Separate unit tests and integration tests

Author: mattsoulanille

coral-tflite-delegate/jasmine-integration.json

@@ -0,0 +1,9 @@
+{
+  "spec_files": [
+    "dist/integration_test/*_test.js"
+  ],
+  "env": {
+    "stopSpecOnExpectationFailure": false,
+    "random": true
+  }
+}

coral-tflite-delegate/jasmine.json

@@ -1,6 +1,6 @@
 {
   "spec_files": [
-    "dist/**/*_test.js"
+    "dist/*_test.js"
   ],
   "env": {
     "stopSpecOnExpectationFailure": false,

coral-tflite-delegate/package.json

@@ -8,6 +8,8 @@
     "build": "tsc",
     "test-dev": "jasmine --config=jasmine.json",
     "test": "yarn build && yarn test-dev",
+    "test-integration-dev": "jasmine --config=jasmine-integration.json",
+    "test-integration": "yarn build && yarn test-integration-dev",
     "lint": "tslint -p . -t verbose"
   },
   "devDependencies": {

coral-tflite-delegate/src/index.ts

@@ -25,21 +25,20 @@ import * as os from 'os';
 //
 // The following names are from the python implementation.
 // https://github.com/google-coral/pycoral/blob/9972f8ec6dbb8b2f46321e8c0d2513e0b6b152ce/pycoral/utils/edgetpu.py#L34-L38
-const libNames = new Map([
+const libNames = new Map<NodeJS.Platform, string>([
   ['linux', 'libedgetpu.so.1'],
   ['darwin', 'libedgetpu.1.dylib'],
-  ['windows', 'edgetpu.dll'],
+  ['win32', 'edgetpu.dll'],
 ]);
 
 export class CoralDelegate implements TFLiteDelegatePlugin {
-  readonly name: 'CoralDelegate';
+  readonly name = 'CoralDelegate';
   readonly tfliteVersion: '2.7';
   readonly node: TFLiteDelegatePlugin['node'];
 
   constructor(readonly options: Array<[string, string]> = [],
-    libPath?: string) {
+              libPath?: string, platform = os.platform()) {
     if (!libPath) {
-      const platform = os.platform();
       libPath = libNames.get(platform);
       if (!libPath) {
         throw new Error(`Unknown platform ${platform}`);

coral-tflite-delegate/src/index_test.ts

@@ -15,49 +15,33 @@
  * =============================================================================
  */
 
-import {CoralDelegate} from './index';
-import {loadTFLiteModel} from 'tfjs-tflite-node';
-import type {TFLiteModel} from 'tfjs-tflite-node/dist/tflite_model';
-import * as fs from 'fs';
-import * as tf from '@tensorflow/tfjs-core';
 import '@tensorflow/tfjs-backend-cpu';
-import * as jpeg from 'jpeg-js';
+import {CoralDelegate} from './index';
 
 describe('coral delegate', () => {
-  const modelPath = './test_model/mobilenet_v2_1.0_224_inat_bird_quant_edgetpu.tflite';
-  let model: TFLiteModel;
-  let parrot: tf.Tensor;
-  let labels: string[];
-
-  beforeEach(async () => {
-    model = await loadTFLiteModel(modelPath, {
-      delegates: [new CoralDelegate([['device', 'usb']])],
-    });
-
-    // Load the input image of a parrot.
-    const parrotJpeg = jpeg.decode(
-      fs.readFileSync('./test_model/parrot-small.jpg'));
+  it('has the name "CoralDelegate"', () => {
+    expect(new CoralDelegate().name).toEqual('CoralDelegate');
+  });
 
-    // Create an RGB array of the parrot's pixels (no Alpha channel).
-    const {width, height, data} = parrotJpeg;
-    const parrotRGB = new Uint8Array(width * height * 3);
-    for (let i = 0; i < width * height; i++) {
-      const i3 = i * 3;
-      const i4 = i * 4;
-      parrotRGB[i3] = data[i4];
-      parrotRGB[i3 + 1] = data[i4 + 1];
-      parrotRGB[i3 + 2] = data[i4 + 2];
-    }
+  it('stores options', () => {
+    const options: Array<[string, string]> = [['foo', 'bar'], ['123', '456']];
+    const coralDelegate = new CoralDelegate(options);
+    expect(coralDelegate.options).toEqual(options);
+  });
 
