pytest-interactive: select tests to run using a REPL

This plugin allows for the selection and run of pytest tests using the command line facilities available in IPython. This includes tab completion along the pytest node hierarchy and test callspec ids as well as the use of standard Python subscript and slice syntax for selection.

Upon invocation with either the --interactive or shorthand --ia arguments, you will enter an IPython shell which allows for navigation of the test tree built during the pytest collection phase.

Enjoy and feel free to submit PRs, ask for features, or report bugs on the github page.


To invoke the interactive plugin simply run pytest as normal from the top of your test tree like so

$ py.test -vvvs --interactive example_test_set/

or more compactly

$ py.test -vvvs --ia example_test_set/

Pytest will execute normally up until the end of the collection phase at which point it will enter a slightly customized IPython shell:

============================ test session starts ============================
platform linux -- Python 3.4.1 -- py-1.4.20 -- pytest-2.5.2 -- /usr/bin/python
plugins: interactive
collected 63 items
Building test tree...
Entering IPython shell...

Welcome to pytest-interactive, the pytest + IPython sensation!
Please explore the test (collection) tree using tt.<TAB>
When finished tabbing to a test node, simply __call__ it to have
pytest invoke all tests collected under that node.
'0' selected >>>

'0' selected >>> tt.<TAB>
tt.test_pinky_strength  tt.test_dogs_breath
tt.test_manliness       tt.test_cats_meow

'0' selected >>> tt.testsdir.test_pinky_strength.<TAB>

That’s right, jacked pinky finger here you come…

Select tests via tab-completion

Basic tab completion should allow you to navigate to the test(s) of interest as well as aid in exploring the overall pytest collection tree.

Tab completion works along python packages, modules, classes and test functions. The latter three types are collected as nodes by pytest out of the box but as an extra aid, intermediary nodes are created for packages containing tests as well. This is helpful to distinguish between different groups of tests in the file system.


The binding tt (abbreviation for test tree) is a reference to the root of a tree of nodes which roughly corresponds to the collection tree gathered by pytest.

If you’d like to see all tests included by a particular node simply evaluate it on the shell to trigger a pretty print __repr__() method:

'0' selected >>> tt.tests.subsets.test_setA
Out [0]:
   <Module 'example_test_set/tests/subsets/'>
0    <Function 'test_modes[a]'>
1    <Function 'test_modes[b]'>
2    <Function 'test_modes[c]'>
3    <Function 'test_inputs[1]'>
4    <Function 'test_inputs[2]'>
5    <Function 'test_inputs[3]'>
     <Class 'TestBoth'>
6      <Function 'test_m[a-1]'>
7      <Function 'test_m[a-2]'>
8      <Function 'test_m[a-3]'>
9      <Function 'test_m[b-1]'>
10     <Function 'test_m[b-2]'>
11     <Function 'test_m[b-3]'>
12     <Function 'test_m[c-1]'>
13     <Function 'test_m[c-2]'>
14     <Function 'test_m[c-3]'>

Total 15 tests

When ready to run pytest, simply __call__() the current node to exit the shell and invoke all tests below it in the tree:

'0' selected >>> tt.test_setB.test_modes()

You have selected the above 3 test(s) to be run.
Would you like to run pytest now? ([y]/n)?

example_test_set/tests/subsets/subsubset/ test_modes[a] PASSED
example_test_set/tests/subsets/subsubset/ test_modes[b] PASSED
example_test_set/tests/subsets/subsubset/ test_modes[c] FAILED

Selection by index or slice

Tests can also be selected by slice or subscript notation. This is handy if you see the test(s) you’d like to run in the pretty print output but don’t feel like tab-completing all the way down the tree to the necessary leaf node.

Take the following subtree of tests for example:

'0' selected >>> tt.test_setB.test_modes
  <Module 'example_test_set/tests/subsets/subsubset/'>
0   <Function 'test_modes[a]'>
1   <Function 'test_modes[b]'>
2   <Function 'test_modes[c]'>

Total 3 tests

Now let’s select the last test

'0' selected >>> tt.test_setB.test_modes[-1]
  <Module 'example_test_set/tests/subsets/subsubset/'>
0   <Function 'test_modes[c]'>

Total 1 tests

Or how about the first two

'0' selected >>> tt.test_setB.test_modes[:2]
  <Module 'example_test_set/tests/subsets/subsubset/'>
0   <Function 'test_modes[a]'>
1   <Function 'test_modes[b]'>

Total 2 tests

You can of course __call__() the indexed node as well to immediately run all tests in the selection.

