@tensorflow/tfjs-node - npm Package Compare versions

Comparing version 0.1.7 to 0.1.8

dist/index.js

@@ -9,3 +9,3 @@ "use strict";
     return new nodejs_kernel_backend_1.NodeJSKernelBackend(bindings('tfjs_binding.node'));
-});
+}, 3);
 if (tfc.ENV.findBackend('tensorflow') != null) {
@@ -12,0 +12,0 @@ tfc.setBackend('tensorflow');

dist/io/file_system_test.js

@@ -45,3 +45,2 @@ "use strict";
 var util_1 = require("util");
-var io_utils_1 = require("./io_utils");
 describe('File system IOHandler', function () {
@@ -252,3 +251,3 @@ var mkdtemp = util_1.promisify(fs.mkdtemp);
                     _a.sent();
-                    weightsData2 = io_utils_1.toBuffer(new Float32Array([-7.7]).buffer);
+                    weightsData2 = Buffer.from(new Float32Array([-7.7]).buffer);
                     return [4, writeFile(path.join(testDir, 'weights.2.bin'), weightsData2, 'binary')];
@@ -255,0 +254,0 @@ case 3:

dist/io/file_system.js

@@ -95,5 +95,5 @@ "use strict";
         return __awaiter(this, void 0, void 0, function () {
-            var exists, stat, readFile, modelJSON, _a, _b, modelArtifacts, dirName, buffers, weightSpecs, _i, _c, group, _d, _e, path, weightFilePath, buffer, _f;
-            return __generator(this, function (_g) {
-                switch (_g.label) {
+            var exists, stat, readFile, modelJSON, _a, _b, modelArtifacts, dirName, buffers, weightSpecs, _i, _c, group, _d, _e, path, weightFilePath, buffer;
+            return __generator(this, function (_f) {
+                switch (_f.label) {
                     case 0:
@@ -106,3 +106,3 @@ if (Array.isArray(this.path)) {
                     case 1:
-                        if (!(_g.sent())) {
+                        if (!(_f.sent())) {
                             throw new Error("Path " + this.path + " does not exist: loading failed.");
@@ -113,3 +113,3 @@ }
                     case 2:
-                        if (!(_g.sent()).isFile()) return [3, 13];
+                        if (!(_f.sent()).isFile()) return [3, 13];
                         readFile = util_1.promisify(fs.readFile);
@@ -119,3 +119,3 @@ _b = (_a = JSON).parse;
                     case 3:
-                        modelJSON = _b.apply(_a, [_g.sent()]);
+                        modelJSON = _b.apply(_a, [_f.sent()]);
                         modelArtifacts = {
@@ -129,3 +129,3 @@ modelTopology: modelJSON.modelTopology,
                         _i = 0, _c = modelJSON.weightsManifest;
-                        _g.label = 4;
+                        _f.label = 4;
                     case 4:
@@ -135,3 +135,3 @@ if (!(_i < _c.length)) return [3, 11];
                         _d = 0, _e = group.paths;
-                        _g.label = 5;
+                        _f.label = 5;
                     case 5:
@@ -143,11 +143,10 @@ if (!(_d < _e.length)) return [3, 9];
                     case 6:
-                        if (!(_g.sent())) {
+                        if (!(_f.sent())) {
                             throw new Error("Weight file " + weightFilePath + " does not exist: loading failed");
                         }
-                        _f = Buffer.bind;
                         return [4, readFile(weightFilePath)];
                     case 7:
-                        buffer = new (_f.apply(Buffer, [void 0, _g.sent()]))();
+                        buffer = _f.sent();
                         buffers.push(buffer);
-                        _g.label = 8;
+                        _f.label = 8;
                     case 8:
@@ -158,3 +157,3 @@ _d++;
                         weightSpecs.push.apply(weightSpecs, group.weights);
-                        _g.label = 10;
+                        _f.label = 10;
                     case 10:
@@ -166,3 +165,3 @@ _i++;
                         modelArtifacts.weightData = io_utils_1.toArrayBuffer(buffers);
-                        _g.label = 12;
+                        _f.label = 12;
                     case 12: return [2, modelArtifacts];
@@ -169,0 +168,0 @@ case 13: throw new Error('The path to load from must be a file. Loading from a directory ' +

