171:54: error: cannot convert 'int16x8_t' to 'uint8x16_t'
171 | Vec128<uint8_t, 16>(vreinterpretq_s16_u8(exp16)))
| ^~~~~
| |
| int16x8_t
In file included from /usr/include/hwy/ops/arm_neon-inl.h:18,
from /usr/include/hwy/highway.h:282,
from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:26:
/usr/lib/gcc/aarch64-redhat-linux/11/include/arm_neon.h:5281:34: note: initializing argument 1 of 'int16x8_t vreinterpretq_s16_u8(uint8x16_t)'
5281 | vreinterpretq_s16_u8 (uint8x16_t __a)
| ~~~~~~~~~~~^~~
In file included from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:36:
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h:174:30: error: cannot convert 'uint8x16_t' to 'int16x8_t'
174 | vreinterpretq_u8_s16(pow_low), vreinterpretq_u8_s16(pow_high), 8));
| ^~~~~~~
| |
| uint8x16_t
In file included from /usr/include/hwy/ops/arm_neon-inl.h:18,
from /usr/include/hwy/highway.h:282,
from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:26:
/usr/lib/gcc/aarch64-redhat-linux/11/include/arm_neon.h:5631:33: note: initializing argument 1 of 'uint8x16_t vreinterpretq_u8_s16(int16x8_t)'
5631 | vreinterpretq_u8_s16 (int16x8_t __a)
| ~~~~~~~~~~^~~
In file included from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:36:
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h:174:61: error: cannot convert 'uint8x16_t' to 'int16x8_t'
174 | vreinterpretq_u8_s16(pow_low), vreinterpretq_u8_s16(pow_high), 8));
| ^~~~~~~~
| |
| uint8x16_t
In file included from /usr/include/hwy/ops/arm_neon-inl.h:18,
from /usr/include/hwy/highway.h:282,
from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:26:
/usr/lib/gcc/aarch64-redhat-linux/11/include/arm_neon.h:5631:33: note: initializing argument 1 of 'uint8x16_t vreinterpretq_u8_s16(int16x8_t)'
5631 | vreinterpretq_u8_s16 (int16x8_t __a)
| ~~~~~~~~~~^~~
In file included from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:36:
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h: In function 'void jxl::N_NEON::{anonymous}::FastXYBTosRGB8(const Image3F&, const jxl::Rect&, const jxl::Rect&, uint8_t*, size_t)':
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h:296:29: error: cannot convert '__Int16x8_t' to 'uint16x8_t' in initialization
296 | uint16x8_t r = srgb_tf(linear_r16);
| ~~~~~~~^~~~~~~~~~~~
| |
| __Int16x8_t
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h:297:29: error: cannot convert '__Int16x8_t' to 'uint16x8_t' in initialization
297 | uint16x8_t g = srgb_tf(linear_g16);
| ~~~~~~~^~~~~~~~~~~~
| |
| __Int16x8_t
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h:298:29: error: cannot convert '__Int16x8_t' to 'uint16x8_t' in initialization
298 | uint16x8_t b = srgb_tf(linear_b16);
| ~~~~~~~^~~~~~~~~~~~
| |
| __Int16x8_t
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h:301:61: error: cannot convert 'uint16x8_t' to 'int16x8_t'
301 | vqmovun_s16(vrshrq_n_s16(vsubq_s16(r, vshrq_n_s16(r, 8)), 6));
| ^
| |
| uint16x8_t
In file included from /usr/include/hwy/ops/arm_neon-inl.h:18,
from /usr/include/hwy/highway.h:282,
from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:26:
/usr/lib/gcc/aarch64-redhat-linux/11/include/arm_neon.h:25724:24: note: initializing argument 1 of 'int16x8_t vshrq_n_s16(int16x8_t, int)'
25724 | vshrq_n_s16 (int16x8_t __a, const int __b)
| ~~~~~~~~~~^~~
In file included from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:36:
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h:303:61: error: cannot convert 'uint16x8_t' to 'int16x8_t'
303 | vqmovun_s16(vrshrq_n_s16(vsubq_s16(g, vshrq_n_s16(g, 8)), 6));
| ^
| |
| uint16x8_t
In file included from /usr/include/hwy/ops/arm_neon-inl.h:18,
from /usr/include/hwy/highway.h:282,
from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:26:
/usr/lib/gcc/aarch64-redhat-linux/11/include/arm_neon.h:25724:24: note: initializing argument 1 of 'int16x8_t vshrq_n_s16(int16x8_t, int)'
25724 | vshrq_n_s16 (int16x8_t __a, const int __b)
| ~~~~~~~~~~^~~
In file included from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:36:
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_xyb-inl.h:305:61: error: cannot convert 'uint16x8_t' to 'int16x8_t'
305 | vqmovun_s16(vrshrq_n_s16(vsubq_s16(b, vshrq_n_s16(b, 8)), 6));
| ^
| |
| uint16x8_t
In file included from /usr/include/hwy/ops/arm_neon-inl.h:18,
from /usr/include/hwy/highway.h:282,
from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/lib/jxl/dec_reconstruct.cc:26:
/usr/lib/gcc/aarch64-redhat-linux/11/include/arm_neon.h:25724:24: note: initializing argument 1 of 'int16x8_t vshrq_n_s16(int16x8_t, int)'
25724 | vshrq_n_s16 (int16x8_t __a, const int __b)
| ~~~~~~~~~~^~~
```
**To Reproduce**
```
%cmake -DENABLE_CCACHE=1 \
-DBUILD_TESTING=OFF \
-DINSTALL_GTEST:BOOL=OFF \
-DJPEGXL_ENABLE_BENCHMARK:BOOL=OFF \
-DJPEGXL_ENABLE_PLUGINS:BOOL=ON \
-DJPEGXL_FORCE_SYSTEM_BROTLI:BOOL=ON \
-DJPEGXL_FORCE_SYSTEM_GTEST:BOOL=ON \
-DJPEGXL_FORCE_SYSTEM_HWY:BOOL=ON \
-DJPEGXL_WARNINGS_AS_ERRORS:BOOL=OFF \
-DBUILD_SHARED_LIBS:BOOL=OFF
%cmake_build -- all doc
```
Flags:
```
CFLAGS='-O2 -flto=auto -ffat-lto-objects -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -fstack-protector-strong -specs=/usr/lib/rpm/redhat/redhat-annobin-cc1 -mbranch-protection=standard -fasynchronous-unwind-tables -fstack-clash-protection'
LDFLAGS='-Wl,-z,relro -Wl,--as-needed -Wl,-z,now -specs=/usr/lib/rpm/redhat/redhat-hardened-ld '
```
**Environment**
- OS: Fedora Rawhide
- Compiler version: GCC 11.1.1
- CPU type:
```
Architecture: aarch64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 5
On-line CPU(s) list: 0-4
Thread(s) per core: 1
Core(s) per socket: 5
Socket(s): 1
NUMA node(s): 1
Vendor ID: APM
Model: 2
Model name: X-Gene
Stepping: 0x3
BogoMIPS: 80.00
NUMA node0 CPU(s): 0-4
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Mitigation; PTI
Vulnerability Spec store bypass: Vulnerable
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Vulnerable
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 cpuid
```
- libhwy: version 0.12.1
- cjxl/djxl version string: v0.3.7
Full log available at https://koji.fedoraproject.org/koji/taskinfo?taskID=69038231 | Build failure on aarch64 with GCC | https://api.github.com/repos/libjxl/libjxl/issues/64/comments | 5 | 2021-05-31T16:58:34Z | 2021-06-01T15:54:46Z | https://github.com/libjxl/libjxl/issues/64 | 907,613,431 | 64 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
The build fails on armv7hl:
```
In file included from /builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/third_party/skcms/skcms.cc:2071:
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/third_party/skcms/src/Transform_inl.h: In function 'baseline::F baseline::F_from_Half(baseline::U16)':
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/third_party/skcms/src/Transform_inl.h:158:26: error: 'float16x4_t' was not declared in this scope; did you mean 'bfloat16x4_t'?
158 | return vcvt_f32_f16((float16x4_t)half);
| ^~~~~~~~~~~
| bfloat16x4_t
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/third_party/skcms/src/Transform_inl.h:158:12: error: 'vcvt_f32_f16' was not declared in this scope; did you mean 'vcvt_f32_bf16'?
158 | return vcvt_f32_f16((float16x4_t)half);
| ^~~~~~~~~~~~
| vcvt_f32_bf16
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/third_party/skcms/src/Transform_inl.h: In function 'baseline::U16 baseline::Half_from_F(baseline::F)':
/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/third_party/skcms/src/Transform_inl.h:187:17: error: 'vcvt_f16_f32' was not declared in this scope; did you mean 'vcvt_bf16_f32'?
187 | return (U16)vcvt_f16_f32(f);
| ^~~~~~~~~~~~
| vcvt_bf16_f32
gmake[2]: *** [third_party/CMakeFiles/skcms.dir/build.make:79: third_party/CMakeFiles/skcms.dir/skcms/skcms.cc.o] Error 1
gmake[2]: Leaving directory '/builddir/build/BUILD/jpeg-xl-v0.3.7-9e9bce86164dc4d01c39eeeb3404d6aed85137b2/armv7hl-redhat-linux-gnueabi'
gmake[1]: *** [CMakeFiles/Makefile2:345: third_party/CMakeFiles/skcms.dir/all] Error 2
```
**To Reproduce**
```
%cmake -DENABLE_CCACHE=1 \
-DBUILD_TESTING=OFF \
-DINSTALL_GTEST:BOOL=OFF \
-DJPEGXL_ENABLE_BENCHMARK:BOOL=OFF \
-DJPEGXL_ENABLE_PLUGINS:BOOL=ON \
-DJPEGXL_FORCE_SYSTEM_BROTLI:BOOL=ON \
-DJPEGXL_FORCE_SYSTEM_GTEST:BOOL=ON \
-DJPEGXL_FORCE_SYSTEM_HWY:BOOL=ON \
-DJPEGXL_WARNINGS_AS_ERRORS:BOOL=OFF \
-DBUILD_SHARED_LIBS:BOOL=OFF
%cmake_build -- all doc
```
Flags:
```
CFLAGS='-O2 -flto=auto -ffat-lto-objects -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -fstack-protector-strong -specs=/usr/lib/rpm/redhat/redhat-annobin-cc1 -march=armv7-a -mfpu=vfpv3-d16 -mtune=generic-armv7-a -mabi=aapcs-linux -mfloat-abi=hard'
LDFLAGS='-Wl,-z,relro -Wl,--as-needed -Wl,-z,now -specs=/usr/lib/rpm/redhat/redhat-hardened-ld '
```
**Environment**
- OS: Fedora Rawhide
- Compiler version: GCC 11.1.1
- CPU type:
```
CPU info:
Architecture: armv7l
Byte Order: Little Endian
CPU(s): 5
On-line CPU(s) list: 0-4
Thread(s) per core: 1
Core(s) per socket: 5
Socket(s): 1
Vendor ID: APM
Model: 2
Model name: X-Gene
Stepping: 0x3
BogoMIPS: 80.00
Flags: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae evtstrm aes pmull sha1 sha2 crc32
```
- libhwy: version 0.12.1
- cjxl/djxl version string: v0.3.7
Full log available at https://koji.fedoraproject.org/koji/taskinfo?taskID=69038231 | Build failure on armv7hl with GCC | https://api.github.com/repos/libjxl/libjxl/issues/63/comments | 12 | 2021-05-31T16:55:52Z | 2021-11-21T22:57:43Z | https://github.com/libjxl/libjxl/issues/63 | 907,612,159 | 63 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
We have a mutex in the LCMS color_management which can probably be removed after an update in LCMS. See https://github.com/libjxl/libjxl/pull/23#discussion_r641966901 for context. | Remove LCMS mutex | https://api.github.com/repos/libjxl/libjxl/issues/53/comments | 2 | 2021-05-31T12:19:50Z | 2022-03-11T20:11:25Z | https://github.com/libjxl/libjxl/issues/53 | 907,419,968 | 53 |
[
"libjxl",
"libjxl"
] | **Summary**
For GIF and JPEG, libjxl will use lossless mode (GIF) / lossless JPEG transcode (JPEG) by default.
However, if you explicitly ask libjxl to use lossless mode by `-q 100` or `-d 0`, the results would be different.
**Steps to reproduce**
* `cjxl -s 3 $infile $outfile`
* `cjxl -s 3 -d 0 $infile $outfile`
* `cjxl -s 3 -q 100 $infile $outfile`
**Observed behavior**
For JPEG transcode, no parameter or `-q 100` produces 280305, but `-d 0` produces 280246.
For GIF, no parameter produces 47431412, while `-d 0` or `-q 100` produces output 26518874.
**Expected behavior**
The output files should be the same.
**Test files**
* https://www.ganganonline.com/contents/slime/img/slime_7cover.gif
* https://jpegxl.info/fallbacklogo.jpg
**Environment**
- OS: Linux
- Compiler version: clang
- CPU type: x86_64
- cjxl/djxl version string: `cjxl [v0.3.7 | SIMD supported: SSE4,Scalar]`
**Additional context**
For the test JPEG file, the difference is small, but for the GIF file, the difference is very large.
| GIF / JPEG -> Lossless JXL: Different results with and without `-q 100` / `-d 0` | https://api.github.com/repos/libjxl/libjxl/issues/44/comments | 0 | 2021-05-29T08:37:37Z | 2021-05-31T12:42:03Z | https://github.com/libjxl/libjxl/issues/44 | 906,419,615 | 44 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
On [GitLab releases page](https://gitlab.com/wg1/jpeg-xl/-/releases), there is changelog (and notes) for every release, but [GitHub releases page](https://github.com/libjxl/libjxl/releases) does not.
**To Reproduce**
Go to GitHub releases page, and click "..." to expand the details of 0.3.7. It only contains:
> Update JPEG-XL with latest changes.
> This includes all changes up to 2021-03-29 10:37:25 +0000.
**Expected behavior**
For 0.3.7:
> * Bump JPEG XL version to 0.3.7.
> * Fix a rounding issue in 8-bit decoding.
>
> Note: This release is for evaluation purposes and may contain bugs, including security bugs, that will not be individually documented when fixed. Always prefer to use the latest release. Please provide feedback and report bugs here. | Releases page has no changelog | https://api.github.com/repos/libjxl/libjxl/issues/43/comments | 1 | 2021-05-29T02:26:52Z | 2021-05-31T14:06:22Z | https://github.com/libjxl/libjxl/issues/43 | 906,301,512 | 43 |
[
"libjxl",
"libjxl"
] | **Describe the solution you'd like**
I hope `cjxl` can offer presets for different types of source material, so that average users can have optimized outputs (in size and in image quality), without the need to know / try the advanced parameters.
It would be best if the presets be applied on both lossy and mathematically lossless mode.
**Additional context**
`cwebp` has `default, photo, picture, drawing, icon, text` presets.
| Presets for different types of source material | https://api.github.com/repos/libjxl/libjxl/issues/42/comments | 0 | 2021-05-29T02:14:40Z | 2022-03-29T10:28:15Z | https://github.com/libjxl/libjxl/issues/42 | 906,297,162 | 42 |
[
"libjxl",
"libjxl"
] | Right now, this repo here has no tags, but the one on gitlab does: https://gitlab.com/wg1/jpeg-xl/-/tags | Missing git tags from gitlab | https://api.github.com/repos/libjxl/libjxl/issues/40/comments | 1 | 2021-05-27T21:17:25Z | 2021-05-27T23:03:43Z | https://github.com/libjxl/libjxl/issues/40 | 904,199,293 | 40 |
[
"libjxl",
"libjxl"
] | I think oss-fuzz has found an assert failure in libjxl. This file:
http://www.rollthepotato.net/~john/clusterfuzz-testcase-minimized-pngsave_buffer_fuzzer-6695474309496832.fuzz
Triggers this:
```
| /src/libjxl/lib/jxl/image_ops.h:25: JXL_ASSERT: SameSize(from, *to)
| AddressSanitizer:DEADLYSIGNAL
| =================================================================
| ==484==ERROR: AddressSanitizer: ILL on unknown address 0x00000250cd19 (pc 0x00000250cd19 bp 0x7f7603b124d0 sp 0x7f7603b124d0 T4)
| #0 0x250cd19 in jxl::Abort() libjxl/lib/jxl/base/status.cc:42:3
| #1 0x28dd2ff in CopyImageTo<float> libjxl/lib/jxl/image_ops.h:25:3
| #2 0x28dd2ff in jxl::ImageBlender::PrepareBlending(jxl::PassesDecoderState*, jxl::FrameOrigin, unsigned long, unsigned long, jxl::ColorEncoding const&, jxl::ImageBundle*) libjxl/lib/jxl/blending.cc:130:9
| #3 0x2d5931d in jxl::FinalizeFrameDecoding(jxl::ImageBundle*, jxl::PassesDecoderState*, jxl::ThreadPool*, bool, bool) libjxl/lib/jxl/dec_reconstruct.cc:1081:5
| #4 0x2bc9b4a in jxl::FrameDecoder::Flush() libjxl/lib/jxl/dec_frame.cc:817:3
| #5 0x2bbe2b2 in jxl::FrameDecoder::FinalizeFrame() libjxl/lib/jxl/dec_frame.cc:849:3
| #6 0x2537b44 in jxl::(anonymous namespace)::JxlDecoderProcessInternal(JxlDecoderStruct*, unsigned char const*, unsigned long) libjxl/lib/jxl/decode.cc:1155:30
| #7 0x2532671 in JxlDecoderProcessInput libjxl/lib/jxl/decode.cc:1668:14
...
