@tensorflow/tfjs-core
Version:
Hardware-accelerated JavaScript library for machine intelligence
998 lines • 52.5 kB
JavaScript
"use strict";
var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
return new (P || (P = Promise))(function (resolve, reject) {
function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
function step(result) { result.done ? resolve(result.value) : new P(function (resolve) { resolve(result.value); }).then(fulfilled, rejected); }
step((generator = generator.apply(thisArg, _arguments || [])).next());
});
};
var __generator = (this && this.__generator) || function (thisArg, body) {
var _ = { label: 0, sent: function() { if (t[0] & 1) throw t[1]; return t[1]; }, trys: [], ops: [] }, f, y, t, g;
return g = { next: verb(0), "throw": verb(1), "return": verb(2) }, typeof Symbol === "function" && (g[Symbol.iterator] = function() { return this; }), g;
function verb(n) { return function (v) { return step([n, v]); }; }
function step(op) {
if (f) throw new TypeError("Generator is already executing.");
while (_) try {
if (f = 1, y && (t = op[0] & 2 ? y["return"] : op[0] ? y["throw"] || ((t = y["return"]) && t.call(y), 0) : y.next) && !(t = t.call(y, op[1])).done) return t;
if (y = 0, t) op = [op[0] & 2, t.value];
switch (op[0]) {
case 0: case 1: t = op; break;
case 4: _.label++; return { value: op[1], done: false };
case 5: _.label++; y = op[1]; op = [0]; continue;
case 7: op = _.ops.pop(); _.trys.pop(); continue;
default:
if (!(t = _.trys, t = t.length > 0 && t[t.length - 1]) && (op[0] === 6 || op[0] === 2)) { _ = 0; continue; }
if (op[0] === 3 && (!t || (op[1] > t[0] && op[1] < t[3]))) { _.label = op[1]; break; }
if (op[0] === 6 && _.label < t[1]) { _.label = t[1]; t = op; break; }
if (t && _.label < t[2]) { _.label = t[2]; _.ops.push(op); break; }
if (t[2]) _.ops.pop();
_.trys.pop(); continue;
}
op = body.call(thisArg, _);
} catch (e) { op = [6, e]; y = 0; } finally { f = t = 0; }
if (op[0] & 5) throw op[1]; return { value: op[0] ? op[1] : void 0, done: true };
}
};
var _this = this;
Object.defineProperty(exports, "__esModule", { value: true });
var tf = require("../index");
var jasmine_util_1 = require("../jasmine_util");
var test_util_1 = require("../test_util");
var browser_http_1 = require("./browser_http");
var modelTopology1 = {
'class_name': 'Sequential',
'keras_version': '2.1.4',
'config': [{
'class_name': 'Dense',
'config': {
'kernel_initializer': {
'class_name': 'VarianceScaling',
'config': {
'distribution': 'uniform',
'scale': 1.0,
'seed': null,
'mode': 'fan_avg'
}
},
'name': 'dense',
'kernel_constraint': null,
'bias_regularizer': null,
'bias_constraint': null,
'dtype': 'float32',
'activation': 'linear',
'trainable': true,
'kernel_regularizer': null,
'bias_initializer': { 'class_name': 'Zeros', 'config': {} },
'units': 1,
'batch_input_shape': [null, 3],
'use_bias': true,
'activity_regularizer': null
}
}],
'backend': 'tensorflow'
};
jasmine_util_1.describeWithFlags('browserHTTPRequest-load fetch', test_util_1.NODE_ENVS, function () {
var requestInits;
beforeEach(function () {
global.fetch = function () { };
requestInits = [];
});
afterAll(function () {
delete global.fetch;
});
var fakeResponse = function (body) { return ({
ok: true,
json: function () {
return Promise.resolve(JSON.parse(body));
},
arrayBuffer: function () {
var buf = body.buffer ?
