Mark all paths underneath a not in cache or delete parent with the same status (#1783)

Author: mattseddon

extension/src/repository/model/collect.ts

@@ -150,3 +150,33 @@ export const collectTrackedPaths = async (
   }
   return acc
 }
+
+const isUncollectedChild = (
+  deleted: Set<string>,
+  deletedPath: string,
+  trackedPath: string
+): boolean => {
+  return !deleted.has(trackedPath) && isSameOrChild(deletedPath, trackedPath)
+}
+
+const collectIfDeletedChild = (
+  deleted: Set<string>,
+  deletedPath: string,
+  trackedPath: string
+): void => {
+  if (isUncollectedChild(deleted, deletedPath, trackedPath)) {
+    deleted.add(trackedPath)
+  }
+}
+
+export const collectDeleted = (
+  deleted: Set<string>,
+  tracked: Set<string>
+): Set<string> => {
+  for (const trackedPath of tracked) {
+    for (const deletedPath of deleted) {
+      collectIfDeletedChild(deleted, deletedPath, trackedPath)
+    }
+  }
+  return deleted
+}

extension/src/repository/model/index.test.ts

@@ -539,7 +539,67 @@ describe('RepositoryState', () => {
           resolve(dvcDemoPath, 'predictions.json'),
           resolve(dvcDemoPath, 'training_metrics'),
           resolve(dvcDemoPath, 'training_metrics.json'),
-          resolve(dvcDemoPath, 'data', 'MNIST', 'raw')
+          resolve(dvcDemoPath, 'data', 'MNIST', 'raw'),
+          resolve(
+            dvcDemoPath,
+            'data',
+            'MNIST',
+            'raw',
+            't10k-images-idx3-ubyte'
+          ),
+          resolve(
+            dvcDemoPath,
+            'data',
+            'MNIST',
+            'raw',
+            't10k-images-idx3-ubyte.gz'
+          ),
+          resolve(
+            dvcDemoPath,
+            'data',
+            'MNIST',
+            'raw',
+            't10k-labels-idx1-ubyte'
+          ),
+          resolve(
+            dvcDemoPath,
+            'data',
+            'MNIST',
+            'raw',
+            't10k-labels-idx1-ubyte.gz'
+          ),
+          resolve(
+            dvcDemoPath,
+            'data',
+            'MNIST',
+            'raw',
+            'train-images-idx3-ubyte'
+          ),
+          resolve(
+            dvcDemoPath,
+            'data',
+            'MNIST',
+            'raw',
+            'train-images-idx3-ubyte.gz'
+          ),
+          resolve(
+            dvcDemoPath,
+            'data',
+            'MNIST',
+            'raw',
+            'train-labels-idx1-ubyte'
+          ),
+          resolve(
+            dvcDemoPath,
+            'data',
+            'MNIST',
+            'raw',
+            'train-labels-idx1-ubyte.gz'
+          ),
+          resolve(dvcDemoPath, 'training_metrics', 'scalars'),
+          resolve(dvcDemoPath, 'training_metrics', 'report.html'),
+          resolve(dvcDemoPath, 'training_metrics', 'scalars', 'acc.tsv'),
+          resolve(dvcDemoPath, 'training_metrics', 'scalars', 'loss.tsv')
         ]),
         renamed: emptySet,
         tracked: new Set([

extension/src/repository/model/index.ts

@@ -1,6 +1,7 @@
 import { dirname, join, resolve } from 'path'
 import omit from 'lodash.omit'
 import {
+  collectDeleted,
   collectModifiedAgainstHead,
   collectTrackedNonLeafs,
   collectTrackedOuts,
@@ -220,21 +221,21 @@ export class RepositoryModel
       modifiedAgainstCache
     )
 
+    const deletedWithChildren = collectDeleted(
+      this.getStateFromDiff(diffOutput[Status.DELETED]),
+      this.getTracked()
+    )
+
     this.state.notInCache = new Set([
       ...this.getStateFromDiff(diffOutput[Status.NOT_IN_CACHE]),
       ...this.filterStatuses(
-        [
-          ...this.getStateFromDiff(diffOutput[Status.DELETED]),
-          ...modifiedAgainstHead
-        ],
+        [...deletedWithChildren, ...modifiedAgainstHead],
         path => this.pathInSet(path, notInCache)
       )
     ])
 
     this.state.deleted = new Set(
-      this.mapToTrackedPaths(diffOutput.deleted || []).filter(
-        path => !this.state.notInCache.has(path)
-      )
+      [...deletedWithChildren].filter(path => !this.state.notInCache.has(path))
     )
   }
 

extension/src/test/suite/repository/index.test.ts

@@ -211,7 +211,9 @@ suite('Repository Test Suite', () => {
 
       const deleted = new Set([
         join(dvcDemoPath, model),
-        join(dvcDemoPath, dataDir)
+        join(dvcDemoPath, dataDir),
+        join(dvcDemoPath, compressedDataset),
+        join(dvcDemoPath, dataset)
       ])
 
       const tracked = new Set([