dist/io/io_utils_test.js

@@ -13,3 +13,3 @@ "use strict";
     it('Single Buffer', function () {
-        var buf = new Buffer([10, 20, 30]);
+        var buf = Buffer.from([10, 20, 30]);
         var ab = io_utils_1.toArrayBuffer(buf);
@@ -19,4 +19,4 @@ expect(new Uint8Array(ab)).toEqual(new Uint8Array([10, 20, 30]));
     it('Two Buffers', function () {
-        var buf1 = new Buffer([10, 20, 30]);
-        var buf2 = new Buffer([40, 50, 60]);
+        var buf1 = Buffer.from([10, 20, 30]);
+        var buf2 = Buffer.from([40, 50, 60]);
         var ab = io_utils_1.toArrayBuffer([buf1, buf2]);
@@ -28,5 +28,5 @@ expect(new Uint8Array(ab)).toEqual(new Uint8Array([
     it('Three Buffers', function () {
-        var buf1 = new Buffer([10, 20, 30]);
-        var buf2 = new Buffer([40, 50, 60]);
-        var buf3 = new Buffer([3, 2, 1]);
+        var buf1 = Buffer.from([10, 20, 30]);
+        var buf2 = Buffer.from([40, 50, 60]);
+        var buf3 = Buffer.from([3, 2, 1]);
         var ab = io_utils_1.toArrayBuffer([buf1, buf2, buf3]);
@@ -33,0 +33,0 @@ expect(new Uint8Array(ab)).toEqual(new Uint8Array([

dist/io/io_utils.js

 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 function toBuffer(ab) {
-    var buf = new Buffer(ab.byteLength);
     var view = new Uint8Array(ab);
-    view.forEach(function (value, i) {
-        buf[i] = value;
-    });
-    return buf;
+    return Buffer.from(view);
 }
@@ -14,24 +10,19 @@ exports.toBuffer = toBuffer;
     if (Array.isArray(buf)) {
-        var totalLength_1 = 0;
-        buf.forEach(function (buffer) {
-            totalLength_1 += buffer.length;
-        });
-        var ab = new ArrayBuffer(totalLength_1);
-        var view_1 = new Uint8Array(ab);
-        var pos_1 = 0;
-        buf.forEach(function (buffer) {
-            for (var i = 0; i < buffer.length; ++i) {
-                view_1[pos_1++] = buffer[i];
-            }
-        });
-        return ab;
-    }
-    else {
-        var ab = new ArrayBuffer(buf.length);
+        var totalLength = 0;
+        for (var _i = 0, buf_1 = buf; _i < buf_1.length; _i++) {
+            var buffer = buf_1[_i];
+            totalLength += buffer.length;
+        }
+        var ab = new ArrayBuffer(totalLength);
         var view = new Uint8Array(ab);
-        for (var i = 0; i < buf.length; ++i) {
-            view[i] = buf[i];
+        var pos = 0;
+        for (var _a = 0, buf_2 = buf; _a < buf_2.length; _a++) {
+            var buffer = buf_2[_a];
+            pos += buffer.copy(view, pos);
         }
         return ab;
     }
+    else {
+        return buf.buffer.slice(buf.byteOffset, buf.byteOffset + buf.byteLength);
+    }
 }
@@ -38,0 +29,0 @@ exports.toArrayBuffer = toArrayBuffer;

dist/nodejs_kernel_backend_test.js

@@ -21,1 +21,6 @@ "use strict";
 });
+describe('type casting', function () {
+    it('exp support int32', function () {
+        tf.exp(tf.scalar(2, 'int32'));
+    });
+});