```
With this version of libjxl: https://gitlab.com/wg1/jpeg-xl/-/compare/040eae8105b61b312a67791213091103f4c0d034...30ea86ab4c1f1b98c21967a2e3d72a51fe77e454 | oss-fuzz reports an assert failure in libjxl | https://api.github.com/repos/libjxl/libjxl/issues/37/comments | 0 | 2021-05-27T15:30:40Z | 2021-06-10T17:04:35Z | https://github.com/libjxl/libjxl/issues/37 | 903,914,222 | 37 |
[
"libjxl",
"libjxl"
] | Hi, this image:
www.rollthepotato.net/~john/clusterfuzz-testcase-minimized-pngsave_buffer_fuzzer-5360982477111296
Produces this error in oss-fuzz:
```
| /src/jpeg-xl/lib/jxl/dec_modular.cc:471:47: runtime error: shift exponent -5 is negative
| #0 0x1434953 in jxl::ModularFrameDecoder::FinalizeDecoding(jxl::PassesDecoderState*, jxl::ThreadPool*, jxl::ImageBundle*) jpeg-xl/lib/jxl/dec_modular.cc:0
| #1 0x1387e5e in jxl::FrameDecoder::Flush() jpeg-xl/lib/jxl/dec_frame.cc:814:3
| #2 0x137f603 in jxl::FrameDecoder::FinalizeFrame() jpeg-xl/lib/jxl/dec_frame.cc:849:3
| #3 0x104eadf in jxl::(anonymous namespace)::JxlDecoderProcessInternal(JxlDecoderStruct*, unsigned char const*, unsigned long) jpeg-xl/lib/jxl/decode.cc:1155:30
| #4 0x104c287 in JxlDecoderProcessInput jpeg-xl/lib/jxl/decode.cc:1668:14
...
```
Not very important, but it should probably be fixed. | oss-fuzz has found an undefined shift (by -5) in libjxl | https://api.github.com/repos/libjxl/libjxl/issues/29/comments | 4 | 2021-05-27T09:06:11Z | 2021-05-31T11:01:44Z | https://github.com/libjxl/libjxl/issues/29 | 903,439,945 | 29 |
[
"libjxl",
"libjxl"
] | Mozilla uses clang-5.0 for some builds; but it doesn't work when compiling in C++17 mode (it is fine in C++11 mode) for what appears to be a clang-5 compiler bug.
We should add a check that libjxl compiles in release mode with clang-5 to not regress here.
Details: https://gitlab.com/wg1/jpeg-xl/-/issues/227 @saschanaz FYI | Add check that libjxl builds with clang-5.0 | https://api.github.com/repos/libjxl/libjxl/issues/28/comments | 5 | 2021-05-26T22:04:42Z | 2021-11-22T16:41:24Z | https://github.com/libjxl/libjxl/issues/28 | 902,963,008 | 28 |
[
"libjxl",
"libjxl"
] | hello,
I'm on aarch64 and am trying to build a standalone of jpeg-xl to detect breakages. I have already done this with one of your dependencies called thirdparty/highway: https://github.com/google/highway/issues/93
the fix for aarch64 was published in v0.12.1 ; so can you please pull in the new version with your submodule magic?
thanks :-) | please update thirdparty/highway to v0.12.1 to unbreak aarch64 and possibly armv7 | https://api.github.com/repos/libjxl/libjxl/issues/21/comments | 10 | 2021-05-26T13:35:10Z | 2021-06-25T22:27:41Z | https://github.com/libjxl/libjxl/issues/21 | 902,403,536 | 21 |
[
"libjxl",
"libjxl"
] | Hi
I used a test file from here...
`$ wget http://www.r0k.us/graphics/kodak/kodak/kodim20.png
`
Create a jxl file with speed 9...
`$ cjxl kodim20.png kodim20.jxl -s 9 -d 3
`
Create a progressive jxl file with speed 9...
`$ cjxl kodim20.png prog_kodim20.jxl -s 9 -d 3 -p
`
Compare the file sizes...
`$ ls -l kodim20.jxl | awk '{print $5}' && ls -l prog_kodim20.jxl | awk '{print $5}'
`
**24893
72467**
The progressive file is much larger than the non-progressive file.
I'm using `cjxl v0.3.7-30ea86ab` | Big file size with progressive + tortoise. | https://api.github.com/repos/libjxl/libjxl/issues/17/comments | 10 | 2021-05-26T10:41:05Z | 2021-05-30T13:15:05Z | https://github.com/libjxl/libjxl/issues/17 | 902,183,177 | 17 |
[
"libjxl",
"libjxl"
] | Thanks for starting the move to full open source!
However, to understand the codebase better, it would be hugely beneficial to have the full git commit history available, and not the squashed ones from the current public GitLab repo. | Full commit history | https://api.github.com/repos/libjxl/libjxl/issues/8/comments | 5 | 2021-05-26T07:48:49Z | 2021-05-26T16:43:29Z | https://github.com/libjxl/libjxl/issues/8 | 901,928,577 | 8 |
[
"libjxl",
"libjxl"
] | If the input files are JPEGs, `cjxl` creates `jpe????.tmp` files in the system `%temp%` folder and doesn't delete them after the encoding is finished.
_Windows 10 20H2 x64, cjxl v0.3.7-12-g04267a8_ | cjxl - temp JPEG files remain | https://api.github.com/repos/libjxl/libjxl/issues/6/comments | 18 | 2021-05-26T03:41:40Z | 2021-07-09T06:03:27Z | https://github.com/libjxl/libjxl/issues/6 | 901,705,435 | 6 |
[
"alexw994",
"eziod"
] | We found a malicious backdoor in versions 0.0.1 of this project, and its malicious backdoor is the request package. Even if the request package was removed by pypi, many mirror sites did not completely delete this package, so it could still be installed.When using pip install eziod==0.0.1 -i http://pypi.doubanio.com/simple --trusted-host pypi.doubanio.com, the request malicious plugin can be successfully installed.
Repair suggestion: delete version 0.0.1 in PyPI | code execution backdoor | https://api.github.com/repos/alexw994/eziod/issues/1/comments | 0 | 2022-06-29T07:05:34Z | 2022-06-29T07:05:34Z | https://github.com/alexw994/eziod/issues/1 | 1,288,265,493 | 1 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hello,
I am importing an XML file with asyncua version 1.1.0. The XML contains many elements as shown below.
<Value>
<uax:ExtensionObject>
<uax:TypeId>
<uax:Identifier>ns=2;i=1005</uax:Identifier>
</uax:TypeId>
<uax:Body>
<uax:ByteString>ADKJAKKAGSKGKDUWGKW==</uax:ByteString>
</uax:Body>
</uax:ExtensionObject>
</Value>
The import gives an error "('Error val should be a list, this is a python-asyncua bug', 'ByteString', <class 'str'>, 'ADKJAKKAGSKGKDUWGKW==')
".
How can this be solved ?
Note: Only asyncua 1.1.0 is available to be installed in my system. | Import XML with ByteString in ExtensionObject Error | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1846/comments | 0 | 2025-06-23T14:11:28Z | 2025-06-23T14:11:28Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1846 | 3,168,355,898 | 1,846 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
UaExpert nor TwinCat shows any names for parameters of methods registered with asyncua. This causes major confusion, especially when imported into something like TwinCat, where you can't make heads or tails of which parameter is which.
**To Reproduce**<br />
Run `examples/server-methods.py`. Produces the screenshot above for `func_async`.
**Expected behavior**<br />
There to be names associated with parameters, such as with `Server.RequestServerStateChange` in the OPC tree:
Or `Server.GetMonitoredItems` which also has named Output Arguments:
**Screenshots**<br />
One of my methods as it is imported into TwinCat 3 4026:
**Version**<br />
Python-Version: 3.12.11<br />
opcua-asyncio Version (e.g. master branch, 0.9): 1.1.6
| Unable to set parameter names on methods | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1845/comments | 3 | 2025-06-20T19:51:03Z | 2025-06-21T21:22:34Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1845 | 3,164,243,901 | 1,845 |
[
"FreeOpcUa",
"opcua-asyncio"
] | `async def task(loop):
url = "opc.tcp://myserver:4840"
try:
client = Client(url=url)
client.set_user("User")
client.set_password("Test")
await client.connect()
print("connected to OPC UA Server")
except Exception:
_logger.exception("error")
finally:
await client.disconnect()
`
URI and Creds are Dummy, Iam just using the standard [auth with no security example](https://github.com/FreeOpcUa/opcua-asyncio/blob/master/examples/client-minimal-auth.py). I am able connect to server using "UA Expert" also using the .net opc ua.
Getting following error
> INFO:asyncua.client.client:connect
INFO:asyncua.client.ua_client.UaClient:opening connection
INFO:asyncua.uaprotocol:updating client limits to: TransportLimits(max_recv_buffer=65535, max_send_buffer=65535, max_chunk_count=0, max_message_size=0)
INFO:asyncua.client.ua_client.UASocketProtocol:open_secure_channel
INFO:asyncua.client.ua_client.UaClient:create_session
INFO:asyncua.client.ua_client.UASocketProtocol:close_secure_channel
INFO:asyncua.client.ua_client.UASocketProtocol:Request to close socket received
ERROR:asyncua:error
Traceback (most recent call last):
File "/Users/lalitm/Work/Python/opcaua/client_connect.py", line 17, in task
await client.connect()
File "/Users/lalitm/Work/Python/opcaua/.venv/lib/python3.12/site-packages/asyncua/client/client.py", line 321, in connect
await self.create_session()
File "/Users/lalitm/Work/Python/opcaua/.venv/lib/python3.12/site-packages/asyncua/client/client.py", line 510, in create_session
response = await self.uaclient.create_session(params)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/lalitm/Work/Python/opcaua/.venv/lib/python3.12/site-packages/asyncua/client/ua_client.py", line 367, in create_session
response.ResponseHeader.ServiceResult.check()
File "/Users/lalitm/Work/Python/opcaua/.venv/lib/python3.12/site-packages/asyncua/ua/uatypes.py", line 383, in check
raise UaStatusCodeError(self.value)
asyncua.ua.uaerrors._auto.BadCertificateInvalid: The certificate provided as a parameter is not valid.(BadCertificateInvalid)
INFO:asyncua.client.client:disconnect
INFO:asyncua.client.ua_client.UaClient:close_session
WARNING:asyncua.client.ua_client.UaClient:close_session but connection wasn't established
WARNING:asyncua.client.ua_client.UaClient:close_secure_channel was called but connection is closed
INFO:asyncua.client.ua_client.UASocketProtocol:Socket has closed connection | Cannot connect for server without security | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1844/comments | 6 | 2025-06-18T12:32:13Z | 2025-06-19T04:20:24Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1844 | 3,156,700,786 | 1,844 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hi, I was looking for support of certificate chain for client side both for user authentication and secure channel, but could not find any direct mentions.
Is this feature implemented? If not, is it planned or at least known about? Are there any implementations or PoCs?
Thank you in advance for the answers. | Client certificate chain support | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1843/comments | 0 | 2025-06-12T12:30:41Z | 2025-06-12T12:30:41Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1843 | 3,140,109,606 | 1,843 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
On several devices I get this warning:
_Requested session timeout to be 3600000ms, got 30000ms instead_
This happens while connecting this way:
```python
async with Client(url=opc_url) as client:
...
```
Setting
`client.session_timeout = 30000`
in this context has no effect, of course.
**To Reproduce**<br />
Steps to reproduce the behavior incl code:
Connecting this way to a Siemens S7-1500 device:
```python
from asyncua import Client
async with Client(url=opc_url) as client:
...
```
**Expected behavior**<br />
The connection should establish without a warning.
For achieving this, explicitly setting and optional parameter would help:
```python
from asyncua import Client
async with Client(url=opc_url, session_timeout=30000) as client:
...
```
but this is not supported, in the `Client` class, the `session_timeout` parameter is statically set to:
```python
self.session_timeout = 3600000 # 1 hour
```
**Version**<br />
Python-Version: 3.11<br />
opcua-asyncio Version: 1.1.6:
| Missing optional session_timeout during connecting in context manager | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1842/comments | 2 | 2025-06-11T08:36:50Z | 2025-06-11T13:10:02Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1842 | 3,135,976,487 | 1,842 |
[
"FreeOpcUa",
"opcua-asyncio"
] |
**Describe the bug** <br />
When creating a Server, some of the ReferenceTypes in the Base Information Model have wrong attribute values.
The attributes isAbstract and Symmetric are not consistently set correctly. This is because the default values are set to True for both, even though they should be set to False for both.
These dafault attributes are set in ua/uaprotocol_auto.py
Lines: 6258, 6259
``` python
data_type = NodeId(ObjectIds.ReferenceTypeAttributes)
SpecifiedAttributes: UInt32 = 0
DisplayName: LocalizedText = field(default_factory=LocalizedText)
Description: LocalizedText = field(default_factory=LocalizedText)
WriteMask: UInt32 = 0
UserWriteMask: UInt32 = 0
isAbstract: Boolean = True
Symmetric: Boolean = True
InverseName: LocalizedText = field(default_factory=LocalizedText)
```
**To Reproduce**<br />
Run the Example Server and a use Client to inspect the Reference "organizes" for example.
According to the OPC UA specification, both isAbstract and Symmetric should be false.
**Version**<br />
Python-Version: 3.10.7
opcua-asyncio Version (e.g. master branch, 0.9): 1.1.6
| Default Attributes for isAbstract and Symmetric of ReferenceTypes are wrong | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1840/comments | 0 | 2025-05-22T13:00:36Z | 2025-05-22T13:00:36Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1840 | 3,083,304,945 | 1,840 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hello,
i'm building custom NodeSets for testing different OPC UA stacks for a simulation. I currently have the issue, that my recent models fail to correctly load into the server to access the variables needed for the simulation.
I found the models to be working with open62541, the UA ReferenceStack and Node-OPCUA. I would like to confirm wether this is an issue with the model generation (or the models), if there is an issue with the XML importer or wether i am missing the proper instantiation required to load the models into the server.
**Describe the bug** <br />
The required variables should be exposed writable and accessable, however on import through load_xml the UserAccessLevel of the entire session is dropped to "CurrentRead". I believe i am having issues with the XML import of the model, since the value attributes for custom variables are not even accessible after import. Custom Properties are not affected by this for some reason.
An example of a broken Variable Type:
To run the server i am using poetry, docker and UA Model Compiler to generate my files.
For generation i used the Schema linked in the Submodules: [Ref](https://github.com/OPCFoundation/UA-Nodeset/tree/526c5e59e7ba9e54f9dc7848ce2baa249395d3ef)
I tried crosschecking the generated files using the available schema files from the foundation using xmllint:
```
xmllint --noout --schema Opc.Ua.ModelDesign.xsd Model.xml
xmllint --noout --schema UANodeSetv105.xsd Reference.NodeSet2.xml
```
I did get a single warning for the models:
```
/model/input/TankModel.xml:255: element Children: Schemas validity error : Element '{http://opcfoundation.org/UA/ModelDesign.xsd}Children': This element is not expected.
```
For the other OPC stacks this did not cause any issue so far.
**To Reproduce**<br />
I ran some of the provided examples to test my import issues. Mainly server-woodworking and server-robotics.
I just changed the import to use the custom files. The provided models do not require DI or any other models as a dependency.
I added one of the model files+NodeSet2 here: https://gist.github.com/Frozenbitz/1506cfdbd097cc7819275b50aba80d25
A note on the publication date: I did have issues with the publication date generated by UA Model Compiler. Currently the import fails because the original files are missing the publication date key, which i usually manually add. I did not find a way to edit how the compiler generates the required model definitions to use the dates from the provided files.
```
<Model ModelUri="urn:open62541.server.application" Version="1.05.02" PublicationDate="2024-04-12T00:00:00Z" ModelVersion="1.5.2">
<RequiredModel ModelUri="http://opcfoundation.org/UA/" XmlSchemaUri="http://opcfoundation.org/UA/2008/02/Types.xsd" PublicationDate="2024-04-12T00:00:00Z" />
</Model>
```
To setup a simple server:
```python
def __init__(self, endpoint, name, model_filepath):
self.server = Server()
self.model_filepath = model_filepath
self.server.set_server_name(name)
self.server.set_endpoint(endpoint)
async def init(self):
await self.server.init()
# This need to be imported at the start or else it will overwrite the data
#await self.server.import_xml(os.path.join(self.model_filepath, "../nodeset/DI/Opc.Ua.Di.NodeSet2.xml"))
await self.server.import_xml(
os.path.join(self.model_filepath, "/devel/meta/demo-nodeset2/Tank.NodeSet2.xml")
)
....
```
Logging on Startup:
```
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=18979, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=18974, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=18981, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=18980, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=18982, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=18980, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=18983, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=18980, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=18984, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=18980, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=18989, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=18988, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=18990, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=18988, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
...
