def list(self):
"""
List the contents of the repo
"""
self._check()
self._repoEmit("Listing")
dataRepo = datarepo.FileSystemDataRepository(self._repoPath)
self._emit(self.referenceSetsDirName)
for referenceSet in dataRepo.getReferenceSets():
self._emitIndent(referenceSet.getLocalId())
for reference in referenceSet.getReferences():
self._emitIndent(reference.getLocalId(), 2)
self._emit(self.ontologiesDirName)
for ontologyMap in dataRepo.getOntologyMaps():
self._emitIndent(ontologyMap.getLocalId(), 1)
self._emit(self.datasetsDirName)
for dataset in dataRepo.getDatasets():
self._emitIndent(dataset.getLocalId())
self._emitIndent(self.readsDirName, 2)
for readGroupSet in dataset.getReadGroupSets():
self._emitIndent(readGroupSet.getLocalId(), 3)
self._emitIndent(self.variantsDirName, 2)
for variantSet in dataset.getVariantSets():
self._emitIndent(variantSet.getLocalId(), 3)
for chromFile in sorted(variantSet._chromFileMap.keys()):
self._emitIndent(chromFile, 4)
self._emitIndent(self.featuresDirName, 2)
for featureSet in dataset.getFeatureSets():
self._emitIndent(featureSet.getLocalId(), 3)
def list(self):
"""
List the contents of the repo
"""
self._check()
self._repoEmit("Listing")
dataRepo = datarepo.FileSystemDataRepository(self._repoPath,
doConsistencyCheck=False)
self._emit(self.referenceSetsDirName)
for referenceSet in dataRepo.getReferenceSets():
self._emitIndent(referenceSet.getLocalId())
for reference in referenceSet.getReferences():
self._emitIndent(reference.getLocalId(), 2)
self._emit(self.datasetsDirName)
for dataset in dataRepo.getDatasets():
self._emitIndent(dataset.getLocalId())
self._emitIndent(self.readsDirName, 2)
for readGroupSet in dataset.getReadGroupSets():
self._emitIndent(readGroupSet.getLocalId(), 3)
self._emitIndent(self.variantsDirName, 2)
for variantSet in dataset.getVariantSets():
self._emitIndent(variantSet.getLocalId(), 3)
for chromFile in sorted(variantSet._chromFileMap.keys()):
self._emitIndent(chromFile, 4)
def setUp(self):
self._maxDiff = None
self._dataDir = paths.testDataDir
self._dataUrl = "file://{}".format(self._dataDir)
dataRepository = datarepo.FileSystemDataRepository(self._dataDir)
self._backend = backend.Backend(dataRepository)
self._client = client.LocalClient(self._backend)
def check(self, doConsistencyCheck=False):
"""
Check the repository for well-formedness
"""
self._check()
if doConsistencyCheck:
dataRepo = datarepo.FileSystemDataRepository(self._repoPath)
dataRepo.checkConsistency()
self._repoEmit("Well-formed".format(self._repoPath))
def __init__(self, featureSetLocalName, dataPath):
"""
:param localId: Name of the GFF3 resource corresponding to a pair
of files, .db and .gff3
:param dataPath: string representing full path to the .db file
:return:
"""
self._dataset = datasets.AbstractDataset(_datasetName)
self._datarepo = datarepo.FileSystemDataRepository("tests/data")
featureSetLocalName = featureSetLocalName[:-3] # remove '.db'
self._testData = _testDataForFeatureSetName[featureSetLocalName]
super(FeatureSetTests, self).__init__(featureSetLocalName, dataPath)
def setUp(self):
self._variantSetName = "testVariantSet"
self._backend = datarepo.FileSystemDataRepository("tests/data")
self._dataset = datasets.AbstractDataset(self._backend)
self._variantSet = variants.AbstractVariantSet(
self._dataset, self._variantSetName)
self._variantAnnotationSet = \
variants.HtslibVariantAnnotationSet(
self._dataset,
"vas",
"tests/data/datasets/dataset1/variants/WASH7P_annotation",
self._backend, self._variantSet)
def __init__(self, variantAnnotationSetId, baseDir):
self._dataset = datasets.AbstractDataset("ds")
self._datarepo = datarepo.FileSystemDataRepository("tests/data")
super(VariantAnnotationSetTest, self).__init__(variantAnnotationSetId,
baseDir)
self._variantSet = variants.HtslibVariantSet(self._dataset, "vs",
self._dataPath, None)
self._variantRecords = []
self._referenceNames = set()
