def test_write(self):
self.nex = NexusReader()
self.nex.read_string("""
#NEXUS
begin data;
Dimensions ntax=2 nchar=2;
Format datatype=standard gap=- symbols="01";
Matrix
Harry 1-
Simon 0?
;
End;
""")
expected_patterns = [
'^begin data;$',
'^\s+dimensions ntax=2 nchar=2;$',
'^\s+format datatype=standard gap=- symbols="01";$',
"^matrix$",
"^Harry\s+1-",
"^Simon\s+0\?$",
'^\s+;$',
'^end;$',
]
written = self.nex.write()
for expected in expected_patterns:
assert re.search(expected, written, re.MULTILINE), \
'Expected "%s"' % expected
class Test_DataHandler_NoMissingInSymbols(unittest.TestCase):
"""
Regression:
Test that the missing or gap symbols are NOT in the SYMBOLS format string
"""
def test_write(self):
self.nex = NexusReader()
self.nex.read_string("""
#NEXUS
begin data;
Dimensions ntax=2 nchar=2;
Format datatype=standard gap=- symbols="01";
Matrix
Harry 1-
Simon 0?
;
End;
""")
expected_patterns = [
'^begin data;$',
'^\s+dimensions ntax=2 nchar=2;$',
'^\s+format datatype=standard gap=- symbols="01";$',
"^matrix$",
"^Harry\s+1-",
"^Simon\s+0\?$",
'^\s+;$',
'^end;$',
]
written = self.nex.write()
for expected in expected_patterns:
assert re.search(expected, written, re.MULTILINE), \
'Expected "%s"' % expected
def test_write_produces_end(self):
nex = NexusReader()
nex.read_string("""
begin assumptions;
A = 1;
end;
""")
assert "end;" in nex.write()
def test_read_string(self):
handle = open(os.path.join(EXAMPLE_DIR, 'example.nex'))
data = handle.read()
handle.close()
nex = NexusReader()
nex.read_string(data)
assert 'data' in nex.blocks
assert 'Simon' in nex.blocks['data'].matrix
def test_run_detranslate(self):
nex = NexusReader(os.path.join(EXAMPLE_DIR,
'example-translated.trees'))
assert not nex.trees._been_detranslated
nex = run_detranslate(nex)
# should NOW be the same as tree 0 in example.trees
other_tree_file = NexusReader(
os.path.join(EXAMPLE_DIR, 'example.trees'))
assert other_tree_file.trees[0] == nex.trees[0]
def test_read_file(nex, examples):
"""Test the Core functionality of NexusReader"""
assert 'data' in nex.blocks
assert 'Simon' in nex.blocks['data'].matrix
with warnings.catch_warnings(record=True) as w:
n = NexusReader()
n.read_file(examples / 'example.nex')
assert len(w) == 1
def test_write_to_file(self):
tmp = NamedTemporaryFile(delete=False, suffix=".nex")
tmp.close()
nex = NexusReader(os.path.join(EXAMPLE_DIR, 'example.nex'))
nex.write_to_file(tmp.name)
assert os.path.isfile(tmp.name)
n2 = NexusReader(tmp.name)
assert n2.data.matrix == nex.data.matrix
assert sorted(n2.data.taxa) == sorted(nex.data.taxa)
os.unlink(tmp.name) # cleanup
def test_error_on_incorrect_dimensions():
with pytest.raises(NexusFormatException):
NexusReader.from_string("""
#NEXUS
begin taxa;
dimensions ntax=2;
taxlabels A B C;
end;
""")
def test_generic_readwrite(self):
expected = [
"begin sets;",
" A = 1;",
" B = 2;",
"end;",
]
nex = NexusReader()
nex.read_string("\n".join(expected))
for line in nex.sets.write().split("\n"):
e = expected.pop(0).strip()
assert line.strip() == e
def test_error_on_duplicate_taxa():
with pytest.raises(NexusFormatException):
NexusReader.from_string("""
#NEXUS
begin trees;
translate
0 Tom,
1 Simon,
2 Tom;
tree tree = (0,1,2)
end;
""")
def test_treelabel(self):
nex = NexusReader()
nex.read_string("""
#NEXUS
begin trees;
translate
0 Tom,
1 Simon,
2 Fred;
tree TREEONE = (0,1,2);
end;
""")
