626 timestamps

constraints.txt

@@ -24,6 +24,6 @@
 # ga4gh-client==0.1.0
 # ga4gh-schemas==0.1.0
 #
-git+git://github.com/ga4gh/schemas.git@master#egg=ga4gh_schemas
+git+git://github.com/ejacox/schemas.git@626_timestamps#egg=ga4gh_schemas
 git+git://github.com/ga4gh/ga4gh-client.git@master#egg=ga4gh_client
 

dev-requirements.txt

@@ -8,7 +8,6 @@
 # that are dependencies of libraries listed in the next section
 # and are not already listed in requirements.txt
 
-python-dateutil==2.5.1
 PyYAML==3.11
 
 ### This section is for the actual libraries ###

docs/datarepo.rst

@@ -122,11 +122,11 @@ dataset has a name, which is used to identify it in the repository manager.
 
 .. code-block:: bash
 
-    $ ga4gh_repo add-dataset registry.db 1kg -d 'Example dataset using 1000 genomes data'
+    $ ga4gh_repo add-dataset registry.db 1kg -d 'Example dataset using 1000 genomes data' -c '2007-04-05T12:30-02:00' -u '2008-04-25T18:30Z'
 
-Adds the dataset with the name ``1kg`` and description
-``'Example dataset using 1000 genomes data'`` to the
-registry database ``registry.db``.
+Adds the dataset with the name ``1kg``, description
+``'Example dataset using 1000 genomes data'``, and a creation and update time
+(in ISO 8601 format) to the registry database ``registry.db``.
 
 ----------------
 add-referenceset

ga4gh/server/cli/repomanager.py

@@ -156,6 +156,8 @@ def addDataset(self):
         self._openRepo()
         dataset = datasets.Dataset(self._args.datasetName)
         dataset.setDescription(self._args.description)
+        dataset.setCreateDateTime(self._args.createDateTime)
+        dataset.setUpdateDateTime(self._args.updateDateTime)
         dataset.setInfo(json.loads(self._args.info))
         self._updateRepo(self._repo.insertDataset, dataset)
 
@@ -585,6 +587,18 @@ def addDatasetNameArgument(cls, subparser):
         subparser.add_argument(
             "datasetName", help="the name of the dataset")
 
+    @classmethod
+    def addDatasetCreateDateTimeArgument(cls, subparser):
+        subparser.add_argument(
+            "-c", "--createDateTime", default=None,
+            help="the creation datetime of the dataset")
+
+    @classmethod
+    def addDatasetUpdateDateTimeArgument(cls, subparser):
+        subparser.add_argument(
+            "-u", "--updateDateTime", default=None,
+            help="the update datetime of the dataset")
+
     @classmethod
     def addDatasetInfoArgument(cls, subparser):
         subparser.add_argument(
@@ -742,6 +756,8 @@ def getParser(cls):
         cls.addRepoArgument(addDatasetParser)
         cls.addDatasetNameArgument(addDatasetParser)
         cls.addDatasetInfoArgument(addDatasetParser)
+        cls.addDatasetCreateDateTimeArgument(addDatasetParser)
+        cls.addDatasetUpdateDateTimeArgument(addDatasetParser)
         cls.addDescriptionOption(addDatasetParser, "dataset")
 
         removeDatasetParser = common_cli.addSubparser(

ga4gh/server/datamodel/datasets.py

@@ -5,6 +5,8 @@
 from __future__ import print_function
 from __future__ import unicode_literals
 
+from datetime import datetime
+
 import ga4gh.server.datamodel as datamodel
 import ga4gh.server.datamodel.reads as reads
 import ga4gh.server.datamodel.sequence_annotations as sequence_annotations
@@ -15,9 +17,8 @@
 import ga4gh.server.datamodel.genotype_phenotype as g2p
 import ga4gh.server.datamodel.rna_quantification as rnaQuantification
 import json
-
 import google.protobuf.struct_pb2 as struct_pb2
-
+import dateutil.parser as datetimeParser
 import ga4gh.schemas.pb as pb
 
 
@@ -59,6 +60,8 @@ def __init__(self, localId):
         self._rnaQuantificationSetIdMap = {}
         self._rnaQuantificationSetNameMap = {}
         self._info = {}
+        self._createDateTime = None
+        self._updateDateTime = None
 
     def populateFromRow(self, row):
         """
@@ -67,13 +70,51 @@ def populateFromRow(self, row):
         """
         self._description = row[b'description']
         self._info = json.loads(row[b'info'])
+        self._createDateTime = row[b'createDateTime']
+        self._updateDateTime = row[b'updateDateTime']
 
     def setDescription(self, description):
         """
         Sets the description for this dataset to the specified value.
         """
         self._description = description
 
+    def setCreateDateTime(self, datetimeStr):
+        """
+        Sets the creation DateTime for this dataset to the specified value.
+
+        The datetime parser accepts a wide variety of time formats, which
+        are converted to ISO format.
+        """
+        if datetimeStr:
+            try:
+                datetimeObj = datetimeParser.parse(datetimeStr)
+            except ValueError:
+                raise exceptions.TimestampFormatException(
+                        datetimeStr, self.getLocalId())
+            except OverflowError:
+                raise exceptions.TimestampFormatException(
+                        datetimeStr, self.getLocalId())
+            self._createDateTime = datetimeObj.isoformat()
+
+    def setUpdateDateTime(self, datetimeStr):
+        """
+        Sets the update DateTime for this dataset to the specified value.
+
+        The datetime parser accepts a wide variety of time formats, which
+        are converted to ISO format.
+        """
+        if datetimeStr:
+            try:
+                datetimeObj = datetimeParser.parse(datetimeStr)
+            except ValueError:
+                raise exceptions.TimestampFormatException(
+                        datetimeStr, self.getLocalId())
+            except OverflowError:
+                raise exceptions.TimestampFormatException(
+                        datetimeStr, self.getLocalId())
+            self._updateDateTime = datetimeObj.isoformat()
+
     def setInfo(self, info):
         """
         Sets the info for this dataset to the specified value.
@@ -143,6 +184,8 @@ def toProtocolElement(self):
         dataset.description = pb.string(self.getDescription())
         for key in self.getInfo():
             dataset.info[key].values.extend(_encodeValue(self._info[key]))
+        dataset.createDateTime = pb.string(self.getCreateDateTime())
+        dataset.updateDateTime = pb.string(self.getUpdateDateTime())
         return dataset
 