# Only read in VCF files with a JSON sidecar saying they're annotated.
for vcfFile in glob.glob(os.path.join(self._dataPath, "*.vcf.gz")):
if self._isAnnotated():
self._readVcf(vcfFile)
self._isCsq = self._hasConsequenceField()
def __init__(self, args):
self._key = args.key
# TODO this is an experimental addition which is useful for
# testing. We should think about this and document it if we
# this it's a useful feature. There is an argument for pushing
# the backend instantiation into the client, and letting the
# client be a factory, instantiating the correct Client class
# depending on the prefix.
filePrefix = "file://"
if args.baseUrl.startswith(filePrefix):
dataDir = args.baseUrl[len(filePrefix):]
theBackend = backend.Backend(
datarepo.FileSystemDataRepository(dataDir))
self._client = client.LocalClient(theBackend)
else:
self._client = client.HttpClient(args.baseUrl,
verbosityToLogLevel(args.verbose),
self._key)
def __init__(self, dataDir):
dataRepository = datarepo.FileSystemDataRepository(dataDir)
super(CpuProfilerBackend, self).__init__(dataRepository)
self.profiler = cProfile.Profile()
def __init__(self, dataDir):
dataRepository = datarepo.FileSystemDataRepository(dataDir)
super(HeapProfilerBackend, self).__init__(dataRepository)
self.profiler = guppy.hpy()
metavar='N',
help='how many pages (max) to load '
'from each test case (default: %(default)s)')
parser.add_argument("--callSetIds",
"-c",
default=[],
help="""Return variant calls which belong to call sets
with these IDs. Pass in IDs as a comma separated list (no spaces),
or '*' (with the single quotes!) to indicate 'all call sets'.
Omit this option to indicate 'no call sets'.
""")
args = parser.parse_args()
dataDir = "ga4gh-example-data"
backend = backend.Backend(datarepo.FileSystemDataRepository(dataDir))
if args.profile == 'heap':
backendClass = HeapProfilerBackend
backend = backendClass(dataDir)
args.repeatLimit = 1
args.pageLimit = 1
elif args.profile == 'cpu':
backendClass = CpuProfilerBackend
backend = backendClass(dataDir)
# Get our list of callSetids
callSetIds = args.callSetIds
if callSetIds != []:
callSetIds = None
if args.callSetIds != "*":
callSetIds = args.callSetIds.split(",")
def configure(configFile=None, baseConfig="ProductionConfig",
port=8000, extraConfig={}):
"""
TODO Document this critical function! What does it do? What does
it assume?
"""
configStr = 'ga4gh.serverconfig:{0}'.format(baseConfig)
app.config.from_object(configStr)
if os.environ.get('GA4GH_CONFIGURATION') is not None:
app.config.from_envvar('GA4GH_CONFIGURATION')
if configFile is not None:
app.config.from_pyfile(configFile)
app.config.update(extraConfig.items())
# Setup file handle cache max size
datamodel.fileHandleCache.setMaxCacheSize(
app.config["FILE_HANDLE_CACHE_MAX_SIZE"])
# Setup CORS
cors.CORS(app, allow_headers='Content-Type')
app.serverStatus = ServerStatus()
