feat: prevent users from initiating export of unfeasibly large anvil datasets (#4953)

__tests__/viewModelBuilders/dataset-export.test.ts

@@ -4,9 +4,17 @@ import { FileFacet } from "@databiosphere/findable-ui/lib/hooks/useFileManifest/
 import { FileManifestState } from "@databiosphere/findable-ui/lib/providers/fileManifestState";
 import { DatasetsResponse } from "../../app/apis/azul/anvil-cmg/common/responses";
 import {
+  TERRA_EXPORT_UNAVAILABLE_DATASET_MESSAGE,
+  TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES,
+} from "../../app/viewModelBuilders/azul/anvil-cmg/common/constants";
+import {
+  buildDatasetExportMethodTerra,
   buildDatasetExportToPlatform,
   buildDatasetExportToPlatformHero,
   buildDatasetExportToPlatformMethod,
+  isDatasetTerraExportUnavailable,
+  renderDatasetTerraExport,
+  renderDatasetTerraExportUnavailable,
 } from "../../app/viewModelBuilders/azul/anvil-cmg/common/viewModelBuilders";
 import { ANVIL_CMG_CATEGORY_KEY } from "../../site-config/anvil-cmg/category";
 import {
@@ -380,3 +388,79 @@ describe("buildDatasetExportToPlatformMethod", () => {
     });
   });
 });
+
+/**
+ * Creates a mock DatasetsResponse with the given dataset title.
+ * @param title - Dataset title.
+ * @returns Mock DatasetsResponse.
+ */
+const createMockDatasetsResponseWithTitle = (title: string): DatasetsResponse =>
+  createMockDatasetsResponse({
+    datasets: [
+      {
+        accessible: true,
+        dataset_id: "test-dataset-id",
+        title,
+      },
+    ],
+  } as Partial<DatasetsResponse>);
+
+describe("isDatasetTerraExportUnavailable", () => {
+  it.each(TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES)(
+    "returns true for %s",
+    (title) => {
+      const response = createMockDatasetsResponseWithTitle(title);
+
+      expect(isDatasetTerraExportUnavailable(response)).toBe(true);
+    }
+  );
+
+  it("returns false for any other dataset", () => {
+    const response = createMockDatasetsResponse();
+
+    expect(isDatasetTerraExportUnavailable(response)).toBe(false);
+  });
+});
+
+describe("buildDatasetExportMethodTerra", () => {
+  it("returns inaccessible with footnote when Terra export is unavailable", () => {
+    const response = createMockDatasetsResponseWithTitle(
+      TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES[0]
+    );
+
+    const result = buildDatasetExportMethodTerra(response);
+
+    expect(result.isAccessible).toBe(false);
+    expect(result.footnote).toBe(TERRA_EXPORT_UNAVAILABLE_DATASET_MESSAGE);
+  });
+
+  it("returns accessible without footnote when Terra export is available", () => {
+    const response = createMockDatasetsResponse();
+
+    const result = buildDatasetExportMethodTerra(response);
+
+    expect(result.isAccessible).toBe(true);
+    expect(result.footnote).toBeNull();
+    expect(result.route).toBe(
+      "/datasets/test-dataset-id/export/export-to-terra"
+    );
+  });
+});
+
+describe("renderDatasetTerraExport and renderDatasetTerraExportUnavailable", () => {
+  it("renders the warning, not the Terra export, when Terra export is unavailable", () => {
+    const response = createMockDatasetsResponseWithTitle(
+      TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES[0]
+    );
+
+    expect(renderDatasetTerraExport(response).isIn).toBe(false);
+    expect(renderDatasetTerraExportUnavailable(response).isIn).toBe(true);
+  });
+
+  it("renders the Terra export, not the warning, when Terra export is available", () => {
+    const response = createMockDatasetsResponse();
+
+    expect(renderDatasetTerraExport(response).isIn).toBe(true);
+    expect(renderDatasetTerraExportUnavailable(response).isIn).toBe(false);
+  });
+});

