htsworkflow/submission/test/test_submission.py

   1
   2 import os
   3 import shutil
   4 from unittest import TestCase, TestSuite, defaultTestLoader
   5
   6 from htsworkflow.util.rdfhelp import \
   7      dafTermOntology, \
   8      get_turtle_header, \
   9      load_string_into_model, \
  10      get_model
  11 from htsworkflow.submission.submission import list_submissions, Submission
  12 from htsworkflow.submission.results import ResultMap
  13 from .submission_test_common import  (
  14     generate_sample_results_tree,
  15     S1_NAME,
  16     S2_NAME,
  17     S1_TURTLE,
  18     S2_TURTLE,
  19     MockAddDetails,
  20 )
  21
  22 import RDF
  23 #import logging
  24 #logging.basicConfig(level=logging.DEBUG)
  25
  26 class TestSubmissionModule(TestCase):
  27     def test_empty_list_submission(self):
  28         model = get_model()
  29         self.assertEqual(len(list(list_submissions(model))), 0)
  30
  31     def test_one_submission(self):
  32         model = get_model()
  33         load_string_into_model(model, "turtle",
  34             """
  35             @prefix subns: <http://jumpgate.caltech.edu/wiki/UcscSubmissionOntology#> .
  36             @prefix test: <http://jumpgate.caltech.edu/wiki/SubmissionsLog/test#> .
  37
  38             <http://jumpgate.caltech.edu/wiki/SubmissionsLog/test#>
  39                subns:has_submission test:lib1 ;
  40                subns:has_submission test:lib2.
  41             """)
  42         submissions = list(list_submissions(model))
  43         self.assertEqual(len(submissions), 1)
  44         self.assertEqual(submissions[0], "test")
  45
  46     def test_two_submission(self):
  47         model = get_model()
  48         load_string_into_model(model, "turtle",
  49             """
  50             @prefix subns: <http://jumpgate.caltech.edu/wiki/UcscSubmissionOntology#> .
  51             @prefix test: <http://jumpgate.caltech.edu/wiki/SubmissionsLog/test#> .
  52
  53             <http://jumpgate.caltech.edu/wiki/SubmissionsLog/test1#>
  54                subns:has_submission test:lib1 .
  55             <http://jumpgate.caltech.edu/wiki/SubmissionsLog/test2#>
  56                subns:has_submission test:lib2 .
  57             """)
  58         submissions = list(list_submissions(model))
  59         self.assertEqual(len(submissions), 2)
  60         truth = set(["test1", "test2"])
  61         testset = set()
  62         for name in submissions:
  63             testset.add(name)
  64         self.assertEqual(testset, truth)
  65
  66 class TestSubmission(TestCase):
  67     def setUp(self):
  68         generate_sample_results_tree(self, 'submission_test')
  69         self.model = get_model()
  70
  71     def tearDown(self):
  72         shutil.rmtree(self.tempdir)
  73
  74     def test_create_submission(self):
  75         model = get_model()
  76         s = Submission('foo', self.model, 'http://localhost')
  77         self.assertEqual(str(s.submissionSet),
  78                          "http://jumpgate.caltech.edu/wiki/SubmissionsLog/foo")
  79         self.assertEqual(str(s.submissionSetNS['']),
  80                          str(RDF.NS(str(s.submissionSet) + '#')['']))
  81         self.assertEqual(str(s.libraryNS['']),
  82                          str(RDF.NS('http://localhost/library/')['']))
  83
  84     def test_scan_submission_dirs(self):
  85         turtle = get_turtle_header() + """
  86 @prefix thisView: <http://jumpgate.caltech.edu/wiki/SubmissionsLog/test/view/> .
  87 thisView:Fastq ucscDaf:filename_re ".*[^12]\\.fastq\\.bz2$" ;
  88                a geoSoft:raw ;
  89                geoSoft:fileTypeLabel "fastq" ;
  90                ucscDaf:output_type "read" .
  91 thisView:FastqRead1 ucscDaf:filename_re ".*r1\\.fastq\\.bz2$" ;
  92                a geoSoft:raw ;
  93                geoSoft:fileTypeLabel "fastq" ;
  94                ucscDaf:output_type "read1" .