dist/nodejs_kernel_backend.d.ts

@@ -23,3 +23,3 @@ import { BackendTimingInfo, DataType, KernelBackend, Rank, ShapeMap, Tensor, Tensor1D, Tensor2D, Tensor3D, Tensor4D } from '@tensorflow/tfjs-core';
     matMul(a: Tensor2D, b: Tensor2D, transposeA: boolean, transposeB: boolean): Tensor2D;
-    stridedSlice<T extends Tensor<Rank>>(x: T, begin: number[], end: number[], strides: number[], beginMask: number, endMask: number): T;
+    stridedSlice<T extends Tensor>(x: T, begin: number[], end: number[], strides: number[], beginMask: number, endMask: number): T;
     slice<T extends Tensor>(x: T, begin: number[], size: number[]): T;
@@ -32,5 +32,6 @@ reverse<T extends Tensor>(a: T, axis: number[]): T;
     multiply(a: Tensor, b: Tensor): Tensor;
-    realDivide(a: Tensor<Rank>, b: Tensor<Rank>): Tensor<Rank>;
-    floorDiv(a: Tensor<Rank>, b: Tensor<Rank>): Tensor<Rank>;
+    realDivide(a: Tensor, b: Tensor): Tensor;
+    floorDiv(a: Tensor, b: Tensor): Tensor;
     divide(a: Tensor, b: Tensor): Tensor;
+    unsortedSegmentSum<T extends Tensor>(x: T, segmentIds: Tensor1D, numSegments: number): Tensor;
     sum(x: Tensor, axes: number[]): Tensor;
@@ -55,2 +56,4 @@ argMin(x: Tensor, axis: number): Tensor;
     maximum(a: Tensor, b: Tensor): Tensor;
+    all(x: Tensor, axes: number[]): Tensor;
+    any(x: Tensor, axes: number[]): Tensor;
     ceil<T extends Tensor>(x: T): T;
@@ -98,4 +101,4 @@ floor<T extends Tensor>(x: T): T;
     conv2dDerFilter(x: Tensor4D, dy: Tensor4D, convInfo: Conv2DInfo): Tensor4D;
-    depthwiseConv2DDerInput(dy: Tensor<Rank.R4>, filter: Tensor<Rank.R4>, convInfo: Conv2DInfo): Tensor4D;
-    depthwiseConv2DDerFilter(x: Tensor<Rank.R4>, dY: Tensor<Rank.R4>, convInfo: Conv2DInfo): Tensor<Rank.R4>;
+    depthwiseConv2DDerInput(dy: Tensor4D, filter: Tensor4D, convInfo: Conv2DInfo): Tensor4D;
+    depthwiseConv2DDerFilter(x: Tensor4D, dY: Tensor4D, convInfo: Conv2DInfo): Tensor4D;
     depthwiseConv2D(input: Tensor4D, filter: Tensor4D, convInfo: Conv2DInfo): Tensor4D;
@@ -113,4 +116,5 @@ maxPool(x: Tensor4D, convInfo: Conv2DInfo): Tensor4D;
     resizeBilinear(x: Tensor4D, newHeight: number, newWidth: number, alignCorners: boolean): Tensor4D;
-    resizeBilinearBackprop(dy: Tensor<Rank.R4>, x: Tensor<Rank.R4>, alignCorners: boolean): Tensor<Rank.R4>;
+    resizeBilinearBackprop(dy: Tensor4D, x: Tensor4D, alignCorners: boolean): Tensor4D;
     resizeNearestNeighbor(x: Tensor4D, newHeight: number, newWidth: number, alignCorners: boolean): Tensor4D;
+    resizeNearestNeighborBackprop(dy: Tensor4D, x: Tensor4D, alignCorners: boolean): Tensor4D;
     batchNormalization(x: Tensor4D, mean: Tensor1D | Tensor4D, variance: Tensor1D | Tensor4D, varianceEpsilon: number, scale?: Tensor1D | Tensor4D, offset?: Tensor1D | Tensor4D): Tensor4D;
@@ -120,3 +124,3 @@ localResponseNormalization4D(x: Tensor4D, radius: number, bias: number, alpha: number, beta: number): Tensor4D;
     oneHot(indices: Tensor1D, depth: number, onValue: number, offValue: number): Tensor2D;
-    cumsum(x: Tensor<Rank>, axis: number, exclusive: boolean, reverse: boolean): Tensor<Rank>;
+    cumsum(x: Tensor, axis: number, exclusive: boolean, reverse: boolean): Tensor;
     fromPixels(pixels: ImageData | HTMLImageElement | HTMLCanvasElement | HTMLVideoElement, numChannels: number): Tensor3D;
@@ -123,0 +127,0 @@ memory(): {