```
```
INFO:asyncua.common.xmlimporter:Importing XML file /devel/modules/model-compiler/output/KRITIS3M.Reference.NodeSet2.xml
INFO:asyncua.common.xmlimporter:namespace map: {1: 2}
INFO:asyncua.common.xmlimporter:Importing xml node (QualifiedName(NamespaceIndex=2, Name='DeviceInfoType'), NodeId(Identifier=1, NamespaceIndex=2, NodeIdType=<NodeIdType.Numeric: 2>)) as (QualifiedName(NamespaceIndex=2, Name='DeviceInfoType') NodeId(Identifier=1, NamespaceIndex=2, NodeIdType=<NodeIdType.Numeric: 2>))
INFO:asyncua.common.xmlimporter:Importing xml node (QualifiedName(NamespaceIndex=2, Name='DeviceID'), NodeId(Identifier=2, NamespaceIndex=2, NodeIdType=<NodeIdType.Numeric: 2>)) as (QualifiedName(NamespaceIndex=2, Name='DeviceID') NodeId(Identifier=2, NamespaceIndex=2, NodeIdType=<NodeIdType.Numeric: 2>))
INFO:asyncua.common.xmlimporter:Importing xml node (QualifiedName(NamespaceIndex=2, Name='Location'), NodeId(Identifier=3, NamespaceIndex=2, NodeIdType=<NodeIdType.Numeric: 2>)) as (QualifiedName(NamespaceIndex=2, Name='Location') NodeId(Identifier=3, NamespaceIndex=2, NodeIdType=<NodeIdType.Numeric: 2>))
...
```
```
INFO:asyncua.common.instantiate_util:Instantiate: Skip optional node QualifiedName(NamespaceIndex=0, Name='NamespaceFile') as part of QualifiedName(NamespaceIndex=2, Name='urn:open62541.server.application')
INFO:asyncua.common.instantiate_util:Instantiate: Skip optional node QualifiedName(NamespaceIndex=0, Name='DefaultRolePermissions') as part of QualifiedName(NamespaceIndex=2, Name='urn:open62541.server.application')
INFO:asyncua.common.instantiate_util:Instantiate: Skip optional node QualifiedName(NamespaceIndex=0, Name='DefaultUserRolePermissions') as part of QualifiedName(NamespaceIndex=2, Name='urn:open62541.server.application')
INFO:asyncua.common.instantiate_util:Instantiate: Skip optional node QualifiedName(NamespaceIndex=0, Name='DefaultAccessRestrictions') as part of QualifiedName(NamespaceIndex=2, Name='urn:open62541.server.application')
INFO:asyncua.common.instantiate_util:Instantiate: Skip optional node QualifiedName(NamespaceIndex=0, Name='ConfigurationVersion') as part of QualifiedName(NamespaceIndex=2, Name='urn:open62541.server.application')
INFO:asyncua.common.instantiate_util:Instantiate: Skip optional node QualifiedName(NamespaceIndex=0, Name='ModelVersion') as part of QualifiedName(NamespaceIndex=2, Name='urn:open62541.server.application')
```
**Expected behavior**<br />
The server should start and accept client sessions with "CurrentRead" and "CurrentWrite" enabled.
**Version**<br />
Python-Version: 3.10, 3.11, 3.12 (tested with Docker and python-base) <br />
opcua-asyncio Version master, 1.6, 1.4:
| Import of custom NodeSet causes some Nodes to be readonly | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1839/comments | 0 | 2025-05-19T13:40:58Z | 2025-05-19T13:40:58Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1839 | 3,073,899,173 | 1,839 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hi,
I think there is an error with the typing of function "new_struct" on this line:
https://github.com/FreeOpcUa/opcua-asyncio/blob/5b1091795dc7745efb94acd94381db17274779e8/asyncua/common/structures104.py#L61
I suspect the correct code would be:
```python
name: Union[ua.QualifiedName, str],
```
| Probable typing error in "new_struct" function | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1837/comments | 0 | 2025-05-02T08:03:26Z | 2025-05-02T08:03:26Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1837 | 3,035,431,047 | 1,837 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
Calls to _to_node_id() fail with int arguments > 255.
A typical example is a call like
`self.get_referenced_nodes()`
where the type of the reference is provided in parameter refs as integer argument.
Meanwhile a lot of standard OPC UA references have integer IDs >> 255 (see example):
**To Reproduce**<br />
The example shows a search for references of type HasDictionaryEntry, a standard reference type of OPC UA.
It fails because the integer id of the type is HasDictionaryEntry = 17597.<br />
```
nodes = await self.get_referenced_nodes(refs = ua.ObjectIds.HasDictionaryEntry, direction = ua.BrowseDirection.Forward, nodeclassmask = ua.NodeClass.Object)
```
The root cause of the failure is, that the current implementation of the function _to_nodeid(), which is utilized when issuing the call in the example, only supports TwoByteNodeIds for input arguments of type int.<br />
**Expected behavior**<br />
The implementation of _to_nodeid() should check the numeric range of the input argument and return an appropriate nodeid object, e.g. as depicted in the following code snippet with additional range check:<br />
```
def _to_nodeid(nodeid: Union["Node", ua.NodeId, str, int]) -> ua.NodeId:
if isinstance(nodeid, int):
if nodeid <= 255:
return ua.TwoByteNodeId(nodeid)
elif nodeid <= 65535:
return ua.FourByteNodeId(nodeid)
else:
return ua.NumericNodeId(nodeid)
if isinstance(nodeid, Node):
return nodeid.nodeid
if isinstance(nodeid, ua.NodeId):
return nodeid
if isinstance(nodeid, str):
return ua.NodeId.from_string(nodeid)
raise ua.UaError(f"Could not resolve '{nodeid}' to a type id")
```
**Version**<br />
Python-Version: 3.13.3 <br />
opcua-asyncio Version: master branch, 1.1.6
| _to_nodeid() fails with int arguments > 255 | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1831/comments | 0 | 2025-04-28T09:11:51Z | 2025-04-28T12:36:13Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1831 | 3,024,266,714 | 1,831 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hi!
I would like to report 3 bugs related to importing and exporting XML and dependent namespaces.
I will also post a PR with test cases to reproduce and a proposal to fix them 😊
**Describe the bug** <br />
*No. 1: Exporting a node with a data type from another namespace leads to an invalid XML*
The namespace URI of the data type is missing in the exported XML file, since the XML exporter only checks for referenced nodes but not the data type attribute.
*No. 2: Exporting a node without a valid namespace URI leads to an invalid XML*
If you export a node that was added to the server with a namespace index that doesn't relate to a namespace URI, consequently the namespace URI is also missing in the XML file.
*No. 3: Importing such invalid XMLs does not report any errors*
Without the knowledge about which URIs these namespace indices from the XMLs belong to, they cannot be added correctly to a server without making mistakes like adding them to the wrong namespace or referencing nodes from an unwanted namespace.
I'm not fully sure what the current code does in these cases, if I understood it correctly it looks quite random and depends on the existence of the namespace indices on the server when importing the XML.
**To Reproduce**<br />
See test cases in PR.
**Expected behavior**<br />
- No. 1: The missing namespace URI should be mentioned in the XML
- No. 2: It should not be possible to export nodes without a valid namespace URI
- No. 3: An error should be reported if nodes in a XML file cannot be related to a namespace URI
For no. 3 I'm not sure how many "strictly speaking invalid" XML files are out there, so I considered one exception that should keep most of them intact:
If a XML file has only nodes from ns=1 and a namespace URI is missing, allow them to be added anyway, as they will just be added to the local server's namespace.
**Version**<br />
Python-Version: 3.12.3<br />
opcua-asyncio Version: f12f3e19b3e7a3bc3db38add30176eef0cecd300
| Issues with missing namespace URIs when importing and exporting XMLs | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1829/comments | 1 | 2025-04-27T14:25:22Z | 2025-04-30T09:39:11Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1829 | 3,023,115,186 | 1,829 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
server.import_xml leads to node duplicates. The function that causes the duplicates is found in the XmlImporter.import_xml() and this line:
`
await self._add_references(remaining_refs)`
The nodes already got added including their refs through this prior called function:
`node = await self._add_node_data(nodedata, no_namespace_migration=True)`
with this code in the function, which contains the same nodes again, the duplication happens internally:
```
self.refs, remaining_refs = [], self.refs
await self._add_references(remaining_refs)
```
Why is this funtion call necessary, if the nodes and their refs already get registered in address space through the prior function? I cannot determine how the duplication happens, but without the second call of _add_references the nodes are not duplicated under the namespace index. **UAExpert** will **not** show the duplication of nodes. Connect to the server through the library client and read the imported xml nodes. Here the server will report duplicate nodes (identical from my perspective - same namespace index and identifier - which should not be possible at all). The duplicates start right after the imported namespace node (level 2).
```
root = server.get_root_node()
nodes = await (root.get_children()) #root layer
nodes = await (nodes[0].get_children()) # namespace layer / nodes
nodes = await (nodes[3].get_children()) # duplicated notes
```
I do not have any clue on how to fix the bug. The internal server calls are somehow challenging to follow. I do not know where to start my search, or what is causing the duplication after one day of debugging.
**To Reproduce**<br />
import an xml. e.g:
```
<?xml version='1.0' encoding='UTF-8'?>
<UANodeSet xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:uax="http://opcfoundation.org/UA/2008/02/Types.xsd" xmlns:xsd="http://www.w3.org/2001/XMLSchema" xmlns="http://opcfoundation.org/UA/2011/03/UANodeSet.xsd">
<NamespaceUris>
<Uri>urn:server</Uri>
<Uri>SYM:</Uri>
</NamespaceUris>
<Aliases>
<Alias Alias="Boolean">i=1</Alias>
<Alias Alias="SByte">i=2</Alias>
<Alias Alias="Byte">i=3</Alias>
<Alias Alias="Int16">i=4</Alias>
<Alias Alias="UInt16">i=5</Alias>
<Alias Alias="Int32">i=6</Alias>
<Alias Alias="UInt32">i=7</Alias>
<Alias Alias="Int64">i=8</Alias>
<Alias Alias="UInt64">i=9</Alias>
<Alias Alias="Float">i=10</Alias>
<Alias Alias="Double">i=11</Alias>
<Alias Alias="String">i=12</Alias>
<Alias Alias="DateTime">i=13</Alias>
<Alias Alias="ByteString">i=15</Alias>
<Alias Alias="Organizes">i=35</Alias>
<Alias Alias="HasTypeDefinition">i=40</Alias>
</Aliases>
<UAObject NodeId="ns=1;s=SYM:" BrowseName="1:SYM:">
<DisplayName>SYM:</DisplayName>
<References>
<Reference ReferenceType="HasTypeDefinition">i=61</Reference>
<Reference ReferenceType="Organizes" IsForward="false">i=85</Reference>
<Reference ReferenceType="Organizes">ns=2;s=S71500ET200MP-Station_2</Reference>
<Reference ReferenceType="Organizes">ns=2;s=S71500ET200MP-Station_1</Reference>
</References>
</UAObject>
<UAObject NodeId="ns=2;s=S71500ET200MP-Station_2" BrowseName="2:S71500ET200MP-Station_2" ParentNodeId="ns=1;s=SYM:">
<DisplayName>S71500ET200MP-Station_2</DisplayName>
<References>
<Reference ReferenceType="HasTypeDefinition">i=61</Reference>
</References>
</UAObject>
<UAObject NodeId="ns=2;s=S71500ET200MP-Station_1" BrowseName="2:S71500ET200MP-Station_1" ParentNodeId="ns=1;s=SYM:">
<DisplayName>S71500ET200MP-Station_1</DisplayName>
<References>
<Reference ReferenceType="HasTypeDefinition">i=61</Reference>
</References>
</UAObject>
</UANodeSet>
```
after the import, just do:
```
root = server.get_root_node()
nodes = await (root.get_children()) #root layer
nodes = await (nodes[0].get_children()) # namespace layer / nodes
nodes = await (nodes[3].get_children()) # duplicated notes
```
**Expected behavior**<br />
import_xml and _add_references should not add any duplicates if the nodes are already there.
**Screenshots**<br />
after importing the above example xml. the following can be observed if the imported nodes are requested:
**Version**<br />
Python-Version: 3.11
opcua-asyncio Version (e.g. master branch, 0.9): 1.1.6
| Server import XML leads to opcua node duplications | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1828/comments | 0 | 2025-04-25T20:30:53Z | 2025-04-25T20:39:16Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1828 | 3,020,976,534 | 1,828 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
Hello together,
I'm trying to read an array within an ExtensionObject from a B&R PLC.
The array on its own can be correctly read. The array within an extension object is treated as a single integer.
This issue is possibly related to: https://github.com/FreeOpcUa/opcua-asyncio/issues/1388
**To Reproduce**<br />
Python Code
```
from asyncua.sync import Client
url = "opc.tcp://127.0.0.1:4840"
client = Client(url)
client.connect()
client.load_type_definitions()
# client.load_data_type_definitions() # Does not load the required definitions
# client.load_enums()
array_node = client.get_node("ns=6;s=::AsGlobalPV:global_structure.nested_Struct1.my_array")
structure_node = client.get_node("ns=6;s=::AsGlobalPV:global_structure")
array_node.get_value()
structure_node.get_value()
```
Output:
```
array_node.get_value()
Out[6]: [11, 12, 13, 14, 15, 16, 17, 0]
structure_node.get_value()
Out[7]: Structure1(bool2=True, bool1=True, bool3=False, nested_Struct1=Nested_struct(bool_nest2=False, bool_nest1=False, bool_nest3=True, _my_array=8, my_array=11))
```
You can see that the array is correctly read if it is directly accessed. If the array is within an extension object it is not recognized.
The bsd (from OPC Binary -> BR.Default) when converted to text is:
```
<opc:TypeDictionary xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:tns="http://br-automation.com/OpcUa/PLC/PV/" DefaultByteOrder="LittleEndian" xmlns:opc="http://opcfoundation.org/BinarySchema/" xmlns:ua="http://opcfoundation.org/UA/" TargetNamespace="http://br-automation.com/OpcUa/PLC/PV/"> <opc:Import Namespace="http://opcfoundation.org/UA/" />
<opc:StructuredType BaseType="ua:ExtensionObject" Name="Structure1">
<opc:Field TypeName="opc:Boolean" Name="bool2"/>
<opc:Field TypeName="opc:Boolean" Name="bool1"/>
<opc:Field TypeName="opc:Boolean" Name="bool3"/>
<opc:Field TypeName="tns:Nested_struct" Name="nested_Struct1"/>
</opc:StructuredType>
<opc:StructuredType BaseType="ua:ExtensionObject" Name="Nested_struct">
<opc:Field TypeName="opc:Boolean" Name="bool_nest2"/>
<opc:Field TypeName="opc:Boolean" Name="bool_nest1"/>
<opc:Field TypeName="opc:Boolean" Name="bool_nest3"/>
<opc:Field TypeName="opc:Int32" Name="#my_array"/>
<opc:Field LengthField="#my_array" TypeName="opc:Int32" Name="my_array"/>
</opc:StructuredType>
</opc:TypeDictionary>
```
**Screenshots**<br />
The corresponding structure from UaExpert which correctly loads the arrays in the extension object:
<img width="395" alt="Image" src="https://github.com/user-attachments/assets/8f5894b8-69f8-4a4f-a928-e59bd4226f00" /><br />
**Version**<br />
Python-Version: 3.11.8
opcua-asyncio Version: 1.1.6
Is there anything else I can supply to troubleshoot this issue?
Thanks in advance for your work!
| Reading an array within an extension object is not possible. | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1825/comments | 2 | 2025-04-23T07:55:49Z | 2025-05-05T11:17:42Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1825 | 3,013,054,264 | 1,825 |
[
"FreeOpcUa",
"opcua-asyncio"
] | ## Problem Statement
I need to model a large namespace that will not support browsing.
I would like to create my nodes for only as long as at least one subscription exists.
## What I've tried
I've tried subscribing to the `ItemSubscriptionCreated` callback, and to use the callback as a chance to create my nodes.
```python
# during setup
self.server.subscribe_server_callback(
CallbackType.ItemSubscriptionCreated, self._dispatch_item_subscription
)
...
# _dispatch_item_subscription
# Create a new node, and set the status of the `ItemSubscriptionCreated` response_params.
event.response_params[index].StatusCode = (
await component.on_item_subscription(item)
)
```
This almost works.
The Good: The nodes are created and the client receives whatever status code I return.
The Bad: The client's subscription does not seem to work (i.e. receive values). Only after creating a second subscription does it work as expected.
I believe this is because although I'm setting the `StatusCode`, my callback is being run after the subscription creation handling logic, meaning the Node doesn't exist yet to properly create a subscription.
## Questions
1. Does this approach make any sense?
2. Is there another way to accomplish a dynamic address space?
I believe Milo supports this in the form of a [ManagedAddressSpace](https://javadoc.io/doc/org.eclipse.milo/sdk-server/0.3.3/org/eclipse/milo/opcua/sdk/server/api/ManagedAddressSpace.html) that gives you exact control over the handling of service calls. | Create Nodes During Subscription | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1824/comments | 4 | 2025-04-23T06:29:38Z | 2025-04-25T15:31:30Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1824 | 3,012,843,125 | 1,824 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hi! Im try to connect to prosys opc ua server with example:
[(https://github.com/FreeOpcUa/opcua-asyncio/blob/master/examples/client_to_prosys.py)]
but have error:
"
................
File "C:\Users\unreg\AppData\Local\Programs\Python\Python311\Lib\enum.py", line 695, in __call__
return cls.__new__(cls, value)
^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\unreg\AppData\Local\Programs\Python\Python311\Lib\enum.py", line 1111, in __new__
raise ve_exc
ValueError: 36 is not a valid NodeIdType"
Prosys Server version 5.5.2-362
SDK version 5.2.8-159
This is very strange, because about six months ago the connection was fine (to the old version of the server).
The client UAexpert 1.6.3 448 work fine
How i can fix it? Thanks!
| Connect to prosys opc ua simulation server failed | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1823/comments | 6 | 2025-04-19T14:03:33Z | 2025-05-12T12:49:18Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1823 | 3,006,451,745 | 1,823 |
[
"FreeOpcUa",
"opcua-asyncio"
] | We're using asyncua as the server part for an integration test setup connecting from an open62541 based client.
