[BUG] EnaSearch failing depending on search term ordering

[xapple]

[BUG] EnaSearch failing depending on search term ordering

[saketkc]

Thanks for catching this! I can confirm that this is a bug.
cc @bscrow

[bscrow]

I could not reproduce this bug in Python 3.7 (see colab example) where the query "cancer human" returns no search results with a message as expected:

No results found for the following search query: {'query': 'CANCER HUMAN', 'accession': None, 'organism': None, 'layout': None, 'mbases': None, 'publication_date': None, 'platform': None, 'selection': None, 'source': None, 'strategy': None, 'title': None}

The traceback suggests an error when calling r.json() on an empty query result. Can I check what version of requests library are you using?

[xapple]

Version '2.27.1' which is the latest at time of writing. But I'm on python 3.9.9, so I'm going to guess something changed with the built-in json module between versions.

[saketkc]

I just tried this with Python 3.10.2 and requests 2.27.1 and was unable to replicate as well:

$ python
Python 3.10.2 | packaged by conda-forge | (main, Feb  1 2022, 19:29:00) [GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from pysradb.search import EnaSearch
>>> instance = EnaSearch(query="cancer human")
>>> instance.search()
No results found for the following search query:
 {'query': 'CANCER HUMAN', 'accession': None, 'organism': None, 'layout': None, 'mbases': None, 'publication_date': None, 'platform': None, 'selection': None, 'source': None, 'strategy': None, 'title': None}
>>> df = instance.get_df()
>>> print(df)
Empty DataFrame
Columns: []
Index: []

[saketkc]

Can you try again @xapple? Not sure if it might be due to an intermittent connectivity issue?

[xapple]

I installed and upgraded the latest pythons on my computer:

• Python 3.9.10
• Python 3.10.2

I can confirm that the problem is present only in 3.9 series and not in 3.10

[saketkc]

Indeed, we see this on the tests also: https://github.com/saketkc/pysradb/runs/5465673995?check_suite_focus=true#step:7:76. But looks like a recent problem (like a couple of weeks old), since the earlier tests pass with 3.9: https://github.com/saketkc/pysradb/runs/5309144937?check_suite_focus=true