dist/nodejs_kernel_backend.js

@@ -122,5 +122,3 @@ "use strict";
     };
-    NodeJSKernelBackend.prototype.dispose = function () {
-        throw new Error('Method not implemented.');
-    };
+    NodeJSKernelBackend.prototype.dispose = function () { };
     NodeJSKernelBackend.prototype.read = function (dataId) {
@@ -241,2 +239,10 @@ return __awaiter(this, void 0, void 0, function () {
     };
+    NodeJSKernelBackend.prototype.unsortedSegmentSum = function (x, segmentIds, numSegments) {
+        var opAttrs = [
+            this.createTypeOpAttr('T', x.dtype),
+            this.createTypeOpAttr('Tindices', 'int32'),
+            this.createTypeOpAttr('Tnumsegments', 'int32')
+        ];
+        return this.executeSingleOutput('UnsortedSegmentSum', opAttrs, [x, segmentIds, tfjs_core_1.scalar(numSegments, 'int32')]);
+    };
     NodeJSKernelBackend.prototype.sum = function (x, axes) {
@@ -323,2 +329,18 @@ var axisTensor = tfjs_core_1.tensor1d(axes, 'int32');
     };
+    NodeJSKernelBackend.prototype.all = function (x, axes) {
+        var opAttrs = [
+            { name: 'keep_dims', type: this.binding.TF_ATTR_BOOL, value: false },
+            this.createTypeOpAttr('Tidx', 'int32')
+        ];
+        var axesTensor = tfjs_core_1.tensor1d(axes, 'int32');
+        return this.executeSingleOutput('All', opAttrs, [x, axesTensor]);
+    };
+    NodeJSKernelBackend.prototype.any = function (x, axes) {
+        var opAttrs = [
+            { name: 'keep_dims', type: this.binding.TF_ATTR_BOOL, value: false },
+            this.createTypeOpAttr('Tidx', 'int32')
+        ];
+        var axesTensor = tfjs_core_1.tensor1d(axes, 'int32');
+        return this.executeSingleOutput('Any', opAttrs, [x, axesTensor]);
+    };
     NodeJSKernelBackend.prototype.ceil = function (x) {
@@ -331,7 +353,9 @@ return this.executeSingleInput('Ceil', x);
     NodeJSKernelBackend.prototype.pow = function (a, b) {
-        var opAttrs = [this.createTypeOpAttr('T', types_1.upcastType(a.dtype, b.dtype))];
-        return this.executeSingleOutput('Pow', opAttrs, [a, b]);
+        var dtype = types_1.upcastType(a.dtype, b.dtype);
+        var opAttrs = [this.createTypeOpAttr('T', dtype)];
+        return this.executeSingleOutput('Pow', opAttrs, [a.cast(dtype), b.cast(dtype)]);
     };
     NodeJSKernelBackend.prototype.exp = function (x) {
-        return this.executeSingleInput('Exp', x);
+        var xTensor = x.dtype === 'int32' ? x.toFloat() : x;
+        return this.executeSingleInput('Exp', xTensor);
     };
@@ -748,2 +772,14 @@ NodeJSKernelBackend.prototype.log = function (x) {
     };
+    NodeJSKernelBackend.prototype.resizeNearestNeighborBackprop = function (dy, x, alignCorners) {
+        var opAttrs = [
+            this.createTypeOpAttr('T', x.dtype), {
+                name: 'align_corners',
+                type: this.binding.TF_ATTR_BOOL,
+                value: alignCorners
+            }
+        ];
+        var _a = x.shape, origHeight = _a[1], origWidth = _a[2];
+        var size = tfjs_core_1.tensor1d([origHeight, origWidth], 'int32');
+        return this.executeSingleOutput('ResizeNearestNeighborGrad', opAttrs, [dy, size]);
+    };
     NodeJSKernelBackend.prototype.batchNormalization = function (x, mean, variance, varianceEpsilon, scale, offset) {
@@ -830,3 +866,28 @@ if (mean.rank > 1) {
     NodeJSKernelBackend.prototype.fromPixels = function (pixels, numChannels) {
-        throw new Error('Method not implemented.');
+        if (pixels == null) {
+            throw new Error('pixels passed to tf.fromPixels() can not be null');
+        }
+        if (pixels.getContext == null) {
+            throw new Error('When running in node, pixels must be an HTMLCanvasElement ' +
+                'like the one returned by the `canvas` npm package');
+        }
+        var vals = pixels
+            .getContext('2d')
+            .getImageData(0, 0, pixels.width, pixels.height)
+            .data;
+        var values;
+        if (numChannels === 4) {
+            values = new Int32Array(vals);
+        }
+        else {
+            var numPixels = pixels.width * pixels.height;
+            values = new Int32Array(numPixels * numChannels);
+            for (var i = 0; i < numPixels; i++) {
+                for (var channel = 0; channel < numChannels; ++channel) {
+                    values[i * numChannels + channel] = vals[i * 4 + channel];
+                }
+            }
+        }
+        var outShape = [pixels.height, pixels.width, numChannels];
+        return tfjs_core_1.tensor3d(values, outShape, 'int32');
     };
@@ -833,0 +894,0 @@ NodeJSKernelBackend.prototype.memory = function () {

