Merge pull request #236 from flatironinstitute/indicate-data-origin

Indicate data origin

gui/src/app/Project/ProjectDataModel.ts

@@ -90,13 +90,26 @@ const isProjectBase = (x: any): x is ProjectBase => {
   return true;
 };
 
+export enum DataSource {
+  GENERATED_BY_R = "generated_by_r",
+  GENERATED_BY_STALE_R = "generated_by_stale_r",
+  GENERATED_BY_PYTHON = "generated_by_python",
+  GENERATED_BY_STALE_PYTHON = "generated_by_stale_python",
+}
+
 type ProjectMetadata = {
   title: string;
+  dataSource?: DataSource;
 };
 
 export const isProjectMetaData = (x: any): x is ProjectMetadata => {
   if (!baseObjectCheck(x)) return false;
   if (typeof x.title !== "string") return false;
+  if (
+    x.dataSource !== undefined && // undefined = manually edited or unknown provenance
+    !Object.values(DataSource).includes(x.dataSource)
+  )
+    return false;
   return true;
 };
 
@@ -128,7 +141,7 @@ export const isProjectDataModel = (x: any): x is ProjectDataModel => {
 export type ProjectPersistentDataModel = Omit<ProjectDataModel, "ephemera">;
 
 export const initialDataModel: ProjectDataModel = {
-  meta: { title: "Untitled" },
+  meta: { title: "Untitled", dataSource: undefined },
   ephemera: {
     stanFileContent: "",
     dataFileContent: "",

gui/src/app/Project/ProjectReducer.ts

@@ -1,5 +1,6 @@
 import { FieldsContentsMap } from "@SpCore/FileMapping";
 import {
+  DataSource,
   initialDataModel,
   ProjectDataModel,
   ProjectKnownFiles,
@@ -24,6 +25,7 @@ export type ProjectReducerAction =
       type: "retitle";
       title: string;
     }
+  | { type: "generateData"; content: string; dataSource: DataSource }
   | {
       type: "editFile";
       content: string;
@@ -69,9 +71,27 @@ const ProjectReducer = (s: ProjectDataModel, a: ProjectReducerAction) => {
     }
     case "commitFile": {
       const newState = { ...s };
+      const newDataSource = confirmDataSourceForCommit(
+        s.meta.dataSource,
+        a.filename,
+      );
+      if (newDataSource !== s.meta.dataSource) {
+        newState.meta = { ...s.meta, dataSource: newDataSource };
+      }
       newState[a.filename] = s.ephemera[a.filename];
       return newState;
     }
+    case "generateData": {
+      return {
+        ...s,
+        [ProjectKnownFiles.DATAFILE]: a.content,
+        ephemera: {
+          ...s.ephemera,
+          [ProjectKnownFiles.DATAFILE]: a.content,
+        },
+        meta: { ...s.meta, dataSource: a.dataSource },
+      };
+    }
     case "setSamplingOpts": {
       return { ...s, samplingOpts: { ...s.samplingOpts, ...a.opts } };
     }
@@ -86,4 +106,25 @@ const ProjectReducer = (s: ProjectDataModel, a: ProjectReducerAction) => {
   }
 };
 
+const confirmDataSourceForCommit = (
+  currentSource: DataSource | undefined,
+  editedFile: ProjectKnownFiles,
+): DataSource | undefined => {
+  if (editedFile === ProjectKnownFiles.DATAFILE) return undefined;
+  if (
+    editedFile === ProjectKnownFiles.DATAPYFILE &&
+    currentSource === DataSource.GENERATED_BY_PYTHON
+  ) {
+    return DataSource.GENERATED_BY_STALE_PYTHON;
+  }
+  if (
+    editedFile === ProjectKnownFiles.DATARFILE &&
+    currentSource === DataSource.GENERATED_BY_R
+  ) {
+    return DataSource.GENERATED_BY_STALE_R;
+  }
+
+  return currentSource;
+};
+
 export default ProjectReducer;

gui/src/app/Scripting/DataGeneration/DataPyWindow.tsx

@@ -22,7 +22,7 @@ const handleHelp = () =>
   );
 
