Add experiment checking before modifying the dataset

DashAI/back/api/api_v1/endpoints/datasets.py

@@ -25,7 +25,7 @@
     to_dashai_dataset,
     update_columns_spec,
 )
-from DashAI.back.dependencies.database.models import Dataset
+from DashAI.back.dependencies.database.models import Dataset, Experiment
 from DashAI.back.dependencies.registry import ComponentRegistry
 
 logger = logging.getLogger(__name__)
@@ -182,6 +182,47 @@ async def get_info(
     return info
 
 
+@router.get("/{dataset_id}/experiments-exist")
+@inject
+async def get_experiments_exist(
+    dataset_id: int,
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+):
+    """Get a boolean indicating if there are experiments associated with the dataset.
+    
+    Parameters
+    ----------
+    dataset_id : int
+        id of the dataset to query.
+
+    Returns
+    -------
+    bool
+        True if there are experiments associated with the dataset, False otherwise.
+    """
+    with session_factory() as db:
+        try:
+            dataset = db.get(Dataset, dataset_id)
+            if not dataset:
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail="Dataset not found",
+                )
+            # Check if there are any experiments associated with the dataset
+            experiments_exist = db.query(Experiment).filter(
+                Experiment.dataset_id == dataset_id
+            ).first() is not None
+
+            return experiments_exist
+
+        except exc.SQLAlchemyError as e:
+            logger.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Internal database error",
+            ) from e
+
+
 @router.get("/{dataset_id}/types")
 @inject
 async def get_types(
@@ -344,6 +385,75 @@ async def upload_dataset(
     logger.debug("Dataset creation sucessfully finished.")
     return new_dataset
 
+@router.post("/copy", status_code=status.HTTP_201_CREATED)
+@inject
+async def copy_dataset(
+    dataset: Dict[str, int],
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+    config: Dict[str, Any] = Depends(lambda: di["config"]),
+):
+    """Copy an existing dataset to create a new one.
+
+    Parameters
+    ----------
+    dataset_id : int
+        ID of the dataset to copy.
+
+    Returns
+    -------
+    Dataset
+        The newly created dataset.
+    """
+    dataset_id = dataset["dataset_id"]
+    print("COPYING DATASET", dataset_id)
+    logger.debug(f"Copying dataset with ID {dataset_id}.")
+
+    with session_factory() as db:
+        # Retrieve the existing dataset
+        original_dataset = db.query(Dataset).filter(Dataset.id == dataset_id).first()
+        if not original_dataset:
+            raise HTTPException(
+                status_code=status.HTTP_404_NOT_FOUND,
+                detail="Original dataset not found.",
+            )
+
+        # Create a new folder for the copied dataset
+        new_name = f"{original_dataset.name}_copy"
+        new_folder_path = config["DATASETS_PATH"] / new_name
+        try:
+            shutil.copytree(original_dataset.file_path, new_folder_path)
+        except FileExistsError:
+            raise HTTPException(
+                status_code=status.HTTP_409_CONFLICT,
+                detail=f"A dataset with the name '{new_name}' already exists.",
+            )
+        except Exception as e:
+            logger.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Failed to copy dataset files.",
+            ) from e
+
+        # Save metadata for the new dataset
+        try:
+            new_dataset = Dataset(
+                name=new_name,
+                file_path=str(new_folder_path),
+            )
+            db.add(new_dataset)
+            db.commit()
+            db.refresh(new_dataset)
+        except exc.SQLAlchemyError as e:
+            logger.exception(e)
+            shutil.rmtree(new_folder_path, ignore_errors=True)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Internal database error.",
+            ) from e
+
+    logger.debug(f"Dataset copied successfully to '{new_name}'.")
+    return new_dataset
+
 
 @router.delete("/{dataset_id}")
 @inject

DashAI/front/src/api/datasets.ts

@@ -8,6 +8,11 @@ export const uploadDataset = async (formData: object): Promise<object> => {
   return response.data;
 };
 
+export const copyDataset = async (formData: object): Promise<object> => {
+  const response = await api.post<object>(`${datasetEndpoint}/copy`, formData);
+  return response.data;
+};
+
 export const getDatasets = async (): Promise<IDataset[]> => {
   const response = await api.get<IDataset[]>(datasetEndpoint);
   return response.data;
@@ -28,6 +33,13 @@ export const getDatasetInfo = async (id: number): Promise<object> => {
   return response.data;
 };
 
+export const getExperimentsExist = async (id: number): Promise<object> => {
+  const response = await api.get<object>(
+    `${datasetEndpoint}/${id}/experiments-exist`,
+  );
+  return response.data;
+};
+
 export const updateDataset = async (
   id: number,
   formData: object,

DashAI/front/src/components/datasets/ConvertDatasetModal.jsx

@@ -26,12 +26,16 @@ import {
   getDatasetConverterList,
 } from "../../api/converter";
 import { ConverterListStatus } from "../../types/converter";
+import { getExperimentsExist } from "../../api/datasets";
+import CopyDatasetModal from "./CopyDatasetModal";
 
 function ConvertDatasetModal({ datasetId }) {
   const { enqueueSnackbar } = useSnackbar();
   const [open, setOpen] = useState(false);
   const [targetColumnIndex, setTargetColumnIndex] = useState(null);
   const [convertersToApply, setConvertersToApply] = useState([]);
+  const [openCopyModal, setOpenCopyModal] = useState(false);
+  const [datasetIdToModify, setDatasetIdToModify] = useState(datasetId);
   const [converterListId, setConverterListId] = useState(null);
   const [converterListStatus, setConverterListStatus] = useState(null);
   const [running, setRunning] = useState(false);
@@ -73,11 +77,11 @@ function ConvertDatasetModal({ datasetId }) {
     }
   };
 
