make `ls` more resilient in case of metadata loading fails with pynwb for some reason #282

t-b · 2020-11-24T00:31:27Z

Follow up from NeurodataWithoutBorders/pynwb#1321.

Can someone explain why HardwareTests-V2-IP8.nwb is good with

$ dandi --log-level DEBUG ls HardwareTests-V2-IP8.nwb
2020-11-24 01:20:38,661 [   DEBUG] No newer (than 0.7.2) version of dandi/dandi-cli found available
2020-11-24 01:20:41,977 [   DEBUG] Calling memoized version of <function get_metadata at 0x0000000008A4A678> for E:\projekte\mies-igor\tools\unit-testing\HardwareTests-V2-IP8.nwb
- age: null
  date_of_birth: null
  experiment_description: null
  experimenter: null
  genotype: null
  identifier: ac24acc942a5b87538bf15d140e06b4576481565b77b114877c4d26ba23fc09e
  institution: null
  keywords: null
  lab: null
  nd_types:
  - Device (7)
  - IntracellularElectrode (6)
  - LabNotebook
  - LabNotebookDevice
  - StimulusSets
  - Subject
  - SweepTable
  - Testpulse
  - TestpulseDevice
  - TimeSeries (4)
  - UserComment
  - UserCommentDevice
  - VoltageClampSeries (2)
  - VoltageClampStimulusSeries (2)
  number_of_electrodes: 0
  number_of_units: 0
  nwb_version: 2.2.4
  path: HardwareTests-V2-IP8.nwb
  related_publications: null
  session_description: PLACEHOLDER
  session_id: null
  session_start_time: 2020-11-21 20:42:02.721000+00:00
  sex: null
  size: 9180392
  species: null
  subject_id: null
(

but bad.nwb gives

$ dandi --log-level DEBUG ls bad.nwb
2020-11-24 12:18:13,126 [   DEBUG] No newer (than 0.7.2) version of dandi/dandi-cli found available
2020-11-24 12:18:17,903 [   DEBUG] Calling memoized version of <function get_metadata at 0x0000000008A49678> for E:\projekte\mies-igor\tools\unit-testing\bad.nwb
2020-11-24 12:18:17,917 [   DEBUG] Running original <function get_metadata at 0x0000000008A49678> on 'E:\\projekte\\mies-igor\\tools\\unit-testing\\bad.nwb'
2020-11-24 12:18:20,087 [   DEBUG] Call to get_metadata on bad.nwb failed: Could not construct IntracellularElectrode object due to: IntracellularElectrode.__init__: incorrect type for 'description' (got 'bytes', expected 'str')
2020-11-24 12:18:20,087 [   DEBUG] Problem obtaining metadata for bad.nwb: 'NoneType' object has no attribute 'items'
2020-11-24 12:18:20,087 [ WARNING] Failed to operate on some paths (empty records were listed):
 AttributeError: 1 paths
- path: bad.nwb
  size: 31985390

Both have the exact same HDF5 type

E:\projekte\mies-igor\tools\unit-testing>h5dump -d "/general/intracellular_ephys/electrode_0/description" bad.nwb HardwareTests-V2-IP8.nwb
HDF5 "bad.nwb" {
DATASET "/general/intracellular_ephys/electrode_0/description" {
   DATATYPE  H5T_STRING {
      STRSIZE H5T_VARIABLE;
      STRPAD H5T_STR_NULLTERM;
      CSET H5T_CSET_UTF8;
      CTYPE H5T_C_S1;
   }
   DATASPACE  SIMPLE { ( 1 ) / ( 1 ) }
   DATA {
   (0): "Headstage 0"
   }
}
}
HDF5 "HardwareTests-V2-IP8.nwb" {
DATASET "/general/intracellular_ephys/electrode_0/description" {
   DATATYPE  H5T_STRING {
      STRSIZE H5T_VARIABLE;
      STRPAD H5T_STR_NULLTERM;
      CSET H5T_CSET_UTF8;
      CTYPE H5T_C_S1;
   }
   DATASPACE  SIMPLE { ( 1 ) / ( 1 ) }
   DATA {
   (0): "Headstage 0"
   }
}
}

files.zip

The text was updated successfully, but these errors were encountered:

yarikoptic · 2020-11-24T01:54:19Z

Wrong second paste? (Looks like hdf5dump and no error in any of those pasted)

t-b · 2020-11-24T11:18:53Z

@yarikoptic Thanks. Fixed.

yarikoptic · 2020-11-24T16:26:32Z

I cannot tell what is wrong, and most likely an underlying pynwb issue, not dandi (but I could be wrong) for

Could not construct IntracellularElectrode object due to: IntracellularElectrode.__init__: incorrect type for 'description' (got 'bytes', expected 'str')

but you could try with env variable DANDI_CACHE=ignore to see if that is not somehow due to serialization during caching of the results (unlikely but possible).

