Upsert dataset

pyscicat/client.py

@@ -226,6 +226,124 @@ def upload_derived_dataset(self, dataset: Dataset) -> str:
         logger.info(f"new dataset created {new_pid}")
         return new_pid
 
+    def upsert_dataset(self, dataset: Dataset, filter_fields) -> str:
+        """Upsert a dataset
+
+        Parameters
+        ----------
+        dataset : Dataset
+            Dataset to load
+
+        filter_fields
+            Filters to locate where to upsert dataset
+
+        Returns
+        -------
+        str
+            pid (or unique identifier) of the dataset
+
+        Raises
+        ------
+        ScicatCommError
+            Raises if a non-20x message is returned
+        """
+        query_results = self.get_datasets(filter_fields)
+        if query_results:
+            filter_fields = json.dumps(filter_fields)
+            if isinstance(dataset, RawDataset):
+                dataset_url = f'{self._base_url}/RawDatasets/upsertWithWhere?where={{"where":{filter_fields}}}'
+            elif isinstance(dataset, DerivedDataset):
+                dataset_url = f'{self._base_url}/DerivedDatasets/upsertWithWhere?where={{"where":{filter_fields}}}'
+            else:
+                logging.error(
+                    "Dataset type not recognized, not Raw or Derived type"
+                )
+            resp = self._send_to_scicat(dataset_url, dataset.dict(exclude_none=True))
+            if not resp.ok:
+                err = resp.json()["error"]
+                raise ScicatCommError(f"Error upserting dataset {err}")
+            new_pid = resp.json().get("pid")
+            logger.info(f"dataset updated {new_pid}")
+            return new_pid
+        else:
+            logger.info("dataset does not exist, could not upsert")
+            raise ScicatCommError("Dataset does not exist, could not upsert.")
+
+    def upsert_raw_dataset(self, dataset: Dataset, filter_fields) -> str:
+        """Upsert a raw dataset
+
+        Parameters
+        ----------
+        dataset : Dataset
+            Dataset to load
+
+        filter_fields
+            Filters to locate where to upsert dataset
+
+        Returns
+        -------
+        str
+            pid (or unique identifier) of the dataset
+
+        Raises
+        ------
+        ScicatCommError
+            Raises if a non-20x message is returned
+        """
+        query_results = self.get_datasets(filter_fields)
+        if query_results:
+            filter_fields = json.dumps(filter_fields)
+            raw_dataset_url = f'{self._base_url}/RawDatasets/upsertWithWhere?where={{"where":{filter_fields}}}'
+            resp = self._send_to_scicat(raw_dataset_url, dataset.dict(exclude_none=True))
+            if not resp.ok:
+                err = resp.json()["error"]
+                raise ScicatCommError(f"Error upserting raw dataset {err}")
+            new_pid = resp.json().get("pid")
+            logger.info(f"dataset updated {new_pid}")
+            return new_pid
+        else:
+            logger.info("dataset does not exist, could not upsert")
+            raise ScicatCommError("Dataset does not exist, could not upsert.")
+
+    def upsert_derived_dataset(self, dataset: Dataset, filter_fields) -> str:
+        """Upsert a derived dataset
+
+        Parameters
+        ----------
+        dataset : Dataset
+            Dataset to upsert
+
+        filter_fields
+            Filters to locate where to upsert dataset
+
+        Returns
+        -------
+        str
+            pid (or unique identifier) of the dataset
+
+        Raises
+        ------
+        ScicatCommError
+            Raises if a non-20x message is returned
+        """
+
+        query_results = self.get_datasets(filter_fields)
+        if query_results:
+            filter_fields = json.dumps(filter_fields)
+            derived_dataset_url = f'{self._base_url}/DerivedDatasets/upsertWithWhere?where={{"where":{filter_fields}}}'
+            resp = self._send_to_scicat(
+                derived_dataset_url, dataset.dict(exclude_none=True)
+            )
+            if not resp.ok:
+                err = resp.json()["error"]
+                raise ScicatCommError(f"Error upserting derived dataset {err}")
+            new_pid = resp.json().get("pid")
+            logger.info(f"dataset updated {new_pid}")
+            return new_pid
+        else:
+            logger.info("dataset does not exist, could not upsert")
+            raise ScicatCommError("Dataset does not exist, could not upsert.")
+
     def upload_datablock(self, datablock: Datablock, datasetType: str = "RawDatasets"):
         """Upload a Datablock
 

pyscicat/tests/test_client.py

@@ -15,6 +15,7 @@
     Datablock,
     DataFile,
     Dataset,
+    RawDataset,
     Ownable,
 )
 
@@ -28,6 +29,10 @@ def add_mock_requests(mock_request):
     )
     mock_request.post(local_url + "Samples", json={"sampleId": "dataset_id"})
     mock_request.post(local_url + "RawDatasets/replaceOrCreate", json={"pid": "42"})
+    mock_request.get(local_url + "/Datasets/?filter=%7B%22where%22:%7B%22sampleId%22:%20%22gargleblaster%22%7D%7D", 
+        json = {"response": "random"})
+    mock_request.post(local_url + "/RawDatasets/upsertWithWhere?where=%7B%22where%22:%7B%22sampleId%22:%20%22gargleblaster%22%7D%7D", 
+        json={"pid": "42"})
     mock_request.post(
         local_url + "RawDatasets/42/origdatablocks",
         json={"response": "random"},
@@ -59,7 +64,7 @@ def test_scicate_ingest():
         assert size is not None
 
         # RawDataset
-        dataset = Dataset(
+        dataset = RawDataset(
             path="/foo/bar",
             size=42,
             owner="slartibartfast",
@@ -78,6 +83,28 @@ def test_scicate_ingest():
         )
         dataset_id = scicat.upload_raw_dataset(dataset)
 
+        # new dataset
+        dataset = Dataset(
+            path="/foo/bar",
+            size=42,
+            owner="slartibartfast",
+            contactEmail="slartibartfast@magrathea.org",
+            creationLocation="magrathea",
+            creationTime=str(datetime.now()),
+            type="raw",
+            instrumentId="earth",
+            proposalId="deepthought",
+            dataFormat="planet",
+            principalInvestigator="A. Mouse",
+            sourceFolder="/foo/bar",
+            scientificMetadata={"a": "newfield"},
+            sampleId="gargleblaster",
+            **ownable.dict()
+        )
+
+        dataset_id = scicat.upsert_raw_dataset(dataset, {"sampleId": "gargleblaster"})
+        assert dataset_id == "42"
+
         # Datablock with DataFiles
         data_file = DataFile(path="/foo/bar", size=42)
         data_block = Datablock(