Example #1
0
 def testInitializedRead(self):
     """
     It must be possible to read from a StringIO instance that is
     initialized on creation.
     """
     s = StringIO('hey')
     self.assertEqual('hey', s.getvalue())
Example #2
0
 def testInitializedRead(self):
     """
     It must be possible to read from a StringIO instance that is
     initialized on creation.
     """
     s = StringIO('hey')
     self.assertEqual('hey', s.getvalue())
Example #3
0
 def testWriteRead(self):
     """
     It must be possible to write and read to/from a StringIO instance as
     normal.
     """
     s = StringIO()
     s.write('hey')
     self.assertEqual('hey', s.getvalue())
Example #4
0
 def testWriteRead(self):
     """
     It must be possible to write and read to/from a StringIO instance as
     normal.
     """
     s = StringIO()
     s.write('hey')
     self.assertEqual('hey', s.getvalue())
Example #5
0
    def testOpenNotCalledOnRepeatedCall(self):
        """
        If a repeated call to pathogenSampleFiles.add is made with the same
        arguments, no file should be read because the original result value is
        cached.
        """
        class Open(object):
            def __init__(self, test, manager):
                self.test = test
                self.manager = manager
                self.count = 0

            def sideEffect(self, filename, *args, **kwargs):
                if self.count == 0:
                    self.test.assertEqual('out/0.fasta', filename)
                    self.count += 1
                    return StringIO('>id1\nACTG\n')
                elif self.count == 1:
                    self.test.assertEqual('out/pathogen-0-sample-0.fasta',
                                          filename)
                    self.count += 1
                    return self.manager
                else:
                    self.test.fail(
                        'We are only supposed to be called twice. '
                        'Filename: %r, Args: %r, Keyword args: %r.' %
                        (filename, args, kwargs))

        fp = StringIO(
            '0.63 41.3 44.2 9 9 12 gi|327410| protein 77 [Lausannevirus]\n'
        )
        fastaIO = StringIO()

        @contextmanager
        def manager():
            yield fastaIO

        pg = ProteinGrouper()
        pg.addFile('filename-1', fp)
        pathogenSampleFiles = PathogenSampleFiles(pg)

        sideEffect = Open(self, manager()).sideEffect
        with patch.object(builtins, 'open') as mockMethod:
            mockMethod.side_effect = sideEffect
            filename = pathogenSampleFiles.add('Lausannevirus', 'filename-1')
            self.assertEqual('out/pathogen-0-sample-0.fasta', filename)
            self.assertEqual('>id1\nACTG\n', fastaIO.getvalue())

            # Repeated call. The side effect open will fail if open is
            # called at this point.
            filename = pathogenSampleFiles.add('Lausannevirus', 'filename-1')
            self.assertEqual('out/pathogen-0-sample-0.fasta', filename)
Example #6
0
    def testIdenticalReadsRemoved(self):
        """
        If two proteins in the same pathogen are matched by the same read, the
        de-duplicated FASTA for the pathogen must have only one copy of the
        duplicated read.
        """
        class Open(object):
            def __init__(self, test, manager):
                self.test = test
                self.manager = manager
                self.expectedFilenames = {'out/0.fasta', 'out/1.fasta',
                                          'out/pathogen-0-sample-0.fasta'}

            def sideEffect(self, filename, *args, **kwargs):
                try:
                    self.expectedFilenames.remove(filename)
                except KeyError:
                    self.test.fail(
                        'Open called with unexpected filename: %r, Args: %r, '
                        'Keyword args: %r.' % (filename, args, kwargs))
                else:
                    if filename == 'out/0.fasta':
                        return StringIO('>id1\nACTG\n')
                    elif filename == 'out/1.fasta':
                        return StringIO('>id1\nACTG\n>id2\nCAGT\n')
                    else:
                        return self.manager

        fp = StringIO(
            '0.63 41.3 44.2 9 9 12 gi|327410| protein 77 [Lausannevirus]\n'
            '0.77 46.6 48.1 5 6 74 gi|327409| ubiquitin [Lausannevirus]\n'
        )
        fastaIO = StringIO()