Then there is a dandi ls issue with

Problem obtaining metadata for bad.nwb: 'NoneType' object has no attribute 'items'

is due to the failure above, I will probably (yet deciding on either it is worth it at this stage -- we are yet to do big RF #249 TODO which would touch on it more harmoniously) submit a PR shortly to handle it a bit better.

And that bad.nwb - it is not the one from https://github.com/NeurodataWithoutBorders/pynwb/files/5578038/HardwareTests-V2-MIES-187d9143.zip , is it available to look "deeper"?

t-b · 2020-11-24T17:38:47Z

@yarikoptic I have attached files.zip to the issue of course.

t-b · 2020-11-24T17:39:55Z

Same error with DANDI_CACHE=ignore.

yarikoptic · 2020-11-25T15:13:40Z

eh, sorry -- missed that link.

for me both files ls fine with pynwb 1.4.0.post.dev3 "installed" straight from upstream repo at 1.4.0-3-g6d6ea6c9 (hdmf was 2.1.0 and then upgraded to 2.2.0) at NeurodataWithoutBorders/pynwb#1282

output

$> dandi  ls /tmp/bad.nwb /tmp/HardwareTests-V2-IP8.nwb      
PATH                          SIZE    IDENTIFIER                                                       NWB   ND_TYPES                                                                                                      SESSION_DESCRIPTION SESSION_START_TIME  
/tmp/bad.nwb                  32.0 MB 2ae7afd1a09f78c3d7c3311d71990095010fab706d91f9048986eef429991a70 2.2.4 CurrentClampSeries (73), CurrentClampStimulusSeries (73), Device (148), IntracellularElectrode (147), LabN... PLACEHOLDER         2019-11-08/18:46:09 
/tmp/HardwareTests-V2-IP8.nwb 9.2 MB  ac24acc942a5b87538bf15d140e06b4576481565b77b114877c4d26ba23fc09e 2.2.4 Device (7), IntracellularElectrode (6), LabNotebook, LabNotebookDevice, StimulusSets, Subject, SweepTable,... PLACEHOLDER         2020-11-21/20:42:02 
Summary:                      41.2 MB                                                                                                                                                                                                          2019-11-08/18:46:09>

