woolball-client/dist/providers/worker-string.d.ts

Client-side library for Woolball enabling secure browser resource sharing for distributed AI task processing

declare const workerCode = "\"use strict\";\n(() => {\n  var __create = Object.create;\n  var __defProp = Object.defineProperty;\n  var __getOwnPropDesc = Object.getOwnPropertyDescriptor;\n  var __getOwnPropNames = Object.getOwnPropertyNames;\n  var __getProtoOf = Object.getPrototypeOf;\n  var __hasOwnProp = Object.prototype.hasOwnProperty;\n  var __defNormalProp = (obj, key, value) => key in obj ? __defProp(obj, key, { enumerable: true, configurable: true, writable: true, value }) : obj[key] = value;\n  var __require = /* @__PURE__ */ ((x3) => typeof require !== \"undefined\" ? require : typeof Proxy !== \"undefined\" ? new Proxy(x3, {\n    get: (a2, b4) => (typeof require !== \"undefined\" ? require : a2)[b4]\n  }) : x3)(function(x3) {\n    if (typeof require !== \"undefined\") return require.apply(this, arguments);\n    throw Error('Dynamic require of \"' + x3 + '\" is not supported');\n  });\n  var __esm = (fn2, res) => function __init() {\n    return fn2 && (res = (0, fn2[__getOwnPropNames(fn2)[0]])(fn2 = 0)), res;\n  };\n  var __commonJS = (cb, mod) => function __require2() {\n    return mod || (0, cb[__getOwnPropNames(cb)[0]])((mod = { exports: {} }).exports, mod), mod.exports;\n  };\n  var __export = (target, all) => {\n    for (var name in all)\n      __defProp(target, name, { get: all[name], enumerable: true });\n  };\n  var __copyProps = (to2, from, except, desc) => {\n    if (from && typeof from === \"object\" || typeof from === \"function\") {\n      for (let key of __getOwnPropNames(from))\n        if (!__hasOwnProp.call(to2, key) && key !== except)\n          __defProp(to2, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });\n    }\n    return to2;\n  };\n  var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__getProtoOf(mod)) : {}, __copyProps(\n    // If the importer is in node compatibility mode or this is not an ESM\n    // file that has been converted to a CommonJS file using a Babel-\n    // compatible transform (i.e. \"__esModule\" has not been set), then set\n    // \"default\" to the CommonJS \"module.exports\" for node compatibility.\n    isNodeMode || !mod || !mod.__esModule ? __defProp(target, \"default\", { value: mod, enumerable: true }) : target,\n    mod\n  ));\n  var __publicField = (obj, key, value) => __defNormalProp(obj, typeof key !== \"symbol\" ? key + \"\" : key, value);\n\n  // node_modules/onnxruntime-common/dist/esm/backend-impl.js\n  var backends, backendsSortedByPriority, registerBackend, tryResolveAndInitializeBackend, resolveBackendAndExecutionProviders;\n  var init_backend_impl = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/backend-impl.js\"() {\n      backends = /* @__PURE__ */ new Map();\n      backendsSortedByPriority = [];\n      registerBackend = (name, backend, priority) => {\n        if (backend && typeof backend.init === \"function\" && typeof backend.createInferenceSessionHandler === \"function\") {\n          const currentBackend = backends.get(name);\n          if (currentBackend === void 0) {\n            backends.set(name, { backend, priority });\n          } else if (currentBackend.priority > priority) {\n            return;\n          } else if (currentBackend.priority === priority) {\n            if (currentBackend.backend !== backend) {\n              throw new Error(`cannot register backend \"${name}\" using priority ${priority}`);\n            }\n          }\n          if (priority >= 0) {\n            const i4 = backendsSortedByPriority.indexOf(name);\n            if (i4 !== -1) {\n              backendsSortedByPriority.splice(i4, 1);\n            }\n            for (let i5 = 0; i5 < backendsSortedByPriority.length; i5++) {\n              if (backends.get(backendsSortedByPriority[i5]).priority <= priority) {\n                backendsSortedByPriority.splice(i5, 0, name);\n                return;\n              }\n            }\n            backendsSortedByPriority.push(name);\n          }\n          return;\n        }\n        throw new TypeError(\"not a valid backend\");\n      };\n      tryResolveAndInitializeBackend = async (backendName) => {\n        const backendInfo = backends.get(backendName);\n        if (!backendInfo) {\n          return \"backend not found.\";\n        }\n        if (backendInfo.initialized) {\n          return backendInfo.backend;\n        } else if (backendInfo.aborted) {\n          return backendInfo.error;\n        } else {\n          const isInitializing = !!backendInfo.initPromise;\n          try {\n            if (!isInitializing) {\n              backendInfo.initPromise = backendInfo.backend.init(backendName);\n            }\n            await backendInfo.initPromise;\n            backendInfo.initialized = true;\n            return backendInfo.backend;\n          } catch (e3) {\n            if (!isInitializing) {\n              backendInfo.error = `${e3}`;\n              backendInfo.aborted = true;\n            }\n            return backendInfo.error;\n          } finally {\n            delete backendInfo.initPromise;\n          }\n        }\n      };\n      resolveBackendAndExecutionProviders = async (options) => {\n        const eps = options.executionProviders || [];\n        const backendHints = eps.map((i4) => typeof i4 === \"string\" ? i4 : i4.name);\n        const backendNames = backendHints.length === 0 ? backendsSortedByPriority : backendHints;\n        let backend;\n        const errors = [];\n        const availableBackendNames = /* @__PURE__ */ new Set();\n        for (const backendName of backendNames) {\n          const resolveResult = await tryResolveAndInitializeBackend(backendName);\n          if (typeof resolveResult === \"string\") {\n            errors.push({ name: backendName, err: resolveResult });\n          } else {\n            if (!backend) {\n              backend = resolveResult;\n            }\n            if (backend === resolveResult) {\n              availableBackendNames.add(backendName);\n            }\n          }\n        }\n        if (!backend) {\n          throw new Error(`no available backend found. ERR: ${errors.map((e3) => `[${e3.name}] ${e3.err}`).join(\", \")}`);\n        }\n        for (const { name, err } of errors) {\n          if (backendHints.includes(name)) {\n            console.warn(`removing requested execution provider \"${name}\" from session options because it is not available: ${err}`);\n          }\n        }\n        const filteredEps = eps.filter((i4) => availableBackendNames.has(typeof i4 === \"string\" ? i4 : i4.name));\n        return [\n          backend,\n          new Proxy(options, {\n            get: (target, prop) => {\n              if (prop === \"executionProviders\") {\n                return filteredEps;\n              }\n              return Reflect.get(target, prop);\n            }\n          })\n        ];\n      };\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/backend.js\n  var init_backend = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/backend.js\"() {\n      init_backend_impl();\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/version.js\n  var version;\n  var init_version = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/version.js\"() {\n      version = \"1.21.0\";\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/env-impl.js\n  var logLevelValue, env;\n  var init_env_impl = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/env-impl.js\"() {\n      init_version();\n      logLevelValue = \"warning\";\n      env = {\n        wasm: {},\n        webgl: {},\n        webgpu: {},\n        versions: { common: version },\n        set logLevel(value) {\n          if (value === void 0) {\n            return;\n          }\n          if (typeof value !== \"string\" || [\"verbose\", \"info\", \"warning\", \"error\", \"fatal\"].indexOf(value) === -1) {\n            throw new Error(`Unsupported logging level: ${value}`);\n          }\n          logLevelValue = value;\n        },\n        get logLevel() {\n          return logLevelValue;\n        }\n      };\n      Object.defineProperty(env, \"logLevel\", { enumerable: true });\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/env.js\n  var env2;\n  var init_env = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/env.js\"() {\n      init_env_impl();\n      env2 = env;\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/tensor-conversion-impl.js\n  var tensorToDataURL, tensorToImageData;\n  var init_tensor_conversion_impl = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/tensor-conversion-impl.js\"() {\n      tensorToDataURL = (tensor, options) => {\n        const canvas = typeof document !== \"undefined\" ? document.createElement(\"canvas\") : new OffscreenCanvas(1, 1);\n        canvas.width = tensor.dims[3];\n        canvas.height = tensor.dims[2];\n        const pixels2DContext = canvas.getContext(\"2d\");\n        if (pixels2DContext != null) {\n          let width;\n          let height;\n          if (options?.tensorLayout !== void 0 && options.tensorLayout === \"NHWC\") {\n            width = tensor.dims[2];\n            height = tensor.dims[3];\n          } else {\n            width = tensor.dims[3];\n            height = tensor.dims[2];\n          }\n          const inputformat = options?.format !== void 0 ? options.format : \"RGB\";\n          const norm = options?.norm;\n          let normMean;\n          let normBias;\n          if (norm === void 0 || norm.mean === void 0) {\n            normMean = [255, 255, 255, 255];\n          } else {\n            if (typeof norm.mean === \"number\") {\n              normMean = [norm.mean, norm.mean, norm.mean, norm.mean];\n            } else {\n              normMean = [norm.mean[0], norm.mean[1], norm.mean[2], 0];\n              if (norm.mean[3] !== void 0) {\n                normMean[3] = norm.mean[3];\n              }\n            }\n          }\n          if (norm === void 0 || norm.bias === void 0) {\n            normBias = [0, 0, 0, 0];\n          } else {\n            if (typeof norm.bias === \"number\") {\n              normBias = [norm.bias, norm.bias, norm.bias, norm.bias];\n            } else {\n              normBias = [norm.bias[0], norm.bias[1], norm.bias[2], 0];\n              if (norm.bias[3] !== void 0) {\n                normBias[3] = norm.bias[3];\n              }\n            }\n          }\n          const stride = height * width;\n          let rTensorPointer = 0, gTensorPointer = stride, bTensorPointer = stride * 2, aTensorPointer = -1;\n          if (inputformat === \"RGBA\") {\n            rTensorPointer = 0;\n            gTensorPointer = stride;\n            bTensorPointer = stride * 2;\n            aTensorPointer = stride * 3;\n          } else if (inputformat === \"RGB\") {\n            rTensorPointer = 0;\n            gTensorPointer = stride;\n            bTensorPointer = stride * 2;\n          } else if (inputformat === \"RBG\") {\n            rTensorPointer = 0;\n            bTensorPointer = stride;\n            gTensorPointer = stride * 2;\n          }\n          for (let i4 = 0; i4 < height; i4++) {\n            for (let j3 = 0; j3 < width; j3++) {\n              const R3 = (tensor.data[rTensorPointer++] - normBias[0]) * normMean[0];\n              const G4 = (tensor.data[gTensorPointer++] - normBias[1]) * normMean[1];\n              const B3 = (tensor.data[bTensorPointer++] - normBias[2]) * normMean[2];\n              const A4 = aTensorPointer === -1 ? 255 : (tensor.data[aTensorPointer++] - normBias[3]) * normMean[3];\n              pixels2DContext.fillStyle = \"rgba(\" + R3 + \",\" + G4 + \",\" + B3 + \",\" + A4 + \")\";\n              pixels2DContext.fillRect(j3, i4, 1, 1);\n            }\n          }\n          if (\"toDataURL\" in canvas) {\n            return canvas.toDataURL();\n          } else {\n            throw new Error(\"toDataURL is not supported\");\n          }\n        } else {\n          throw new Error(\"Can not access image data\");\n        }\n      };\n      tensorToImageData = (tensor, options) => {\n        const pixels2DContext = typeof document !== \"undefined\" ? document.createElement(\"canvas\").getContext(\"2d\") : new OffscreenCanvas(1, 1).getContext(\"2d\");\n        let image;\n        if (pixels2DContext != null) {\n          let width;\n          let height;\n          let channels;\n          if (options?.tensorLayout !