        @contextmanager
        def manager():
            yield fastaIO

        pg = ProteinGrouper()
        pg.addFile('filename-1', fp)
        pathogenSampleFiles = PathogenSampleFiles(pg)

        opener = Open(self, manager())
        with patch.object(builtins, 'open') as mockMethod:
            mockMethod.side_effect = opener.sideEffect
            filename = pathogenSampleFiles.add('Lausannevirus', 'filename-1')

        self.assertEqual('out/pathogen-0-sample-0.fasta', filename)
        self.assertEqual('>id1\nACTG\n>id2\nCAGT\n', fastaIO.getvalue())
        # Make sure all expected filenames were seen by the mocked open.
        self.assertEqual(set(), opener.expectedFilenames)
Example #7
0
    def testWritePathogenIndex(self):
        """
        The writePatogenIndex function must write a file with the expected
        content.
        """
        pathogenSampleFiles = PathogenSampleFiles(None)
        pathogenSampleFiles._pathogens = {
            'virus b': 4,
            'virus a': 3,
            'virus c': 9,
        }

        fp = StringIO()
        pathogenSampleFiles.writePathogenIndex(fp)
        self.assertEqual('3 virus a\n4 virus b\n9 virus c\n', fp.getvalue())
Example #8
0
    def testWriteSampleIndex(self):
        """
        The writeSampleIndex function must write a file with the expected
        content.
        """
        pathogenSampleFiles = PathogenSampleFiles(None)
        pathogenSampleFiles._samples = {
            'NEO11': 500,
            'NEO33': 24,
            'NEO66': 333,
        }

        fp = StringIO()
        pathogenSampleFiles.writeSampleIndex(fp)
        self.assertEqual('24 NEO33\n333 NEO66\n500 NEO11\n', fp.getvalue())
Example #9
0
    def testTwoReads(self):
        """
        It must be possible to access a FASTA file with two reads like a dict.
        """

        pyfaidxIndex = StringIO()

        class Open(object):
            def __init__(self, test, manager):
                self.test = test
                self.manager = manager
                self.count = 0

            def sideEffect(self, filename, *args, **kwargs):
                if self.count == 0:
                    self.test.assertEqual('filename.fasta', filename)
                    self.count += 1
                    return BytesIO(b'>id1\nACTG\n>id2\nAACCTTGG\n')
                elif self.count == 1:
                    self.test.assertEqual('filename.fasta', filename)
                    self.count += 1
                    return StringIO('>id1\nACTG\n>id2\nAACCTTGG\n')
                elif self.count == 2:
                    self.count += 1
                    return self.manager
                elif self.count == 3:
                    self.count += 1
                    return StringIO(pyfaidxIndex.getvalue())
                else:
                    self.test.fail(
                        'Open called too many times. Filename: %r, Args: %r, '
                        'Keyword args: %r.' % (filename, args, kwargs))

        @contextmanager
        def manager():
            yield pyfaidxIndex

        sideEffect = Open(self, manager()).sideEffect
        with patch.object(builtins, 'open') as mockMethod:
            mockMethod.side_effect = sideEffect
            reads = FastaFaiReads('filename.fasta')
            self.assertEqual(DNARead('id1', 'ACTG'), reads['id1'])
            self.assertEqual(DNARead('id2', 'AACCTTGG'), reads['id2'])
            # Check that the fai index was built correctly.
            self.assertEqual(pyfaidxIndex.getvalue(),
                             'id1\t4\t5\t4\t5\nid2\t8\t15\t8\t9\n')
Example #10
0
    def testTwoReads(self):
        """
        It must be possible to access a FASTA file with two reads like a dict.
        """

        pyfaidxIndex = StringIO()

        class Open(object):
            def __init__(self, test, manager):
                self.test = test
                self.manager = manager
                self.count = 0

            def sideEffect(self, filename, *args, **kwargs):
                if self.count == 0:
                    self.test.assertEqual('filename.fasta', filename)
                    self.count += 1
                    return BytesIO(b'>id1\nACTG\n>id2\nAACCTTGG\n')
                elif self.count == 1:
                    self.test.assertEqual('filename.fasta', filename)
                    self.count += 1
                    return StringIO('>id1\nACTG\n>id2\nAACCTTGG\n')
                elif self.count == 2:
                    self.count += 1
                    return self.manager
                elif self.count == 3:
                    self.count += 1
                    return StringIO(pyfaidxIndex.getvalue())
                else:
                    self.test.fail(
                        'Open called too many times. Filename: %r, Args: %r, '
                        'Keyword args: %r.' % (filename, args, kwargs))

        @contextmanager
        def manager():
            yield pyfaidxIndex

        sideEffect = Open(self, manager()).sideEffect
        with patch.object(builtins, 'open') as mockMethod:
            mockMethod.side_effect = sideEffect
            reads = FastaFaiReads('filename.fasta')
            self.assertEqual(DNARead('id1', 'ACTG'), reads['id1'])
            self.assertEqual(DNARead('id2', 'AACCTTGG'), reads['id2'])
            # Check that the fai index was built correctly.
            self.assertEqual(pyfaidxIndex.getvalue(),
                             'id1\t4\t5\t4\t5\nid2\t8\t15\t8\t9\n')