so I then did pip install --upgrade --force-reinstall pynwb==1.4.0

which pulled in everything including a kitchensink

(git)lena:~/proj/dandi/dandi-cli[tags/0.7.2^0]
$> pip install --upgrade --force-reinstall pynwb==1.4.0
Collecting pynwb==1.4.0
  Using cached pynwb-1.4.0-py2.py3-none-any.whl (94 kB)
Collecting numpy>=1.16
  Downloading numpy-1.19.4-cp38-cp38-manylinux2010_x86_64.whl (14.5 MB)
     |████████████████████████████████| 14.5 MB 4.9 MB/s 
Collecting h5py>=2.9
  Downloading h5py-3.1.0-cp38-cp38-manylinux1_x86_64.whl (4.4 MB)
     |████████████████████████████████| 4.4 MB 5.9 MB/s 
Collecting pandas>=0.23
  Downloading pandas-1.1.4-cp38-cp38-manylinux1_x86_64.whl (9.3 MB)
     |████████████████████████████████| 9.3 MB 5.3 MB/s 
Collecting hdmf<3,>=2.1.0
  Using cached hdmf-2.2.0-py2.py3-none-any.whl (132 kB)
Collecting python-dateutil>=2.7
  Using cached python_dateutil-2.8.1-py2.py3-none-any.whl (227 kB)
Collecting pytz>=2017.2
  Downloading pytz-2020.4-py2.py3-none-any.whl (509 kB)
     |████████████████████████████████| 509 kB 5.9 MB/s 
Collecting ruamel.yaml>=0.15
  Using cached ruamel.yaml-0.16.12-py2.py3-none-any.whl (111 kB)
Collecting scipy>=1.1
  Downloading scipy-1.5.4-cp38-cp38-manylinux1_x86_64.whl (25.8 MB)
     |████████████████████████████████| 25.8 MB 5.6 MB/s 
Collecting jsonschema>=2.6.0
  Using cached jsonschema-3.2.0-py2.py3-none-any.whl (56 kB)
Collecting six>=1.5
  Using cached six-1.15.0-py2.py3-none-any.whl (10 kB)
Collecting ruamel.yaml.clib>=0.1.2; platform_python_implementation == "CPython" and python_version < "3.9"
  Using cached ruamel.yaml.clib-0.2.2-cp38-cp38-manylinux1_x86_64.whl (578 kB)
Collecting setuptools
  Using cached setuptools-50.3.2-py3-none-any.whl (785 kB)
Collecting pyrsistent>=0.14.0
  Downloading pyrsistent-0.17.3.tar.gz (106 kB)
     |████████████████████████████████| 106 kB 7.3 MB/s 
Collecting attrs>=17.4.0
  Downloading attrs-20.3.0-py2.py3-none-any.whl (49 kB)
     |████████████████████████████████| 49 kB 3.2 MB/s 
Building wheels for collected packages: pyrsistent
  Building wheel for pyrsistent (setup.py) ... done
  Created wheel for pyrsistent: filename=pyrsistent-0.17.3-cp38-cp38-linux_x86_64.whl size=108030 sha256=35d9b62b6ae1719f89cdc792f8a5a1a519fc6b56cab22e086e3fadaebe431c40
  Stored in directory: /home/yoh/.cache/pip/wheels/3d/22/08/7042eb6309c650c7b53615d5df5cc61f1ea9680e7edd3a08d2
Successfully built pyrsistent
ERROR: launchpadlib 1.10.13 requires testresources, which is not installed.
Installing collected packages: numpy, h5py, six, python-dateutil, pytz, pandas, ruamel.yaml.clib, ruamel.yaml, scipy, setuptools, pyrsistent, attrs, jsonschema, hdmf, pynwb
  Attempting uninstall: numpy
    Found existing installation: numpy 1.19.3
    Not uninstalling numpy at /usr/lib/python3/dist-packages, outside environment /home/yoh/proj/dandi/dandi-cli/venvs/dev3
    Can't uninstall 'numpy'. No files were found to uninstall.
  Attempting uninstall: h5py
    Found existing installation: h5py 2.10.0
    Not uninstalling h5py at /usr/lib/python3/dist-packages, outside environment /home/yoh/proj/dandi/dandi-cli/venvs/dev3
    Can't uninstall 'h5py'. No files were found to uninstall.
  Attempting uninstall: six
    Found existing installation: six 1.14.0
    Uninstalling six-1.14.0:
      Successfully uninstalled six-1.14.0
  Attempting uninstall: python-dateutil
    Found existing installation: python-dateutil 2.8.1
    Not uninstalling python-dateutil at /usr/lib/python3/dist-packages, outside environment /home/yoh/proj/dandi/dandi-cli/venvs/dev3
    Can't uninstall 'python-dateutil'. No files were found to uninstall.
  Attempting uninstall: pytz
    Found existing installation: pytz 2020.1
    Not uninstalling pytz at /usr/lib/python3/dist-packages, outside environment /home/yoh/proj/dandi/dandi-cli/venvs/dev3
    Can't uninstall 'pytz'. No files were found to uninstall.
  Attempting uninstall: pandas
    Found existing installation: pandas 1.1.3
    Not uninstalling pandas at /usr/lib/python3/dist-packages, outside environment /home/yoh/proj/dandi/dandi-cli/venvs/dev3
    Can't uninstall 'pandas'. No files were found to uninstall.
  Attempting uninstall: ruamel.yaml.clib
    Found existing installation: ruamel.yaml.clib 0.2.2
    Not uninstalling ruamel.yaml.clib at /usr/lib/python3/dist-packages, outside environment /home/yoh/proj/dandi/dandi-cli/venvs/dev3
    Can't uninstall 'ruamel.yaml.clib'. No files were found to uninstall.
  Attempting uninstall: ruamel.yaml
    Found existing installation: ruamel.yaml 0.16.12
    Not uninstalling ruamel.yaml at /usr/lib/python3/dist-packages, outside environment /home/yoh/proj/dandi/dandi-cli/venvs/dev3
    Can't uninstall 'ruamel.yaml'. No files were found to uninstall.
  Attempting uninstall: scipy
    Found existing installation: scipy 1.5.2
    Not uninstalling scipy at /usr/lib/python3/dist-packages, outside environment /home/yoh/proj/dandi/dandi-cli/venvs/dev3
    Can't uninstall 'scipy'. No files were found to uninstall.
  Attempting uninstall: setuptools
    Found existing installation: setuptools 44.0.0
    Uninstalling setuptools-44.0.0:
      Successfully uninstalled setuptools-44.0.0
  Attempting uninstall: pyrsistent
    Found existing installation: pyrsistent 0.15.5
    Not uninstalling pyrsistent at /usr/lib/python3/dist-packages, outside environment /home/yoh/proj/dandi/dandi-cli/venvs/dev3
    Can't uninstall 'pyrsistent'. No files were found to uninstall.
  Attempting uninstall: attrs
    Found existing installation: attrs 19.3.0
    Not uninstalling attrs at /usr/lib/python3/dist-packages, outside environment /home/yoh/proj/dandi/dandi-cli/venvs/dev3
    Can't uninstall 'attrs'. No files were found to uninstall.
  Attempting uninstall: jsonschema
    Found existing installation: jsonschema 3.2.0
    Uninstalling jsonschema-3.2.0:
      Successfully uninstalled jsonschema-3.2.0
  Attempting uninstall: hdmf
    Found existing installation: hdmf 2.2.0
    Uninstalling hdmf-2.2.0:
      Successfully uninstalled hdmf-2.2.0
  Attempting uninstall: pynwb
    Found existing installation: pynwb 1.4.0
    Uninstalling pynwb-1.4.0:
      Successfully uninstalled pynwb-1.4.0
Successfully installed attrs-20.3.0 h5py-3.1.0 hdmf-2.2.0 jsonschema-3.2.0 numpy-1.19.4 pandas-1.1.4 pynwb-1.4.0 pyrsistent-0.17.3 python-dateutil-2.8.1 pytz-2020.4 ruamel.yaml-0.16.12 ruamel.yaml.clib-0.2.2 scipy-1.5.4 setuptools-50.3.2 six-1.15.0
pip install --upgrade --force-reinstall pynwb==1.4.0  14.34s user 2.57s system 68% cpu 24.512 total