== void 0 && options.tensorLayout === \"NHWC\") {\n            width = tensor.dims[2];\n            height = tensor.dims[1];\n            channels = tensor.dims[3];\n          } else {\n            width = tensor.dims[3];\n            height = tensor.dims[2];\n            channels = tensor.dims[1];\n          }\n          const inputformat = options !== void 0 ? options.format !== void 0 ? options.format : \"RGB\" : \"RGB\";\n          const norm = options?.norm;\n          let normMean;\n          let normBias;\n          if (norm === void 0 || norm.mean === void 0) {\n            normMean = [255, 255, 255, 255];\n          } else {\n            if (typeof norm.mean === \"number\") {\n              normMean = [norm.mean, norm.mean, norm.mean, norm.mean];\n            } else {\n              normMean = [norm.mean[0], norm.mean[1], norm.mean[2], 255];\n              if (norm.mean[3] !== void 0) {\n                normMean[3] = norm.mean[3];\n              }\n            }\n          }\n          if (norm === void 0 || norm.bias === void 0) {\n            normBias = [0, 0, 0, 0];\n          } else {\n            if (typeof norm.bias === \"number\") {\n              normBias = [norm.bias, norm.bias, norm.bias, norm.bias];\n            } else {\n              normBias = [norm.bias[0], norm.bias[1], norm.bias[2], 0];\n              if (norm.bias[3] !== void 0) {\n                normBias[3] = norm.bias[3];\n              }\n            }\n          }\n          const stride = height * width;\n          if (options !== void 0) {\n            if (options.format !== void 0 && channels === 4 && options.format !== \"RGBA\" || channels === 3 && options.format !== \"RGB\" && options.format !== \"BGR\") {\n              throw new Error(\"Tensor format doesn't match input tensor dims\");\n            }\n          }\n          const step = 4;\n          let rImagePointer = 0, gImagePointer = 1, bImagePointer = 2, aImagePointer = 3;\n          let rTensorPointer = 0, gTensorPointer = stride, bTensorPointer = stride * 2, aTensorPointer = -1;\n          if (inputformat === \"RGBA\") {\n            rTensorPointer = 0;\n            gTensorPointer = stride;\n            bTensorPointer = stride * 2;\n            aTensorPointer = stride * 3;\n          } else if (inputformat === \"RGB\") {\n            rTensorPointer = 0;\n            gTensorPointer = stride;\n            bTensorPointer = stride * 2;\n          } else if (inputformat === \"RBG\") {\n            rTensorPointer = 0;\n            bTensorPointer = stride;\n            gTensorPointer = stride * 2;\n          }\n          image = pixels2DContext.createImageData(width, height);\n          for (let i4 = 0; i4 < height * width; rImagePointer += step, gImagePointer += step, bImagePointer += step, aImagePointer += step, i4++) {\n            image.data[rImagePointer] = (tensor.data[rTensorPointer++] - normBias[0]) * normMean[0];\n            image.data[gImagePointer] = (tensor.data[gTensorPointer++] - normBias[1]) * normMean[1];\n            image.data[bImagePointer] = (tensor.data[bTensorPointer++] - normBias[2]) * normMean[2];\n            image.data[aImagePointer] = aTensorPointer === -1 ? 255 : (tensor.data[aTensorPointer++] - normBias[3]) * normMean[3];\n          }\n        } else {\n          throw new Error(\"Can not access image data\");\n        }\n        return image;\n      };\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/tensor-factory-impl.js\n  var bufferToTensor, tensorFromImage, tensorFromTexture, tensorFromGpuBuffer, tensorFromMLTensor, tensorFromPinnedBuffer;\n  var init_tensor_factory_impl = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/tensor-factory-impl.js\"() {\n      init_tensor_impl();\n      bufferToTensor = (buffer, options) => {\n        if (buffer === void 0) {\n          throw new Error(\"Image buffer must be defined\");\n        }\n        if (options.height === void 0 || options.width === void 0) {\n          throw new Error(\"Image height and width must be defined\");\n        }\n        if (options.tensorLayout === \"NHWC\") {\n          throw new Error(\"NHWC Tensor layout is not supported yet\");\n        }\n        const { height, width } = options;\n        const norm = options.norm ?? { mean: 255, bias: 0 };\n        let normMean;\n        let normBias;\n        if (typeof norm.mean === \"number\") {\n          normMean = [norm.mean, norm.mean, norm.mean, norm.mean];\n        } else {\n          normMean = [norm.mean[0], norm.mean[1], norm.mean[2], norm.mean[3] ?? 255];\n        }\n        if (typeof norm.bias === \"number\") {\n          normBias = [norm.bias, norm.bias, norm.bias, norm.bias];\n        } else {\n          normBias = [norm.bias[0], norm.bias[1], norm.bias[2], norm.bias[3] ?? 0];\n        }\n        const inputformat = options.format !== void 0 ? options.format : \"RGBA\";\n        const outputformat = options.tensorFormat !== void 0 ? options.tensorFormat !== void 0 ? options.tensorFormat : \"RGB\" : \"RGB\";\n        const stride = height * width;\n        const float32Data = outputformat === \"RGBA\" ? new Float32Array(stride * 4) : new Float32Array(stride * 3);\n        let step = 4, rImagePointer = 0, gImagePointer = 1, bImagePointer = 2, aImagePointer = 3;\n        let rTensorPointer = 0, gTensorPointer = stride, bTensorPointer = stride * 2, aTensorPointer = -1;\n        if (inputformat === \"RGB\") {\n          step = 3;\n          rImagePointer = 0;\n          gImagePointer = 1;\n          bImagePointer = 2;\n          aImagePointer = -1;\n        }\n        if (outputformat === \"RGBA\") {\n          aTensorPointer = stride * 3;\n        } else if (outputformat === \"RBG\") {\n          rTensorPointer = 0;\n          bTensorPointer = stride;\n          gTensorPointer = stride * 2;\n        } else if (outputformat === \"BGR\") {\n          bTensorPointer = 0;\n          gTensorPointer = stride;\n          rTensorPointer = stride * 2;\n        }\n        for (let i4 = 0; i4 < stride; i4++, rImagePointer += step, bImagePointer += step, gImagePointer += step, aImagePointer += step) {\n          float32Data[rTensorPointer++] = (buffer[rImagePointer] + normBias[0]) / normMean[0];\n          float32Data[gTensorPointer++] = (buffer[gImagePointer] + normBias[1]) / normMean[1];\n          float32Data[bTensorPointer++] = (buffer[bImagePointer] + normBias[2]) / normMean[2];\n          if (aTensorPointer !