With 1.1.6, we get an error at this point:
```
Traceback (most recent call last):
File "c:\workspace\tools\Python312-32\Lib\site-packages\asyncua\server\uaprocessor.py", line 147, in process_message
return await self._process_message(typeid, requesthdr, seqhdr, body)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "c:\workspace\tools\Python312-32\Lib\site-packages\asyncua\server\uaprocessor.py", line 213, in _process_message
data = self._connection.security_policy.peer_certificate + params.ClientNonce
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~
TypeError: can't concat NoneType to bytes
```
While debugging into it, I've found out, that the check
```python
if self._connection.security_policy.peer_certificate is None:
data = params.ClientNonce
else:
data = self._connection.security_policy.peer_certificate + params.ClientNonce
```
is not working, since `self._connection.security_policy.peer_certificate` is `b''`, but not `None`.
We're using SecurityPolicyNone, so I assume due to the change of the initialization of the security_policies between 1.1.5 and 1.1.6, the peer_certificate is accidently set, but the Nonce is always None.
When I add
```python
if not len(self.peer_certificate):
self.peer_certificate = None
```
after line
https://github.com/FreeOpcUa/opcua-asyncio/blob/4a11975af723be18f6889819aa5a161e0c343541/asyncua/crypto/security_policies.py#L519
it works flawlessly again | Possible regression 1.1.5 -> 1.1.6 | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1821/comments | 3 | 2025-04-17T07:35:59Z | 2025-04-29T12:41:20Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1821 | 3,001,694,882 | 1,821 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Both the `call_method` and `call_method_full` functions throw an exception if the status is not Ok, which means that it is not possible to see the return value of an Uncertain result. As far as I understand the spec if a method returns an Uncertain status it indicates that the return values *might* be unreliable, but it could still be useful to examine their values.
Since most people probably only care about if the status is Ok or not I don't think it makes sense to change `call_method`, but maybe since `call_method_full` returns a `CallMethodResult` object, which contains the status code, maybe it makes sense for that method not to throw an exception on a not-Ok status? Or, if that breaks backwards compatibility, would it be possible to add either an additional parameter (with a default value) to `call_method_full` which disables the check of the return value, or a completely new function to do this?
For a real world use case of this, see the "Joining Systems Base" companion spec. The methods defined there have `status` and `statusMessage` return values, and in case of error the the OPC UA status is set to Uncertain and these return values describe the error in more detail. See eg the [EnableAsset](https://reference.opcfoundation.org/IJT/Base/v100/docs/7.4.3) method and the discussion about the [status](https://reference.opcfoundation.org/IJT/Base/v100/docs/7.2.3)
I'm happy to provide a pull request with the above proposed changes, but first want some input about what sort of API the developers would prefer | Accessing return values when a method return Uncertain | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1815/comments | 0 | 2025-04-04T12:07:10Z | 2025-04-04T12:07:10Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1815 | 2,972,179,304 | 1,815 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
I'm attempting to import a variable into TwinCat 3 (4026) which is an array of ExtensionObjects. No matter the size of the array, TwinCat 3 refuses to import it and reports a "Something odd happened" message with no further details.
This error only occurs when the variable is an array. If the variable is a singleton ExtensionObject--there is no issue. If I create this ExtensionObject and variable in another TwinCat 3 server, and import it with the same TwinCat 3 client that throws the error for `asyncua`, it works just fine. However, I am able to import this variable--as an array created with `asyncua`--into UaExpert and work with it just fine. Therefore, this appears to be a metadata incompatibility issue between asyncua and TwinCat 3.
Here's a screenshot of the Attributes panel in UaExpert of the variable:
Of the Structure type:
Of the Value editor of the variable:
I am reaching out to Beckhoff support for their input on this issue. I wanted to also open this report here in case we're able to find the solution while Beckhoff works this from their end. I'll report any notable information from them here as well.
**To Reproduce**<br />
Here is my node adding code. It supports folders, objects, variables, and structure definitions. For variables, it also supports primitives, structures, and arrays.
```python
async def _add_node(self, record: NodeRecord) -> bool:
# Do not operate on DELETE events.
if record.operation == SchemaChangeOperation.DELETE:
return False
# Do not operate on existing nodes.
with suppress(BadNodeIdUnknown, UaStringParsingError):
node = self._server.get_node(record.node_id)
await node.read_display_name() # Test if node exists
return False # No error was raised, so node exists
logger.info(
"Node '{}/{}' does not exist, adding...",
record.display_name.Text,
record.node_id.to_string(),
)
# Retrieve the parent node.
parent_node = (
self._server.get_node(record.parent_node_id)
if record.parent_node_id
else self._server.get_objects_node()
)
logger.debug(
"Setting parent of node to -> '{}/{}'",
(await parent_node.read_display_name()).Text,
record.parent_node_id.to_string(),
)
# Assign a new node id if required.
if record.type_definition != ObjectIds.StructureDefinition:
await self._assign_node_id(record)
# Update the browse name. It is influenced by node id changes.
record.browse_name = QualifiedName(
record.browse_name.Name,
record.node_id.NamespaceIndex,
)
# Add the node to the tree.
match record.node_class:
case NodeClass.Object:
if record.type_definition == ObjectIds.FolderType:
node = await parent_node.add_folder(
record.node_id,
record.browse_name,
)
elif record.type_definition == ObjectIds.BaseObjectType:
node = await parent_node.add_object(
record.node_id,
record.browse_name,
)
case NodeClass.Variable:
if record.type_definition == ObjectIds.StructureDefinition:
# Construct the structure.
# Structures are registered with the `ua` module,
# and can be accessed with
# `ua.<structure browse name>`.
fields = []
for field_dict in record.fields:
fields.append(
new_struct_field(
name=field_dict["name"],
dtype=field_dict["data_type"],
array=field_dict["is_array"],
)
)
await new_struct(
self._server,
self._namespace_idx,
record.browse_name.Name, # Namespace may be set to 999, so just use 'Name'.
fields,
)
await self._server.load_data_type_definitions()
return
elif record.type_definition == ObjectIds.StructureType:
structure_cls = getattr(ua, record.structure_cls_name)
if record.value.is_array:
value = [structure_cls(**value) for value in record.value.Value]
else:
value = structure_cls(**record.value.Value)
node = await parent_node.add_variable(
record.node_id,
record.browse_name,
Variant(
Value=value,
VariantType=VariantType.ExtensionObject,
Dimensions=record.value.Dimensions,
is_array=record.value.is_array,
),
)
elif record.type_definition == ObjectIds.BaseDataVariableType:
node = await parent_node.add_variable(
record.node_id,
record.browse_name,
record.value,
)
# Enable writability.
await node.set_writable(writable=record.writable)
logger.debug(
"Node '{}/{}' writable: {}",
record.display_name.Text,
record.node_id.to_string(),
record.writable,
)
# Configure array dimensions.
if isinstance(record.value.Value, (list, tuple, set)):
await node.write_array_dimensions([len(record.value.Value)])
await node.write_value_rank(ValueRank.OneDimension)
# Enable historization.
if record.historizing:
with suppress(UaNodeAlreadyHistorizedError, UaError):
await self._enable_historization(node)
logger.debug(
"Historization enabled for node '{}/{}': {}",
record.display_name.Text,
record.node_id.to_string(),
record.historizing,
)
# Set display name.
await node.write_attribute(
AttributeIds.DisplayName,
DataValue(Value=Variant(Value=record.display_name)),
)
logger.success(
"Node '{}/{}' added successfully",
record.display_name.Text,
record.node_id.to_string(),
)
return True
```
```python
# NodeRecord definition used above. It is essentially a dataclass.
class NodeRecord:
"""
Configuration class for managing nodes in an asyncua OPC UA server.
Supports add, change, and delete events for folders, variables, objects,
and custom extension objects, with child node tracking and array support.
"""
# Operation
operation: SchemaChangeOperation = SchemaChangeOperation.ADD
# Database Identification
database_id: int | None = None
# Node Identification
node_id: NodeId | None = None # Specific NodeId, autogenerated if None
parent_node_id: NodeId = NodeId.from_string("i=85") # Objects folder
namespace_idx: int = 0 # Namespace index
# Node Metadata
display_name: LocalizedText = LocalizedText("UnnamedNode")
browse_name: QualifiedName = QualifiedName("Unnamed", 0)
description: LocalizedText | None = None
# Node Type and Structure
node_class: NodeClass = NodeClass.Object
type_definition: ObjectIds | None = None # e.g., FolderType (i=61)
# Variable-specific Settings
value: Variant | list[Any] | None = None # Scalar or array value
writable: bool = False # UserWriteMask bit
historizing: bool = False # Enable history
# Custom Extension Object Settings
structure_cls_name: str | None = None
fields: list[dict] | None = None
# Child Nodes
children: set[str] = set() # Direct child nodes by NodeId
```
**Expected behavior**<br />
No warning/error from TwinCat 3.
**Version**<br />
Python-Version: 3.12.9
opcua-asyncio Version (e.g. master branch, 0.9): 1.1.5
| Arrays of ExtensionObjects incompatible with TwinCat 3 | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1813/comments | 7 | 2025-04-03T16:37:25Z | 2025-05-28T12:01:57Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1813 | 2,970,151,384 | 1,813 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
Cannot export nodes with nodeIds containing german umlaute / vowel mutations. In the old library python-opcua there was an issue that German umlaute in the nodeid would break the library from working; this was somehow fixed with this merge (here https://github.com/FreeOpcUa/opcua-asyncio/issues/621 and here https://github.com/FreeOpcUa/opcua-asyncio/issues/1207#issue-1585979228). However, this fix was only a "read"-related fix, because if you try to get the displayname, or browsename or some other information from that node, that contain the mutated vowel e.g. ä,ü,ö, the .readAttribute function that calls the server will result in an BadNodeIdUnknown, because the called now converted nodeid with 0x"hex" in it do not exist after the quick'n'dirty conversion introduced through quick fixing the read-related problem.
If I try to use the well-coded XMLExporter, the additional calls to the server to get every information of the node that contains umlaute in their nodeid will result in the aforementioned BadNodeIdUnknown error because the nodeid is practically wrong after the conversion.
**To Reproduce**<br />
add a node containing umlaute "äüö" in the nodeid
read the node with the opcua-asyncio client, e.g. root.getChildren()
create the XmlExporter(client)
use await exporter.build_etree([node]) <- a lot of BadNodeIdUnkown errors will appear
**Expected behavior**<br />
convert the nodeid meaningfully in that way that it is possible to use them for requests
After that every subsequent request for any information should be successful
**Version**<br />
Python-Version: 3.11.9<br />
opcua-asyncio Version : 1.1.5
| Cannot export nodes with german umlaute / vowel mutations. BadNodeIdUnknown if implemented conversion of the nodeid is used for requests.. | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1811/comments | 0 | 2025-04-03T08:34:03Z | 2025-04-03T08:38:26Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1811 | 2,968,754,144 | 1,811 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hi all,
Can you please help me to understand this situation ?
I have 3 custom enums:
opcua_server | root:Enum trouvée : enumSungrowInverterDeviceTypeCode
opcua_server | root:Enum trouvée : enumSungrowInverterOutputType
opcua_server | root:Enum trouvée : enumSungrowInverterWorkState
They are linked to this class:
opcua_server | class typSungrowInverter:
opcua_server |
opcua_server | '''
opcua_server | typSungrowInverter structure autogenerated from StructureDefinition object
opcua_server | '''
opcua_server |
opcua_server | data_type = ua.NodeId.from_string('''ns=4;s=|tprop|wago_opcua.Application.typSungrowInverter''')
opcua_server | sSN: ua.String = ua.String()
opcua_server | uiDeviceTypeCode: ua.Enumeration = **field(default_factory=ua.Enumeration)**
opcua_server | rP_Nominal: ua.Float = ua.Float(0)
opcua_server | uiOutputType: ua.Enumeration = **field(default_factory=ua.Enumeration)**
opcua_server | rYieldDaily: ua.Float = ua.Float(0)
opcua_server | rYieldTotal: ua.Float = ua.Float(0)
opcua_server | udiRunningTime: ua.UInt32 = ua.UInt32(0)
opcua_server | rTempeInternal: ua.Float = ua.Float(0)
opcua_server | rS: ua.Float = ua.Float(0)
opcua_server | arrrMPPT_U: typing.List[ua.Float] = field(default_factory=list)
opcua_server | arrrMPPT_I: typing.List[ua.Float] = field(default_factory=list)
opcua_server | rP_DC: ua.Float = ua.Float(0)
opcua_server | rU_L1_L2: ua.Float = ua.Float(0)
opcua_server | rU_L2_L3: ua.Float = ua.Float(0)
opcua_server | rU_L3_L1: ua.Float = ua.Float(0)
opcua_server | rI_L1: ua.Float = ua.Float(0)
opcua_server | rI_L2: ua.Float = ua.Float(0)
opcua_server | rI_L3: ua.Float = ua.Float(0)
opcua_server | rP: ua.Float = ua.Float(0)
opcua_server | rQ: ua.Float = ua.Float(0)
opcua_server | rPF: ua.Float = ua.Float(0)
opcua_server | rFreq: ua.Float = ua.Float(0)
opcua_server | uiWorkState: ua.Enumeration = **field(default_factory=ua.Enumeration)**
opcua_server | uiAlarmCode: ua.UInt16 = ua.UInt16(0)
The problem is in bold. I don't understand why the generated class refers to ua.Enumeration here ?
It generates the following error when I want to access to ua.typSungrowInverter()
opcua_server | AttributeError: module 'asyncua.ua' has no attribute 'Enumeration'
It seems logic as Enumeration is not created. Only other enums but they are not correctly linked to the Class
Do I miss something ?
Thank you in advance for your help
Michael | load_data_type_definitions() does not generate correctly a custom class | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1809/comments | 1 | 2025-04-01T11:37:39Z | 2025-04-06T01:37:42Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1809 | 2,963,176,104 | 1,809 |
[
"FreeOpcUa",
"opcua-asyncio"
] | To save some time, please provide us following informations, if possible:
**Description** <br />
When I start the script examples/statemachine-example.py, I encounter multiple errors related to node addition and type mismatches. The script fails to complete successfully.
**To Reproduce**<br />
Start the script examples/statemachine-example.py.
Observe the following errors in the console output:
INFO:asyncua.server.internal_server:No user manager specified. Using default permissive manager instead.
INFO:asyncua.server.internal_session:Created internal session Internal
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=11715, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15958, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15959, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15960, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15961, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15962, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15963, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15964, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=16134, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=16135, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=16136, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.common.instantiate_util:Instantiate: Skip node without modelling rule QualifiedName(NamespaceIndex=0, Name='FiniteStateMachineType') as part of QualifiedName(NamespaceIndex=2, Name='StateMachine')
INFO:asyncua.common.instantiate_util:Instantiate: Skip node without modelling rule QualifiedName(NamespaceIndex=0, Name='InitialStateType') as part of QualifiedName(NamespaceIndex=2, Name='Idle')
INFO:asyncua.common.instantiate_util:Instantiate: Skip node without modelling rule QualifiedName(NamespaceIndex=0, Name='ChoiceStateType') as part of QualifiedName(NamespaceIndex=2, Name='Idle')
INFO:asyncua.common.instantiate_util:Instantiate: Skip node without modelling rule QualifiedName(NamespaceIndex=0, Name='InitialStateType') as part of QualifiedName(NamespaceIndex=2, Name='Loading')
INFO:asyncua.common.instantiate_util:Instantiate: Skip node without modelling rule QualifiedName(NamespaceIndex=0, Name='ChoiceStateType') as part of QualifiedName(NamespaceIndex=2, Name='Loading')
INFO:asyncua.common.instantiate_util:Instantiate: Skip node without modelling rule QualifiedName(NamespaceIndex=0, Name='InitialStateType') as part of QualifiedName(NamespaceIndex=2, Name='Initializing')
INFO:asyncua.common.instantiate_util:Instantiate: Skip node without modelling rule QualifiedName(NamespaceIndex=0, Name='ChoiceStateType') as part of QualifiedName(NamespaceIndex=2, Name='Initializing')
INFO:asyncua.common.instantiate_util:Instantiate: Skip node without modelling rule QualifiedName(NamespaceIndex=0, Name='InitialStateType') as part of QualifiedName(NamespaceIndex=2, Name='Processing')
INFO:asyncua.common.instantiate_util:Instantiate: Skip node without modelling rule QualifiedName(NamespaceIndex=0, Name='ChoiceStateType') as part of QualifiedName(NamespaceIndex=2, Name='Processing')
INFO:asyncua.common.instantiate_util:Instantiate: Skip node without modelling rule QualifiedName(NamespaceIndex=0, Name='InitialStateType') as part of QualifiedName(NamespaceIndex=2, Name='Finished')
INFO:asyncua.common.instantiate_util:Instantiate: Skip node without modelling rule QualifiedName(NamespaceIndex=0, Name='ChoiceStateType') as part of QualifiedName(NamespaceIndex=2, Name='Finished')
WARNING:asyncua.server.address_space:Write refused: Variant: Variant(Value=NodeId(Identifier=6, NamespaceIndex=2, NodeIdType=<NodeIdType.FourByte: 1>), VariantType=<VariantType.NodeId: 17>, Dimensions=None, is_array=False) with type 17 does not have expected type: 0
Traceback (most recent call last):
File "C:\workspace_examples\opcua\statemachine-example.py", line 88, in <module>
asyncio.run(main())
File "C:\Python311\Lib\asyncio\runners.py", line 190, in run
return runner.run(main)
^^^^^^^^^^^^^^^^
File "C:\Python311\Lib\asyncio\runners.py", line 118, in run
return self._loop.run_until_complete(task)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Python311\Lib\asyncio\base_events.py", line 650, in run_until_complete
return future.result()
^^^^^^^^^^^^^^^
File "C:\workspace_examples\opcua\statemachine-example.py", line 55, in main
await mystatemachine.change_state(state1, trans1, f"{mystatemachine._name}: Idle", 300)
File "C:\workspace_examples\opcua\.venv\Lib\site-packages\asyncua\common\statemachine.py", line 183, in change_state
await self._write_state(state)
File "C:\workspace_examples\opcua\.venv\Lib\site-packages\asyncua\common\statemachine.py", line 206, in _write_state
await self._current_state_id_node.write_value(state.node.nodeid, varianttype=ua.VariantType.NodeId)
File "C:\workspace_examples\opcua\.venv\Lib\site-packages\asyncua\common\node.py", line 269, in write_value
await self.write_attribute(ua.AttributeIds.Value, dv)
File "C:\workspace_examples\opcua\.venv\Lib\site-packages\asyncua\common\node.py", line 323, in write_attribute
result[0].check()
File "C:\workspace_examples\opcua\.venv\Lib\site-packages\asyncua\ua\uatypes.py", line 377, in check
raise UaStatusCodeError(self.value)
asyncua.ua.uaerrors._auto.BadTypeMismatch: The value supplied for the attribute is not of the same type as the attribute"s value.(BadTypeMismatch)
Process finished with exit code 1
**Version**<br />
Python-Version: 3.11<br />
opcua-asyncio Version: 1.1.5 | Problems running statemachine-example.py | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1807/comments | 2 | 2025-03-31T20:38:35Z | 2025-04-01T07:03:14Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1807 | 2,961,527,067 | 1,807 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
When I subscribe using this library, to my equipment's OPC/UA server (which is not based on FreeOpcUa), it sends a `PublishResponse` with a `NotificationData` containing an `ExtensionObject` of length -1 (0xffff). As a result, I get an exception: `asyncua.common.utils.NotEnoughData: Not enough data left in buffer, request for 4, we have 0`
I believe the equipment's OPC/UA server is non-compliant - a length of -1 is not valid. However, it appears to be a common enough bug with OPC/UA servers that the [OPC Foundation's reference implementation](https://github.com/OPCFoundation/UA-.NETStandard) have implemented a [interop fix](https://github.com/OPCFoundation/UA-.NETStandard/commit/a887f909f1d314cfa7c32989628afa754984c4f1).