and now it doesn't work on either of those:

$> DANDI_CACHE=ignore dandi -l debug ls -f yaml /tmp/bad.nwb /tmp/HardwareTests-V2-IP8.nwb
2020-11-25 10:11:05,197 [   DEBUG] No newer (than 0.7.2) version of dandi/dandi-cli found available
2020-11-25 10:11:06,412 [   DEBUG] Calling memoized version of <function get_metadata at 0x7f2119214550> for /tmp/bad.nwb
2020-11-25 10:11:06,412 [   DEBUG] Running original <function get_metadata at 0x7f2119214550> on '/tmp/bad.nwb'
2020-11-25 10:11:07,503 [   DEBUG] Call to get_metadata on /tmp/bad.nwb failed: Could not construct IntracellularElectrode object due to: IntracellularElectrode.__init__: incorrect type for 'description' (got 'bytes', expected 'str')
2020-11-25 10:11:07,503 [   DEBUG] Problem obtaining metadata for /tmp/bad.nwb: 'NoneType' object has no attribute 'items'
2020-11-25 10:11:07,503 [   DEBUG] Calling memoized version of <function get_metadata at 0x7f2119214550> for /tmp/HardwareTests-V2-IP8.nwb
2020-11-25 10:11:07,504 [   DEBUG] Running original <function get_metadata at 0x7f2119214550> on '/tmp/HardwareTests-V2-IP8.nwb'
2020-11-25 10:11:08,175 [   DEBUG] Call to get_metadata on /tmp/HardwareTests-V2-IP8.nwb failed: Could not construct IntracellularElectrode object due to: IntracellularElectrode.__init__: incorrect type for 'description' (got 'bytes', expected 'str')
2020-11-25 10:11:08,175 [   DEBUG] Problem obtaining metadata for /tmp/HardwareTests-V2-IP8.nwb: 'NoneType' object has no attribute 'items'
2020-11-25 10:11:08,175 [ WARNING] Failed to operate on some paths (empty records were listed):
 AttributeError: 2 paths
- path: /tmp/bad.nwb
  size: 31985390
- path: /tmp/HardwareTests-V2-IP8.nwb
  size: 9180392

so at least there is some consistency ;) note that the upgrade pulled in h5py-3.1.0 which hdmf is not compatible with yet.

