Python unittest ignores numpy

Question

I am writing python tests using unittest and running tests from the command line with

nosetests --with-coverage -x 

When I include numpy in one of my tests, it also tries to test numpy packages. Output Example:

 ... Name Stmts Miss Cover Missing ------------------------------------------------------------- CLOCK 39 33 15% 3, 7-13, 17, 20-25, 28-47 LFU 42 1 98% 52 LRU 95 9 91% 12, 64, 68, 101, 115-118, 131 LRU10 54 1 98% 68 LRU3 54 1 98% 68 argparse 1177 1177 0% 3-2361 cache 86 33 62% 36-47, 86-89, 95-116 common 87 54 38% 17, 20, 23, 28, 31-32, 35-36, 39, 42, 46, 48, 50, 52, 54, 57-64, 67-68, 72-89, 95-96, 102-107, 112-118, 123 ctypes 341 341 0% 4-555 ctypes._endian 35 35 0% 4-64 numpy 56 56 0% 107-197 numpy.__config__ 27 27 0% 3-32 numpy._import_tools 224 224 0% 1-348 numpy.add_newdocs 275 275 0% 11-7459 numpy.compat 8 8 0% 11-20 numpy.compat._inspect 106 106 0% 8-221 numpy.compat.py3k 57 57 0% 5-89 numpy.core 56 56 0% 1-74 numpy.core._internal 350 350 0% 7-570 numpy.core._methods 72 72 0% 6-130 numpy.core.arrayprint 354 354 0% 6-752 numpy.core.defchararray 364 364 0% 18-2686 numpy.core.fromnumeric 310 310 0% 4-2918 numpy.core.function_base 24 24 0% 1-173 numpy.core.getlimits 132 132 0% 4-306 numpy.core.info 3 3 0% 84-87 numpy.core.machar 186 186 0% 8-338 numpy.core.memmap 83 83 0% 1-305 numpy.core.numeric 523 523 0% 1-2730 numpy.core.numerictypes 381 381 0% 83-1035 numpy.core.records 356 356 0% 37-808 numpy.core.shape_base 50 50 0% 1-277 numpy.ctypeslib 187 187 0% 52-426 numpy.fft 7 7 0% 1-11 numpy.fft.fftpack 109 109 0% 33-1119 numpy.fft.helper 51 51 0% 5-223 numpy.fft.info 2 2 0% 177-179 numpy.lib 38 38 0% 1-45 numpy.lib._datasource 178 178 0% 34-656 numpy.lib._iotools 372 372 0% 4-874 numpy.lib.arraypad 383 383 0% 6-1469 numpy.lib.arraysetops 98 98 0% 27-450 numpy.lib.arrayterator 72 72 0% 10-224 numpy.lib.financial 112 112 0% 11-735 numpy.lib.format 178 178 0% 137-614 numpy.lib.function_base 889 889 0% 1-3555 numpy.lib.index_tricks 250 250 0% 1-849 numpy.lib.info 3 3 0% 148-151 numpy.lib.nanfunctions 138 138 0% 17-838 numpy.lib.npyio 729 729 0% 1-1899 numpy.lib.polynomial 386 386 0% 5-1266 numpy.lib.scimath 55 55 0% 18-560 numpy.lib.shape_base 200 200 0% 1-834 numpy.lib.stride_tricks 48 48 0% 8-121 numpy.lib.twodim_base 116 116 0% 4-929 numpy.lib.type_check 104 104 0% 4-605 numpy.lib.ufunclike 23 23 0% 6-177 numpy.lib.utils 517 517 0% 1-1134 numpy.linalg 6 6 0% 45-54 numpy.linalg.info 2 2 0% 35-37 numpy.linalg.linalg 530 530 0% 11-2131 numpy.ma 15 15 0% 39-58 numpy.ma.core 2324 2324 0% 23-7243 numpy.ma.extras 610 610 0% 11-1885 numpy.matrixlib 6 6 0% 4-12 numpy.matrixlib.defmatrix 286 286 0% 1-1094 numpy.polynomial 11 11 0% 16-29 numpy.polynomial.chebyshev 432 432 0% 88-2015 numpy.polynomial.hermite 382 382 0% 60-1750 numpy.polynomial.hermite_e 379 379 0% 60-1746 numpy.polynomial.laguerre 379 379 0% 60-1742 numpy.polynomial.legendre 386 386 0% 84-1768 numpy.polynomial.polynomial 302 302 0% 56-1493 numpy.polynomial.polytemplate 4 4 0% 12-17 numpy.polynomial.polyutils 67 67 0% 34-384 numpy.random 13 13 0% 89-114 numpy.random.info 3 3 0% 85-89 numpy.version 7 7 0% 3-10 statistics 19 2 89% 24-25 statistics.countingghost 84 44 48% 43-60, 66-69, 75-80, 83-89, 93-104, 107, 110-113 statistics.rounder 89 17 81% 29, 40, 70-83, 112-113 ------------------------------------------------------------- TOTAL 17686 17145 3% ---------------------------------------------------------------------- Ran 3 tests in 3.279s OK 

How can I properly ignore numpy tests with -I -m or -e options? I could pass this grep -v numpy, but it's not cool.

Answer 1

There is a special plug-in for the nose - nose-exclude :

nose-exclude is a Nose plugin that makes it easy to specify directories that should be excluded from testing.

Another option is to specify the omit configuration parameter in the .coveragerc file:

omit (multi-string): a list of file name templates, files that need to be left out of measurement or reporting.

See also: How to exclude package layout from python coverage report using nosetests

Answer 2

You can also specify in which package you want to run the tests when you call the nose with the --cover-package flag.

eg.

 nosetests --with-coverage --cover-package=the-name-of-your-package -x