body.buffer :
body;
return Promise.resolve(buf);
}
}); };
var setupFakeWeightFiles = function (fileBufferMap) {
spyOn(global, 'fetch')
.and.callFake(function (path, init) {
requestInits.push(init);
return fakeResponse(fileBufferMap[path]);
});
};
it('1 group, 2 weights, 1 path', function () { return __awaiter(_this, void 0, void 0, function () {
var weightManifest1, floatData, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightManifest1 = [{
paths: ['weightfile0'],
weights: [
{
name: 'dense/kernel',
shape: [3, 1],
dtype: 'float32',
},
{
name: 'dense/bias',
shape: [2],
dtype: 'float32',
}
]
}];
floatData = new Float32Array([1, 3, 3, 7, 4]);
setupFakeWeightFiles({
'./model.json': JSON.stringify({ modelTopology: modelTopology1, weightsManifest: weightManifest1 }),
'./weightfile0': floatData,
});
handler = tf.io.browserHTTPRequest('./model.json');
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelTopology1);
expect(modelArtifacts.weightSpecs).toEqual(weightManifest1[0].weights);
expect(new Float32Array(modelArtifacts.weightData)).toEqual(floatData);
expect(requestInits).toEqual([{}, {}]);
return [2];
}
});
}); });
it('throw exception if no fetch polyfill', function () {
delete global.fetch;
try {
tf.io.browserHTTPRequest('./model.json');
}
catch (err) {
expect(err.message)
.toMatch(/not supported outside the web browser without a fetch polyfill/);
}
});
});
jasmine_util_1.describeWithFlags('browserHTTPRequest-save', test_util_1.CHROME_ENVS, function () {
var weightSpecs1 = [
{
name: 'dense/kernel',
shape: [3, 1],
dtype: 'float32',
},
{
name: 'dense/bias',
shape: [1],
dtype: 'float32',
}
];
var weightData1 = new ArrayBuffer(16);
var artifacts1 = {
modelTopology: modelTopology1,
weightSpecs: weightSpecs1,
weightData: weightData1,
};
var requestInits = [];
beforeEach(function () {
requestInits = [];
spyOn(window, 'fetch').and.callFake(function (path, init) {
if (path === 'model-upload-test' || path === 'http://model-upload-test') {
requestInits.push(init);
return new Response(null, { status: 200 });
}
else {
return new Response(null, { status: 404 });
}
});
});
it('Save topology and weights, default POST method', function (done) {
var testStartDate = new Date();
var handler = tf.io.getSaveHandlers('http://model-upload-test')[0];
handler.save(artifacts1)
.then(function (saveResult) {
expect(saveResult.modelArtifactsInfo.dateSaved.getTime())
.toBeGreaterThanOrEqual(testStartDate.getTime());
expect(saveResult.modelArtifactsInfo.modelTopologyBytes)
.toEqual(JSON.stringify(modelTopology1).length);
expect(saveResult.modelArtifactsInfo.weightSpecsBytes)
.toEqual(JSON.stringify(weightSpecs1).length);
expect(saveResult.modelArtifactsInfo.weightDataBytes)
.toEqual(weightData1.byteLength);
expect(requestInits.length).toEqual(1);
var init = requestInits[0];
expect(init.method).toEqual('POST');
var body = init.body;
var jsonFile = body.get('model.json');
var jsonFileReader = new FileReader();
jsonFileReader.onload = function (event) {
var modelJSON = JSON.parse(event.target.result);
expect(modelJSON.modelTopology).toEqual(modelTopology1);
expect(modelJSON.weightsManifest.length).toEqual(1);
expect(modelJSON.weightsManifest[0].weights).toEqual(weightSpecs1);
var weightsFile = body.get('model.weights.bin');
var weightsFileReader = new FileReader();
weightsFileReader.onload = function (event) {
var weightData = event.target.result;
expect(new Uint8Array(weightData))
.toEqual(new Uint8Array(weightData1));
done();
};
weightsFileReader.onerror = function (ev) {
done.fail(weightsFileReader.error.message);
};
weightsFileReader.readAsArrayBuffer(weightsFile);
};
jsonFileReader.onerror = function (ev) {
done.fail(jsonFileReader.error.message);