what h5py do you have ? (we need WTF #57)

yarikoptic · 2020-11-25T15:17:32Z

downgrade pip install --upgrade 'h5py==2.10.0' made it work again

output

$> DANDI_CACHE=ignore dandi ls  /tmp/bad.nwb /tmp/HardwareTests-V2-IP8.nwb     
PATH                          SIZE    IDENTIFIER                                                       SESSION_START_TIME   ND_TYPES                                                                                                      SESSION_DESCRIPTION NWB  
/tmp/bad.nwb                  32.0 MB 2ae7afd1a09f78c3d7c3311d71990095010fab706d91f9048986eef429991a70 2019-11-08/18:46:09  CurrentClampSeries (73), CurrentClampStimulusSeries (73), Device (148), IntracellularElectrode (147), LabN... PLACEHOLDER         2.2.4
/tmp/HardwareTests-V2-IP8.nwb 9.2 MB  ac24acc942a5b87538bf15d140e06b4576481565b77b114877c4d26ba23fc09e 2020-11-21/20:42:02  Device (7), IntracellularElectrode (6), LabNotebook, LabNotebookDevice, StimulusSets, Subject, SweepTable,... PLACEHOLDER         2.2.4
Summary:                      41.2 MB                                                                  2019-11-08/18:46:09>                                                                                                                                        
                                                                                                       2020-11-21/20:42:02<

h5py 3.0 introduced breaking changes (see e.g. #282) and hdmf yet to account for them. We better cache that version too to improve consistency of outputs etc

yarikoptic · 2020-11-25T15:59:21Z

So, nothing really to "fix" on dandi-cli side, besides improving that ls behavior in case of pynwb etc puking. I will retitle for that

t-b · 2020-11-26T20:07:19Z

Thanks for digging. Once I downgrade h5py to "2.10.0" it indeed works. I thought I checked that but looks like I got lost in tracking the various issues involved in that.

t-b mentioned this issue Nov 24, 2020

IVSCC export issues AllenInstitute/MIES#747

Closed

yarikoptic added a commit that referenced this issue Nov 25, 2020

BF: add h5py.__version__ into the list of tokens for caching

142e21e

h5py 3.0 introduced breaking changes (see e.g. #282) and hdmf yet to account for them. We better cache that version too to improve consistency of outputs etc

yarikoptic mentioned this issue Nov 25, 2020

BF: add h5py.__version__ into the list of tokens for caching #284

Merged

yarikoptic added the severity-minor label Nov 25, 2020

yarikoptic mentioned this issue Dec 1, 2020

ls - avoid workaround, more consistent reporting of errors #293

Merged

1 task

yarikoptic closed this as completed in #293 Dec 2, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

make `ls` more resilient in case of metadata loading fails with pynwb for some reason #282

make `ls` more resilient in case of metadata loading fails with pynwb for some reason #282

t-b commented Nov 24, 2020 •

edited

Loading

yarikoptic commented Nov 24, 2020

t-b commented Nov 24, 2020

yarikoptic commented Nov 24, 2020

t-b commented Nov 24, 2020

t-b commented Nov 24, 2020

yarikoptic commented Nov 25, 2020

yarikoptic commented Nov 25, 2020

yarikoptic commented Nov 25, 2020

t-b commented Nov 26, 2020

make ls more resilient in case of metadata loading fails with pynwb for some reason #282

make ls more resilient in case of metadata loading fails with pynwb for some reason #282

Comments

t-b commented Nov 24, 2020 • edited Loading

yarikoptic commented Nov 24, 2020

t-b commented Nov 24, 2020

yarikoptic commented Nov 24, 2020

t-b commented Nov 24, 2020

t-b commented Nov 24, 2020

yarikoptic commented Nov 25, 2020

yarikoptic commented Nov 25, 2020

yarikoptic commented Nov 25, 2020

t-b commented Nov 26, 2020

make `ls` more resilient in case of metadata loading fails with pynwb for some reason #282

make `ls` more resilient in case of metadata loading fails with pynwb for some reason #282

t-b commented Nov 24, 2020 •

edited

Loading