Health data set upload

api/data_ingestion/models/file_upload.py

@@ -60,10 +60,13 @@ def filename(self) -> str:
             country = self.country
 
         if self.dataset == "structured":
-            # For structured datasets, use original filename with upload ID
             original_name = Path(self.original_filename).stem
             filename = f"{original_name}_{self.id}"
             return f"{filename}{ext}"
+        if self.dataset == "health":
+            # {ISO3}_{original_stem}_{timestamp}.csv under health-master/<ISO3>/
+            stem = Path(self.original_filename).stem or "health_upload"
+            return f"{country}_{stem}_{timestamp}{ext}"
         else:
             filename_elements = [self.id, country, self.dataset]
             if self.source is not None and self.dataset != "geolocation":
@@ -80,6 +83,15 @@ def upload_path(self) -> str:
 
             return f"{settings.LAKEHOUSE_PATH}/raw/custom-dataset/{self.filename}"
 
+        if self.dataset == "health":
+            # Blob path within AZURE_BLOB_CONTAINER_NAME (no leading slash):
+            # updated_master_schema/health-master/<ISO3 or $NA>/<ISO3>_<stem>_<timestamp>.csv
+            country_segment = "$NA" if self.country == "N/A" else self.country
+            return (
+                f"updated_master_schema/health-master/"
+                f"{country_segment}/{self.filename}"
+            )
+
         # For other datasets, use the uploads path
         if self.dataset == "unstructured":
             dataset_path = "unstructured"

api/data_ingestion/routers/upload.py

@@ -512,6 +512,13 @@ async def upload_unstructured(  # noqa: C901
     )
     db.add(file_upload)
     await db.commit()
+    await db.refresh(file_upload)
+
+    # Keep parity with school uploads: persist a sidecar JSON metadata path.
+    metadata_file_path = get_metadata_path(file_upload.upload_path)
+    file_upload.metadata_json_path = metadata_file_path
+    db.add(file_upload)
+    await db.commit()
 
     client = storage_client.get_blob_client(file_upload.upload_path)
 
@@ -531,6 +538,11 @@ async def upload_unstructured(  # noqa: C901
             metadata=metadata,
             content_settings=ContentSettings(content_type=file_type),
         )
+        metadata_blob_client = storage_client.get_blob_client(
+            file_upload.metadata_json_path
+        )
+        metadata_json_bytes = json.dumps(metadata, indent=2).encode()
+        metadata_blob_client.upload_blob(metadata_json_bytes, overwrite=True)
         response.status_code = status.HTTP_201_CREATED
     except HttpResponseError as err:
         raise HTTPException(
@@ -591,6 +603,17 @@ async def upload_structured(  # noqa: C901
             detail="File extension must be .csv for structured datasets.",
         )
 
+    portal_ds = (form.portal_dataset or "").strip().lower()
+    if portal_ds == "health":
+        dataset_label = "health"
+    elif portal_ds == "":
+        dataset_label = "structured"
+    else:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail="Invalid portal_dataset value.",
+        )
+
     # For structured datasets, always use "N/A" as country
     if form.country == "Global Dataset":
         country_code = "N/A"
@@ -608,7 +631,7 @@ async def upload_structured(  # noqa: C901
         uploader_id=database_user.id,
         uploader_email=database_user.email,
         country=country_code,
-        dataset="structured",
+        dataset=dataset_label,
         original_filename=file.filename,
         column_to_schema_mapping={},
         column_license={},
@@ -624,7 +647,7 @@ async def upload_structured(  # noqa: C901
             **{str(k): str(v) for k, v in orjson.loads(form.metadata).items()},
             "country": form.country,
             "uploader_email": email,
-            "dataset_type": "structured",
+            "dataset_type": dataset_label,
         }
 
         if form.source is not None:

api/data_ingestion/schemas/upload.py

@@ -61,6 +61,8 @@ class UnstructuredFileUploadRequest:
     country: str = Form(...)
     metadata: str = Form(...)
     source: str | None = Form(None)
+    # When "health", stores dataset=health and uses the health raw path (see upload_structured).
+    portal_dataset: str | None = Form(None)
 