dist/run_tests.js

@@ -5,13 +5,16 @@ "use strict";
 var jasmine_util = require("@tensorflow/tfjs-core/dist/jasmine_util");
+Error.stackTraceLimit = Infinity;
 var jasmineCtor = require('jasmine');
-jasmine_util.setBeforeAll(function () { });
-jasmine_util.setAfterAll(function () { });
-jasmine_util.setBeforeEach(function () { });
-jasmine_util.setAfterEach(function () { });
-jasmine_util.setTestEnvFeatures([{ BACKEND: 'tensorflow' }]);
+var bindings = require("bindings");
+var nodejs_kernel_backend_1 = require("./nodejs_kernel_backend");
+jasmine_util.setTestEnvs([{
+        name: 'test-tensorflow',
+        factory: function () {
+            return new nodejs_kernel_backend_1.NodeJSKernelBackend(bindings('tfjs_binding.node'));
+        },
+        features: {}
+    }]);
 var IGNORE_LIST = [
     'depthwiseConv2D',
     'separableConv2d',
-    'IORouterRegistry',
-    'arrayBufferToBase64String', 'stringByteLength'
 ];
@@ -22,4 +25,8 @@ var runner = new jasmineCtor();
         'src/**/*_test.ts', 'node_modules/@tensorflow/tfjs-core/dist/**/*_test.js'
-    ]
+    ],
+    random: false
 });
+if (process.env.JASMINE_SEED) {
+    runner.seed(process.env.JASMINE_SEED);
+}
 var env = jasmine.getEnv();
@@ -26,0 +33,0 @@ env.specFilter = function (spec) {

dist/version.d.ts

@@ -1,2 +0,2 @@
-declare const version = "0.1.7";
+declare const version = "0.1.8";
 export { version };

dist/version.js

 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-var version = '0.1.7';
+var version = '0.1.8';
 exports.version = version;

package.json

 {
   "name": "@tensorflow/tfjs-node",
-  "version": "0.1.7",
+  "version": "0.1.8",
   "main": "dist/index.js",
@@ -20,12 +20,15 @@ "types": "dist/index.d.ts",
     "test": "ts-node src/run_tests.ts",
+    "coverage": "nyc ts-node src/run_tests.ts",
+    "link-local": "yalc link",
     "publish-local": "yarn prep && yarn build && yalc push"
   },
   "devDependencies": {
-    "@tensorflow/tfjs-core": "~0.11.5",
+    "@tensorflow/tfjs-core": "~0.12.1",
     "@types/bindings": "~1.3.0",
     "@types/jasmine": "~2.8.6",
-    "@types/node": "~9.6.2",
+    "@types/node": "^10.5.1",
     "@types/rimraf": "~2.0.2",
     "clang-format": "~1.2.2",
     "jasmine": "~3.1.0",
+    "nyc": "^12.0.2",
     "ts-node": "~5.0.0",
@@ -40,4 +43,4 @@ "tslint": "~5.9.1",
   "peerDependencies": {
-    "@tensorflow/tfjs-core": "~0.11.5"
+    "@tensorflow/tfjs-core": "~0.12.1"
   }
 }

New alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Environment variable access

Supply chain risk

Package accesses environment variables, which may be a sign of credential stuffing or data theft.

Found 1 instance in 1 package

Fixed alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Improved metrics

Total package byte prevSize

233747

increased by49.74%

Number of package files

49

increased by40%

Lines of code

2093

increased by5.71%

Worsened metrics

Dev dependency count

12

increased by9.09%

Number of low supply chain risk alerts

4

increased by100%

No dependency changes