def test_specs__path_is_displayed_in_exception():
def _path_is_displayed_in_exception(spec, value):
assert_raises_regexp(MakefileError, _DUMMY_PATH_STR, spec, _DUMMY_PATH,
value)
yield _path_is_displayed_in_exception, IsInt(), "foo"
yield _path_is_displayed_in_exception, IsUnsignedInt(), -1
yield _path_is_displayed_in_exception, IsFloat(), "abc"
yield _path_is_displayed_in_exception, IsBoolean(), 1
yield _path_is_displayed_in_exception, IsStr(), 1
yield _path_is_displayed_in_exception, IsNone(), 1
yield _path_is_displayed_in_exception, ValueLT(0), 1
yield _path_is_displayed_in_exception, ValueLE(0), 1
yield _path_is_displayed_in_exception, ValueGE(0), -1
yield _path_is_displayed_in_exception, ValueGT(0), -1
yield _path_is_displayed_in_exception, ValueIn([1]), 2
yield _path_is_displayed_in_exception, ValuesIntersect([1]), [2]
yield _path_is_displayed_in_exception, ValuesSubsetOf([1]), [2]
yield _path_is_displayed_in_exception, ValueMissing(), True
yield _path_is_displayed_in_exception, And(IsStr), 1
yield _path_is_displayed_in_exception, Or(IsStr), 1
yield _path_is_displayed_in_exception, Xor(IsStr, IsInt), True
yield _path_is_displayed_in_exception, Not(IsInt), 1
yield _path_is_displayed_in_exception, StringIn("abc"), 1
yield _path_is_displayed_in_exception, StringsIntersect("abc"), [1]
yield _path_is_displayed_in_exception, StringsSubsetOf("abc"), [1]
yield _path_is_displayed_in_exception, StringIsUppercase(), 1
yield _path_is_displayed_in_exception, StringStartsWith("FOO"), 1
yield _path_is_displayed_in_exception, StringEndsWith("FOO"), 1
yield _path_is_displayed_in_exception, IsListOf(IsInt), "foo"
yield _path_is_displayed_in_exception, IsDictOf(IsInt, IsInt), 1
def test_process_makefile__missing_list_default_value():
current = {"A": 1}
expected = {"A": 1, "B": [1, 2, 3]}
specs = {"A": IsInt, "B": IsListOf(IsInt, default=[1, 2, 3])}
result = process_makefile(current, specs)
assert_equal(result, expected)
def test_is_list_of__default_set__valid_value():
spec = IsListOf(IsInt, default=range(5))
assert_equal(spec.default, range(5))
def test_is_list_of__non_list_rejected():
spec = IsListOf(IsInt)
assert_raises(MakefileError, spec, _DUMMY_PATH, {1: 2})
def test_is_list_of__default_not_set():
spec = IsListOf(IsInt)
assert_is(spec.default, DEFAULT_NOT_SET)
def test_is_list_of__mixed_list_rejected():
spec = IsListOf(IsInt)
assert_raises(MakefileError, spec, _DUMMY_PATH, [1, 'b', 3])
def test_is_list_of__default_description():
spec = IsListOf(IsInt, IsFloat)
assert_equal(spec.description, "[(an integer) or (a float), ...]")
def test_is_list_of__list_of_ints_accepted():
spec = IsListOf(IsInt)
spec(_DUMMY_PATH, [1, 2, 3])
def test_is_list_of__list_of_non_ints_rejected():
spec = IsListOf(IsInt)
assert_raises(MakefileError, spec, _DUMMY_PATH, ['a', 'b', 'c'])
def test_is_list_of__empty_list_always_ok():
spec = IsListOf(IsInt)
spec(_DUMMY_PATH, [])
CLI_PARAMETERS,
},
}
_VALIDATION = {
"Project": {
"Title": IsStr(default="Untitled"),
"Samples": _VALIDATION_SAMPLES,
"RegionsOfInterest": {
IsStr: {
"Prefix": IsStr(default=REQUIRED_VALUE),
"Realigned": IsBoolean(default=False),
"ProteinCoding": IsBoolean(default=False),
"IncludeIndels": IsBoolean(default=True),
"HomozygousContigs": {
IsStr: Or(IsNone, IsListOf(IsStr))
},
},
},
"FilterSingletons": {
IsStr: [IsStr],
},
},
"Genotyping": {
"Defaults": _VALIDATION_GENOTYPES,
IsStr: _VALIDATION_GENOTYPES,
},
"MultipleSequenceAlignment": {
"Defaults": _VALIDATION_MSA,
IsStr: _VALIDATION_MSA,
},
_VALID_TARGET_NAME = \
And(_alphanum_check(whitelist="._-"),
ValueGE(2, key=len, description="at least two characters long"))
_VALIDATION_OPTIONS = {
# Sequencing platform, used to tag read-groups.
"Platform":
StringIn(("CAPILLARY", "LS454", "ILLUMINA", "SOLID", "HELICOS",
"IONTORRENT", "PACBIO"),
default="ILLUMINA"),
# Offset for quality scores in FASTQ files.
"QualityOffset":
ValueIn((33, 64, "Solexa"), default=33),
# Split a lane into multiple entries, one for each (pair of) file(s)
"SplitLanesByFilenames":
Or(IsBoolean, IsListOf(IsStr), default=True),
# Format to use when compressing FASTQ files ("gz" or "bz2")
"CompressionFormat":
ValueIn(("gz", "bz2"), default="bz2"),
"AdapterRemoval": {
"Version": ValueIn(("v1.4", "v1.5+"), default="v1.5+"),
"--pcr1": IsStr,
"--pcr2": IsStr,
"--adapter1": IsStr,
"--adapter2": IsStr,
"--maxns": IsUnsignedInt,
"--minquality": IsUnsignedInt,
"--trimns": Or(IsNone, IsBoolean),
"--trimqualities": Or(IsNone, IsBoolean),
"--collapse": Or(IsNone, IsBoolean, default=True),
"--mm": Or(IsFloat, IsUnsignedInt, default=3),
},
},
}
_VALIDATION = {
"Project" : {
"Title" : IsStr,
"Taxa" : _validate_taxa,
"Intervals" : {
IsStr : {
"Genome" : IsStr,
"Protein coding" : IsBoolean,
"Orthology map" : IsStr,
"Homozygous Contigs" : {
IsStr : IsListOf(IsStr),
},
},
},
"Filter Singletons" : {
IsStr : IsListOf(IsStr),
},
},
"Genotyping" : {
"Default" : OneOf("random", "samtools", case_sensitive = False),
"Padding" : IsInt,
"Indels" : IsBoolean,
AnyOf("MPileup", "BCFTools", "Random") : {
IsStrWithPrefix("-") : CLI_PARAMETERS,
},
"VCF_Filter" : {
from pypeline.common.makefile import \
read_makefile, \
IsUnsignedInt, \
IsFloat, \
IsStr, \
IsBoolean, \
And, \
Or, \
ValueIn, \
IsNone, \
StringIn, \
ValueGE, \
ValuesSubsetOf, \
IsListOf
EXCLUDEBED = Or(IsListOf(IsStr), IsStr, IsNone, default=None)
def _alphanum_check(whitelist):
description = "characters a-z, A-Z, 0-9%s allowed"
description %= (", and %r" % whitelist, ) if whitelist else ""
whitelist += string.ascii_letters + string.digits
return And(IsStr(), ValuesSubsetOf(whitelist, description=description))
_VALID_BED_NAME = _VALID_TARGET_NAME = \
And(_alphanum_check(whitelist=".-"),
ValueGE(2, key=len, description="at least two characters long"))