petl-developers / petl / 43495886813

94%

push

github

fix: accept a single tuple or list field selection in valuecounter() and valuecounts() (#812)

This has the objective of accepting a compound field selection passed
as a single tuple or list in `valuecounter()` and `valuecounts()`, the
way `values()` already does.

Changes:
1. Fixed a bug in `valuecounter()`: it passed its `*field` varargs to
`values()` as one argument, double-wrapping a single tuple/list selector
into a nested tuple that `asindices()` rejects with
`FieldSelectionError`. `valuecounter(table, ('foo', 'bar'))` and
`valuecounter(table, ['foo', 'bar'])` now count the `('foo', 'bar')`
compound key, identical to `valuecounter(table, 'foo', 'bar')` /
`valuecounter(table, *fields)`.
2. Fixed the same bug in `valuecounts()`: `ValueCountsView` kept the
wrapped form, so a single tuple/list argument failed with the same error
on iteration. `ValueCountsView.__init__` now unwraps a single container
argument like `ValuesView.__init__` does (only when the element is
itself a tuple/list, since `__iter__` splats `*self.field` — a bare
string element must not be splatted into characters).
3. Repointed `test_valuecounter_invalid_tuple_field_message` at a
genuinely invalid selector (`(('foo', 'bar'),)`), so the tuple rendering
of the `FieldSelectionError` message stays covered; a valid compound
selection no longer raises.
4. Documented the behavior in the `valuecounter` docstring (doctest
included) and in `docs/changes.rst`.

The rest of petl already accepts this form: `values()`, `valuecount()`,
`typecounter()`, `typecounts()`, `stringpatterncounter()`,
`stringpatterns()`, `parsecounter()` and `parsecounts()` all take a
single tuple/list of fields as a compound selection — only
`valuecounter()`/`valuecounts()` rejected it, which is what #641 ran
into.

Closes #641
Co-authored-by: rahul05ranjan <11793455+rahul05ranjan@users.noreply.github.com>

28 of 28 new or added lines in 3 files covered. (100.0%)

16883 of 18005 relevant lines covered (93.77%)

0.94 hits per line