== -1 && aImagePointer !== -1) {\n            float32Data[aTensorPointer++] = (buffer[aImagePointer] + normBias[3]) / normMean[3];\n          }\n        }\n        const outputTensor = outputformat === \"RGBA\" ? new Tensor(\"float32\", float32Data, [1, 4, height, width]) : new Tensor(\"float32\", float32Data, [1, 3, height, width]);\n        return outputTensor;\n      };\n      tensorFromImage = async (image, options) => {\n        const isHTMLImageEle = typeof HTMLImageElement !== \"undefined\" && image instanceof HTMLImageElement;\n        const isImageDataEle = typeof ImageData !== \"undefined\" && image instanceof ImageData;\n        const isImageBitmap = typeof ImageBitmap !== \"undefined\" && image instanceof ImageBitmap;\n        const isString = typeof image === \"string\";\n        let data;\n        let bufferToTensorOptions = options ?? {};\n        const createCanvas = () => {\n          if (typeof document !== \"undefined\") {\n            return document.createElement(\"canvas\");\n          } else if (typeof OffscreenCanvas !== \"undefined\") {\n            return new OffscreenCanvas(1, 1);\n          } else {\n            throw new Error(\"Canvas is not supported\");\n          }\n        };\n        const createCanvasContext = (canvas) => {\n          if (typeof HTMLCanvasElement !== \"undefined\" && canvas instanceof HTMLCanvasElement) {\n            return canvas.getContext(\"2d\");\n          } else if (canvas instanceof OffscreenCanvas) {\n            return canvas.getContext(\"2d\");\n          } else {\n            return null;\n          }\n        };\n        if (isHTMLImageEle) {\n          const canvas = createCanvas();\n          canvas.width = image.width;\n          canvas.height = image.height;\n          const pixels2DContext = createCanvasContext(canvas);\n          if (pixels2DContext != null) {\n            let height = image.height;\n            let width = image.width;\n            if (options !== void 0 && options.resizedHeight !== void 0 && options.resizedWidth !== void 0) {\n              height = options.resizedHeight;\n              width = options.resizedWidth;\n            }\n            if (options !== void 0) {\n              bufferToTensorOptions = options;\n              if (options.tensorFormat !== void 0) {\n                throw new Error(\"Image input config format must be RGBA for HTMLImageElement\");\n              } else {\n                bufferToTensorOptions.tensorFormat = \"RGBA\";\n              }\n              bufferToTensorOptions.height = height;\n              bufferToTensorOptions.width = width;\n            } else {\n              bufferToTensorOptions.tensorFormat = \"RGBA\";\n              bufferToTensorOptions.height = height;\n              bufferToTensorOptions.width = width;\n            }\n            pixels2DContext.drawImage(image, 0, 0);\n            data = pixels2DContext.getImageData(0, 0, width, height).data;\n          } else {\n            throw new Error(\"Can not access image data\");\n          }\n        } else if (isImageDataEle) {\n          let height;\n          let width;\n          if (options !== void 0 && options.resizedWidth !== void 0 && options.resizedHeight !== void 0) {\n            height = options.resizedHeight;\n            width = options.resizedWidth;\n          } else {\n            height = image.height;\n            width = image.width;\n          }\n          if (options !== void 0) {\n            bufferToTensorOptions = options;\n          }\n          bufferToTensorOptions.format = \"RGBA\";\n          bufferToTensorOptions.height = height;\n          bufferToTensorOptions.width = width;\n          if (options !== void 0) {\n            const tempCanvas = createCanvas();\n            tempCanvas.width = width;\n            tempCanvas.height = height;\n            const pixels2DContext = createCanvasContext(tempCanvas);\n            if (pixels2DContext != null) {\n              pixels2DContext.putImageData(image, 0, 0);\n              data = pixels2DContext.getImageData(0, 0, width, height).data;\n            } else {\n              throw new Error(\"Can not access image data\");\n            }\n          } else {\n            data = image.data;\n          }\n        } else if (isImageBitmap) {\n          if (options === void 0) {\n            throw new Error(\"Please provide image config with format for Imagebitmap\");\n          }\n          const canvas = createCanvas();\n          canvas.width = image.width;\n          canvas.height = image.height;\n          const pixels2DContext = createCanvasContext(canvas);\n          if (pixels2DContext != null) {\n            const height = image.height;\n            const width = image.width;\n            pixels2DContext.drawImage(image, 0, 0, width, height);\n            data = pixels2DContext.getImageData(0, 0, width, height).data;\n            bufferToTensorOptions.height = height;\n            bufferToTensorOptions.width = width;\n            return bufferToTensor(data, bufferToTensorOptions);\n          } else {\n            throw new Error(\"Can not access image data\");\n          }\n        } else if (isString) {\n          return new Promise((resolve, reject) => {\n            const canvas = createCanvas();\n            const context = createCanvasContext(canvas);\n            if (!image || !context) {\n              return reject();\n            }\n            const newImage = new Image();\n            newImage.crossOrigin = \"Anonymous\";\n            newImage.src = image;\n            newImage.onload = () => {\n              canvas.width = newImage.width;\n              canvas.height = newImage.height;\n              context.drawImage(newImage, 0, 0, canvas.width, canvas.height);\n              const img = context.getImageData(0, 0, canvas.width, canvas.height);\n              bufferToTensorOptions.height = canvas.height;\n              bufferToTensorOptions.width = canvas.width;\n              resolve(bufferToTensor(img.data, bufferToTensorOptions));\n            };\n          });\n        } else {\n          throw new Error(\"Input data provided is not supported - aborted tensor creation\");\n        }\n        if (data !