__tests__/viewModelBuilders/export.test.ts

@@ -2,10 +2,16 @@ import { Breadcrumb } from "@databiosphere/findable-ui/lib/components/common/Bre
 import { ViewContext } from "@databiosphere/findable-ui/lib/config/entities";
 import { FileFacet } from "@databiosphere/findable-ui/lib/hooks/useFileManifest/common/entities";
 import { FileManifestState } from "@databiosphere/findable-ui/lib/providers/fileManifestState";
+import { TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES } from "../../app/viewModelBuilders/azul/anvil-cmg/common/constants";
 import {
+  buildAlertTerraExportUnavailable,
+  buildExportMethodTerra,
   buildExportToPlatform,
   buildExportToPlatformHero,
   buildExportToPlatformMethod,
+  getSelectedTerraExportUnavailableDatasetTitles,
+  renderCohortTerraExport,
+  renderCohortTerraExportUnavailable,
 } from "../../app/viewModelBuilders/azul/anvil-cmg/common/viewModelBuilders";
 import { ANVIL_CMG_CATEGORY_KEY } from "../../site-config/anvil-cmg/category";
 import {
@@ -339,3 +345,157 @@ describe("buildExportToPlatformMethod", () => {
     });
   });
 });
+
+/**
+ * Creates a mock ViewContext with the given dataset title filter selected.
+ * @param datasetTitles - Selected dataset titles.
+ * @returns Mock ViewContext.
+ */
+const createMockViewContextWithDatasetTitles = (
+  datasetTitles: string[]
+): ViewContext<unknown> => {
+  const viewContext = createMockViewContext();
+  viewContext.exploreState.filterState = [
+    { categoryKey: ANVIL_CMG_CATEGORY_KEY.DATASET_TITLE, value: datasetTitles },
+  ];
+  return viewContext;
+};
+
+describe("getSelectedTerraExportUnavailableDatasetTitles", () => {
+  it("returns selected datasets that cannot be exported to Terra", () => {
+    const [unavailableTitle] = TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES;
+
+    const result = getSelectedTerraExportUnavailableDatasetTitles([
+      {
+        categoryKey: ANVIL_CMG_CATEGORY_KEY.DATASET_TITLE,
+        value: [unavailableTitle, "ANVIL_Other_Dataset"],
+      },
+    ]);
+
+    expect(result).toEqual([unavailableTitle]);
+  });
+
+  it("returns an empty array when no dataset title filter is selected", () => {
+    const result = getSelectedTerraExportUnavailableDatasetTitles([
+      { categoryKey: "test.filter", value: ["value1"] },
+    ]);
+
+    expect(result).toEqual([]);
+  });
+
+  it("returns an empty array when only other datasets are selected", () => {
+    const result = getSelectedTerraExportUnavailableDatasetTitles([
+      {
+        categoryKey: ANVIL_CMG_CATEGORY_KEY.DATASET_TITLE,
+        value: ["ANVIL_Other_Dataset"],
+      },
+    ]);
+
+    expect(result).toEqual([]);
+  });
+});
+
+describe("buildExportMethodTerra", () => {
+  it("ignores the selected datasets before the files facets have loaded", () => {
+    const viewContext = createMockViewContextWithDatasetTitles([
+      TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES[0],
+    ]);
+    viewContext.fileManifestState = createMockFileManifestState({
+      isFacetsSuccess: false,
+    });
+
+    const result = buildExportMethodTerra(undefined, viewContext);
+
+    expect(result.isAccessible).toBe(false);
+    expect(result.footnote).toBeNull();
+  });
+
+  it("returns inaccessible with footnote naming the selected unavailable datasets", () => {
+    const [unavailableTitle] = TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES;
+    const viewContext = createMockViewContextWithDatasetTitles([
+      unavailableTitle,
+    ]);
+
+    const result = buildExportMethodTerra(undefined, viewContext);
+
+    expect(result.isAccessible).toBe(false);
+    expect(result.footnote).toContain(unavailableTitle);
+  });
+
+  it("returns file manifest accessibility when no unavailable dataset is selected", () => {
+    const fileManifestState = createMockFileManifestState({
+      filesFacets: [createMockFileFacet()],