 
 @dataclass

ui/src/components/upload/Health.tsx

@@ -0,0 +1,30 @@
+import {
+  BaseUploadMetadataForm,
+  UploadMetadataFormProps,
+} from "@/components/upload/uploadMetadataFormBase.tsx";
+import {
+  health,
+  healthMetadataDatasetSection,
+  healthMetadataNationalSection,
+} from "@/constants/metadata";
+
+const HEALTH_INTRO = {
+  title: "Add health metadata",
+  paragraphs: [
+    "Provide context for this health dataset: who compiled or uploaded it, what period it covers, and how it was collected.",
+    "Required fields include country, health dataset description, focal point (person uploading or responsible), data owner, and the year the data refers to.",
+  ],
+};
+
+export function Health(props: UploadMetadataFormProps) {
+  return (
+    <BaseUploadMetadataForm
+      {...props}
+      mapping={health}
+      datasetSectionHeading={healthMetadataDatasetSection}
+      nationalPracticesHeading={healthMetadataNationalSection}
+      introTitle={HEALTH_INTRO.title}
+      introParagraphs={HEALTH_INTRO.paragraphs}
+    />
+  );
+}

ui/src/components/upload/UploadLanding.tsx

@@ -37,7 +37,8 @@ function UploadLanding(props: UploadLandingProps) {
   const { hasCoverage, hasGeolocation, isAdmin } = useRoles();
 
   // Tab 0 = Geolocation (source gigasync), 1 = API (source api),
-  // 2 = Coverage (dataset coverage), 3 = Schemaless (dataset structured)
+  // 2 = Coverage (dataset coverage), 3 = Schemaless (dataset structured),
+  // 4 = Health (dataset health)
   const tabFilter = (() => {
     switch (selectedTab) {
       case 0:
@@ -48,6 +49,8 @@ function UploadLanding(props: UploadLandingProps) {
         return { source: null, dataset: "coverage" as const };
       case 3:
         return { source: null, dataset: "structured" as const };
+      case 4:
+        return { source: null, dataset: "health" as const };
       default:
         return { source: null, dataset: null };
     }
@@ -101,7 +104,7 @@ function UploadLanding(props: UploadLandingProps) {
                 shared and in which context.
               </p>
             </div>
-            <div className="grid grid-cols-4">
+            <div className="grid grid-cols-1 gap-3 sm:grid-cols-2 lg:grid-cols-5">
               {(hasGeolocation || isAdmin) && (
                 <Button
                   as={Link}
@@ -145,6 +148,19 @@ function UploadLanding(props: UploadLandingProps) {
               >
                 Schemaless dataset
               </Button>
+              <Button
+                as={Link}
+                to="/upload/$uploadGroup/$uploadType"
+                params={{
+                  uploadGroup: "other",
+                  uploadType: "health",
+                }}
+                className="w-full"
+                size="xl"
+                renderIcon={Add}
+              >
+                Health dataset
+              </Button>
             </div>
           </Stack>
 
@@ -158,6 +174,7 @@ function UploadLanding(props: UploadLandingProps) {
               <Tab>API</Tab>
               <Tab>Coverage</Tab>
               <Tab>Schemaless</Tab>
+              <Tab>Health</Tab>
             </TabList>
 
             <TabPanels>
@@ -189,6 +206,13 @@ function UploadLanding(props: UploadLandingProps) {
                   dataset={tabFilter.dataset}
                 />
               </TabPanel>
+              <TabPanel className="p-0">
+                <UploadsTable
+                  {...props}
+                  source={tabFilter.source}
+                  dataset={tabFilter.dataset}
+                />
+              </TabPanel>
             </TabPanels>
           </Tabs>
         </Stack>