== void 0) {\n          return bufferToTensor(data, bufferToTensorOptions);\n        } else {\n          throw new Error(\"Input data provided is not supported - aborted tensor creation\");\n        }\n      };\n      tensorFromTexture = (texture, options) => {\n        const { width, height, download, dispose } = options;\n        const dims = [1, height, width, 4];\n        return new Tensor({ location: \"texture\", type: \"float32\", texture, dims, download, dispose });\n      };\n      tensorFromGpuBuffer = (gpuBuffer, options) => {\n        const { dataType, dims, download, dispose } = options;\n        return new Tensor({ location: \"gpu-buffer\", type: dataType ?? \"float32\", gpuBuffer, dims, download, dispose });\n      };\n      tensorFromMLTensor = (mlTensor, options) => {\n        const { dataType, dims, download, dispose } = options;\n        return new Tensor({ location: \"ml-tensor\", type: dataType ?? \"float32\", mlTensor, dims, download, dispose });\n      };\n      tensorFromPinnedBuffer = (type, buffer, dims) => new Tensor({ location: \"cpu-pinned\", type, data: buffer, dims: dims ?? [buffer.length] });\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/tensor-impl-type-mapping.js\n  var NUMERIC_TENSOR_TYPE_TO_TYPEDARRAY_MAP, NUMERIC_TENSOR_TYPEDARRAY_TO_TYPE_MAP, isTypedArrayChecked, checkTypedArray;\n  var init_tensor_impl_type_mapping = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/tensor-impl-type-mapping.js\"() {\n      NUMERIC_TENSOR_TYPE_TO_TYPEDARRAY_MAP = /* @__PURE__ */ new Map([\n        [\"float32\", Float32Array],\n        [\"uint8\", Uint8Array],\n        [\"int8\", Int8Array],\n        [\"uint16\", Uint16Array],\n        [\"int16\", Int16Array],\n        [\"int32\", Int32Array],\n        [\"bool\", Uint8Array],\n        [\"float64\", Float64Array],\n        [\"uint32\", Uint32Array],\n        [\"int4\", Uint8Array],\n        [\"uint4\", Uint8Array]\n      ]);\n      NUMERIC_TENSOR_TYPEDARRAY_TO_TYPE_MAP = /* @__PURE__ */ new Map([\n        [Float32Array, \"float32\"],\n        [Uint8Array, \"uint8\"],\n        [Int8Array, \"int8\"],\n        [Uint16Array, \"uint16\"],\n        [Int16Array, \"int16\"],\n        [Int32Array, \"int32\"],\n        [Float64Array, \"float64\"],\n        [Uint32Array, \"uint32\"]\n      ]);\n      isTypedArrayChecked = false;\n      checkTypedArray = () => {\n        if (!isTypedArrayChecked) {\n          isTypedArrayChecked = true;\n          const isBigInt64ArrayAvailable = typeof BigInt64Array !== \"undefined\" && BigInt64Array.from;\n          const isBigUint64ArrayAvailable = typeof BigUint64Array !== \"undefined\" && BigUint64Array.from;\n          const Float16Array2 = globalThis.Float16Array;\n          const isFloat16ArrayAvailable = typeof Float16Array2 !== \"undefined\" && Float16Array2.from;\n          if (isBigInt64ArrayAvailable) {\n            NUMERIC_TENSOR_TYPE_TO_TYPEDARRAY_MAP.set(\"int64\", BigInt64Array);\n            NUMERIC_TENSOR_TYPEDARRAY_TO_TYPE_MAP.set(BigInt64Array, \"int64\");\n          }\n          if (isBigUint64ArrayAvailable) {\n            NUMERIC_TENSOR_TYPE_TO_TYPEDARRAY_MAP.set(\"uint64\", BigUint64Array);\n            NUMERIC_TENSOR_TYPEDARRAY_TO_TYPE_MAP.set(BigUint64Array, \"uint64\");\n          }\n          if (isFloat16ArrayAvailable) {\n            NUMERIC_TENSOR_TYPE_TO_TYPEDARRAY_MAP.set(\"float16\", Float16Array2);\n            NUMERIC_TENSOR_TYPEDARRAY_TO_TYPE_MAP.set(Float16Array2, \"float16\");\n          } else {\n            NUMERIC_TENSOR_TYPE_TO_TYPEDARRAY_MAP.set(\"float16\", Uint16Array);\n          }\n        }\n      };\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/tensor-utils-impl.js\n  var calculateSize, tensorReshape;\n  var init_tensor_utils_impl = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/tensor-utils-impl.js\"() {\n      init_tensor_impl();\n      calculateSize = (dims) => {\n        let size = 1;\n        for (let i4 = 0; i4 < dims.length; i4++) {\n          const dim = dims[i4];\n          if (typeof dim !== \"number\" || !Number.isSafeInteger(dim)) {\n            throw new TypeError(`dims[${i4}] must be an integer, got: ${dim}`);\n          }\n          if (dim < 0) {\n            throw new RangeError(`dims[${i4}] must be a non-negative integer, got: ${dim}`);\n          }\n          size *= dim;\n        }\n        return size;\n      };\n      tensorReshape = (tensor, dims) => {\n        switch (tensor.location) {\n          case \"cpu\":\n            return new Tensor(tensor.type, tensor.data, dims);\n          case \"cpu-pinned\":\n            return new Tensor({\n              location: \"cpu-pinned\",\n              data: tensor.data,\n              type: tensor.type,\n              dims\n            });\n          case \"texture\":\n            return new Tensor({\n              location: \"texture\",\n              texture: tensor.texture,\n              type: tensor.type,\n              dims\n            });\n          case \"gpu-buffer\":\n            return new Tensor({\n              location: \"gpu-buffer\",\n              gpuBuffer: tensor.gpuBuffer,\n              type: tensor.type,\n              dims\n            });\n          case \"ml-tensor\":\n            return new Tensor({\n              location: \"ml-tensor\",\n              mlTensor: tensor.mlTensor,\n              type: tensor.type,\n              dims\n            });\n          default:\n            throw new Error(`tensorReshape: tensor location ${tensor.location} is not supported`);\n        }\n      };\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/tensor-impl.js\n  var Tensor;\n  var init_tensor_impl = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/tensor-impl.js\"() {\n      init_tensor_conversion_impl();\n      init_tensor_factory_impl();\n      init_tensor_impl_type_mapping();\n      init_tensor_utils_impl();\n      Tensor = class {\n        /**\n         * implementation.