 const DataPyWindow: FunctionComponent<Props> = () => {
-  const { consoleRef, status, onStatus, onData } = useDataGenState();
+  const { consoleRef, status, onStatus, onData } = useDataGenState("python");
 
   const callbacks = useMemo(
     () => ({

gui/src/app/Scripting/DataGeneration/DataRWindow.tsx

@@ -19,7 +19,7 @@ const handleHelp = () =>
   );
 
 const DataRWindow: FunctionComponent<Props> = () => {
-  const { consoleRef, status, onStatus, onData } = useDataGenState();
+  const { consoleRef, status, onStatus, onData } = useDataGenState("r");
 
   const { run } = useWebR({ consoleRef, onStatus, onData });
   const handleRun = useCallback(

gui/src/app/Scripting/DataGeneration/useDataGenState.ts

@@ -1,12 +1,12 @@
 import { useCallback, useContext, useRef, useState } from "react";
-import { ProjectKnownFiles } from "@SpCore/ProjectDataModel";
+import { DataSource } from "@SpCore/ProjectDataModel";
 import { writeConsoleOutToDiv } from "@SpScripting/OutputDivUtils";
 import { InterpreterStatus } from "@SpScripting/InterpreterTypes";
 import { ProjectContext } from "@SpCore/ProjectContextProvider";
 
 // A custom hook to share logic between the Python and R data generation windows
 // This contains the output div ref, the interpreter state, and the callback to update the data.
-const useDataGenState = () => {
+const useDataGenState = (source: "python" | "r") => {
   const [status, setStatus] = useState<InterpreterStatus>("idle");
   const consoleRef = useRef<HTMLDivElement>(null);
 
@@ -21,11 +21,13 @@ const useDataGenState = () => {
       if (dataJson !== lastData.current) {
         lastData.current = dataJson;
         update({
-          type: "editFile",
+          type: "generateData",
           content: dataJson,
-          filename: ProjectKnownFiles.DATAFILE,
+          dataSource:
+            source === "python"
+              ? DataSource.GENERATED_BY_PYTHON
+              : DataSource.GENERATED_BY_R,
         });
-        update({ type: "commitFile", filename: ProjectKnownFiles.DATAFILE });
         // Use "stan-playground" prefix to distinguish from console output of the running code
         writeConsoleOutToDiv(
           consoleRef,

gui/src/app/pages/HomePage/HomePage.tsx

@@ -5,20 +5,25 @@ import useMediaQuery from "@mui/material/useMediaQuery";
 import StanFileEditor from "@SpComponents/StanFileEditor";
 import TabWidget from "@SpComponents/TabWidget";
 import TextEditor from "@SpComponents/TextEditor";
+import { ColorOptions, ToolbarItem } from "@SpComponents/ToolBar";
 import { FileNames } from "@SpCore/FileMapping";
 import { ProjectContext } from "@SpCore/ProjectContextProvider";
 import {
+  DataSource,
   modelHasUnsavedChanges,
   ProjectKnownFiles,
 } from "@SpCore/ProjectDataModel";
 import Sidebar, { drawerWidth } from "@SpPages/Sidebar";
 import TopBar from "@SpPages/TopBar";
 import DataPyWindow from "@SpScripting/DataGeneration/DataPyWindow";
 import DataRWindow from "@SpScripting/DataGeneration/DataRWindow";
+import { unreachable } from "@SpUtil/unreachable";
 import {
   FunctionComponent,
+  useCallback,
   useContext,
   useEffect,
+  useMemo,
   useRef,
   useState,
 } from "react";
@@ -92,6 +97,7 @@ type LeftViewProps = {
 