};
jsonFileReader.readAsText(jsonFile);
})
.catch(function (err) {
done.fail(err.stack);
});
});
it('Save topology only, default POST method', function (done) {
var testStartDate = new Date();
var handler = tf.io.getSaveHandlers('http://model-upload-test')[0];
var topologyOnlyArtifacts = { modelTopology: modelTopology1 };
handler.save(topologyOnlyArtifacts)
.then(function (saveResult) {
expect(saveResult.modelArtifactsInfo.dateSaved.getTime())
.toBeGreaterThanOrEqual(testStartDate.getTime());
expect(saveResult.modelArtifactsInfo.modelTopologyBytes)
.toEqual(JSON.stringify(modelTopology1).length);
expect(saveResult.modelArtifactsInfo.weightSpecsBytes).toEqual(0);
expect(saveResult.modelArtifactsInfo.weightDataBytes).toEqual(0);
expect(requestInits.length).toEqual(1);
var init = requestInits[0];
expect(init.method).toEqual('POST');
var body = init.body;
var jsonFile = body.get('model.json');
var jsonFileReader = new FileReader();
jsonFileReader.onload = function (event) {
var modelJSON = JSON.parse(event.target.result);
expect(modelJSON.modelTopology).toEqual(modelTopology1);
expect(body.get('model.weights.bin')).toEqual(null);
done();
};
jsonFileReader.onerror = function (event) {
done.fail(jsonFileReader.error.message);
};
jsonFileReader.readAsText(jsonFile);
})
.catch(function (err) {
done.fail(err.stack);
});
});
it('Save topology and weights, PUT method, extra headers', function (done) {
var testStartDate = new Date();
var handler = tf.io.browserHTTPRequest('model-upload-test', {
method: 'PUT',
headers: { 'header_key_1': 'header_value_1', 'header_key_2': 'header_value_2' }
});
handler.save(artifacts1)
.then(function (saveResult) {
expect(saveResult.modelArtifactsInfo.dateSaved.getTime())
.toBeGreaterThanOrEqual(testStartDate.getTime());
expect(saveResult.modelArtifactsInfo.modelTopologyBytes)
.toEqual(JSON.stringify(modelTopology1).length);
expect(saveResult.modelArtifactsInfo.weightSpecsBytes)
.toEqual(JSON.stringify(weightSpecs1).length);
expect(saveResult.modelArtifactsInfo.weightDataBytes)
.toEqual(weightData1.byteLength);
expect(requestInits.length).toEqual(1);
var init = requestInits[0];
expect(init.method).toEqual('PUT');
expect(init.headers).toEqual({
'header_key_1': 'header_value_1',
'header_key_2': 'header_value_2'
});
var body = init.body;
var jsonFile = body.get('model.json');
var jsonFileReader = new FileReader();
jsonFileReader.onload = function (event) {
var modelJSON = JSON.parse(event.target.result);
expect(modelJSON.modelTopology).toEqual(modelTopology1);
expect(modelJSON.weightsManifest.length).toEqual(1);
expect(modelJSON.weightsManifest[0].weights).toEqual(weightSpecs1);
var weightsFile = body.get('model.weights.bin');
var weightsFileReader = new FileReader();
weightsFileReader.onload = function (event) {
var weightData = event.target.result;
expect(new Uint8Array(weightData))
.toEqual(new Uint8Array(weightData1));
done();
};
weightsFileReader.onerror = function (event) {
done.fail(weightsFileReader.error.message);
};
weightsFileReader.readAsArrayBuffer(weightsFile);
};
jsonFileReader.onerror = function (event) {
done.fail(jsonFileReader.error.message);
};
jsonFileReader.readAsText(jsonFile);
})
.catch(function (err) {
done.fail(err.stack);
});
});
it('404 response causes Error', function (done) {
var handler = tf.io.getSaveHandlers('http://invalid/path')[0];
handler.save(artifacts1)
.then(function (saveResult) {
done.fail('Calling browserHTTPRequest at invalid URL succeeded ' +
'unexpectedly');
})
.catch(function (err) {
done();
});
});
it('getLoadHandlers with one URL string', function () {
var handlers = tf.io.getLoadHandlers('http://foo/model.json');
expect(handlers.length).toEqual(1);