\n         */\n        constructor(arg0, arg1, arg2) {\n          checkTypedArray();\n          let type;\n          let dims;\n          if (typeof arg0 === \"object\" && \"location\" in arg0) {\n            this.dataLocation = arg0.location;\n            type = arg0.type;\n            dims = arg0.dims;\n            switch (arg0.location) {\n              case \"cpu-pinned\": {\n                const expectedTypedArrayConstructor = NUMERIC_TENSOR_TYPE_TO_TYPEDARRAY_MAP.get(type);\n                if (!expectedTypedArrayConstructor) {\n                  throw new TypeError(`unsupported type \"${type}\" to create tensor from pinned buffer`);\n                }\n                if (!(arg0.data instanceof expectedTypedArrayConstructor)) {\n                  throw new TypeError(`buffer should be of type ${expectedTypedArrayConstructor.name}`);\n                }\n                this.cpuData = arg0.data;\n                break;\n              }\n              case \"texture\": {\n                if (type !== \"float32\") {\n                  throw new TypeError(`unsupported type \"${type}\" to create tensor from texture`);\n                }\n                this.gpuTextureData = arg0.texture;\n                this.downloader = arg0.download;\n                this.disposer = arg0.dispose;\n                break;\n              }\n              case \"gpu-buffer\": {\n                if (type !== \"float32\" && type !== \"float16\" && type !== \"int32\" && type !== \"int64\" && type !== \"uint32\" && type !== \"uint8\" && type !== \"bool\" && type !== \"uint4\" && type !== \"int4\") {\n                  throw new TypeError(`unsupported type \"${type}\" to create tensor from gpu buffer`);\n                }\n                this.gpuBufferData = arg0.gpuBuffer;\n                this.downloader = arg0.download;\n                this.disposer = arg0.dispose;\n                break;\n              }\n              case \"ml-tensor\": {\n                if (type !== \"float32\" && type !== \"float16\" && type !== \"int32\" && type !== \"int64\" && type !== \"uint32\" && type !== \"uint64\" && type !== \"int8\" && type !== \"uint8\" && type !== \"bool\" && type !== \"uint4\" && type !== \"int4\") {\n                  throw new TypeError(`unsupported type \"${type}\" to create tensor from MLTensor`);\n                }\n                this.mlTensorData = arg0.mlTensor;\n                this.downloader = arg0.download;\n                this.disposer = arg0.dispose;\n                break;\n              }\n              default:\n                throw new Error(`Tensor constructor: unsupported location '${this.dataLocation}'`);\n            }\n          } else {\n            let data;\n            let maybeDims;\n            if (typeof arg0 === \"string\") {\n              type = arg0;\n              maybeDims = arg2;\n              if (arg0 === \"string\") {\n                if (!Array.isArray(arg1)) {\n                  throw new TypeError(\"A string tensor's data must be a string array.\");\n                }\n                data = arg1;\n              } else {\n                const typedArrayConstructor = NUMERIC_TENSOR_TYPE_TO_TYPEDARRAY_MAP.get(arg0);\n                if (typedArrayConstructor === void 0) {\n                  throw new TypeError(`Unsupported tensor type: ${arg0}.`);\n                }\n                if (Array.isArray(arg1)) {\n                  if (arg0 === \"float16\" && typedArrayConstructor === Uint16Array || arg0 === \"uint4\" || arg0 === \"int4\") {\n                    throw new TypeError(`Creating a ${arg0} tensor from number array is not supported. Please use ${typedArrayConstructor.name} as data.`);\n                  } else if (arg0 === \"uint64\" || arg0 === \"int64\") {\n                    data = typedArrayConstructor.from(arg1, BigInt);\n                  } else {\n                    data = typedArrayConstructor.from(arg1);\n                  }\n                } else if (arg1 instanceof typedArrayConstructor) {\n                  data = arg1;\n                } else if (arg1 instanceof Uint8ClampedArray) {\n                  if (arg0 === \"uint8\") {\n                    data = Uint8Array.from(arg1);\n                  } else {\n                    throw new TypeError(`A Uint8ClampedArray tensor's data must be type of uint8`);\n                  }\n                } else if (arg0 === \"float16\" && arg1 instanceof Uint16Array && typedArrayConstructor !== Uint16Array) {\n                  data = new globalThis.Float16Array(arg1.buffer, arg1.byteOffset, arg1.length);\n                } else {\n                  throw new TypeError(`A ${type} tensor's data must be type of ${typedArrayConstructor}`);\n                }\n              }\n            } else {\n              maybeDims = arg1;\n              if (Array.isArray(arg0)) {\n                if (arg0.length === 0) {\n                  throw new TypeError(\"Tensor type cannot be inferred from an empty array.\");\n                }\n                const firstElementType = typeof arg0[0];\n                if (firstElementType === \"string\") {\n                  type = \"string\";\n                  data = arg0;\n                } else if (firstElementType === \"boolean\") {\n                  type = \"bool\";\n                  data = Uint8Array.from(arg0);\n                } else {\n                  throw new TypeError(`Invalid element type of data array: ${firstElementType}.`);\n                }\n              } else if (arg0 instanceof Uint8ClampedArray) {\n                type = \"uint8\";\n                data = Uint8Array.from(arg0);\n              } else {\n                const mappedType = NUMERIC_TENSOR_TYPEDARRAY_TO_TYPE_MAP.get(arg0.constructor);\n                if (mappedType === void 0) {\n                  throw new TypeError(`Unsupported type for tensor data: ${arg0.constructor}.`);\n                }\n                type = mappedType;\n                data = arg0;\n              }\n            }\n            if (maybeDims === void 0) {\n              maybeDims = [data.length];\n            } else if (!Array.isArray(maybeDims)) {\n              throw new TypeError(\"A tensor's dims must be a number array\");\n            }\n            dims = maybeDims;\n            this.cpuData = data;\n            this.dataLocation = \"cpu\";\n          }\n          const size = calculateSize(dims);\n          if (this.cpuData && size !