     def getVariantSets(self):
@@ -369,6 +412,18 @@ def getDescription(self):
         """
         return self._description
 
+    def getCreateDateTime(self):
+        """
+        Returns the free text creation DateTime of this dataset.
+        """
+        return self._createDateTime
+
+    def getUpdateDateTime(self):
+        """
+        Returns the free text update DateTime of this dataset.
+        """
+        return self._updateDateTime
+
     def getNumRnaQuantificationSets(self):
         """
         Returns the number of rna quantification sets in this dataset.
@@ -422,6 +477,7 @@ def __init__(
             numExpressionLevels=2):
         super(SimulatedDataset, self).__init__(localId)
         self._description = "Simulated dataset {}".format(localId)
+        self.setCreateDateTime(datetime(2015, 1, 1).isoformat())
 
         for i in range(numPhenotypeAssociationSets):
             localId = "simPas{}".format(i)

ga4gh/server/datarepo.py

@@ -808,6 +808,8 @@ def _createDatasetTable(self, cursor):
                 name TEXT NOT NULL,
                 description TEXT,
                 info TEXT,
+                createDateTime TEXT,
+                updateDateTime TEXT,
                 UNIQUE (name)
             );
         """
@@ -818,15 +820,17 @@ def insertDataset(self, dataset):
         Inserts the specified dataset into this repository.
         """
         sql = """
-            INSERT INTO Dataset (id, name, description, info)
-            VALUES (?, ?, ?, ?);
+            INSERT INTO Dataset (id, name, description, info, createDateTime,
+                updateDateTime)
+            VALUES (?, ?, ?, ?, ?, ?);
         """
         cursor = self._dbConnection.cursor()
         try:
             cursor.execute(sql, (
                 dataset.getId(), dataset.getLocalId(),
                 dataset.getDescription(),
-                json.dumps(dataset.getInfo())))
+                json.dumps(dataset.getInfo()), dataset.getCreateDateTime(),
+                dataset.getUpdateDateTime()))
         except sqlite3.IntegrityError:
             raise exceptions.DuplicateNameException(dataset.getLocalId())
 

ga4gh/server/exceptions.py

@@ -571,6 +571,15 @@ def __init__(self, filename, message):
             filename, message)
 
 
+class TimestampFormatException(DataException):
+    """
+    Exception thrown for invalid datetimes.
+    """
+    def __init__(self, timestamp, source):
+        self.message = "Error reading timestamp '{}' from {}".format(
+            timestamp, source)
+
+
 class MalformedException(DataException):
     """
     A base exception class for exceptions thrown when faulty VCF file

requirements.txt

@@ -52,6 +52,7 @@ requests==2.7.0
 oic==0.7.6
 pyOpenSSL==0.15.1
 lxml==3.4.4
+python-dateutil==2.5.1
 
 # We need sphinx-argparse to build on readthedocs.
 sphinx-argparse==0.1.15

scripts/prepare_compliance_data.py

@@ -138,6 +138,8 @@ def run(self):
         dataset = datasets.Dataset("brca1")
         # Some info is set, it isn't important what
         dataset.setInfo({"version": ga4gh.server.__version__})
+        dataset.setCreateDateTime('2008-04-25T18:30Z')
+        dataset.setUpdateDateTime('2016-05-15T17:34-01:00')
         self.repo.insertDataset(dataset)
 
         hg00096Individual = biodata.Individual(dataset, "HG00096")

tests/unit/test_datasets.py

@@ -6,8 +6,10 @@
 from __future__ import unicode_literals
 
 import unittest
+from datetime import datetime
 
 import ga4gh.server.datamodel.datasets as datasets
+import ga4gh.server.exceptions as exceptions
 
 
 class TestDatasets(unittest.TestCase):
@@ -18,7 +20,19 @@ def testToProtocolElement(self):
         datasetId = 'ds1'
         dataset = datasets.SimulatedDataset(datasetId, 1, 2, 3, 4, 5)
         dataset.setInfo({"test": "test"})
+        createDT = '2015-01-01T00:00:00'  # set in simulated
+        updateDT = datetime.utcnow().isoformat()
+        dataset.setUpdateDateTime(updateDT)
         gaDataset = dataset.toProtocolElement()
         self.assertIsNotNone(gaDataset.info)
         self.assertEqual(gaDataset.info['test'].values[0].string_value, "test")
         self.assertEqual(dataset.getId(), gaDataset.id)
+        self.assertEqual(gaDataset.createDateTime, createDT)
+        self.assertEqual(gaDataset.updateDateTime, updateDT)
+
+    def testBadDateTime(self):
+        dataset = datasets.SimulatedDataset('ds1', 1, 2, 3, 4, 5)
+        badDatetime = '215010100:00:00'
+        self.assertRaises(
+                exceptions.TimestampFormatException,
+                dataset.setUpdateDateTime, badDatetime)