expect(handlers[0] instanceof browser_http_1.BrowserHTTPRequest).toEqual(true);
});
it('getLoadHandlers with two URL strings', function () {
var handlers = tf.io.getLoadHandlers(['https://foo/graph.pb', 'https://foo/weights_manifest.json']);
expect(handlers.length).toEqual(1);
expect(handlers[0] instanceof browser_http_1.BrowserHTTPRequest).toEqual(true);
});
it('Existing body leads to Error', function () {
expect(function () { return tf.io.browserHTTPRequest('model-upload-test', {
body: 'existing body'
}); }).toThrowError(/requestInit is expected to have no pre-existing body/);
});
it('Empty, null or undefined URL paths lead to Error', function () {
expect(function () { return tf.io.browserHTTPRequest(null); })
.toThrowError(/must not be null, undefined or empty/);
expect(function () { return tf.io.browserHTTPRequest(undefined); })
.toThrowError(/must not be null, undefined or empty/);
expect(function () { return tf.io.browserHTTPRequest(''); })
.toThrowError(/must not be null, undefined or empty/);
});
it('router', function () {
expect(browser_http_1.httpRequestRouter('http://bar/foo') instanceof browser_http_1.BrowserHTTPRequest)
.toEqual(true);
expect(browser_http_1.httpRequestRouter('https://localhost:5000/upload') instanceof
browser_http_1.BrowserHTTPRequest)
.toEqual(true);
expect(browser_http_1.httpRequestRouter('localhost://foo')).toBeNull();
expect(browser_http_1.httpRequestRouter('foo:5000/bar')).toBeNull();
});
});
jasmine_util_1.describeWithFlags('parseUrl', test_util_1.BROWSER_ENVS, function () {
it('should parse url with no suffix', function () {
var url = 'http://google.com/file';
var _a = browser_http_1.parseUrl(url), prefix = _a[0], suffix = _a[1];
expect(prefix).toEqual('http://google.com/');
expect(suffix).toEqual('');
});
it('should parse url with suffix', function () {
var url = 'http://google.com/file?param=1';
var _a = browser_http_1.parseUrl(url), prefix = _a[0], suffix = _a[1];
expect(prefix).toEqual('http://google.com/');
expect(suffix).toEqual('?param=1');
});
it('should parse url with multiple serach params', function () {
var url = 'http://google.com/a?x=1/file?param=1';
var _a = browser_http_1.parseUrl(url), prefix = _a[0], suffix = _a[1];
expect(prefix).toEqual('http://google.com/a?x=1/');
expect(suffix).toEqual('?param=1');
});
});
jasmine_util_1.describeWithFlags('browserHTTPRequest-load', test_util_1.BROWSER_ENVS, function () {
describe('JSON model', function () {
var requestInits;
var windowFetchSpy;
var setupFakeWeightFiles = function (fileBufferMap) {
windowFetchSpy = spyOn(window, 'fetch')
.and.callFake(function (path, init) {
requestInits.push(init);
return new Response(fileBufferMap[path]);
});
};
beforeEach(function () {
requestInits = [];
});
it('1 group, 2 weights, 1 path', function () { return __awaiter(_this, void 0, void 0, function () {
var weightManifest1, floatData, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightManifest1 = [{
paths: ['weightfile0'],
weights: [
{
name: 'dense/kernel',
shape: [3, 1],
dtype: 'float32',
},
{
name: 'dense/bias',
shape: [2],
dtype: 'float32',
}
]
}];
floatData = new Float32Array([1, 3, 3, 7, 4]);
setupFakeWeightFiles({
'./model.json': JSON.stringify({ modelTopology: modelTopology1, weightsManifest: weightManifest1 }),
'./weightfile0': floatData,
});
handler = tf.io.browserHTTPRequest('./model.json');
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelTopology1);
expect(modelArtifacts.weightSpecs).toEqual(weightManifest1[0].weights);
expect(new Float32Array(modelArtifacts.weightData)).toEqual(floatData);
expect(requestInits).toEqual([{}, {}]);
expect(windowFetchSpy.calls.mostRecent().object).toEqual(window);