Can we apply the same interop fix here?
I believe the change is simply to change [ua_binary.py;546-547](https://github.com/FreeOpcUa/opcua-asyncio/blob/e6d646d12f1db1e59229034439f9b0baf9d2da6e/asyncua/ua/ua_binary.py#L546-L547) from:
```python
if length < 1:
body = Buffer(b"")
```
...to...
```python
if length == -1:
# Interop fix for old OPC/UA implementations that omit to fill in the length
body = data
elif length < 1:
body = Buffer(b"")
```
**To Reproduce**<br />
I can reproduce this using the built-in `uasubscribe` tool. The OPC/UA server sends back an `ExtensionObject` of the following form:
```
0000 01 00 2b 03 01 ff ff ff ff 05 00 00 00 c9 00 00
0010 00 0d 0a 54 37 8e 3c 80 29 12 ab 53 95 db 01 80
0020 29 12 ab 53 95 db 01 c9 00 00 00 0d 0b 00 00 00
0030 00 98 40 8b 3f 40 15 25 ab 53 95 db 01 40 15 25
0040 ab 53 95 db 01 c9 00 00 00 0d 0b 00 00 00 a0 b7
0050 b3 91 3f d0 13 4b ab 53 95 db 01 d0 13 4b ab 53
0060 95 db 01 c9 00 00 00 0d 0b 00 00 00 a0 09 15 8c
0070 3f 60 9a 84 ab 53 95 db 01 60 9a 84 ab 53 95 db
0080 01 c9 00 00 00 0d 0b 00 00 00 e0 43 a6 91 3f d0
0090 86 a0 ab 53 95 db 01 d0 86 a0 ab 53 95 db 01 00
00a0 00 00 00
```
Octets 0-3 are the `TypeId`, octet 4 is the `EncodingMask` and octet 5-8 are the length (-1).
I see the following exception:
```
WARNING:asyncua.client.client:Requested session timeout to be 3600000ms, got 100000ms instead
WARNING:asyncua.client.client:Revised values returned differ from subscription values: CreateSubscriptionResult(SubscriptionId=4, RevisedPublishingInterval=500.0, RevisedLifetimeCount=9900, RevisedMaxKeepAliveCount=150)
Type Ctr-C to exit
ERROR:asyncua.client.ua_client.UaClient:Error parsing notification from server
Traceback (most recent call last):
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/client/ua_client.py", line 579, in publish
response = struct_from_binary(ua.PublishResponse, data)
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/ua/ua_binary.py", line 696, in struct_from_binary
return _create_dataclass_deserializer(objtype)(data)
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/ua/ua_binary.py", line 687, in decode
kwargs[field.name] = deserialize_field(data)
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/ua/ua_binary.py", line 687, in decode
kwargs[field.name] = deserialize_field(data)
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/ua/ua_binary.py", line 687, in decode
kwargs[field.name] = deserialize_field(data)
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/ua/ua_binary.py", line 254, in deserialize
return list(unpack_array(data, length))
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/ua/ua_binary.py", line 247, in <genexpr>
return (deserialize_element(data) for _ in range(length))
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/ua/ua_binary.py", line 542, in extensionobject_from_binary
return from_binary(cls, body)
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/ua/ua_binary.py", line 629, in from_binary
return _create_type_deserializer(uatype, type(None))(data)
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/ua/ua_binary.py", line 687, in decode
kwargs[field.name] = deserialize_field(data)
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/ua/ua_binary.py", line 589, in _deserialize
size = Primitives.Int32.unpack(data)
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/ua/ua_binary.py", line 131, in unpack
return struct.unpack(self.format, data.read(self.size))[0]
File "/home/mattwilliams/.local/lib/python3.8/site-packages/asyncua/common/utils.py", line 62, in read
raise NotEnoughData(f"Not enough data left in buffer, request for {size}, we have {self._size}")
asyncua.common.utils.NotEnoughData: Not enough data left in buffer, request for 4, we have 0
```
**Expected behavior**<br />
I expect the `ExtensionObject` to be parsed correctly, by assuming that -1 means the rest of the data buffer.
**Version**<br />
Python-Version: 3.8<br />
opcua-asyncio Version (e.g. master branch, 0.9): 1.1.5
| asyncua.common.utils.NotEnoughData parsing an ExtensionObject with length -1 | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1801/comments | 1 | 2025-03-29T06:17:47Z | 2025-03-31T14:02:43Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1801 | 2,957,806,022 | 1,801 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
When a remote server returns a value whose type is a Structure subtype that is encoded by a non-root `DataTypeEncodingType` (like `GDS`'s `ApplicationRecordDataType`, which idiotically defines its own `Default Binary` (`id=134`)),
and asyncua has already previously loaded that nodeset with a different `NamespaceIndex` (like if you start a server, and then start a client),
then asyncua will fail to decode that value, and leave it as an unparseable ExtensionObject, like:
```
ExtensionObject(TypeId=NodeId(Identifier=134, NamespaceIndex=3, NodeIdType=NodeIdType.FourByte), Body=b'\x04\x02\x00\x07\xf7A\xc8\xba[\x10K\x85\x01\xaa\xe1\xe4\xe2\xf7\x11 \x00\x00\x00urn:app-asr-01:Codesys:eUAServer\x00\x00\x00\x00\x01\x00\x00\x00\x02\x11\x00\x00\x00eUAServer@Codesys\x15\x00\x00\x00urn:Codesys:eUAServer\x01\x00\x00\x00\x19\x00\x00\x00opc.tcp://10.10.4.78:4840\x01\x00\x00\x00\x02\x00\x00\x00DA')
```
There is now no way in asyncua's API to decode this, even if we constructed a new `ExtensionObject` with the correct `TypeId` with `NamespaceIndex=5`
**To Reproduce**<br />
1. Create a `Server` and import a few nodeset XMLs including GDS (`http://opcfoundation.org/UA/GDS/`).
* Let's say the resulting GDS `NamespaceIndex` is `5`
2. `server.load_data_type_definitions()`
3. Create a `Client` and connect to an OPC UA service with GDS, which has a **different** `NamespaceIndex` for its `http://opcfoundation.org/UA/GDS/` (let's say `3`)
* You can call `client.load_data_type_definitions()` here if you want, or not, it makes no difference.
4. Invoke `call_method` `FindApplications` (GDS `i=143`) on the `Directory` (GDS `i=141`) node, with an argument that should return at least one value (of type `ApplicationRecordDataType` (GDS `i=1`).
**Expected behavior**<br />
You should get a (list of) `ApplicationRecordDataType`.
**Actual Behavior**<br />
asyncua fails to decode the `ExtensionObject` (like above).
If you skip steps 1 & 2, and instead do call `client.load_data_type_definitions()`, it will work correctly.
The fundamental problem seems to be that asyncua uses one global set of namespaces for encoders, and does not wire up data type encoding definitions from different clients/servers correctly.
**Version**<br />
Python-Version: 3.13.2
opcua-asyncio Version (e.g. master branch, 0.9): 1.1.5 | Failure to decode ExtensionObject if remote encoding NodeId does not match previously loaded local definition | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1800/comments | 3 | 2025-03-26T21:31:09Z | 2025-03-27T21:30:45Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1800 | 2,950,827,425 | 1,800 |
[
"FreeOpcUa",
"opcua-asyncio"
] |
**Describe the bug** <br />
Using opcua-asyncio github repo with tag v1.1.5.
We are trying to run the example scripts - `server-with-encryption.py` and `client-with-encryption.py`
We are getting `ValueError: Decryption failed` when client is trying to connect to the server. Below are the logs -
```
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=16136, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.internal_server:starting internal server
INFO:asyncua.server.binary_server_asyncio:Listening on 0.0.0.0:4840
INFO:asyncua.server.binary_server_asyncio:New connection from ('127.0.0.1', 44910)
INFO:asyncua.uaprotocol:updating server limits to: TransportLimits(max_recv_buffer=65535, max_send_buffer=65535, max_chunk_count=1601, max_message_size=104857600)
ERROR:asyncua.server.binary_server_asyncio:Exception raised while processing message from client
Traceback (most recent call last):
File "/home/eiid-d1-l3t015/opcua/examples/../asyncua/server/binary_server_asyncio.py", line 99, in _process_received_message_loop
await self._process_one_msg(header, buf)
File "/home/eiid-d1-l3t015/opcua/examples/../asyncua/server/binary_server_asyncio.py", line 105, in _process_one_msg
ret = await self.processor.process(header, buf)
File "/home/eiid-d1-l3t015/opcua/examples/../asyncua/server/uaprocessor.py", line 104, in process
msg = self._connection.receive_from_header_and_body(header, body)
File "/home/eiid-d1-l3t015/opcua/examples/../asyncua/common/connection.py", line 415, in receive_from_header_and_body
chunk = MessageChunk.from_header_and_body(self.security_policy, header, body, use_prev_key=False)
File "/home/eiid-d1-l3t015/opcua/examples/../asyncua/common/connection.py", line 125, in from_header_and_body
decrypted = crypto.decrypt(data.read(len(data)))
File "/home/eiid-d1-l3t015/opcua/examples/../asyncua/crypto/security_policies.py", line 198, in decrypt
return self.Decryptor.decrypt(data)
File "/home/eiid-d1-l3t015/opcua/examples/../asyncua/crypto/security_policies.py", line 309, in decrypt
decrypted += self.decryptor(self.client_pk,
File "/home/eiid-d1-l3t015/opcua/examples/../asyncua/crypto/uacrypto.py", line 200, in decrypt_rsa_oaep
text = private_key.decrypt(
ValueError: Decryption failed
INFO:asyncua.server.binary_server_asyncio:Lost connection from ('127.0.0.1', 44910), None
INFO:asyncua.server.uaprocessor:Cleanup client connection: ('127.0.0.1', 44910)
```
**To Reproduce**<br />
Go to the examples directory and run the python scripts `server-with-encryption.py` and `client-with-encryption.py`
```
python3 server-with-encryption.py
```
```
python3 client-with-encryption.py
```
**Expected behavior**<br />
We should not get the Decryption failed error, client should be able to connect to server.
**Version**<br />
Python-Version: 3.10<br />
opcua-asyncio Version (e.g. master branch, 0.9): v1.1.5
| server-with-encryption.py giving ValueError: Decryption failed | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1797/comments | 0 | 2025-03-17T09:09:49Z | 2025-03-17T09:09:49Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1797 | 2,924,286,729 | 1,797 |
[
"FreeOpcUa",
"opcua-asyncio"
] | With prior versions of asyncua (<1.0.x) I was able to access nodes with as ``var = client.get_node(ua.NodeId(1002, 2))`` or ``var = client.get_node("ns=3;i=2002")``. How these always return ``BadNodeIdUnknown``!
My server has many layers of nodes and using ``get_child()`` is not convenient...
TIA
| How to access nodes without using get_child method? | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1795/comments | 3 | 2025-03-10T14:47:17Z | 2025-03-10T16:33:07Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1795 | 2,907,528,971 | 1,795 |
[
"FreeOpcUa",
"opcua-asyncio"
] | I am trying to understand how OPC compatibility works, but maybe you can help us with this.
In our organization, we have a simulator that uses the [Traeger](https://www.traeger.de/en/products/development/opcua/opcua-sdk) as the OPC UA Server. When we try to connect to it using this library, it is throwing a `TimeoutError`. The error happens on the `open_secure_channel()` method of the `us_client.py` file.
Traeger OPC Server: https://www.traeger.de/en/products/development/opcua/opcua-sdk
As I understand it, for as long the OPC Server is within the same UA protocol, this library should work, irregardless to which language the OPC Server is written. Does this library supports the Traeger as the OPC UA server?
Can you help us understand or verify whether this library can support them? Thanks a lot. | Traeger as the OPC Server | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1793/comments | 9 | 2025-03-06T09:29:52Z | 2025-03-12T09:52:44Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1793 | 2,899,856,519 | 1,793 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
There are the following problems with the processing of from_string/to_string of NodeId
- Missing percent encoding for NamespaceUri when from_string/to_string
- Identifier cannot be parsed correctly when it contains ";"
- Also, unlike the specifications, nsu= and srv= are written after the Identifier, so if the Identifier happens to contain a confusing string like ;nsu=, parsing will be impossible.
https://reference.opcfoundation.org/Core/Part6/v105/docs/5.1.12
**To Reproduce**<br />
```py
from asyncua import ua
before_node = ua.ExpandedNodeId(Identifier="foo;nsu=bar", NamespaceUri="http://example.com/q?=foo;bar")
print(before_node) # ExpandedNodeId(Identifier='foo;nsu=bar', NamespaceIndex=0, NodeIdType=<NodeIdType.String: 3>, NamespaceUri='http://example.com/q?=foo;bar', ServerIndex=0)
after_node = ua.NodeId.from_string(before_node.to_string())
print(after_node) # UaStringParsingError
```
**Expected behavior**<br />
Probably the fix will be something like this:
```py
from urllib.parse import unquote, quote
class NodeId:
@staticmethod
def _from_string(string):
elements = string
identifier = None
namespace = 0
ntype = None
srv = None
nsu = None
while elements:
k, v = elements.split("=", 1)
k = k.strip()
# Split only if not a string
elements = ""
if k != "s":
if ";" in v:
v, elements = v.split(";", 1)
v = v.strip()
if k == "ns":
namespace = int(v)
elif k == "i":
ntype = NodeIdType.Numeric
identifier = int(v)
elif k == "s":
ntype = NodeIdType.String
identifier = v
elif k == "g":
ntype = NodeIdType.Guid
identifier = uuid.UUID(f"urn:uuid:{v}")
elif k == "b":
ntype = NodeIdType.ByteString
identifier = bytes(v, 'utf-8')
elif k == "srv":
srv = int(v)
elif k == "nsu":
nsu = unquote(v) # Decode parcent encoded string
if identifier is None:
raise UaStringParsingError(f"Could not find identifier in string: {string}")
if nsu is not None or srv is not None:
return ExpandedNodeId(identifier, namespace, ntype, NamespaceUri=nsu, ServerIndex=srv)
return NodeId(identifier, namespace, ntype)
class ExpandedNodeId(NodeId):
def to_string(self):
string = []
if self.ServerIndex:
string.append(f"srv={self.ServerIndex}")
if self.NamespaceUri:
string.append(f"nsu={quote(self.NamespaceUri)}") # Percent encode
string.append(NodeId.to_string(self)) # Add after processing
return ";".join(string)
``` | Bad from_string/to_string result for NodeId | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1792/comments | 1 | 2025-03-06T04:14:32Z | 2025-03-06T04:30:24Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1792 | 2,899,284,390 | 1,792 |
[
"FreeOpcUa",
"opcua-asyncio"
] |
**Describe the bug** <br />
Callling run_until_complete() consistently causes a TimeoutError after around ~22 minutes of function running. The exact code is:
```
# ------------------------------------------------------------------------------
def exec_loop(opcua_obj, file_name):
'''
Asyncio Loop wrapper function.
