Default string output of TableSet with multiple layers of nesting throws AttributeError: 'TableSet' object has no attribute 'rows'

[vergenzt]

Example below. I encountered this while exploring a dataset in an ipdb debugging session (debugging a dbt macro). For some reason ipdb wasn't showing me the full stack traces for errors -- just the exception type and description.

So the only output I got looked like this:

ipdb> mytable.group_by('col1').group_by('col2')
*** AttributeError: 'TableSet' object has no attribute 'rows'
ipdb> _

making this issue more confusing than it may have otherwise been.

Note, now that I understand this is simply an issue with the string output on the REPL and not an issue with the underlying data (and I can still save it to disk), this issue does not block me.

Example to reproduce

Input:

>>> import agate
>>> mytable = agate.Table([('foo', 'FOO', 1), ('foo', 'FOO', 2), ('bar', 'BOZ', 1), ('bar', 'BAZ', 2), ('bar', 'BIZ')])
<stdin>:1: RuntimeWarning: Column names not specified. "('a', 'b', 'c')" will be used as names.

Actual string output:

>>> mytable.group_by('a').group_by('b')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File ".../.venv/lib/python3.8/site-packages/agate/mapped_sequence.py", line 87, in __repr__
    return self.__str__()
  File ".../.venv/lib/python3.8/site-packages/agate/tableset/__init__.py", line 91, in __str__
    self.print_structure(output=structure)
  File ".../.venv/lib/python3.8/site-packages/agate/tableset/print_structure.py", line 24, in print_structure
    type_column = [str(len(table.rows)) for key, table in self.items()[0:max_length]]
  File ".../.venv/lib/python3.8/site-packages/agate/tableset/print_structure.py", line 24, in <listcomp>
    type_column = [str(len(table.rows)) for key, table in self.items()[0:max_length]]
AttributeError: 'TableSet' object has no attribute 'rows'

JSON output for clarity:

>>> mytable.group_by('a').group_by('b').to_json('/dev/stdout', nested=True)
{"foo": {"FOO": [{"a": "foo", "b": "FOO", "c": 1.0}, {"a": "foo", "b": "FOO", "c": 2.0}]}, "bar": {"BOZ": [{"a": "bar", "b": "BOZ", "c": 1.0}], "BAZ": [{"a": "bar", "b": "BAZ", "c": 2.0}], "BIZ": [{"a": "bar", "b": "BIZ", "c": null}]}}

Expected string output:

???

(I have no ideas here other than "probably not an exception". 🙂 Figured I'd still go ahead and share.)

[jpmckinney]

Thank you for reporting!

The issue was that group_by – uniquely – returns a TableSet instead of a Table. Calling group_by a second time yielded a TableSet containing TableSets (instead of directly containing Tables), causing the error.

[jpmckinney]

Ah, my solution worked for group_by, but it messed up pivot (which calls group_by and aggregate) and aggregate (which contains logic for nested table sets).

[jpmckinney]

Does this make any sense for that input?

| table   | rows |
| ------- | ---- |
| foo.FOO | 2    |
| bar.BOZ | 1    |
| bar.BAZ | 1    |
| bar.BIZ | 1    |

[jpmckinney]

The patch: #776