919 Add DOI support

backend/app/config.py

@@ -88,5 +88,10 @@ class Settings(BaseSettings):
     # defautl listener heartbeat time interval in seconds 5 minutes
     listener_heartbeat_interval = 5 * 60
 
+    # DOI datacite details
+    DOI_ENABLED = True
+    DATACITE_TEST_URL = "https://api.test.datacite.org/dois"
+    DATACITE_URL = "https://api.datacite.org/dois"
+
 
 settings = Settings()

backend/app/models/datasets.py

@@ -44,6 +44,7 @@ class DatasetBaseCommon(DatasetBase):
     origin_id: Optional[PydanticObjectId] = None
     standard_license: bool = True
     license_id: Optional[str] = None
+    doi: Optional[str] = None
 
 
 class DatasetPatch(BaseModel):

backend/app/routers/datasets.py

@@ -52,6 +52,7 @@
 from app.models.users import UserOut
 from app.rabbitmq.listeners import submit_dataset_job
 from app.routers.authentication import get_admin, get_admin_mode
+from app.routers.doi import DataCiteClient
 from app.routers.files import add_file_entry, add_local_file_entry
 from app.routers.licenses import delete_license
 from app.search.connect import delete_document_by_id
@@ -482,9 +483,52 @@ async def delete_dataset(
     raise HTTPException(status_code=404, detail=f"Dataset {dataset_id} not found")
 
 
+@router.post("/{dataset_id}/doi", response_model=DatasetOut)
+async def mint_doi(
+    dataset_id: str,
+    user=Depends(get_current_user),
+    fs: Minio = Depends(dependencies.get_fs),
+    es: Elasticsearch = Depends(dependencies.get_elasticsearchclient),
+    allow: bool = Depends(Authorization(RoleType.OWNER)) and settings.DOI_ENABLED,
+):
+    if (dataset := await DatasetFreezeDB.get(PydanticObjectId(dataset_id))) is not None:
+        metadata = {
+            "data": {
+                "type": "dois",
+                "attributes": {
+                    "prefix": os.getenv("DATACITE_PREFIX"),
+                    "url": f"{settings.frontend_url}/datasets/{dataset_id}",
+                    "titles": [{"title": dataset.name}],
+                    "creators": [
+                        {
+                            "name": dataset.creator.first_name
+                            + " "
+                            + dataset.creator.last_name
+                        }
+                    ],
+                    "publisher": "DataCite e.V.",
+                    "publicationYear": datetime.datetime.now().year,
+                },
+            }
+        }
+        dataCiteClient = DataCiteClient()
+        response = dataCiteClient.create_doi(metadata)
+        print("doi created:", response.get("data").get("id"))
+        dataset.doi = response.get("data").get("id")
+        dataset.modified = datetime.datetime.utcnow()
+        await dataset.save()
+
+        # TODO: if we ever index freeze datasets
+        # await index_dataset(es, DatasetOut(**dataset_db), update=True)
+        return dataset.dict()
+    else:
+        raise HTTPException(status_code=404, detail=f"Dataset {dataset_id} not found")
+
+
 @router.post("/{dataset_id}/freeze", response_model=DatasetFreezeOut)
 async def freeze_dataset(
     dataset_id: str,
+    publish_doi: bool = False,
     user=Depends(get_current_user),
     fs: Minio = Depends(dependencies.get_fs),
     es: Elasticsearch = Depends(dependencies.get_elasticsearchclient),
@@ -545,6 +589,9 @@ async def freeze_dataset(
 
         # TODO thumbnails, visualizations
 
+        if publish_doi:
+            return await mint_doi(frozen_dataset.id)
+
         return frozen_dataset.dict()
 
     raise HTTPException(status_code=404, detail=f"Dataset {dataset_id} not found")

backend/app/routers/doi.py

@@ -0,0 +1,51 @@
+import os
+
+import requests
+from requests.auth import HTTPBasicAuth
+
+
+class DataCiteClient:
+    def __init__(self, test_mode=False):
+        self.auth = HTTPBasicAuth(
+            os.getenv("DATACITE_USERNAME"), os.getenv("DATACITE_PASSWORD")
+        )
+        self.headers = {"Content-Type": "application/vnd.api+json"}
+        self.base_url = (
+            "https://api.test.datacite.org/"
+            if test_mode
+            else "https://api.datacite.org/"
+        )
+
+    def create_doi(self, metadata):
+        url = f"{self.base_url}dois"
+        response = requests.post(
+            url, auth=self.auth, headers=self.headers, json=metadata
+        )
+        return response.json()
+
+    def get_all_dois(self):
+        url = f"{self.base_url}dois"
+        response = requests.get(url, auth=self.auth, headers=self.headers)
+        return response.json()
+
+    def get_doi(self, doi):
+        url = f"{self.base_url}dois/{doi}"
+        response = requests.get(url, auth=self.auth, headers=self.headers)
+        return response.json()
+
+    def update_doi(self, doi, metadata):
+        url = f"{self.base_url}dois/{doi}"
+        response = requests.put(
+            url, auth=self.auth, headers=self.headers, json=metadata
+        )
+        return response.json()
+
+    def delete_doi(self, doi):
+        url = f"{self.base_url}dois/{doi}"
+        response = requests.delete(url, auth=self.auth, headers=self.headers)
+        return response.status_code == 204
+
+    def get_doi_activity_status(self, doi):
+        url = f"{self.base_url}events?doi={doi}"
+        response = requests.get(url, auth=self.auth, headers=self.headers)
+        return response.json()

backend/app/tests/test_datasets.py

@@ -23,6 +23,15 @@ def test_get_one(client: TestClient, headers: dict):
     assert response.json().get("id") is not None
 