return [2];
}
});
}); });
it('1 group, 2 weights, 1 path, with requestInit', function () { return __awaiter(_this, void 0, void 0, function () {
var weightManifest1, floatData, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightManifest1 = [{
paths: ['weightfile0'],
weights: [
{
name: 'dense/kernel',
shape: [3, 1],
dtype: 'float32',
},
{
name: 'dense/bias',
shape: [2],
dtype: 'float32',
}
]
}];
floatData = new Float32Array([1, 3, 3, 7, 4]);
setupFakeWeightFiles({
'./model.json': JSON.stringify({ modelTopology: modelTopology1, weightsManifest: weightManifest1 }),
'./weightfile0': floatData,
});
handler = tf.io.browserHTTPRequest('./model.json', { headers: { 'header_key_1': 'header_value_1' } });
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelTopology1);
expect(modelArtifacts.weightSpecs).toEqual(weightManifest1[0].weights);
expect(new Float32Array(modelArtifacts.weightData)).toEqual(floatData);
expect(requestInits).toEqual([
{ headers: { 'header_key_1': 'header_value_1' } },
{ headers: { 'header_key_1': 'header_value_1' } }
]);
expect(windowFetchSpy.calls.mostRecent().object).toEqual(window);
return [2];
}
});
}); });
it('1 group, 2 weight, 2 paths', function () { return __awaiter(_this, void 0, void 0, function () {
var weightManifest1, floatData1, floatData2, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightManifest1 = [{
paths: ['weightfile0', 'weightfile1'],
weights: [
{
name: 'dense/kernel',
shape: [3, 1],
dtype: 'float32',
},
{
name: 'dense/bias',
shape: [2],
dtype: 'float32',
}
]
}];
floatData1 = new Float32Array([1, 3, 3]);
floatData2 = new Float32Array([7, 4]);
setupFakeWeightFiles({
'./model.json': JSON.stringify({ modelTopology: modelTopology1, weightsManifest: weightManifest1 }),
'./weightfile0': floatData1,
'./weightfile1': floatData2,
});
handler = tf.io.browserHTTPRequest('./model.json');
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelTopology1);
expect(modelArtifacts.weightSpecs).toEqual(weightManifest1[0].weights);
expect(new Float32Array(modelArtifacts.weightData))
.toEqual(new Float32Array([1, 3, 3, 7, 4]));
return [2];
}
});
}); });
it('2 groups, 2 weight, 2 paths', function () { return __awaiter(_this, void 0, void 0, function () {
var weightsManifest, floatData1, floatData2, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightsManifest = [
{
paths: ['weightfile0'],
weights: [{
name: 'dense/kernel',
shape: [3, 1],
dtype: 'float32',
}]
},
{
paths: ['weightfile1'],
weights: [{
name: 'dense/bias',
shape: [2],
dtype: 'float32',
}],
}
];
floatData1 = new Float32Array([1, 3, 3]);
floatData2 = new Float32Array([7, 4]);
setupFakeWeightFiles({
'./model.json': JSON.stringify({ modelTopology: modelTopology1, weightsManifest: weightsManifest }),
'./weightfile0': floatData1,
'./weightfile1': floatData2,
});
handler = tf.io.browserHTTPRequest('./model.json');
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelTopology1);
expect(modelArtifacts.weightSpecs)
.toEqual(weightsManifest[0].weights.concat(weightsManifest[1].weights));
expect(new Float32Array(modelArtifacts.weightData))
.toEqual(new Float32Array([1, 3, 3, 7, 4]));
return [2];
}
});
}); });
it('2 groups, 2 weight, 2 paths, Int32 and Uint8 Data', function () { return __awaiter(_this, void 0, void 0, function () {
var weightsManifest, floatData1, floatData2, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightsManifest = [
{
paths: ['weightfile0'],
weights: [{
name: 'fooWeight',
shape: [3, 1],
dtype: 'int32',
}]
},
{
paths: ['weightfile1'],
weights: [{
name: 'barWeight',
shape: [2],
dtype: 'bool',
}],
}
];
floatData1 = new Int32Array([1, 3, 3]);
floatData2 = new Uint8Array([7, 4]);
setupFakeWeightFiles({