+      isFacetsSuccess: true,
+      summary: { fileCount: 10 },
+    });
+    const viewContext = createMockViewContext({ fileManifestState });
+
+    const result = buildExportMethodTerra(undefined, viewContext);
+
+    expect(result.isAccessible).toBe(true);
+    expect(result.footnote).toBeNull();
+  });
+});
+
+describe("renderCohortTerraExport and renderCohortTerraExportUnavailable", () => {
+  it("renders neither the warning nor the Terra export before the file manifest is enabled", () => {
+    const viewContext = createMockViewContextWithDatasetTitles([
+      TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES[0],
+    ]);
+    viewContext.fileManifestState = createMockFileManifestState({
+      isEnabled: false,
+    });
+
+    expect(renderCohortTerraExport(undefined, viewContext).isIn).toBe(false);
+    expect(
+      renderCohortTerraExportUnavailable(undefined, viewContext).isIn
+    ).toBe(false);
+  });
+
+  it("renders the warning, not the Terra export, when an unavailable dataset is selected", () => {
+    const viewContext = createMockViewContextWithDatasetTitles([
+      TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES[0],
+    ]);
+
+    expect(renderCohortTerraExport(undefined, viewContext).isIn).toBe(false);
+    expect(
+      renderCohortTerraExportUnavailable(undefined, viewContext).isIn
+    ).toBe(true);
+  });
+
+  it("renders the Terra export, not the warning, when no unavailable dataset is selected", () => {
+    const viewContext = createMockViewContext();
+
+    expect(renderCohortTerraExport(undefined, viewContext).isIn).toBe(true);
+    expect(
+      renderCohortTerraExportUnavailable(undefined, viewContext).isIn
+    ).toBe(false);
+  });
+});
+
+describe("buildAlertTerraExportUnavailable", () => {
+  it("uses singular wording when one unavailable dataset is selected", () => {
+    const [title] = TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES;
+    const viewContext = createMockViewContextWithDatasetTitles([title]);
+
+    const result = buildAlertTerraExportUnavailable(undefined, viewContext);
+
+    expect(result.content).toBe(
+      `Export to Terra is unavailable for ${title} because its metadata is too large for Terra to import. Remove this dataset from the selection to export to Terra.`
+    );
+  });
+
+  it("uses plural wording when several unavailable datasets are selected", () => {
+    const [firstTitle, secondTitle] = TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES;
+    const viewContext = createMockViewContextWithDatasetTitles([
+      firstTitle,
+      secondTitle,
+    ]);
+
+    const result = buildAlertTerraExportUnavailable(undefined, viewContext);
+
+    expect(result.content).toBe(
+      `Export to Terra is unavailable for ${firstTitle}, ${secondTitle} because their metadata is too large for Terra to import. Remove these datasets from the selection to export to Terra.`
+    );
+  });
+});

app/viewModelBuilders/azul/anvil-cmg/common/constants.ts

@@ -0,0 +1,22 @@
+/**
+ * Titles of datasets that cannot be exported to Terra. Their verbatim PFB manifests are too large to generate,
+ * and Terra is unable to import them.
+ * See https://github.com/DataBiosphere/azul/issues/8134.
+ */
+export const TERRA_EXPORT_UNAVAILABLE_DATASET_TITLES = [
+  "ANVIL_eMERGE_GWAS_GRU",
+  "ANVIL_eMERGE_GWAS_HMB_GSO",
+  "ANVIL_eMERGE_GWAS_HMB_MDS",
+  "ANVIL_eMERGE_GWAS_HMB_PUB_GSO",
+];
+
+/**
+ * Reason Terra export is unavailable, shared by the dataset and cohort explanations.
+ */
+export const TERRA_EXPORT_UNAVAILABLE_REASON =
+  "metadata is too large for Terra to import";
+
+/**
+ * Explanation displayed when Terra export is unavailable for a dataset.
+ */
+export const TERRA_EXPORT_UNAVAILABLE_DATASET_MESSAGE = `Export to Terra is unavailable for this dataset because its ${TERRA_EXPORT_UNAVAILABLE_REASON}.`;