-  const handleSaveConfig = async () => {
+  const saveAndEnqueueConverterList = async (id) => {
     try {
       // Save the list of converters to apply
       const response = await saveDatasetConverterList(
-        datasetId,
+        id,
         convertersToApply.reduce((acc, { name, params, scope, pipelineId }) => {
           acc[name] = {
             params: params,
@@ -117,14 +121,31 @@ function ConvertDatasetModal({ datasetId }) {
     }
   };
 
+  const handleSaveConfig = async () => {
+    // Check if there are experiments associated with the dataset
+    try {
+      const hasExperiments = await getExperimentsExist(datasetIdToModify);
+      if (hasExperiments) {
+        setOpenCopyModal(true);
+      } else {
+        await saveAndEnqueueConverterList(datasetIdToModify);
+      }
+    } catch (error) {
+      enqueueSnackbar(
+        "Error while trying to check if there are experiments associated with the dataset",
+        {
+          variant: "error",
+        },
+      );
+    }
+  };
+
   const getConverterListStatus = async () => {
     getDatasetConverterList(converterListId)
       .then((convertersFromDB) => {
-        console.log(convertersFromDB);
         setConverterListStatus(convertersFromDB.status);
       })
       .catch((error) => {
-        console.log(error);
         enqueueSnackbar("Error while trying to fetch converters", {
           variant: "error",
         });
@@ -193,7 +214,7 @@ function ConvertDatasetModal({ datasetId }) {
                   Dataset summary
                 </Typography>
               </Grid>
-              <DatasetSummaryTable datasetId={datasetId} />
+              <DatasetSummaryTable datasetId={datasetIdToModify} />
 
               {/* Converter selector */}
               <Grid item xs={12} display={"flex"} alignItems={"center"} gap={2}>
@@ -244,7 +265,7 @@ function ConvertDatasetModal({ datasetId }) {
               </Grid>
               {/* Selected converters table */}
               <ConverterTable
-                datasetId={datasetId}
+                datasetId={datasetIdToModify}
                 convertersToApply={convertersToApply}
                 setConvertersToApply={setConvertersToApply}
               />
@@ -269,6 +290,14 @@ function ConvertDatasetModal({ datasetId }) {
           </Button>
         </DialogActions>
       </Dialog>
+      {/* Modal to make a copy of the dataset */}
+      <CopyDatasetModal
+        datasetId={datasetIdToModify}
+        updateDatasetId={setDatasetIdToModify}
+        open={openCopyModal}
+        setOpen={setOpenCopyModal}
+        modifyDataset={saveAndEnqueueConverterList}
+      />
     </React.Fragment>
   );
 }

DashAI/front/src/components/datasets/CopyDatasetModal.jsx

@@ -0,0 +1,81 @@
+import React from "react";
+import PropTypes from "prop-types";
+import { useSnackbar } from "notistack";
+import {
+  Button,
+  Dialog,
+  DialogActions,
+  DialogContent,
+  DialogContentText,
+  DialogTitle,
+} from "@mui/material";
+import { copyDataset } from "../../api/datasets";
+
+function CopyDatasetModal({
+  datasetId,
+  updateDatasetId,
+  open,
+  setOpen,
+  modifyDataset,
+}) {
+  const { enqueueSnackbar } = useSnackbar();
+
+  const handleDatasetModification = async (id) => {
+    modifyDataset(id);
+    setOpen(false);
+  };
+
+  const handleDatasetCopyModification = async () => {
+    try {
+      const datasetCopy = await copyDataset({
+        dataset_id: datasetId,
+      });
+      updateDatasetId(datasetCopy.id);
+      enqueueSnackbar("Dataset copied successfully", {
+        variant: "success",
+      });
+
+      handleDatasetModification(datasetCopy.id);
+    } catch (error) {
+      enqueueSnackbar("Error while trying to create a copy of the dataset.");
+    } finally {
+      setOpen(false);
+    }
+  };
+  return (
+    <Dialog open={open} onClose={() => setOpen(false)}>
+      <DialogTitle>Existing experiments</DialogTitle>
+      <DialogContent>
+        <DialogContentText>
+          This dataset is currently used in existing experiments. Modifying it
+          may impact the results of re-running those experiments. Would you like
+          to create and modify a copy of this dataset instead?
+        </DialogContentText>
+      </DialogContent>
+      <DialogActions>
+        <Button onClick={() => setOpen(false)} autoFocus>
+          Cancel
+        </Button>
+        <Button onClick={() => handleDatasetModification(datasetId)}>
+          Modify anyway
+        </Button>
+        <Button
+          onClick={handleDatasetCopyModification}
+          variant="contained"
+          color="primary"
+        >
+          Make a copy
+        </Button>
+      </DialogActions>
+    </Dialog>
+  );
+}
+CopyDatasetModal.propTypes = {
+  datasetId: PropTypes.number.isRequired,
+  updateDatasetId: PropTypes.func.isRequired,
+  open: PropTypes.bool.isRequired,
+  setOpen: PropTypes.func.isRequired,
+  modifyDataset: PropTypes.func.isRequired,
+};
+
+export default CopyDatasetModal;