'path1/model.json': JSON.stringify({ modelTopology: modelTopology1, weightsManifest: weightsManifest }),
'path1/weightfile0': floatData1,
'path1/weightfile1': floatData2,
});
handler = tf.io.browserHTTPRequest('path1/model.json');
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelTopology1);
expect(modelArtifacts.weightSpecs)
.toEqual(weightsManifest[0].weights.concat(weightsManifest[1].weights));
expect(new Int32Array(modelArtifacts.weightData.slice(0, 12)))
.toEqual(new Int32Array([1, 3, 3]));
expect(new Uint8Array(modelArtifacts.weightData.slice(12, 14)))
.toEqual(new Uint8Array([7, 4]));
return [2];
}
});
}); });
it('topology only', function () { return __awaiter(_this, void 0, void 0, function () {
var handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
setupFakeWeightFiles({
'./model.json': JSON.stringify({ modelTopology: modelTopology1 }),
});
handler = tf.io.browserHTTPRequest('./model.json');
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelTopology1);
expect(modelArtifacts.weightSpecs).toBeUndefined();
expect(modelArtifacts.weightData).toBeUndefined();
return [2];
}
});
}); });
it('weights only', function () { return __awaiter(_this, void 0, void 0, function () {
var weightsManifest, floatData1, floatData2, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightsManifest = [
{
paths: ['weightfile0'],
weights: [{
name: 'fooWeight',
shape: [3, 1],
dtype: 'int32',
}]
},
{
paths: ['weightfile1'],
weights: [{
name: 'barWeight',
shape: [2],
dtype: 'float32',
}],
}
];
floatData1 = new Int32Array([1, 3, 3]);
floatData2 = new Float32Array([-7, -4]);
setupFakeWeightFiles({
'path1/model.json': JSON.stringify({ weightsManifest: weightsManifest }),
'path1/weightfile0': floatData1,
'path1/weightfile1': floatData2,
});
handler = tf.io.browserHTTPRequest('path1/model.json');
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toBeUndefined();
expect(modelArtifacts.weightSpecs)
.toEqual(weightsManifest[0].weights.concat(weightsManifest[1].weights));
expect(new Int32Array(modelArtifacts.weightData.slice(0, 12)))
.toEqual(new Int32Array([1, 3, 3]));
expect(new Float32Array(modelArtifacts.weightData.slice(12, 20)))
.toEqual(new Float32Array([-7, -4]));
return [2];
}
});
}); });
it('Missing modelTopology and weightsManifest leads to error', function (done) { return __awaiter(_this, void 0, void 0, function () {
var handler;
return __generator(this, function (_a) {
setupFakeWeightFiles({ 'path1/model.json': JSON.stringify({}) });
handler = tf.io.browserHTTPRequest('path1/model.json');
handler.load()
.then(function (modelTopology1) {
done.fail('Loading from missing modelTopology and weightsManifest ' +
'succeeded expectedly.');
})
.catch(function (err) {
expect(err.message)
.toMatch(/contains neither model topology or manifest/);
done();
});
return [2];
});
}); });
});
describe('Binary model', function () {
var requestInits;
var modelData;
var setupFakeWeightFiles = function (fileBufferMap) {
spyOn(window, 'fetch').and.callFake(function (path, init) {
requestInits.push(init);
return new Response(fileBufferMap[path]);
});
};
beforeEach(function () {
requestInits = [];
modelData = new ArrayBuffer(5);
});
it('1 group, 2 weights, 1 path', function (done) {
var weightManifest1 = [{
paths: ['weightfile0'],
weights: [
{
name: 'dense/kernel',
shape: [3, 1],
dtype: 'float32',
},
{
name: 'dense/bias',
shape: [2],
dtype: 'float32',
}
]
}];
var floatData = new Float32Array([1, 3, 3, 7, 4]);
setupFakeWeightFiles({
'./model.pb': modelData,
'./weights_manifest.json': JSON.stringify(weightManifest1),
'./weightfile0': floatData,
});
var handler = tf.io.browserHTTPRequest(['./model.pb', './weights_manifest.json']);
handler.load()