 const LeftView: FunctionComponent<LeftViewProps> = () => {
   const { data, update } = useContext(ProjectContext);
+
   return (
     <Split horizontal>
       <StanFileEditor
@@ -114,28 +120,87 @@ const LeftView: FunctionComponent<LeftViewProps> = () => {
         }
         readOnly={false}
       />
-      <TextEditor
-        language="json"
-        label={FileNames.DATAFILE}
-        text={data.dataFileContent}
-        onSaveText={() =>
-          update({
-            type: "commitFile",
-            filename: ProjectKnownFiles.DATAFILE,
-          })
+      <DataEditor />
+    </Split>
+  );
+};
+
+const DataEditor: FunctionComponent<unknown> = () => {
+  const { data, update } = useContext(ProjectContext);
+
+  const dataIsEdited = useMemo(() => {
+    return data.dataFileContent !== data.ephemera.dataFileContent;
+  }, [data.dataFileContent, data.ephemera.dataFileContent]);
+
+  const dataSourceDesc: undefined | { msg: string; color: ColorOptions } =
+    useMemo(() => {
+      if (dataIsEdited) return undefined;
+
+      switch (data.meta.dataSource) {
+        case undefined: {
+          return undefined;
         }
-        editedText={data.ephemera.dataFileContent}
-        onSetEditedText={(content: string) =>
-          update({
-            type: "editFile",
-            content,
-            filename: ProjectKnownFiles.DATAFILE,
-          })
+        case DataSource.GENERATED_BY_PYTHON: {
+          return { msg: "data.py", color: "info.main" };
         }
-        readOnly={false}
-        contentOnEmpty={"Enter JSON data or use the data generation tab"}
-      />
-    </Split>
+        case DataSource.GENERATED_BY_R: {
+          return { msg: "data.R", color: "info.main" };
+        }
+        case DataSource.GENERATED_BY_STALE_PYTHON: {
+          return { msg: "a prior version of data.py.", color: "warning.main" };
+        }
+        case DataSource.GENERATED_BY_STALE_R: {
+          return { msg: "a prior version of data.R.", color: "warning.main" };
+        }
+        default:
+          return unreachable(data.meta.dataSource);
+      }
+    }, [dataIsEdited, data.meta.dataSource]);
+
+  const dataMessage: ToolbarItem[] = useMemo(() => {
+    if (dataSourceDesc === undefined) {
+      return [];
+    } else {
+      return [
+        {
+          type: "text",
+          label: `Data generated by ${dataSourceDesc.msg}`,
+          color: dataSourceDesc.color,
+        },
+      ];
+    }
+  }, [dataSourceDesc]);
+
+  const onSetEditedText = useCallback(
+    (content: string) => {
+      update({
+        type: "editFile",
+        content,
+        filename: ProjectKnownFiles.DATAFILE,
+      });
+    },
+    [update],
+  );
+
+  const onSaveText = useCallback(() => {
+    update({
+      type: "commitFile",
+      filename: ProjectKnownFiles.DATAFILE,
+    });
+  }, [update]);
+
+  return (
+    <TextEditor
+      language="json"
+      label={FileNames.DATAFILE}
+      text={data.dataFileContent}
+      onSaveText={onSaveText}
+      editedText={data.ephemera.dataFileContent}
+      onSetEditedText={onSetEditedText}
+      readOnly={false}
+      toolbarItems={dataMessage}
+      contentOnEmpty={"Enter JSON data or use the data generation tab"}
+    />
   );
 };
 

gui/test/app/Project/ProjectDataModel.test.ts

@@ -1,4 +1,5 @@
 import {
+  DataSource,
   defaultSamplingOpts,
   exportedForTesting,
   getStringKnownFileKeys,
@@ -248,6 +249,23 @@ describe("Project data model type guards", () => {
       expect(isProjectMetaData({ title: 6 })).toBe(false);
       expect(isProjectMetaData({ no_title: "title" })).toBe(false);
     });
+    test("Returns true for valid data source", () => {
+      expect(
+        isProjectMetaData({
+          title: "title",
+          dataSource: DataSource.GENERATED_BY_PYTHON,
+        }),
+      ).toBe(true);
+      expect(
+        isProjectMetaData({
+          title: "title",
+          dataSource: DataSource.GENERATED_BY_R,
+        }),
+      ).toBe(true);
+    });
+    test("Returns false on bad data source", () => {
+      expect(isProjectMetaData({ title: "title", dataSource: 1 })).toBe(false);
+    });
   });
   describe("Project ephemeral-data typeguard", () => {
     test("Returns true for valid project files object", () => {