 
+def test_create_doi(client: TestClient, headers: dict):
+    dataset_id = create_dataset(client, headers).get("id")
+    response = client.get(
+        f"{settings.API_V2_STR}/datasets/{dataset_id}", headers=headers
+    )
+    assert response.status_code == 200
+    assert response.json().get("id") is not None
+
+
 def test_delete(client: TestClient, headers: dict):
     dataset_id = create_dataset(client, headers).get("id")
     response = client.delete(

docs/docs/devs/dataset-versioning.md

@@ -128,8 +128,12 @@ for client consumption. These views include:
     - providing users with greater control over dataset management
 - **Forbidden Modifications**: Prevent modifications to a released dataset.
 
-## Future Enhancements
+## Digital Object Identifier (DOI) Integration
+Currently, the feature to generate DOI through [DataCite](https://datacite.org/) is integrated with Clowder. The user is provided this option
+when they release a dataset. Clowder then talks with the DataCite API to mint a DOI for the released dataset and
+submits some metadata about the dataset like its title, URL, and creator details. The generated DOI is displayed in the
+dataset page in the Details section.
 
-- **Mint DOI**: Integrate DOI support to allow minting Digital Object Identifiers (DOIs) for each dataset version,
-  ensuring unique and persistent
-  identification ([Issue #919](https://github.com/clowder-framework/clowder2/issues/919)).
+## Future Enhancements
+- **Add support for CrossRef when generate DOI**: Currently, Clowder supports DataCite for minting DOIs. We might need to
+  integrate CrossRef to provide users with more options, as some users may already have an account with CrossRef.

frontend/src/actions/dataset.js

@@ -186,10 +186,11 @@ export function updateDataset(datasetId, formData) {
 
 export const FREEZE_DATASET = "FREEZE_DATASET";
 
-export function freezeDataset(datasetId) {
+export function freezeDataset(datasetId, publishDOI = false) {
 	return (dispatch) => {
 		return V2.DatasetsService.freezeDatasetApiV2DatasetsDatasetIdFreezePost(
-			datasetId
+			datasetId,
+			publishDOI
 		)
 			.then((json) => {
 				dispatch({

frontend/src/app.config.ts

@@ -33,6 +33,7 @@ interface Config {
 	defaultExtractionJobs: number;
 	defaultMetadataDefintionPerPage: number;
 	defaultVersionPerPage: number;
+	enableDOI: boolean;
 }
 
 const config: Config = <Config>{};
@@ -100,5 +101,6 @@ config["defaultFeeds"] = 5;
 config["defaultExtractionJobs"] = 5;
 config["defaultMetadataDefintionPerPage"] = 5;
 config["defaultVersionPerPage"] = 3;
+config["enableDOI"] = true;
 
 export default config;

frontend/src/components/datasets/Dataset.tsx

@@ -604,6 +604,21 @@ export const Dataset = (): JSX.Element => {
 						) : (
 							<></>
 						)}
+						<br />
+						{dataset.doi && dataset.doi !== undefined ? (
+							<div>
+								<Typography variant="h5" gutterBottom>
+									DOI
+								</Typography>
+								<Typography>
+									<Link href={`https://doi.org/${dataset.doi}`}>
+										https://doi.org/{dataset.doi}
+									</Link>
+								</Typography>
+							</div>
+						) : (
+							<></>
+						)}
 					</>
 					<DatasetDetails details={dataset} myRole={datasetRole.role} />
 				</Grid>

frontend/src/components/datasets/OtherMenu.tsx

@@ -10,6 +10,7 @@ import {
 } from "@mui/material";
 import ArrowDropDownIcon from "@mui/icons-material/ArrowDropDown";
 import { ActionModal } from "../dialog/ActionModal";
+import { ActionModalWithCheckbox } from "../dialog/ActionModalWithCheckbox";
 import {
 	datasetDeleted,
 	freezeDataset as freezeDatasetAction,
@@ -46,8 +47,10 @@ export const OtherMenu = (props: ActionsMenuProps): JSX.Element => {
 	const datasetRole = useSelector(
 		(state: RootState) => state.dataset.datasetRole
 	);
-	const freezeDataset = (datasetId: string | undefined) =>
-		dispatch(freezeDatasetAction(datasetId));
+	const freezeDataset = (
+		datasetId: string | undefined,
+		publishDOI: boolean | undefined
+	) => dispatch(freezeDatasetAction(datasetId, publishDOI));
 
 	const listGroups = () => dispatch(fetchGroups(0, 21));
 