.then(function (modelArtifacts) {
expect(modelArtifacts.modelTopology).toEqual(modelData);
expect(modelArtifacts.weightSpecs)
.toEqual(weightManifest1[0].weights);
expect(new Float32Array(modelArtifacts.weightData))
.toEqual(floatData);
expect(requestInits).toEqual([{}, {}, {}]);
done();
})
.catch(function (err) { return done.fail(err.stack); });
});
it('1 group, 2 weights, 1 path with suffix', function () { return __awaiter(_this, void 0, void 0, function () {
var weightManifest1, floatData, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightManifest1 = [{
paths: ['weightfile0'],
weights: [
{
name: 'dense/kernel',
shape: [3, 1],
dtype: 'float32',
},
{
name: 'dense/bias',
shape: [2],
dtype: 'float32',
}
]
}];
floatData = new Float32Array([1, 3, 3, 7, 4]);
setupFakeWeightFiles({
'./model.pb?tfjs-format=file': modelData,
'./weights_manifest.json?tfjs-format=file': JSON.stringify(weightManifest1),
'./weightfile0?tfjs-format=file': floatData,
});
handler = tf.io.browserHTTPRequest([
'./model.pb?tfjs-format=file',
'./weights_manifest.json?tfjs-format=file'
]);
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelData);
expect(modelArtifacts.weightSpecs).toEqual(weightManifest1[0].weights);
expect(new Float32Array(modelArtifacts.weightData)).toEqual(floatData);
expect(requestInits).toEqual([{}, {}, {}]);
return [2];
}
});
}); });
it('1 group, 2 weights, 1 path, with requestInit', function () { return __awaiter(_this, void 0, void 0, function () {
var weightManifest1, floatData, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightManifest1 = [{
paths: ['weightfile0'],
weights: [
{
name: 'dense/kernel',
shape: [3, 1],
dtype: 'float32',
},
{
name: 'dense/bias',
shape: [2],
dtype: 'float32',
}
]
}];
floatData = new Float32Array([1, 3, 3, 7, 4]);
setupFakeWeightFiles({
'./model.pb': modelData,
'./weights_manifest.json': JSON.stringify(weightManifest1),
'./weightfile0': floatData,
});
handler = tf.io.browserHTTPRequest(['./model.pb', './weights_manifest.json'], { headers: { 'header_key_1': 'header_value_1' } });
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelData);
expect(modelArtifacts.weightSpecs).toEqual(weightManifest1[0].weights);
expect(new Float32Array(modelArtifacts.weightData)).toEqual(floatData);
expect(requestInits).toEqual([
{ headers: { 'header_key_1': 'header_value_1' } },
{ headers: { 'header_key_1': 'header_value_1' } },
{ headers: { 'header_key_1': 'header_value_1' } },
]);
return [2];
}
});
}); });
it('1 group, 2 weight, 2 paths', function () { return __awaiter(_this, void 0, void 0, function () {
var weightManifest1, floatData1, floatData2, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightManifest1 = [{
paths: ['weightfile0', 'weightfile1'],
weights: [
{
name: 'dense/kernel',
shape: [3, 1],
dtype: 'float32',
},
{
name: 'dense/bias',
shape: [2],
dtype: 'float32',
}
]
}];
floatData1 = new Float32Array([1, 3, 3]);
floatData2 = new Float32Array([7, 4]);
setupFakeWeightFiles({
'./model.pb': modelData,
'./weights_manifest.json': JSON.stringify(weightManifest1),
'./weightfile0': floatData1,
'./weightfile1': floatData2,
});
handler = tf.io.browserHTTPRequest(['./model.pb', './weights_manifest.json']);
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelData);
expect(modelArtifacts.weightSpecs).toEqual(weightManifest1[0].weights);
expect(new Float32Array(modelArtifacts.weightData))
.toEqual(new Float32Array([1, 3, 3, 7, 4]));
return [2];
}
});
}); });
it('2 groups, 2 weight, 2 paths', function () { return __awaiter(_this, void 0, void 0, function () {
var weightsManifest, floatData1, floatData2, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightsManifest = [
{
paths: ['weightfile0'],