== this.cpuData.length) {\n            if ((type === \"uint4\" || type === \"int4\") && Math.ceil(size / 2) === this.cpuData.length) {\n            } else {\n              throw new Error(`Tensor's size(${size}) does not match data length(${this.cpuData.length}).`);\n            }\n          }\n          this.type = type;\n          this.dims = dims;\n          this.size = size;\n        }\n        // #endregion\n        // #region factory\n        static async fromImage(image, options) {\n          return tensorFromImage(image, options);\n        }\n        static fromTexture(texture, options) {\n          return tensorFromTexture(texture, options);\n        }\n        static fromGpuBuffer(gpuBuffer, options) {\n          return tensorFromGpuBuffer(gpuBuffer, options);\n        }\n        static fromMLTensor(mlTensor, options) {\n          return tensorFromMLTensor(mlTensor, options);\n        }\n        static fromPinnedBuffer(type, buffer, dims) {\n          return tensorFromPinnedBuffer(type, buffer, dims);\n        }\n        // #endregion\n        // #region conversions\n        toDataURL(options) {\n          return tensorToDataURL(this, options);\n        }\n        toImageData(options) {\n          return tensorToImageData(this, options);\n        }\n        // #endregion\n        // #region properties\n        get data() {\n          this.ensureValid();\n          if (!this.cpuData) {\n            throw new Error(\"The data is not on CPU. Use `getData()` to download GPU data to CPU, or use `texture` or `gpuBuffer` property to access the GPU data directly.\");\n          }\n          return this.cpuData;\n        }\n        get location() {\n          return this.dataLocation;\n        }\n        get texture() {\n          this.ensureValid();\n          if (!this.gpuTextureData) {\n            throw new Error(\"The data is not stored as a WebGL texture.\");\n          }\n          return this.gpuTextureData;\n        }\n        get gpuBuffer() {\n          this.ensureValid();\n          if (!this.gpuBufferData) {\n            throw new Error(\"The data is not stored as a WebGPU buffer.\");\n          }\n          return this.gpuBufferData;\n        }\n        get mlTensor() {\n          this.ensureValid();\n          if (!this.mlTensorData) {\n            throw new Error(\"The data is not stored as a WebNN MLTensor.\");\n          }\n          return this.mlTensorData;\n        }\n        // #endregion\n        // #region methods\n        async getData(releaseData) {\n          this.ensureValid();\n          switch (this.dataLocation) {\n            case \"cpu\":\n            case \"cpu-pinned\":\n              return this.data;\n            case \"texture\":\n            case \"gpu-buffer\":\n            case \"ml-tensor\": {\n              if (!this.downloader) {\n                throw new Error(\"The current tensor is not created with a specified data downloader.\");\n              }\n              if (this.isDownloading) {\n                throw new Error(\"The current tensor is being downloaded.\");\n              }\n              try {\n                this.isDownloading = true;\n                const data = await this.downloader();\n                this.downloader = void 0;\n                this.dataLocation = \"cpu\";\n                this.cpuData = data;\n                if (releaseData && this.disposer) {\n                  this.disposer();\n                  this.disposer = void 0;\n                }\n                return data;\n              } finally {\n                this.isDownloading = false;\n              }\n            }\n            default:\n              throw new Error(`cannot get data from location: ${this.dataLocation}`);\n          }\n        }\n        dispose() {\n          if (this.isDownloading) {\n            throw new Error(\"The current tensor is being downloaded.\");\n          }\n          if (this.disposer) {\n            this.disposer();\n            this.disposer = void 0;\n          }\n          this.cpuData = void 0;\n          this.gpuTextureData = void 0;\n          this.gpuBufferData = void 0;\n          this.mlTensorData = void 0;\n          this.downloader = void 0;\n          this.isDownloading = void 0;\n          this.dataLocation = \"none\";\n        }\n        // #endregion\n        // #region tensor utilities\n        ensureValid() {\n          if (this.dataLocation === \"none\") {\n            throw new Error(\"The tensor is disposed.\");\n          }\n        }\n        reshape(dims) {\n          this.ensureValid();\n          if (this.downloader || this.disposer) {\n            throw new Error(\"Cannot reshape a tensor that owns GPU resource.\");\n          }\n          return tensorReshape(this, dims);\n        }\n      };\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/tensor.js\n  var Tensor2;\n  var init_tensor = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/tensor.js\"() {\n      init_tensor_impl();\n      Tensor2 = Tensor;\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/trace.js\n  var TRACE, TRACE_FUNC, TRACE_FUNC_BEGIN, TRACE_FUNC_END;\n  var init_trace = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/trace.js\"() {\n      init_env_impl();\n      TRACE = (deviceType, label) => {\n        if (typeof env.trace === \"undefined\" ? !env.wasm.trace : !env.trace) {\n          return;\n        }\n        console.timeStamp(`${deviceType}::ORT::${label}`);\n      };\n      TRACE_FUNC = (msg, extraMsg) => {\n        const stack = new Error().stack?.split(/\\r\\n|\\r|\\n/g) || [];\n        let hasTraceFunc = false;\n        for (let i4 = 0; i4 < stack.length; i4++) {\n          if (hasTraceFunc && !stack[i4].includes(\"TRACE_FUNC\")) {\n            let label = `FUNC_${msg}::${stack[i4].trim().split(\" \")[1]}`;\n            if (extraMsg) {\n              label += `::${extraMsg}`;\n            }\n            TRACE(\"CPU\", label);\n            return;\n          }\n          if (stack[i4].includes(\"TRACE_FUNC\")) {\n            hasTraceFunc = true;\n          }\n        }\n      };\n      TRACE_FUNC_BEGIN = (extraMsg) => {\n        if (typeof env.trace === \"undefined\" ? !env.wasm.trace : !env.trace) {\n          return;\n        }\n        TRACE_FUNC(\"BEGIN\", extraMsg);\n      };\n      TRACE_FUNC_END = (extraMsg) => {\n        if (typeof env.trace === \"undefined\" ? !env.wasm.trace : !env.trace) {\n          return;\n        }\n        TRACE_FUNC(\"END\", extraMsg);\n      };\n    }\n  });\n\n  // node_modules/onnxruntime-common/dist/esm/inference-session-impl.js\n  var InferenceSession;\n  var init_inference_session_impl = __esm({\n    \"node_modules/onnxruntime-common/dist/esm/inference-session-impl.js\"() {\n      init_backend_impl();\n      init_tensor();\n      init_trace();\n      InferenceSession = class _InferenceSession {\n        constructor(handler) {\n          this.handler = handler;\n        }\n        async run(feeds, arg1, arg2) {\n          TRACE_FUNC_BEGIN();\n          const fetches = {};\n          let options = {};\n          if (typeof feeds !