-    parrot = tf.tensor(parrotRGB, [1, 224, 224, 3]);
-    labels = fs.readFileSync('./test_model/inat_bird_labels.txt', 'utf-8')
-      .split('\n');
+  it('allows manually setting lib path', () => {
+    const libPath = 'some lib path';
+    const coralDelegate = new CoralDelegate([], libPath);
+    expect(coralDelegate.node.path).toEqual(libPath);
   });
 
-  it('runs a coral model (will fail without coral device)', () => {
-    const prediction = model.predict(parrot);
-    const argmax = tf.argMax(prediction as tf.Tensor, 1);
-    const label = labels[argmax.dataSync()[0]];
-    expect(label).toEqual('Ara macao (Scarlet Macaw)');
+  it('sets the lib path automatically based on platorm', () => {
+    const coralLinux = new CoralDelegate([], undefined, 'linux');
+    const coralMac = new CoralDelegate([], undefined, 'darwin');
+    const coralWindows = new CoralDelegate([], undefined, 'win32');
+
+    expect(coralLinux.node.path).toEqual('libedgetpu.so.1');
+    expect(coralMac.node.path).toEqual('libedgetpu.1.dylib');
+    expect(coralWindows.node.path).toEqual('edgetpu.dll');
   });
 });

coral-tflite-delegate/src/integration_test/index_test.ts

@@ -0,0 +1,63 @@
+/**
+ * @license
+ * Copyright 2022 Google LLC. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+
+import {CoralDelegate} from '../index';
+import {loadTFLiteModel} from 'tfjs-tflite-node';
+import type {TFLiteModel} from 'tfjs-tflite-node/dist/tflite_model';
+import * as fs from 'fs';
+import * as tf from '@tensorflow/tfjs-core';
+import '@tensorflow/tfjs-backend-cpu';
+import * as jpeg from 'jpeg-js';
+
+describe('coral delegate', () => {
+  const modelPath = './test_model/mobilenet_v2_1.0_224_inat_bird_quant_edgetpu.tflite';
+  let model: TFLiteModel;
+  let parrot: tf.Tensor;
+  let labels: string[];
+
+  beforeEach(async () => {
+    model = await loadTFLiteModel(modelPath, {
+      delegates: [new CoralDelegate([['device', 'usb']])],
+    });
+
+    // Load the input image of a parrot.
+    const parrotJpeg = jpeg.decode(
+      fs.readFileSync('./test_model/parrot-small.jpg'));
+
+    // Create an RGB array of the parrot's pixels (no Alpha channel).
+    const {width, height, data} = parrotJpeg;
+    const parrotRGB = new Uint8Array(width * height * 3);
+    for (let i = 0; i < width * height; i++) {
+      const i3 = i * 3;
+      const i4 = i * 4;
+      parrotRGB[i3] = data[i4];
+      parrotRGB[i3 + 1] = data[i4 + 1];
+      parrotRGB[i3 + 2] = data[i4 + 2];
+    }
+
+    parrot = tf.tensor(parrotRGB, [1, 224, 224, 3]);
+    labels = fs.readFileSync('./test_model/inat_bird_labels.txt', 'utf-8')
+      .split('\n');
+  });
+
+  it('runs a coral model (will fail without coral device)', () => {
+    const prediction = model.predict(parrot);
+    const argmax = tf.argMax(prediction as tf.Tensor, 1);
+    const label = labels[argmax.dataSync()[0]];
+    expect(label).toEqual('Ara macao (Scarlet Macaw)');
+  });
+});

coral-tflite-delegate/tslint.json

@@ -0,0 +1,3 @@
+{
+  "extends": "../tslint.json"
+}

presubmit.ts

@@ -39,13 +39,7 @@ parentDirs.forEach(curParentDir => {
   dirs.push(...curDirs.map(curDir => join(curParentDir, curDir)));
 })
 
-const skippedDirs = new Set(['coral-tflite-delegate']);
-
 dirs.forEach(dir => {
-  if (skippedDirs.has(dir)) {
-    console.log(`~~~~~~~~~~~~ Skipping ${dir} ~~~~~~~~~~~~`);
-    return;
-  }
   shell.cd(dir);
 
   if (!fs.existsSync('./package.json')) {