'''
loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop)
try:
loop.set_debug(True)
loop.run_until_complete(task(opcua_obj, file_name))
except TimeoutError as exp:
print('TimeoutError')
except KeyboardInterrupt as exp:
print('KeyboardInterrupt')
finally:
loop.close()
```
This function should be running until complete so it doesn't make sense that it is causing a TimeoutError. Why would it do that?
Is it because when traversing using node.get_children() the connection is lost and it can't recover? If so is there a way configure the asyncua Client to automatically reconnect?
[example_opcua_async_client.py.txt](https://github.com/user-attachments/files/19091050/example_opcua_async_client.py.txt)
**To Reproduce**<br />
Run the attached Python app with a OPC-UA server that is running.
**Expected behavior**<br />
Expecting not to TimeoutError.
**Screenshots**<br />
The output of the program shows it exits with TimeoutError exception.
**Version**<br />
Python-Version: Python 3.10.12<br />
opcua-asyncio Version (e.g. master branch, 0.9): asyncua==1.1.5
| Call to run_until_complete returning TimeoutError | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1791/comments | 1 | 2025-03-05T14:45:48Z | 2025-03-05T15:38:54Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1791 | 2,897,537,269 | 1,791 |
[
"FreeOpcUa",
"opcua-asyncio"
] |
**Describe the bug** <br />
I am using "asyncua" library to connect to an OPC-UA server and get list of all of the tags available on that server. The Python application takes almost 4 hours to traverse the OPC-UA server, here is the output of the app:
```
Starting OPC-UA Client Application
Started at: 2025-02-09 06:28:42.764038
URL: opc.tcp://172.16.1.5:62541
File (Output): Tags_2025-02-09_062842_UTC.txt
Client(url=url)
Set User
Set Password
Set Application_URI
Client set_security
USE_TRUST_STORE: True
if USE_TRUST_STORE
async with client -> get nodes
Called client.node.root
Called to get_child 0:Objects
Call asyncio.wait_for()
Runtime Information
Start Time: 2025-02-09 06:28:42.764038
End Time: 2025-02-09 10:22:53.466855
Duration: 3:54:10.702817
```
The code calls:
`await extract_node_paths(objects, "/Objects", output_file)`
and the main function looks like this:
```
async def extract_node_paths(node, path, output_file):
children = await node.get_children()
for child in children:
browse_name = await child.read_browse_name()
child_path = f"{path}/{browse_name.Name}"
if await child.read_node_class() == 2: # Variable node class
with open(output_file, 'a') as f:
f.write(child_path + '\n')
await extract_node_paths(child, child_path, output_file)
```
The output has the following syntax:
```
/Objects/Tag Providers/ZLT/ZLT/Dispatcher/Configuration/Limits/LimitBESS
/Objects/Tag Providers/ZLT/ZLT/Dispatcher/Configuration/Limits/LimitEnergyBESS
/Objects/Tag Providers/ZLT/ZLT/Dispatcher/Configuration/Limits/LimitPOI
```
The output file has 104,529 lines. So 104,529 tags.
The question I have is:
1. Is it possible to speed this up where it can do this work in 20 minutes?
2. Is there a built-in API can get a tree list of the tags is much less time?
**To Reproduce**<br />
Code is shown above.
**Expected behavior**<br />
I was expecting that there would be a relatively fast (minutes and not hours) way to get a OPC-UA server tags list.
**Screenshots**<br />
None, but can get some if needed.
**Version**<br />
Python-Version: Python 3.10.12<br />
opcua-asyncio Version (e.g. master branch, 0.9): asyncua==1.1.5
| Speeding up OPC-UA node traversal from hours to minutes | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1787/comments | 14 | 2025-02-12T22:08:10Z | 2025-03-17T17:36:15Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1787 | 2,849,454,516 | 1,787 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
A server method decorated with `@uamethod` does not convert to the dtypes specified when registering the method at the server. Integers are always converted to `Int64` and floats to `Double`.
**To Reproduce**<br />
Adapted from the sync examples:
```python
# %%
from asyncua.sync import Server
from asyncua import ua, uamethod
def print_result(func):
def wrapped(*args, **kwargs):
result = func(*args, **kwargs)
print(result)
return result
return wrapped
@print_result
@uamethod
def multiply(parent, x, y):
print("multiply method call with parameters: ", x, y)
return x * y
# set up our server
server = Server()
server.set_endpoint("opc.tcp://0.0.0.0:4840/freeopcua/server/")
# set up our own namespace, not really necessary but should as spec
uri = "http://examples.freeopcua.github.io"
idx = server.register_namespace(uri)
# populating our address space
myobj = server.nodes.objects.add_method(
idx,
"multiply",
multiply,
[ua.VariantType.Int16, ua.VariantType.Int16],
[ua.VariantType.Int16],
)
# starting!
server.start()
#%%
from asyncua.sync import Client, ThreadLoop
with ThreadLoop() as tloop:
with Client("opc.tcp://localhost:4840/freeopcua/server/", tloop=tloop) as client:
uri = "http://examples.freeopcua.github.io"
idx = client.get_namespace_index(uri)
obj = client.nodes.objects
res = obj.call_method(f"{idx}:multiply", 3, 2)
```
Printed output:
```
multiply method call with parameters: 3 2
[Variant(Value=6, VariantType=<VariantType.Int64: 8>, Dimensions=None, is_array=False)]
```
**Expected behavior**<br />
The return type should be `Int16` as in `add_method` instead of `Int64`.
**Problem**<br/>
The conversion in `uamethod` is performed by calling `_format_call_outputs` and `to_variant` if no `Variant` is returned by the method; a new `Variant` with `VariantType=None` is initialized.
Thereby, the types are guessed and the types from the method registration are ignored.
From methods.py:
```python
def uamethod(func):
"""
Method decorator to automatically convert
arguments and output to and from variants
"""
if iscoroutinefunction(func):
async def wrapper(parent, *args):
func_args = _format_call_inputs(parent, *args)
result = await func(*func_args)
return _format_call_outputs(result)
else:
def wrapper(parent, *args):
func_args = _format_call_inputs(parent, *args)
result = func(*func_args)
return _format_call_outputs(result)
return wrapper
def _format_call_outputs(result):
if result is None:
return []
elif isinstance(result, ua.CallMethodResult):
result.OutputArguments = to_variant(*result.OutputArguments)
return result
elif isinstance(result, ua.StatusCode):
return result
elif isinstance(result, tuple):
return to_variant(*result)
else:
return to_variant(result)
def to_variant(*args: Iterable) -> List[ua.Variant]:
"""Create a list of ua.Variants from a given iterable of arguments."""
uaargs: List[ua.Variant] = []
for arg in args:
if not isinstance(arg, ua.Variant):
arg = ua.Variant(arg)
uaargs.append(arg)
return uaargs
```
**Version**<br />
Python-Version: 3.11
opcua-asyncio Version (e.g. master branch, 0.9): 1.1.5
| Server method decorated with uamethod do not convert to correct dtypes | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1784/comments | 4 | 2025-02-04T14:58:50Z | 2025-02-06T13:05:23Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1784 | 2,830,519,717 | 1,784 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hello everyone,
I'm trying to write a variable in a tag/node of the PLC with asyncio and I can't do it... I don't know what could be happening, if you know something that could help me I would appreciate it.
I can connect, readValues, getNodes etc... but can't write, and tag is writable.
I've tried both with write_values([nodeIds], [valuesToWrite]) and with write_value(nodeId, valueToWrite) and it's not working...
Here is the part of the code, let me explain you:
```
# results = await globals.opc_client.write_values(nodes_to_write, nodesValues_to_write)
for node, value in zip(nodes_to_write, nodesValues_to_write):
print("node in zip", node)
print("value in zip", value)
result = await node.write_value(value)
# for result in results:
if result.StatusCode.is_good():
print("Write successful")
else:
print("Error writing tag dentro de result:", result.StatusCode)
```
**node in zip:**
[ns=3;s="EQ11_PH1"."TO"."TEST1"]
**value in zip:**
DataValue(Value=Variant(Value='1', VariantType=<VariantType.Int16: 4>, Dimensions=None, is_array=False), StatusCode_=StatusCode(value=0), SourceTimestamp=None, ServerTimestamp=None, SourcePicoseconds=None, ServerPicoseconds=None)
**And I get that error:**
Error writing tag: required argument is not an integer
**I try to write into the node directly like this too:**
result = await node.write_value(85)
**And then I get that error:**
"Error writing tag: The server does not support writing the combination of value, status and timestamps provided.(BadWriteNotSupported)"
It seems something happens in the dataValue, that's my code to get it:
`
valorOK2 = ua.DataValue(ua.Variant(value, ua.VariantType(switch_opc_data_type(tag['DATATYPE']))))
`
where datatype is FLOAT,INT,DEC, etc... and I switch to a valid opc_data_type:
```
from asyncua import ua
def switch_opc_data_type(datatype):
datatype = datatype.lower()
if datatype == "int":
return ua.VariantType.Int16.value
elif datatype == "dint":
return ua.VariantType.Int32.value
elif datatype == "str":
return ua.VariantType.String.value
elif datatype == "dec":
return ua.VariantType.Float.value
elif datatype == "ddec":
return ua.VariantType.Double.value
elif datatype == "dat":
return ua.VariantType.DateTime.value
else:
return ua.VariantType.String.value
```
That's all I think... so if someone can help me I would appreciate it!
Thanks! | Can't write values... datatype error? | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1783/comments | 3 | 2025-01-31T10:39:13Z | 2025-01-31T11:16:56Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1783 | 2,823,020,646 | 1,783 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hi! I have an extensionobject of the type
MyObject Array[32]
[0]:
element1 : BOOL;
element2 : FLOAT;
element3: INT;
[1]
element1 : BOOL;
element2 : FLOAT;
element3: INT;
[2]
element1 : BOOL;
element2 : FLOAT;
element3: INT;
.
.
.
[31]
element1 : BOOL;
element2 : FLOAT;
element3: INT;
I have managed to access the node and stored it as 'MyNode'
however I cant access any of the data by name.
If I want to read element1 of the third item I can do that by
print = MyObject[2].Body[0]
I would rather be able to access it by MyObject[2].element1
or at least be able to read the names of all the elements by a method like read_display_name() or similar.
I have been trying to access the names of the elements in some way or another, but with no luck unfortunately.
If any of you could help, I would greatly appreciate it :) | Reading Extensionobjects | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1778/comments | 7 | 2025-01-23T09:15:25Z | 2025-01-28T13:31:17Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1778 | 2,806,349,008 | 1,778 |
[
"FreeOpcUa",
"opcua-asyncio"
] | I'm trying to install the package in my virtual environment using the command line given in your documentation. I get the following error:
I thought python3.8 was enough to install your package as python>=3.7 is required according to your doc.
Am i wrong ?
Thx in advance | Can't install with python 3.8 | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1776/comments | 1 | 2025-01-21T16:48:30Z | 2025-01-22T10:45:07Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1776 | 2,802,344,512 | 1,776 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hi,
I'm trying to share a value between two nodes. I have a "Parent" node which is a custom structure with a value "Acknowledge"
I want to create a node Acknowledge has a child of Parent which shares the value of Acknowledge in the parent, so that when I update Acknowledge in the Parent node it also updates the value of Acknowledge in the node below. As you can see in the screenshots below, the values are not synced :
Here is my code:
 | Sharing values between 2 nodes | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1775/comments | 1 | 2025-01-21T14:06:19Z | 2025-01-21T14:22:11Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1775 | 2,801,953,605 | 1,775 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Issue**:
- When a client creates a subscription to my server, deletes it, waits ~60 seconds, and tries to create it again, the server gets stuck in an infinite loop of trying to publish an answer.
- If the client immediately tries to create the subscription again or within ~60 seconds, its able to without issue.
- Is there some advice of what could be going wrong here?
**Server Console:**
Monitored Item
Node NodeId(Identifier=UUID('000103e9-3001-0001-0000-000000000000'), NamespaceIndex=2, NodeIdType=<NodeIdType.Guid: 4>) was created
INFO:asyncua.server.uaprocessor:Server wants to send publish answer but no publish request is available,enqueuing publish results callback, length of queue is 1
INFO:asyncua.server.uaprocessor:Server wants to send publish answer but no publish request is available,enqueuing publish results callback, length of queue is 1
INFO:asyncua.server.uaprocessor:Server wants to send publish answer but no publish request is available,enqueuing publish results callback, length of queue is 1
INFO:asyncua.server.uaprocessor:Server wants to send publish answer but no publish request is available,enqueuing publish results callback, length of queue is 1
...
continues until I kill the server.... | Server wants to send publish answer but no publish request is available | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1774/comments | 1 | 2025-01-17T23:25:06Z | 2025-01-20T08:45:20Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1774 | 2,796,466,031 | 1,774 |
[
"FreeOpcUa",
"opcua-asyncio"
] | I would like to warn you that at least the lastest 3 versions deployed on PyPi are not aligned to the GitHub code (I checked from 1.1.5 to 1.1.3).
I discovered this since installing the lastest version of the library from pip, authentication with certificates doesn't validate the corresponding private key.
Checking the source code released on PyPi of the aforementioned versions, you can see in internal_session.py:122 that the certificate validation check is missing (was fixed in https://github.com/FreeOpcUa/opcua-asyncio/commit/3c6317be7b1f1e8942ecf9feaab7495b946b70c5#diff-11331c0757144766f90517604c255b8cd7be5f482d28634b997042a5bbffc399R129).
I don't know if it is because of the failing GitHub actions. Could you check it out?
Thanks. | Lastest versions on PyPi are not updated | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1773/comments | 1 | 2025-01-15T16:40:54Z | 2025-01-15T19:09:48Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1773 | 2,790,322,718 | 1,773 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug**
When calling `asyncua_server.delete_nodes([self.node], recursive=True)` there appears message in terminal:
```
.local/lib/python3.12/site-packages/asyncua/server/address_space.py:428: RuntimeWarning: coroutine 'MonitoredItemService.datachange_callback' was never awaited
callback(handle, None, ua.StatusCode(ua.StatusCodes.BadNodeIdUnknown))
RuntimeWarning: Enable tracemalloc to get the object allocation traceback
```
It seems there are some sync/async issues, maybe some functionality doesn't work properly.
**To Reproduce**
Call `asyncua_server.delete_nodes([self.node], recursive=True)`
**Expected behavior**
No warnings
**Version**
Python-Version: 3.12
opcua-asyncio Version: 1.1.5
| Coroutine not awaited while deleting node | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1771/comments | 2 | 2025-01-11T14:21:36Z | 2025-01-12T04:52:52Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1771 | 2,781,903,572 | 1,771 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
I trying to rename object on running opc ua server.
From old DisplayName is 'SNMP' to new DisplayName is 'SNMP changed'.
But object name in tree not changed.
Code received new name from redis subscription.
**To Reproduce**<br />
Here is code example:
`async def object_changed(jval):
org_id = jval["args"]["id"]
query = "SELECT po.id_p, po.address, p.name FROM page_object AS po LEFT JOIN page AS p ON "
query += f"p.id_p=po.id_p WHERE po.id_p={org_id}"
org, = await db.select(query, f"Database error on object {org_id}:")
node_ = server.get_node(ua.NodeId(org_id, 1))
display_name = await node_.read_display_name()
if org["name"] != display_name.Text:
now = datetime.datetime.now()
await node_.write_attribute(ua.AttributeIds.DisplayName, \
ua.DataValue(ua.uatypes.LocalizedText(org["name"].replace(':', ' ')), \
ua.StatusCode(0), now.replace(tzinfo=tz_local).astimezone(tz_utc)))
await node_.write_attribute(ua.AttributeIds.BrowseName, \
ua.DataValue(ua.uatypes.QualifiedName(org["name"].replace(':', ' ')), \
ua.StatusCode(0), now.replace(tzinfo=tz_local).astimezone(tz_utc)))
await node_.write_attribute(ua.AttributeIds.Description, \
ua.DataValue(ua.uatypes.LocalizedText(org["name"].replace(':', ' ')), \
ua.StatusCode(0), now.replace(tzinfo=tz_local).astimezone(tz_utc)))`
**Screenshots**<br />
Screenshot attached
**Version**<br />
Python-Version: 3.11.2 x64<br />
opcua-asyncio Version (e.g. master branch, 0.9): asyncua-1.1.5
| How-to rename object on running opc ua server? | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1766/comments | 4 | 2024-12-23T15:23:48Z | 2025-05-28T17:43:37Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1766 | 2,756,275,126 | 1,766 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
I'm running an OPCUA server on a Raspberry Pi 4B. When I closed the connection from my laptop acting as client an error got logged. It shows that the watchdog should handle cleaning up the sessions but it did not this time for unknown reasons, resulting in a crash in crash in `BinaryServer._close_tasks`.
**To Reproduce**<br />
Not entirely clear. The only thing I know for sure is that it happened when I closed the client session. It never happened before.
**Expected behavior**<br />
I expected the client session to close properly like the logs show:
**Screenshots**<br />
**Version**<br />
Python-Version:3.11.2<br />
opcua-asyncio Version: 1.1.5
| Watchdog didn't clean up sessions properly resulting in crash in BinaryServer._close_tasks | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1763/comments | 0 | 2024-12-19T08:27:22Z | 2024-12-19T08:27:22Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1763 | 2,749,595,262 | 1,763 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
The namespace index of exported nodes will not match up with what is observed in UaExpert. This is because the nodeset only contains the namespaces of nodes in the NodeSet.