  95 thisView:FastqRead2 ucscDaf:filename_re ".*r2\\.fastq\\.bz2$" ;
  96                a geoSoft:raw ;
  97                geoSoft:fileTypeLabel "fastq" ;
  98                ucscDaf:output_type "read2" .
  99 thisView:alignments ucscDaf:filename_re ".*\\.bam$" ;
 100                a geoSoft:supplemental ;
 101                geoSoft:fileTypeLabel "bam" ;
 102                ucscDaf:output_type "alignments" .
 103
 104         """
 105         map = ResultMap()
 106         map['1000'] = os.path.join(self.sourcedir, S1_NAME)
 107         map['2000'] = os.path.join(self.sourcedir, S2_NAME)
 108
 109         s = Submission('foo', self.model, 'http://localhost')
 110         mock = MockAddDetails(self.model, turtle)
 111         mock.add_turtle(S1_TURTLE)
 112         mock.add_turtle(S2_TURTLE)
 113         s._add_library_details_to_model =  mock
 114         s.scan_submission_dirs(map)
 115
 116         nodes = list(s.analysis_nodes(map))
 117         self.assertEqual(len(nodes), 2)
 118         expected = set((
 119             'http://jumpgate.caltech.edu/wiki/SubmissionsLog/foo#1000-sample',
 120             'http://jumpgate.caltech.edu/wiki/SubmissionsLog/foo#2000-sample',
 121         ))
 122         got = set((str(nodes[0]), str(nodes[1])))
 123         self.assertEqual(expected, got)
 124
 125     def test_find_best_match(self):
 126         turtle = get_turtle_header() + """
 127 @prefix thisView: <http://jumpgate.caltech.edu/wiki/SubmissionsLog/test/view/> .
 128 thisView:Fastq ucscDaf:filename_re ".*[^12]\\.fastq\\.bz2$" ;
 129                a geoSoft:raw ;
 130                geoSoft:fileTypeLabel "fastq" ;
 131                ucscDaf:output_type "read" .
 132 thisView:FastqRead1 ucscDaf:filename_re ".*r1\\.fastq\\.bz2$" ;
 133                a geoSoft:raw ;
 134                geoSoft:fileTypeLabel "fastq" ;
 135                ucscDaf:output_type "read1" .
 136 thisView:FastqRead2 ucscDaf:filename_re ".*r2\\.fastq\\.bz2$" ;
 137                a geoSoft:raw ;
 138                geoSoft:fileTypeLabel "fastq" ;
 139                ucscDaf:output_type "read2" .
 140 thisView:alignments ucscDaf:filename_re ".*\\.bam$" ;
 141                a geoSoft:supplemental ;
 142                geoSoft:fileTypeLabel "bam" ;
 143                ucscDaf:output_type "alignments" .
 144
 145         """
 146         load_string_into_model(self.model, 'turtle', turtle)
 147         s = Submission('foo', self.model, 'http://localhost')
 148         q = RDF.Statement(None, dafTermOntology['filename_re'], None)
 149         view_map = s._get_filename_view_map()
 150         self.assertEqual(len(view_map), 4)
 151
 152         fastq = s.find_best_match("asdf.fastq.bz2")
 153         self.assertEqual(
 154             str(fastq),
 155             "http://jumpgate.caltech.edu/wiki/SubmissionsLog/test/view/Fastq")
 156
 157         fastq = s.find_best_match("asdf.r2.fastq.bz2")
 158         self.assertEqual(
 159             str(fastq),
 160             "http://jumpgate.caltech.edu/wiki/SubmissionsLog/test/view/FastqRead2")
 161
 162 def suite():
 163     suite = TestSuite()
 164     suite.addTests(
 165         defaultTestLoader.loadTestsFromTestCase(TestSubmissionModule))
 166     suite.addTests(
 167         defaultTestLoader.loadTestsFromTestCase(TestSubmission))
 168     return suite
 169
 170 if __name__ == "__main__":
 171     from unittest import main
 172     main(defaultTest='suite')