# Allocate the backend
# We use URLs to specify the backend. Currently we have file:// URLs (or
# URLs with no scheme) for the FileSystemBackend, and special empty:// and
# simulated:// URLs for empty or simulated data sources.
dataSource = urlparse.urlparse(app.config["DATA_SOURCE"], "file")
if dataSource.scheme == "simulated":
# Ignore the query string
randomSeed = app.config["SIMULATED_BACKEND_RANDOM_SEED"]
numCalls = app.config["SIMULATED_BACKEND_NUM_CALLS"]
variantDensity = app.config["SIMULATED_BACKEND_VARIANT_DENSITY"]
numVariantSets = app.config["SIMULATED_BACKEND_NUM_VARIANT_SETS"]
numReferenceSets = app.config[
"SIMULATED_BACKEND_NUM_REFERENCE_SETS"]
numReferencesPerReferenceSet = app.config[
"SIMULATED_BACKEND_NUM_REFERENCES_PER_REFERENCE_SET"]
numAlignmentsPerReadGroup = app.config[
"SIMULATED_BACKEND_NUM_ALIGNMENTS_PER_READ_GROUP"]
dataRepository = datarepo.SimulatedDataRepository(
randomSeed=randomSeed, numCalls=numCalls,
variantDensity=variantDensity, numVariantSets=numVariantSets,
numReferenceSets=numReferenceSets,
numReferencesPerReferenceSet=numReferencesPerReferenceSet,
numAlignments=numAlignmentsPerReadGroup)
elif dataSource.scheme == "empty":
dataRepository = datarepo.EmptyDataRepository()
elif dataSource.scheme == "file":
dataRepository = datarepo.FileSystemDataRepository(os.path.join(
dataSource.netloc, dataSource.path))
dataRepository.checkConsistency()
else:
raise exceptions.ConfigurationException(
"Unsupported data source scheme: " + dataSource.scheme)
theBackend = backend.Backend(dataRepository)
theBackend.setRequestValidation(app.config["REQUEST_VALIDATION"])
theBackend.setResponseValidation(app.config["RESPONSE_VALIDATION"])
theBackend.setDefaultPageSize(app.config["DEFAULT_PAGE_SIZE"])
theBackend.setMaxResponseLength(app.config["MAX_RESPONSE_LENGTH"])
app.backend = theBackend
app.secret_key = os.urandom(SECRET_KEY_LENGTH)
app.oidcClient = None
app.tokenMap = None
app.myPort = port
if "OIDC_PROVIDER" in app.config:
# The oic client. If we're testing, we don't want to verify
# SSL certificates
app.oidcClient = oic.oic.Client(
verify_ssl=('TESTING' not in app.config))
app.tokenMap = {}
try:
app.oidcClient.provider_config(app.config['OIDC_PROVIDER'])
except requests.exceptions.ConnectionError:
configResponse = message.ProviderConfigurationResponse(
issuer=app.config['OIDC_PROVIDER'],
authorization_endpoint=app.config['OIDC_AUTHZ_ENDPOINT'],
token_endpoint=app.config['OIDC_TOKEN_ENDPOINT'],
revocation_endpoint=app.config['OIDC_TOKEN_REV_ENDPOINT'])
app.oidcClient.handle_provider_config(configResponse,
app.config['OIDC_PROVIDER'])
# The redirect URI comes from the configuration.
# If we are testing, then we allow the automatic creation of a
# redirect uri if none is configured
redirectUri = app.config.get('OIDC_REDIRECT_URI')
if redirectUri is None and 'TESTING' in app.config:
redirectUri = 'https://{0}:{1}/oauth2callback'.format(
socket.gethostname(), app.myPort)
app.oidcClient.redirect_uris = [redirectUri]
if redirectUri is []:
raise exceptions.ConfigurationException(
'OIDC configuration requires a redirect uri')
# We only support dynamic registration while testing.
if ('registration_endpoint' in app.oidcClient.provider_info and
'TESTING' in app.config):
app.oidcClient.register(
app.oidcClient.provider_info["registration_endpoint"],
redirect_uris=[redirectUri])
else:
response = message.RegistrationResponse(
client_id=app.config['OIDC_CLIENT_ID'],
client_secret=app.config['OIDC_CLIENT_SECRET'],
redirect_uris=[redirectUri],
verify_ssl=False)
app.oidcClient.store_registration_info(response)
def setUp(self):
self._dataDir = os.path.join("tests", "data")
self._backend = backend.Backend(
datarepo.FileSystemDataRepository(self._dataDir))
self._dataRepo = self._backend.getDataRepository()
def setUp(self):
dataRepository = datarepo.FileSystemDataRepository(paths.testDataDir)
self._backend = backend.Backend(dataRepository)
self._client = client.LocalClient(self._backend)