**To Reproduce**<br />
Connect to a server and export the XML using a subset of all nodes, e.g
```python
client = Client(url)
print("Connecting to server...")
async with client as client:
print("Connected to server")
objects_node = client.get_objects_node()
all_nodes: List[Node] = []
await recursively_populate_children(objects_node, all_nodes)
filtered_nodes: List[Node] = []
for node in all_nodes:
if await is_in_namespace(node, ns=3):
filtered_nodes.append(node)
await client.export_xml(filtered_nodes, output_file_path)
```
There will only be one element in the `NamespaceUris` array, e.g.
```xml
<NamespaceUris>
<Uri>http://www.helloworld.com/OPCUA/MyNamespace</Uri>
</NamespaceUris>
```
And the namespace index of nodes will all be `1`, and not `3` like they should be. (I think they are not `0`, because the `NamespaceUris` implicitly contains `http://opcfoundation.org/UA/` as index `0`, even if it looks like the `0` index is `http://www.helloworld.com/OPCUA/MyNamespace`.
I think the issue is related to this function
```python
async def _add_namespaces(self, nodes):
ns_array = await self.server.get_namespace_array()
idxs = await self._get_ns_idxs_of_nodes(nodes)
# now create a dict of idx_in_address_space to idx_in_exported_file
self._addr_idx_to_xml_idx = self._make_idx_dict(idxs, ns_array)
ns_to_export = [ns_array[i] for i in sorted(list(self._addr_idx_to_xml_idx.keys())) if i != 0]
# write namespaces to xml
self._add_namespace_uri_els(ns_to_export)
```
**Version**<br />
Python-Version: 3
opcua-asyncio Version: 1.1.5
| Nodeset exporter should include the full namespace array, not just the namespaces of the nodes that are exported | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1758/comments | 3 | 2024-12-05T10:02:31Z | 2025-04-27T10:26:20Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1758 | 2,719,936,409 | 1,758 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Description:**
I am trying to delete a node with a custom type using the asyncua framework. I have reviewed the client-deleting.py example provided in the repository and attempted to implement node deletion in my application. However, I am encountering an issue: after calling delete_nodes, I am still able to access and read the node's attributes like BrowseName.
**Code Example:**
```
# Attempt to delete a batch of nodes
await self.client.delete_nodes(self.obj_list)
# Attempt to delete a single node
node = self.client.get_node(self.obj_nodeid)
results = await self.client.delete_nodes([node]) # Delete one node
# Verify if the node still exists
node_after = self.client.get_node(self.obj_nodeid)
try:
browse_name = await node_after.read_browse_name()
except Exception:
logger.info("Delete Successfully")
```
**Observed Behavior:**
Even after calling delete_nodes, the node remains accessible. I am still able to retrieve the BrowseName using read_browse_name() without any exceptions.
**Expected Behavior:**
After calling delete_nodes, the node should no longer exist, and attempting to access it should raise an exception.
**Additional Context:**
The node being deleted has a custom type derived from BaseObjectType.
asyncua version: 1.1.5
python version: 3.8.17
I have stored the nodes in a list and attempted both batch deletion and single-node deletion.
Is there something I am missing in the deletion process? Any guidance or example code would be greatly appreciated. | How to Delete a Node with Custom Type | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1757/comments | 0 | 2024-12-05T07:10:23Z | 2024-12-05T07:10:23Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1757 | 2,719,549,270 | 1,757 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hi all,
I'm writing an OPC UA Server using latest version 1.1.5.
The server define a new structure using the:
from asyncua.common.structures104 import new_struct, new_enum, new_struct_field
**method to create and add new structure type to OPC UA Server**
```
async def define_work_order_datatype(server):
res, * = await new_struct(
server,
idx,
"WorkOrderDataType",
[
new_struct_field("order_locked" , ua.VariantType.Boolean),
new_struct_field("order_priority" , ua.VariantType.Byte),
new_struct_field("job_order_code" , ua.VariantType.String),
new_struct_field("customer_code" , ua.VariantType.String),
new_struct_field("item_code" , ua.VariantType.String),
new_struct_field("material_code" , ua.VariantType.String),
new_struct_field("order_notes" , ua.VariantType.String),
new_struct_field("used_deadline_datetime" , ua.VariantType.Boolean),
new_struct_field("deadline_datetime" , ua.VariantType.DateTime),
new_struct_field("file_name_1" , ua.VariantType.String),
new_struct_field("pieces_per_file_1" , ua.VariantType.Int32),
new_struct_field("requested_pieces_1" , ua.VariantType.Int32),
new_struct_field("file_name_2" , ua.VariantType.String),
new_struct_field("pieces_per_file_2" , ua.VariantType.Int32),
new_struct_field("requested_pieces_2" , ua.VariantType.Int32),
new_struct_field("file_name_3" , ua.VariantType.String),
new_struct_field("pieces_per_file_3" , ua.VariantType.Int32),
new_struct_field("requested_pieces_3" , ua.VariantType.Int32),
new_struct_field("file_name_4" , ua.VariantType.String),
new_struct_field("pieces_per_file_4" , ua.VariantType.Int32),
new_struct_field("requested_pieces_4" , ua.VariantType.Int32),
new_struct_field("file_name_5" , ua.VariantType.String),
new_struct_field("pieces_per_file_5" , ua.VariantType.Int32),
new_struct_field("requested_pieces_5" , ua.VariantType.Int32),
new_struct_field("file_name_6" , ua.VariantType.String),
new_struct_field("pieces_per_file_6" , ua.VariantType.Int32),
new_struct_field("requested_pieces_6" , ua.VariantType.Int32),
new_struct_field("file_name_7" , ua.VariantType.String),
new_struct_field("pieces_per_file_7" , ua.VariantType.Int32),
new_struct_field("requested_pieces_7" , ua.VariantType.Int32),
new_struct_field("file_name_8" , ua.VariantType.String),
new_struct_field("pieces_per_file_8" , ua.VariantType.Int32),
new_struct_field("requested_pieces_8" , ua.VariantType.Int32),
],
)
return res
```
**Create new structure data type and method which use it as argument**
```
# create new structure datatype and add it to OPC UA Struct node
work_order_datatype = await define_work_order_datatype(self.server)
# create and add uamethod witch use new structure datatype as argument
a_work_order_order_code = ua.Argument()
a_work_order_order_code.Name = "order_code"
a_work_order_order_code.DataType = ua.NodeId(ua.ObjectIds.String)
a_work_order_order_code.ValueRank = -1
a_work_order_order_code.Description = ua.LocalizedText("work order: order code")
a_work_order_data = ua.Argument()
a_work_order_data.Name = "data"
a_work_order_data.DataType = work_order_datatype.nodeid
a_work_order_data.ValueRank = -1
a_work_order_data.Description = ua.LocalizedText("work order: data")
await self.server.nodes.objects.add_method(
idx,
"work_order_add",
self.api_work_order_add,
[a_work_order_order_code, a_work_order_data],
[ua.VariantType.Boolean]
)
```
**uamethod implementation**
```
@uamethod
async def api_work_order_add(self, parent, order_code: str, data):
# ??? HOW TO GET STRUCTURE DATA ???
return True
```
The data content is:
```
[Dbg]>>> data
ExtensionObject(TypeId=NodeId(Identifier=2, NamespaceIndex=2, NodeIdType=<NodeIdType.FourByte: 1>), Body=b'\x00\x00\x05\x00\x00\x0012341\x04\x00\x00\x001234\x01\x00\x00\x003\x05\x00\x00\x00sderg\x07\x00\x00\x00dfgsdfg\x00\x00@m%\xebS\xbf\x01\x08\x00\x00\x00sdfgsdfg\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff\x00\x00\x00\x00\x00\x00\x00\x00')
```
I was able, in some way, to get data using struct but the code is terrible and complex to maintain:
```
import struct
from asyncua.ua import ua
@uamethod
async def api_work_order_add(self, parent, order_code: str, data):
try:
# Stampa informazioni di debug
print("ExtensionObject TypeId:", data.TypeId)
print("Body hex:", data.Body.hex())
print("Body length:", len(data.Body))
# Punto di partenza per la lettura
offset = 0
# Funzione di supporto per leggere stringhe
def read_string(body, start):
# Legge la lunghezza della stringa (4 byte)
str_len = struct.unpack('<i', body[start:start+4])[0]
start += 4
# Se la lunghezza è -1, è una stringa vuota
if str_len == -1:
return '', start
# Legge la stringa
string_value = body[start:start+str_len].decode('utf-8')
return string_value, start + str_len
# Lettura campi booleani e byte
order_locked = data.Body[offset] == 1
offset += 1
order_priority = data.Body[offset]
offset += 1
# Lettura stringhe
job_order_code, offset = read_string(data.Body, offset)
customer_code, offset = read_string(data.Body, offset)
item_code, offset = read_string(data.Body, offset)
material_code, offset = read_string(data.Body, offset)
order_notes, offset = read_string(data.Body, offset)
# Lettura booleano deadline
used_deadline_datetime = data.Body[offset] == 1
offset += 1
# Lettura datetime (64-bit intero che rappresenta i tick di Windows)
deadline_datetime_raw = struct.unpack('<Q', data.Body[offset:offset+8])[0]
# Converti i tick di Windows in datetime di Python se necessario
deadline_datetime = ua.datetime_from_win_filetime(deadline_datetime_raw) if deadline_datetime_raw != 0 else None
offset += 8
# Lettura dettagli file
file_details = []
for _ in range(8):
file_name, offset = read_string(data.Body, offset)
pieces_per_file = struct.unpack('<i', data.Body[offset:offset+4])[0]
offset += 4
requested_pieces = struct.unpack('<i', data.Body[offset:offset+4])[0]
offset += 4
file_details.append({
'file_name': file_name,
'pieces_per_file': pieces_per_file,
'requested_pieces': requested_pieces
})
# Stampa i dati decodificati
print("Order Locked:", order_locked)
print("Order Priority:", order_priority)
print("Job Order Code:", job_order_code)
print("Customer Code:", customer_code)
print("Item Code:", item_code)
print("Material Code:", material_code)
print("Order Notes:", order_notes)
print("Used Deadline:", used_deadline_datetime)
print("Deadline Datetime:", deadline_datetime)
# Stampa dettagli dei file
for i, file_info in enumerate(file_details, 1):
if file_info['file_name']:
print(f"File {i}:")
print(f" Name: {file_info['file_name']}")
print(f" Pieces per File: {file_info['pieces_per_file']}")
print(f" Requested Pieces: {file_info['requested_pieces']}")
return True
except Exception as e:
print(f"Errore nella decodifica dei dati: {e}")
import traceback
traceback.print_exc()
return False
```
There is some better and ready to use way to get info from "data" using library functions ?
| How to get data from a new_struct datatype received in a @uamethod | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1756/comments | 1 | 2024-12-04T15:08:30Z | 2024-12-04T16:42:54Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1756 | 2,718,045,190 | 1,756 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
When subscribing to a node component on an OPCUA server the SourceTimeStamp is identical to the ServerTimeStamp. When viewing the data with UaExpert or similar software these two are different. The same goes for logging these values with these softwares.
**To Reproduce**<br />
* Setup a server with a node that recieves data with a source timestamp (Double data type) in batch. The time difference between each data point is 2 minutes while the transmission frequency is 1 hour (so every hour 30 values are sent).
* Subscribe to the node using the example subscribe client supplied here (with username and password in my case)
* View the logged output.
The used code
```
#%%
import asyncio
import logging
from asyncua import Client, Node, ua
logging.basicConfig(level=logging.INFO)
_logger = logging.getLogger("asyncua")
class SubscriptionHandler:
"""
The SubscriptionHandler is used to handle the data that is received for the subscription.
"""
def datachange_notification(self, node: Node, val, data):
"""
Callback for asyncua Subscription.
This method will be called when the Client received a data change message from the Server.
"""
server_timestamp = data.monitored_item.Value.ServerTimestamp
source_timestamp = data.monitored_item.Value.SourceTimestamp
print(f"Server Timestamp: {server_timestamp}")
print(f"Source Timestamp: {source_timestamp}")
_logger.info("datachange_notification %r %s", node, val)
async def main(url,username,password):
"""
Main task of this Client-Subscription example.
"""
node_addresses = ["ns=2;s=DL_01.NI_0094.HValue"]
client = Client(url)
client.set_user(username)
client.set_password(password)
async with client:
nodes = [client.get_node(address) for address in node_addresses]
var = nodes[0]
handler = SubscriptionHandler()
# We create a Client Subscription.
subscription = await client.create_subscription(500, handler)
nodes = [
var,
client.get_node(ua.ObjectIds.Server_ServerStatus_CurrentTime),
]
# We subscribe to data changes for two nodes (variables).
await subscription.subscribe_data_change(nodes)
# We let the subscription run for ten seconds
await asyncio.sleep(4)
# We delete the subscription (this un-subscribes from the data changes of the two variables).
# This is optional since closing the connection will also delete all subscriptions.
await subscription.delete()
# After one second we exit the Client context manager - this will close the connection.
await asyncio.sleep(1)
#%%
if __name__ == "__main__":
url = "opc.tcp://servername:4840"
username = "*******"
password = "*******"
await main(url,username,password)
```
Some example output from my testing
```
INFO:asyncua.common.subscription:Publish callback called with result: PublishResult(SubscriptionId=3367183, AvailableSequenceNumbers=[7], MoreNotifications=False, NotificationMessage_=NotificationMessage(SequenceNumber=7,
PublishTime=datetime.datetime(2024, 12, 3, 10, 53, 0, 834848, tzinfo=datetime.timezone.utc), NotificationData=[DataChangeNotification(MonitoredItems=[MonitoredItemNotification(ClientHandle=202, Value=DataValue(Value=Variant(
Value=datetime.datetime(2024, 12, 3, 10, 53, 0, 802766, tzinfo=datetime.timezone.utc), VariantType=<VariantType.DateTime: 13>, Dimensions=None, is_array=False), StatusCode_=StatusCode(value=0),
SourceTimestamp=datetime.datetime(2024, 12, 3, 10, 53, 0, 802766, tzinfo=datetime.timezone.utc),
ServerTimestamp=datetime.datetime(2024, 12, 3, 10, 53, 0, 802766, tzinfo=datetime.timezone.utc), SourcePicoseconds=None, ServerPicoseconds=None))], DiagnosticInfos=[])]), Results=[StatusCode(value=0)], DiagnosticInfos=[])
INFO:asyncua:datachange_notification Node(NodeId(Identifier=2258, NamespaceIndex=0, NodeIdType=<NodeIdType.FourByte: 1>))
2024-12-03 10:53:00.802766+00:00 2024-12-03 10:53:00.802766+00:00
```
**Expected behavior**<br />
The data recieved by the server has the Source timestamps with 2 minute difference, but the recieved data with the python client shows the SourceTimeStamp as identical with the ServerTimeStamp.
**Screenshots**<br />
The data logged using UaExports data logger view shows the logged timestamps being different (expected behavior).
Printout of the python opcua clients subscribe timestamps.
**Version**<br />
Python-Version:<br />
3.11.7
opcua-asyncio Version (e.g. master branch, 0.9):
asyncua == 1.1.5
| SourceTimeStamp is identical to ServerTimeStamp | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1754/comments | 1 | 2024-12-03T11:54:49Z | 2024-12-03T12:47:08Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1754 | 2,714,815,068 | 1,754 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Why we are unable to get data change when nodes get inactive and active again however it handle doesn't get change and also it is present in monitored items dictionary? | Unable to get data change notification | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1751/comments | 0 | 2024-11-21T08:44:28Z | 2024-11-21T08:44:28Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1751 | 2,678,563,887 | 1,751 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
Unable to get client subscribe to work.
**To Reproduce**<br />
See code.… [1]
**Expected behavior**<br />
If Data is changed on the server the client should be notified via the subscription.