== \"object\" || feeds === null || feeds instanceof Tensor2 || Array.isArray(feeds)) {\n            throw new TypeError(\"'feeds' must be an object that use input names as keys and OnnxValue as corresponding values.\");\n          }\n          let isFetchesEmpty = true;\n          if (typeof arg1 === \"object\") {\n            if (arg1 === null) {\n              throw new TypeError(\"Unexpected argument[1]: cannot be null.\");\n            }\n            if (arg1 instanceof Tensor2) {\n              throw new TypeError(\"'fetches' cannot be a Tensor\");\n            }\n            if (Array.isArray(arg1)) {\n              if (arg1.length === 0) {\n                throw new TypeError(\"'fetches' cannot be an empty array.\");\n              }\n              isFetchesEmpty = false;\n              for (const name of arg1) {\n                if (typeof name !== \"string\") {\n                  throw new TypeError(\"'fetches' must be a string array or an object.\");\n                }\n                if (this.outputNames.indexOf(name) === -1) {\n                  throw new RangeError(`'fetches' contains invalid output name: ${name}.`);\n                }\n                fetches[name] = null;\n              }\n              if (typeof arg2 === \"object\" && arg2 !== null) {\n                options = arg2;\n              } else if (typeof arg2 !== \"undefined\") {\n                throw new TypeError(\"'options' must be an object.\");\n              }\n            } else {\n              let isFetches = false;\n              const arg1Keys = Object.getOwnPropertyNames(arg1);\n              for (const name of this.outputNames) {\n                if (arg1Keys.indexOf(name) !== -1) {\n                  const v4 = arg1[name];\n                  if (v4 === null || v4 instanceof Tensor2) {\n                    isFetches = true;\n                    isFetchesEmpty = false;\n                    fetches[name] = v4;\n                  }\n                }\n              }\n              if (isFetches) {\n                if (typeof arg2 === \"object\" && arg2 !== null) {\n                  options = arg2;\n                } else if (typeof arg2 !== \"undefined\") {\n                  throw new TypeError(\"'options' must be an object.\");\n                }\n              } else {\n                options = arg1;\n              }\n            }\n          } else if (typeof arg1 !== \"undefined\") {\n            throw new TypeError(\"Unexpected argument[1]: must be 'fetches' or 'options'.\");\n          }\n          for (const name of this.inputNames) {\n            if (typeof feeds[name] === \"undefined\") {\n              throw new Error(`input '${name}' is missing in 'feeds'.`);\n            }\n          }\n          if (isFetchesEmpty) {\n            for (const name of this.outputNames) {\n              fetches[name] = null;\n            }\n          }\n          const results = await this.handler.run(feeds, fetches, options);\n          const returnValue = {};\n          for (const key in results) {\n            if (Object.hasOwnProperty.call(results, key)) {\n              const result = results[key];\n              if (result instanceof Tensor2) {\n                returnValue[key] = result;\n              } else {\n                returnValue[key] = new Tensor2(result.type, result.data, result.dims);\n              }\n            }\n          }\n          TRACE_FUNC_END();\n          return returnValue;\n        }\n        async release() {\n          return this.handler.dispose();\n        }\n        static async create(arg0, arg1, arg2, arg3) {\n          TRACE_FUNC_BEGIN();\n          let filePathOrUint8Array;\n          let options = {};\n          if (typeof arg0 === \"string\") {\n            filePathOrUint8Array = arg0;\n            if (typeof arg1 === \"object\" && arg1 !== null) {\n              options = arg1;\n            } else if (typeof arg1 !== \"undefined\") {\n              throw new TypeError(\"'options' must be an object.\");\n            }\n          } else if (arg0 instanceof Uint8Array) {\n            filePathOrUint8Array = arg0;\n            if (typeof arg1 === \"object\" && arg1 !== null) {\n              options = arg1;\n            } else if (typeof arg1 !== \"undefined\") {\n              throw new TypeError(\"'options' must be an object.\");\n            }\n          } else if (arg0 instanceof ArrayBuffer || typeof SharedArrayBuffer !== \"undefined\" && arg0 instanceof SharedArrayBuffer) {\n            const buffer = arg0;\n            let byteOffset = 0;\n            let byteLength = arg0.byteLength;\n            if (typeof arg1 === \"object\" && arg1 !== null) {\n              options = arg1;\n            } else if (typeof arg1 === \"number\") {\n              byteOffset = arg1;\n              if (!Number.isSafeInteger(byteOffset)) {\n                throw new RangeError(\"'byteOffset' must be an integer.\");\n              }\n              if (byteOffset < 0 || byteOffset >= buffer.byteLength) {\n                throw new RangeError(`'byteOffset' is out of range [0, ${buffer.byteLength}).`);\n              }\n              byteLength = arg0.byteLength - byteOffset;\n              if (typeof arg2 === \"number\") {\n                byteLength = arg2;\n                if (!Number.isSafeInteger(byteLength)) {\n                  th