Filtering by parameterized test callspec ids

Tests which are generated at runtime (aka parametrized) can be filtered by their callspec ids. Normally the ids are shown inside the braces [...] of the test nodeid which often looks something like:

<Function 'test_some_feature[blah-mode5-ipv6]'>

(i.e. what you get for output when using the --collectonly arg)

To access the available ids use the node’s params attribute.

'0' selected >>> tt.params.<TAB>
tt.params.a      tt.params.b      tt.params.c    tt.params.mouse

'0' selected >>> tt.params.a
   <Module 'example_test_set/tests/'>
0    <Function 'test_modes[a]'>
     <Class 'TestBoth'>
1      <Function 'test_m[a-dog]'>
2      <Function 'test_m[a-cat]'>
3      <Function 'test_m[a-mouse]'>
   <Module 'example_test_set/tests/subsets/'>
4    <Function 'test_modes[a]'>
     <Class 'TestBoth'>
5      <Function 'test_m[a-1]'>
6      <Function 'test_m[a-2]'>
7      <Function 'test_m[a-3]'>
   <Module 'example_test_set/tests/subsets/subsubset/'>
8    <Function 'test_modes[a]'>
     <Class 'TestBoth'>
9      <Function 'test_m[a-1]'>
10     <Function 'test_m[a-2]'>
11     <Function 'test_m[a-3]'>
   <Module 'example_test_set/tests2/'>
12   <Function 'test_modes[a]'>
     <Class 'TestBoth'>
13     <Function 'test_m[a-1]'>
14     <Function 'test_m[a-2]'>
15     <Function 'test_m[a-3]'>

Total 16 tests

You can continue to filter in this way as much as is possible

'0' selected >>> tt.params.a.params.<TAB>

'0' selected >>> tt.params.a.params.mouse
  <Module 'example_test_set/tests/'>
    <Class 'TestBoth'>
0     <Function 'test_m[a-mouse]'>

Total 1 tests


There is one stipulation with using id filtering which is that the id tokens must be valid python literals. Otherwise the __getattr__() overloading of the node will not work. It is recomended that you give your parameterized tests tab completion friendly ids using the ids kwarg as documented on the pytest site.

Multiple selections and magics

So by now I’m sure you’ve thought but what if I want to select tests from totally different parts of the tree?

Well lucky for you some %magics have been added to the shell to help with just that problem:

'0' selected >>> tt.test_setB.test_modes
  <Module 'example_test_set/tests/subsets/subsubset/'>
0   <Function 'test_modes[a]'>
1   <Function 'test_modes[b]'>
2   <Function 'test_modes[c]'>

Total 3 tests

'0' selected >>> add tt.test_setB.test_modes[-2:]

'2' selected >>>

You can easily show the contents of your selection

'2' selected >>> show
  <Module 'example_test_set/tests/subsets/subsubset/'>
0   <Function 'test_modes[b]'>
1   <Function 'test_modes[c]'>

'2' selected >>>

You can also remove tests from the current selection by index

'2' selected >>> remove 1

'1' selected >>> show
 <Module 'example_test_set/tests/subsets/subsubset/'>
0  <Function 'test_modes[b]'>

'1' selected >>>

When ready to run your tests simply exit the shell

'1' selected >>> exit

For additional docs on the above shell %magics simply use the %? magic syntax available in the IPython shell (i.e. add? or remove? or show?).

lastfailed tests

pytest exposes the list of failed test from the most recent run and stores them in its cache. Normally you can select only those test using the --lf flag. pytest-interactive always wraps the last failed test set in its shell’s local namespace using under name lastfailed.

Using the pytest cache

You can store test sets for access across sessions using the pytest cache. A magic %cache is available for creating and deleting entries:

'0' selected >>> cache add tt.test_setB.test_modes setb

'0' selected >>> cache

setb -> 3 items

'0' selected >>> setb  # the local namespace is auto-updated
  <Module 'example_test_set/tests/subsets/subsubset/'>
0   <Function 'test_modes[a]'>
1   <Function 'test_modes[b]'>
2   <Function 'test_modes[c]'>

'0' selected >>> exit

# ... come back in a later session ...

 # the local namespace is auto-populated with cache entries
'0' selected >>> setb
  <Module 'example_test_set/tests/subsets/subsubset/'>
0   <Function 'test_modes[a]'>
1   <Function 'test_modes[b]'>
2   <Function 'test_modes[c]'>

'0' selected >>> cache del setb
Deleted cache entry for 'setb'

See %cache? for full command details.