**Version**<br />
Python-Version: 3.11.9<br />
opcua-asyncio Version (e.g. master branch, 0.9): 1.1.5
[1]
```python
# server.py
import asyncio
import logging
from asyncua import Server, Client, Node
from datetime import datetime
class OPCUAServer:
def __init__(self, endpoint: str = "opc.tcp://0.0.0.0:4840/freeopcua/server/"):
self.endpoint = endpoint
self.server = Server()
self.trigger_var = None
self.part_id_var = None
self.namespace_idx = None
async def init(self):
await self.server.init()
self.server.set_endpoint(self.endpoint)
# Set server name
await self.server.set_application_uri("urn:example:opcua:server")
# Get Objects node
objects = self.server.get_objects_node()
# Create custom namespace
uri = "http://examples.freeopcua.github.io"
self.namespace_idx = await self.server.register_namespace(uri)
# Load XML nodeset
await self.server.import_xml("UA_NodeSet.xml")
# Create a new object
myobj = await objects.add_object(self.namespace_idx, "Process")
# Create variables
self.trigger_var = await myobj.add_variable(
self.namespace_idx, "Trigger", False
)
self.part_id_var = await myobj.add_variable(self.namespace_idx, "PartID", "")
# Set variables writable
await self.trigger_var.set_writable()
await self.part_id_var.set_writable()
print(f"Server namespace index: {self.namespace_idx}")
print(f"Trigger node id: {self.trigger_var.nodeid}")
print(f"PartID node id: {self.part_id_var.nodeid}")
async def start(self):
async with self.server:
while True:
# Simulate trigger and part ID updates
current_time = datetime.now().strftime("%H:%M:%S")
await self.trigger_var.write_value(True)
name = (await self.trigger_var.read_browse_name()).Name
value = (await self.trigger_var.read_value())
print(f"{name} = {value}")
await self.part_id_var.write_value(f"PART_{current_time}")
name = (await self.part_id_var.read_browse_name()).Name
value = (await self.part_id_var.read_value())
print(f"{name} = {value}")
# Wait for 5 seconds before next update
await asyncio.sleep(5)
await self.trigger_var.write_value(False)
name = (await self.trigger_var.read_browse_name()).Name
value = (await self.trigger_var.read_value())
print(f"{name} = {value}")
await asyncio.sleep(5)
# client.py
class SubscriptionHandler:
def datachange_notification(self, node: Node, val, data):
try:
node_name = node
print(f"New value for {node_name}: {val} {data=}")
except Exception as e:
print(f"Error in notification handler: {e}")
class OPCUAClient:
def __init__(self, url: str = "opc.tcp://localhost:4840/freeopcua/server/"):
self.url = url
self.client = Client(url=self.url)
async def subscribe_to_variables(self):
async with self.client:
try:
# Find the namespace index
uri = "http://examples.freeopcua.github.io"
nsidx = await self.client.get_namespace_index(uri)
print(f"Client namespace index: {nsidx}")
# Get the Process node first
objects = self.client.get_objects_node()
process_node = await objects.get_child(f"{nsidx}:Process")
# Get variables using their browse paths
trigger_node = await process_node.get_child(f"{nsidx}:Trigger")
part_id_node = await process_node.get_child(f"{nsidx}:PartID")
print(f"Found trigger node: {trigger_node.nodeid}")
print(f"Found part_id node: {part_id_node.nodeid}")
# Create subscription
handler = SubscriptionHandler()
subscription = await self.client.create_subscription(100, handler=handler)
await subscription.subscribe_data_change([trigger_node, part_id_node], sampling_interval=0)
# Keep the client running
while True:
print("ZZzzzZZzzZ!")
await asyncio.sleep(1)
except Exception as e:
print(f"Error in client: {e}")
raise
# Example XML configuration (UA_NodeSet.xml)
XML_CONTENT = """<?xml version="1.0" encoding="utf-8"?>
<UANodeSet xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:uax="http://opcfoundation.org/UA/2008/02/Types.xsd" xmlns="http://opcfoundation.org/UA/2011/03/UANodeSet.xsd">
<NamespaceUris>
<Uri>http://examples.freeopcua.github.io</Uri>
</NamespaceUris>
<UAObject NodeId="ns=1;i=1" BrowseName="1:Process">
<DisplayName>Process</DisplayName>
<References>
<Reference ReferenceType="HasComponent" IsForward="false">i=85</Reference>
</References>
</UAObject>
<UAVariable NodeId="ns=1;i=2" BrowseName="1:Trigger" DataType="Boolean">
<DisplayName>Trigger</DisplayName>
<References>
<Reference ReferenceType="HasComponent" IsForward="false">ns=1;i=1</Reference>
</References>
</UAVariable>
<UAVariable NodeId="ns=1;i=3" BrowseName="1:PartID" DataType="String">
<DisplayName>PartID</DisplayName>
<References>
<Reference ReferenceType="HasComponent" IsForward="false">ns=1;i=1</Reference>
</References>
</UAVariable>
</UANodeSet>
"""
# main.py
async def main():
# Save XML configuration
with open("UA_NodeSet.xml", "w") as f:
f.write(XML_CONTENT)
# Create and start server
server = OPCUAServer()
await server.init()
# Create and start client
client = OPCUAClient()
# Run both server and client concurrently
await asyncio.gather(server.start(), client.subscribe_to_variables())
if __name__ == "__main__":
# logging.basicConfig(level=logging.INFO)
asyncio.run(main())
```
| Unable to get subscribe example working | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1746/comments | 2 | 2024-11-11T16:01:16Z | 2024-11-14T13:54:30Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1746 | 2,649,750,336 | 1,746 |
[
"FreeOpcUa",
"opcua-asyncio"
] | I took the very basic server example and changed the protocol from `opc.tcp` to `opc.http`. I want to deploy the server to heroku but that apparently does not support tcp endpoints and requires http/s.
This is the server:
```python
import asyncio
from asyncua import Server
async def main():
server = Server()
await server.init()
server.set_endpoint('opc.http://0.0.0.0:4840/')
uri = "http://examples.freeopcua.github.io"
idx = await server.register_namespace(uri)
myobj = await server.nodes.objects.add_object(idx, "MyObject")
myvar = await myobj.add_variable(idx, "MyVariable", 6.7)
await myvar.set_writable()
async with server:
while True:
await asyncio.sleep(1)
new_val = await myvar.get_value() + 0.1
await myvar.write_value(new_val)
if __name__ == '__main__':
print('starting server')
asyncio.run(main())
```
When connecting with a basic client:
```python
from asyncua import Client
url = "opc.http://0.0.0.0:4840/freeopcua/server/"
async with Client(url=url) as client:
_logger.info("Root node is: %r", client.nodes.root)
```
I'm getting this error:
```
[...]
INFO:asyncua.client.client:find_endpoint [EndpointDescription(EndpointUrl='opc.http://0.0.0.0:4840/', Server=ApplicationDescription(ApplicationUri='urn:freeopcua:python:server', ProductUri='urn:freeopcua.github.io:python:server', ApplicationName=LocalizedText(Locale=None, Text='FreeOpcUa Python Server'), ApplicationType_=<ApplicationType.ClientAndServer: 2>, GatewayServerUri=None, DiscoveryProfileUri=None, DiscoveryUrls=['opc.http://0.0.0.0:4840/']), ServerCertificate=None, SecurityMode=<MessageSecurityMode.None_: 1>, SecurityPolicyUri='http://opcfoundation.org/UA/SecurityPolicy#None', UserIdentityTokens=[UserTokenPolicy(PolicyId='anonymous', TokenType=<UserTokenType.Anonymous: 0>, IssuedTokenType=None, IssuerEndpointUrl=None, SecurityPolicyUri=None), UserTokenPolicy(PolicyId='certificate_basic256sha256', TokenType=<UserTokenType.Certificate: 2>, IssuedTokenType=None, IssuerEndpointUrl=None, SecurityPolicyUri=None), UserTokenPolicy(PolicyId='username', TokenType=<UserTokenType.UserName: 1>, IssuedTokenType=None, IssuerEndpointUrl=None, SecurityPolicyUri=None)], TransportProfileUri='http://opcfoundation.org/UA-Profile/Transport/uatcp-uasc-uabinary', SecurityLevel=0)] <MessageSecurityMode.None_: 1> 'http://opcfoundation.org/UA/SecurityPolicy#None'
INFO:asyncua.client.ua_client.UASocketProtocol:close_secure_channel
INFO:asyncua.client.ua_client.UASocketProtocol:Request to close socket received
INFO:asyncua.client.ua_client.UASocketProtocol:Socket has closed connection
Traceback (most recent call last):
File "cloud_demo_app/client/main.py", line 76, in <module>
asyncio.run(main())
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/asyncio/runners.py", line 44, in run
return loop.run_until_complete(main)
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/asyncio/base_events.py", line 642, in run_until_complete
return future.result()
File "cloud_demo_app/client/main.py", line 26, in main
async with Client(url=url) as client:
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/asyncua/client/client.py", line 95, in __aenter__
await self.connect()
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/asyncua/client/client.py", line 311, in connect
await self.create_session()
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/asyncua/client/client.py", line 511, in create_session
ep = Client.find_endpoint(response.ServerEndpoints, self.security_policy.Mode, self.security_policy.URI)
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/asyncua/client/client.py", line 130, in find_endpoint
raise ua.UaError(f"No matching endpoints: {security_mode}, {policy_uri}")
asyncua.ua.uaerrors._base.UaError: No matching endpoints: 1, http://opcfoundation.org/UA/SecurityPolicy#None
```
Do you have a hint how to work with http over tcp? | Use http instead of tcp | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1745/comments | 0 | 2024-11-11T09:08:23Z | 2024-11-11T09:08:36Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1745 | 2,648,662,914 | 1,745 |
[
"FreeOpcUa",
"opcua-asyncio"
] | I took the very basic server example and changed the protocol from `opc.tcp` to `opc.http`. I want to deploy the server to heroku but that apparently does not support tcp endpoints and requires http/s.
This is the server:
```python
import asyncio
from asyncua import Server
async def main():
server = Server()
await server.init()
server.set_endpoint('opc.http://0.0.0.0:4840/freeopcua/server/')
uri = "http://examples.freeopcua.github.io"
idx = await server.register_namespace(uri)
myobj = await server.nodes.objects.add_object(idx, "MyObject")
myvar = await myobj.add_variable(idx, "MyVariable", 6.7)
await myvar.set_writable()
async with server:
while True:
await asyncio.sleep(1)
new_val = await myvar.get_value() + 0.1
await myvar.write_value(new_val)
if __name__ == '__main__':
print('starting server')
asyncio.run(main())
```
When connecting with a basic client:
```python
from asyncua import Client
url = "opc.http://0.0.0.0:4840/freeopcua/server/"
async with Client(url=url) as client:
_logger.info("Root node is: %r", client.nodes.root)
```
I'm getting this error:
```
[...]
INFO:asyncua.client.client:find_endpoint [EndpointDescription(EndpointUrl='opc.http://0.0.0.0:4840/', Server=ApplicationDescription(ApplicationUri='urn:freeopcua:python:server', ProductUri='urn:freeopcua.github.io:python:server', ApplicationName=LocalizedText(Locale=None, Text='FreeOpcUa Python Server'), ApplicationType_=<ApplicationType.ClientAndServer: 2>, GatewayServerUri=None, DiscoveryProfileUri=None, DiscoveryUrls=['opc.http://0.0.0.0:4840/']), ServerCertificate=None, SecurityMode=<MessageSecurityMode.None_: 1>, SecurityPolicyUri='http://opcfoundation.org/UA/SecurityPolicy#None', UserIdentityTokens=[UserTokenPolicy(PolicyId='anonymous', TokenType=<UserTokenType.Anonymous: 0>, IssuedTokenType=None, IssuerEndpointUrl=None, SecurityPolicyUri=None), UserTokenPolicy(PolicyId='certificate_basic256sha256', TokenType=<UserTokenType.Certificate: 2>, IssuedTokenType=None, IssuerEndpointUrl=None, SecurityPolicyUri=None), UserTokenPolicy(PolicyId='username', TokenType=<UserTokenType.UserName: 1>, IssuedTokenType=None, IssuerEndpointUrl=None, SecurityPolicyUri=None)], TransportProfileUri='http://opcfoundation.org/UA-Profile/Transport/uatcp-uasc-uabinary', SecurityLevel=0)] <MessageSecurityMode.None_: 1> 'http://opcfoundation.org/UA/SecurityPolicy#None'
INFO:asyncua.client.ua_client.UASocketProtocol:close_secure_channel
INFO:asyncua.client.ua_client.UASocketProtocol:Request to close socket received
INFO:asyncua.client.ua_client.UASocketProtocol:Socket has closed connection
Traceback (most recent call last):
File "cloud_demo_app/client/main.py", line 76, in <module>
asyncio.run(main())
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/asyncio/runners.py", line 44, in run
return loop.run_until_complete(main)
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/asyncio/base_events.py", line 642, in run_until_complete
return future.result()
File "cloud_demo_app/client/main.py", line 26, in main
async with Client(url=url) as client:
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/asyncua/client/client.py", line 95, in __aenter__
await self.connect()
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/asyncua/client/client.py", line 311, in connect
await self.create_session()
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/asyncua/client/client.py", line 511, in create_session
ep = Client.find_endpoint(response.ServerEndpoints, self.security_policy.Mode, self.security_policy.URI)
File "/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/asyncua/client/client.py", line 130, in find_endpoint
raise ua.UaError(f"No matching endpoints: {security_mode}, {policy_uri}")
asyncua.ua.uaerrors._base.UaError: No matching endpoints: 1, http://opcfoundation.org/UA/SecurityPolicy#None
```
Do you have a hint how to work with http over tcp? | Use http instead of tcp | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1744/comments | 2 | 2024-11-11T09:08:22Z | 2024-11-11T12:26:58Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1744 | 2,648,662,883 | 1,744 |
[
"FreeOpcUa",
"opcua-asyncio"
] | README.md has this command to run to run tests:
`python -m pip install -r dev_requirements.txt`
However dev_requirements.txt is not in repo. | dev_requirements.py is missing | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1740/comments | 2 | 2024-10-29T16:39:03Z | 2024-10-31T09:42:10Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1740 | 2,621,795,494 | 1,740 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Description** <br />
I'm working with a nodeset generated by CODESYS for a PLC, which has been causing several issues. While most seem to stem from the CODESYS side, one issue appears to be an unnecessary limitation in the Asyncua library itself. Specifically, the nodeset generated uses string-based identifiers for ExtensionObjects, as shown in the abbreviated XML example below. When exporting the nodeset to XML, the check for integer identifiers in <code>xmlexporter.py/_value_to_etree()</code> raises an UaInvalidParameterError:
```xml
<UAVariable/>
<DisplayName/>
<Description/>
<References/>
<Value>
<uax:ExtensionObject>
<uax:TypeId>
<uax:Identifier>ns=3;s=|type|My_String_Identifier</uax:Identifier>
</uax:TypeId>
<uax:Body/>
</uax:ExtensionObject>
</Value>
</UAVariable>
```
From UA Part 6: Mappings (V. 1.05.03) / section 5.2.2.15:
> "Server implementers should use namespace-qualified numeric NodeIds for any DataTypeEncoding Objects they define. This will minimize the overhead introduced by packing Structured DataType values into an ExtensionObject."
My understanding of this part of the standard is that it recommends—but does not strictly require—the use of numeric NodeIds for ExtensionObjects. The standard only mandates numeric identifiers explicitly for ExtensionObjects in the UA namespace (section 5.2.2.15). Currently, my nodeset exports correctly if I disable the numeric identifier check.
**To Reproduce**<br />
Unfortunately I can't provide the complete nodeset, but any ExtensionObject with a String TypeId should trigger the behavior.
**Expected behavior**<br />
Allow String IDs for types of Extension objects.
**Screenshots**<br />
UAExpert doesn't have any problems with String IDs for types of ExtensionObjects
**Versions**<br />
Python-Version: 3.12.3<br />
opcua-asyncio Version: 1.1.5 (installed via Anaconda)
| ExtensionObjects with String TypeIds | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1739/comments | 0 | 2024-10-29T09:03:51Z | 2024-10-29T09:07:11Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1739 | 2,620,587,743 | 1,739 |
[
"FreeOpcUa",
"opcua-asyncio"
] | Hi,
I tried starting `./examples/server_minimal.py` and got a wall of INFO about not existing nodes.
I tried starting it on current version from master and version 1.1.5. I tried starting it on Debian 12 and Ubuntu 22.04 - got the same result. Using Python 3.10
`INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=11715, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15958, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15959, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15960, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15961, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15962, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15963, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=15964, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=16134, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=16135, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists
INFO:asyncua.server.address_space:add_node: while adding node NumericNodeId(Identifier=16136, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>), requested parent node NumericNodeId(Identifier=15957, NamespaceIndex=0, NodeIdType=<NodeIdType.Numeric: 2>) does not exists`
| Requested parent node does not exist in server-minimal.py | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1733/comments | 2 | 2024-10-25T11:06:25Z | 2025-06-16T09:09:59Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1733 | 2,613,794,137 | 1,733 |
[
"FreeOpcUa",
"opcua-asyncio"
] | **Describe the bug** <br />
Initializing a server with server.init(Path('cache_file')) will create a cache_file.db in the same directory as the script has been executed. Running the script the second time will recreate the same cache file.
**Solution** <br />
In the internal_server.py file the load_standard_address_space function must be modified, that the Path(...).with_suffix('.db') can be recognized.
Current code snippet:
```
async def load_standard_address_space(self, shelf_file: Optional[Path] = None):
if shelf_file:
is_file = await asyncio.get_running_loop().run_in_executor(
None, Path.is_file, shelf_file
) or await asyncio.get_running_loop().run_in_executor(None, Path.is_file, shelf_file.with_suffix('.db'))
```
Working code snippet:
```
async def load_standard_address_space(self, shelf_file: Optional[Path] = None):
if shelf_file:
is_file = await asyncio.get_running_loop().run_in_executor(
None, Path.is_file, shelf_file
) or await asyncio.get_running_loop().run_in_executor(None, Path.is_file, shelf_file.with_suffix('.db'))
```
**To Reproduce**<br />
Steps to reproduce the behavior incl code:
Calling this script:
```
import asyncio
from pyinstrument import Profiler
from asyncua import Server
from pathlib import Path
async def opcua_startup():
server = Server()
await server.init(Path('cache_file'))
print("Server initialized")
server.set_endpoint('opc.tcp://localhost:4840/freeopcua/server/')
server.set_server_name('FreeOpcUa Example Server')
await server.start()
print("Server started")
return True
async def main():
p = Profiler()
with p:
await opcua_startup()
p.print()
if __name__ == '__main__':
asyncio.run(main())
```
**Expected behavior**<br />
A clear and concise description of what you expected to happen.
If this script would be called twice. The cache file should be catched and used to be much more faster
**Version**<br />
Python-Version:3.9<br />
opcua-asyncio Version: 1.1.5
Hardware: RevolutionPi / RaspberryPi
| Could not load shelf file on server init if it already exists | https://api.github.com/repos/FreeOpcUa/opcua-asyncio/issues/1732/comments | 0 | 2024-10-25T08:34:29Z | 2024-10-25T08:36:15Z | https://github.com/FreeOpcUa/opcua-asyncio/issues/1732 | 2,613,457,147 | 1,732 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.