@@ -87,6 +90,7 @@ export const OtherMenu = (props: ActionsMenuProps): JSX.Element => {
 	};
 
 	const [anchorEl, setAnchorEl] = React.useState<Element | null>(null);
+	const [publishDOI, setPublishDOI] = React.useState<boolean>(false);
 
 	const handleOptionClick = (event: React.MouseEvent<any>) => {
 		setAnchorEl(event.currentTarget);
@@ -124,13 +128,16 @@ export const OtherMenu = (props: ActionsMenuProps): JSX.Element => {
 				}}
 				actionLevel={"error"}
 			/>
-			<ActionModal
+			<ActionModalWithCheckbox
 				actionOpen={freezeDatasetConfirmOpen}
 				actionTitle="Are you ready to release this version of the dataset?"
-				actionText="By proceeding with the release, you will lock in the current content of the dataset, including all associated files, folders, metadata, and visualizations. Once released, these elements will be set as final and cannot be altered. However, you can continue to make edits and improvements on the ongoing version of the dataset."
+				actionText="By proceeding with the release, you will lock in the current content of the dataset, including all associated files, folders, metadata, and visualizations. Once released, these elements will be set as final and cannot be altered. However, you can continue to make edits and improvements on the ongoing version of the dataset. Optionally, you can also generate a Digital Object Identifier (DOI) by selecting the checkbox below. It will be displayed in the dataset page in the Details section."
+				checkboxLabel="Generate a DOI for this version of the dataset."
+				checkboxSelected={publishDOI}
+				setCheckboxSelected={setPublishDOI}
 				actionBtnName="Release"
 				handleActionBtnClick={() => {
-					freezeDataset(datasetId);
+					freezeDataset(datasetId, publishDOI);
 					setFreezeDatasetConfirmOpen(false);
 				}}
 				handleActionCancel={() => {

frontend/src/components/dialog/ActionModalWithCheckbox.tsx

@@ -0,0 +1,82 @@
+import React from "react";
+import {
+	Button,
+	Checkbox,
+	Dialog,
+	DialogActions,
+	DialogContent,
+	DialogContentText,
+	DialogTitle,
+	FormControlLabel,
+} from "@mui/material";
+
+type ActionLevel = "error" | "warning" | "info";
+
+type ActionModalProps = {
+	actionOpen: boolean;
+	actionTitle: string;
+	actionText: string;
+	checkboxLabel: string;
+	checkboxSelected: boolean;
+	publishDOI: boolean;
+	setCheckboxSelected: (value: boolean) => void;
+	actionBtnName: string;
+	handleActionBtnClick: () => void;
+	handleActionCancel: () => void;
+	actionLevel?: ActionLevel;
+};
+
+export const ActionModalWithCheckbox: React.FC<ActionModalProps> = (
+	props: ActionModalProps
+) => {
+	const {
+		actionOpen,
+		actionTitle,
+		actionText,
+		checkboxLabel,
+		checkboxSelected,
+		setCheckboxSelected,
+		actionBtnName,
+		handleActionBtnClick,
+		handleActionCancel,
+		actionLevel,
+	} = props;
+
+	return (
+		<Dialog open={actionOpen} onClose={handleActionCancel} maxWidth={"sm"}>
+			<DialogTitle id="confirmation-dialog-title">{actionTitle}</DialogTitle>
+			<DialogContent>
+				<DialogContentText>{actionText}</DialogContentText>
+				<br />
+				<FormControlLabel
+					value={"end"}
+					control={
+						<Checkbox
+							defaultChecked={checkboxSelected}
+							onChange={() => {
+								setCheckboxSelected(!checkboxSelected);
+							}}
+						/>
+					}
+					label={checkboxLabel}
+				/>
+			</DialogContent>
+			<DialogActions>
+				<Button
+					variant="outlined"
+					onClick={handleActionCancel}
+					color={actionLevel ?? "primary"}
+				>
+					Cancel
+				</Button>
+				<Button
+					variant="contained"
+					onClick={handleActionBtnClick}
+					color={actionLevel ?? "primary"}
+				>
+					{actionBtnName}
+				</Button>
+			</DialogActions>
+		</Dialog>
+	);
+};

frontend/src/openapi/v2/models/DatasetFreezeOut.ts

@@ -30,6 +30,7 @@ export type DatasetFreezeOut = {
     origin_id?: string;
     standard_license?: boolean;
     license_id?: string;
+    doi?: string;
     id?: string;
     frozen?: boolean;
     frozen_version_num: number;

frontend/src/openapi/v2/models/DatasetOut.ts

@@ -30,6 +30,7 @@ export type DatasetOut = {
     origin_id?: string;
     standard_license?: boolean;
     license_id?: string;
+    doi?: string;
     id?: string;
     frozen?: boolean;
     frozen_version_num?: number;