assert len(nex.trees.trees) == 1
assert nex.trees.trees == ['tree TREEONE = (0,1,2);']
def test_labelled_unrooted(self):
nex = NexusReader()
nex.read_string("""
#NEXUS
begin trees;
translate
0 Tom,
1 Simon,
2 Fred;
tree unrooted [U] = (0,1,2);
end;
""")
assert len(nex.trees.trees) == 1
assert nex.trees.trees == ['tree unrooted [U] = (0,1,2);']
def test_BEASTAscertainmentChecker_fail_with_non_empty():
# fail with two
nex = NexusReader.from_string("""
#NEXUS
Begin data;
charstatelabels
1 CHAR_A_ascertainement,
2 CHAR_A,
3 CHAR_B_ascertainement,
4 CHAR_B
;
Dimensions ntax=3 nchar=4;
Format datatype=standard symbols="01" gap=-;
Matrix
A 0101
B 0101
C 10?1
;
begin assumptions;
charset A = 1-2;
charset B = 3-4;
end;
""")
c = BEASTAscertainmentChecker(nex)
assert len(c.errors) == 1 # should ONLY be one
class Test_TreeHandler_Regression_Mesquite(unittest.TestCase):
"""Regression: Test that we can parse MESQUITE taxa blocks"""
def setUp(self):
self.nex = NexusReader(
os.path.join(REGRESSION_DIR, 'mesquite_formatted_branches.trees')
)
def test_attributes(self):
assert len(self.nex.trees.attributes) == 2
assert self.nex.trees.attributes[0] == \
"""Title 'Trees from "temp.trees"';"""
assert self.nex.trees.attributes[1] == \
"""LINK Taxa = Untitled_Block_of_Taxa;"""
def test_found_trees(self):
assert self.nex.trees.ntrees == 1
def test_found_taxa(self):
assert len(self.nex.trees.taxa) == 3
assert 'A' in self.nex.trees.taxa
assert 'B' in self.nex.trees.taxa
assert 'C' in self.nex.trees.taxa
def test_was_translated(self):
assert self.nex.trees.was_translated
def test_translation(self):
assert self.nex.trees.translators['1'] == 'A'
assert self.nex.trees.translators['2'] == 'B'
assert self.nex.trees.translators['3'] == 'C'
def test_write(self):
written = self.nex.write()
assert """Title 'Trees from "temp.trees"';""" in written
assert """LINK Taxa = Untitled_Block_of_Taxa;""" in written
def test_TreeHandler_MrBayes(regression):
"""
Test reading of treefile generated by a 2003-era MrBayes which, for some reason, have
a tab between "tree\t<name>=(...)"
"""
nex = NexusReader(regression / 'mrbayes.trees')
assert len(nex.trees.trees) == 1
def test_write_produces_end():
nex = NexusReader.from_string("""
begin assumptions;
A = 1;
end;
""")
assert "end;" in nex.write()
def setUp(self):
self.nex = NexusReader()
self.nex.read_string("""
#NEXUS
Begin data;
TITLE Untitled_Block_of_Taxa;
LINK Taxa = Untitled_Block_of_Taxa;
Dimensions ntax=2 nchar=2;
Format datatype=standard gap=- symbols="01";
Matrix
Harry 00
Simon 01
;
End;
""")
def test_run_deltree(self):
nex = NexusReader(os.path.join(EXAMPLE_DIR, 'example.trees'))
new_nex = run_deltree('2', nex, do_print=False)
assert len(new_nex.trees.trees) == 2
assert new_nex.trees.ntrees == 2
assert new_nex.trees[0].startswith('tree tree.0.1065.603220')
assert new_nex.trees[1].startswith('tree tree.20000.883.396049')
def test_TreeHandler_Taxon_with_asterisk(regression):
"""
Test reading of treefile that contains a taxon with an asterisk in it.
"""
nex = NexusReader(regression / 'tree_with_asterisk_in_taxa.trees')
assert len(nex.trees.trees) == 1
assert len(nex.trees.taxa) == 38
assert list(nex.trees.taxa)[35 - 1] == "*R35" # zero indexed, so 35-1
def test_TreeHandler_TranslateBlockMismatch(regression):
"""
Test that a warning is generated when a treefile has an incorrectly sized
translate block.