weights: [{
name: 'dense/kernel',
shape: [3, 1],
dtype: 'float32',
}]
},
{
paths: ['weightfile1'],
weights: [{
name: 'dense/bias',
shape: [2],
dtype: 'float32',
}],
}
];
floatData1 = new Float32Array([1, 3, 3]);
floatData2 = new Float32Array([7, 4]);
setupFakeWeightFiles({
'./model.pb': modelData,
'./weights_manifest.json': JSON.stringify(weightsManifest),
'./weightfile0': floatData1,
'./weightfile1': floatData2,
});
handler = tf.io.browserHTTPRequest(['./model.pb', './weights_manifest.json']);
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelData);
expect(modelArtifacts.weightSpecs)
.toEqual(weightsManifest[0].weights.concat(weightsManifest[1].weights));
expect(new Float32Array(modelArtifacts.weightData))
.toEqual(new Float32Array([1, 3, 3, 7, 4]));
return [2];
}
});
}); });
it('2 groups, 2 weight, 2 paths, Int32 and Uint8 Data', function () { return __awaiter(_this, void 0, void 0, function () {
var weightsManifest, floatData1, floatData2, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightsManifest = [
{
paths: ['weightfile0'],
weights: [{
name: 'fooWeight',
shape: [3, 1],
dtype: 'int32',
}]
},
{
paths: ['weightfile1'],
weights: [{
name: 'barWeight',
shape: [2],
dtype: 'bool',
}],
}
];
floatData1 = new Int32Array([1, 3, 3]);
floatData2 = new Uint8Array([7, 4]);
setupFakeWeightFiles({
'path1/model.pb': modelData,
'path2/weights_manifest.json': JSON.stringify(weightsManifest),
'path2/weightfile0': floatData1,
'path2/weightfile1': floatData2,
});
handler = tf.io.browserHTTPRequest(['path1/model.pb', 'path2/weights_manifest.json']);
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelData);
expect(modelArtifacts.weightSpecs)
.toEqual(weightsManifest[0].weights.concat(weightsManifest[1].weights));
expect(new Int32Array(modelArtifacts.weightData.slice(0, 12)))
.toEqual(new Int32Array([1, 3, 3]));
expect(new Uint8Array(modelArtifacts.weightData.slice(12, 14)))
.toEqual(new Uint8Array([7, 4]));
return [2];
}
});
}); });
it('2 groups, 2 weight, weight path prefix, Int32 and Uint8 Data', function () { return __awaiter(_this, void 0, void 0, function () {
var weightsManifest, floatData1, floatData2, handler, modelArtifacts;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
weightsManifest = [
{
paths: ['weightfile0'],
weights: [{
name: 'fooWeight',
shape: [3, 1],
dtype: 'int32',
}]
},
{
paths: ['weightfile1'],
weights: [{
name: 'barWeight',
shape: [2],
dtype: 'bool',
}],
}
];
floatData1 = new Int32Array([1, 3, 3]);
floatData2 = new Uint8Array([7, 4]);
setupFakeWeightFiles({
'path1/model.pb': modelData,
'path2/weights_manifest.json': JSON.stringify(weightsManifest),
'path3/weightfile0': floatData1,
'path3/weightfile1': floatData2,
});
handler = tf.io.browserHTTPRequest(['path1/model.pb', 'path2/weights_manifest.json'], {}, 'path3/');
return [4, handler.load()];
case 1:
modelArtifacts = _a.sent();
expect(modelArtifacts.modelTopology).toEqual(modelData);
expect(modelArtifacts.weightSpecs)
.toEqual(weightsManifest[0].weights.concat(weightsManifest[1].weights));
expect(new Int32Array(modelArtifacts.weightData.slice(0, 12)))
.toEqual(new Int32Array([1, 3, 3]));
expect(new Uint8Array(modelArtifacts.weightData.slice(12, 14)))
.toEqual(new Uint8Array([7, 4]));
return [2];
}
});
}); });
it('the url path length is not 2 should leads to error', function () {
expect(function () { return tf.io.browserHTTPRequest(['path1/model.pb']); }).toThrow();
});
});
it('Overriding BrowserHTTPRequest fetchFunc', function () { return __awaiter(_this, void 0, void 0, function () {
function customFetch(input, init) {
return __awaiter(this, void 0, void 0, function () {
return __generator(this, function (_a) {
fetchInputs.push(input);
fetch