"""
with pytest.raises(TranslateTableException):
nex = NexusReader(regression / 'tree_translate_mismatch.trees')
nex.trees.detranslate()
def test_generic_readwrite(self):
expected = """Begin data;
Dimensions ntax=4 nchar=2;
Format datatype=standard symbols="01" gap=-;
Matrix
Harry 00
Simon 01
Betty 10
Louise 11
;
""".split("\n")
nex = NexusReader()
nex.handlers['data'] = GenericHandler
nex.read_file(os.path.join(EXAMPLE_DIR, 'example.nex'))
for line in nex.data.write().split("\n"):
e = expected.pop(0).strip()
assert line.strip() == e
def test_write_to_file(nex, tmpdir):
tmp = pathlib.Path(str(tmpdir.join('f.nex')))
nex.write_to_file(tmp)
assert tmp.is_file()
n2 = NexusReader(tmp)
assert n2.data.matrix == nex.data.matrix
assert sorted(n2.data.taxa) == sorted(nex.data.taxa)
def test_ok_starting_with_zero(self):
nex = NexusReader()
nex.read_string("""
#NEXUS
begin trees;
translate
0 Tom,
1 Simon,
2 Fred;
tree tree = (0,1,2)
end;
""")
assert len(nex.trees.translators) == 3
assert '0' in nex.trees.translators
assert '1' in nex.trees.translators
assert '2' in nex.trees.translators
def test_detranslate(self):
assert not self.nex.trees._been_detranslated
self.nex.trees.detranslate()
# should NOW be the same as tree 0 in example.trees
other_tree_file = NexusReader(
os.path.join(EXAMPLE_DIR, 'example.trees')
)
assert other_tree_file.trees[0] == self.nex.trees[0]
def test_ok_starting_with_one(self):
nex = NexusReader()
nex.read_string("""
#NEXUS
begin trees;
translate
1 Tom,
2 Simon,
3 Fred;
tree tree = (1,2,3)
end;
""")
assert len(nex.trees.translators) == 3
assert '1' in nex.trees.translators
assert '2' in nex.trees.translators
assert '3' in nex.trees.translators
def test_read_gzip_file(nex_string, tmpdir):
# first, MAKE a gzip file
with gzip.open(str(tmpdir.join('f.gz')), 'wb') as h:
h.write(nex_string.encode('utf8'))
# test it's ok
nex = NexusReader(str(tmpdir.join('f.gz')))
assert 'data' in nex.blocks
assert 'Simon' in nex.blocks['data'].matrix
def test_regression(self):
nex = NexusReader(
os.path.join(REGRESSION_DIR, 'white_space_in_matrix.nex')
)
assert nex.blocks['data'].nchar == 2
assert nex.blocks['data'].matrix['Harry'] == ['0', '0']
assert nex.blocks['data'].matrix['Simon'] == ['0', '1']
assert nex.blocks['data'].matrix['Betty'] == ['1', '0']
assert nex.blocks['data'].matrix['Louise'] == ['1', '1']
def test_run_resample_1(self):
# shouldn't resample anything..
nex = NexusReader(os.path.join(EXAMPLE_DIR, 'example.trees'))
new_nex = run_resample('1', nex, do_print=False)
assert len(new_nex.trees.trees) == 3
assert new_nex.trees.ntrees == 3
assert new_nex.trees[0].startswith('tree tree.0.1065.603220')
assert new_nex.trees[1].startswith('tree tree.10000.874.808756')
assert new_nex.trees[2].startswith('tree tree.20000.883.396049')
class Test_DataHandler_Regression_Mesquite(unittest.TestCase):
"""Regression: Test that we can parse MESQUITE data blocks"""
def setUp(self):
self.nex = NexusReader()
self.nex.read_string("""
#NEXUS
Begin data;
TITLE Untitled_Block_of_Taxa;
LINK Taxa = Untitled_Block_of_Taxa;
Dimensions ntax=2 nchar=2;
Format datatype=standard gap=- symbols="01";
Matrix
Harry 00
Simon 01
;
End;
""")
def test_attributes(self):
assert len(self.nex.data.attributes) == 2
assert self.nex.data.attributes[0] == \
"""TITLE Untitled_Block_of_Taxa;"""
assert self.nex.data.attributes[1] == \
"""LINK Taxa = Untitled_Block_of_Taxa;"""
def test_write(self):
expected_patterns = [
'^begin data;$',
'^\s+TITLE Untitled_Block_of_Taxa;$',
'^\s+LINK Taxa = Untitled_Block_of_Taxa;$',
'^\s+dimensions ntax=2 nchar=2;$',
'^\s+format datatype=standard gap=- symbols="01";$',
"^matrix$",
"^Harry\s+00",
"^Simon\s+01$",
'^\s+;$',
'^end;$',
]
written = self.nex.write()
for expected in expected_patterns:
assert re.search(expected, written, re.MULTILINE), \
'Expected "%s"' % expected
def test_error_on_duplicate_block():
with warnings.catch_warnings(record=True):
with pytest.raises(NexusFormatException):
NexusReader.from_string("""
#NEXUS
Begin data;
Dimensions ntax=5 nchar=1;
Format datatype=standard symbols="01" gap=-;
Matrix
Harry 1
;
Begin data;
Dimensions ntax=5 nchar=1;
Format datatype=standard symbols="01" gap=-;
Matrix
Harry 1
""")
def test_error_on_incorrect_dimensions(self):
with self.assertRaises(NexusFormatException):
NexusReader().read_string("""
#NEXUS
begin taxa;
dimensions ntax=2;
taxlabels A B C;
end;
""")
