Webai-polyfill

README.md

@@ -21,8 +21,8 @@ https://user-images.githubusercontent.com/1011391/230610706-96755450-4a3b-4530-b
 ### ℹ️ Contents
 
 - [Window: use your own AI models on the web](#window-use-your-own-ai-models-on-the-web)
-    - [📺 Demo](#-demo)
-    - [ℹ️ Contents](#ℹ️-contents)
+  - [📺 Demo](#-demo)
+  - [ℹ️ Contents](#ℹ️-contents)
   - [⭐️ Main features](#️-main-features)
   - [⚙️ How it works](#️-how-it-works)
   - [📥 Installation](#-installation)
@@ -149,7 +149,7 @@ handler to improve UX, since not all models and config options support it.
 ### Examples
 
 - [Next.js Window AI](https://github.com/YanniKouloumbis/next-js-window-ai) - A Next.js app that demonstrates how to use Window AI in a chat application. ([Demo](https://next-js-window-ai.vercel.app/))
-- [Robot Companion](https://github.com/zoan37/robot-companion) - An  AI robot that can move, emote, and change facial expressions while chatting. ([Demo](https://robot-companion.vercel.app/))
+- [Robot Companion](https://github.com/zoan37/robot-companion) - An AI robot that can move, emote, and change facial expressions while chatting. ([Demo](https://robot-companion.vercel.app/))
 
 ### Functions
 
@@ -195,16 +195,15 @@ The `BETA_generate3DObject` function allows you to generate 3D objects with a de
 Here's an example request:
 
 ```javascript
-const [ result ] = await window.ai.BETA_generate3DObject(
-  { "prompt": "a glazed donut" }, 
-  { "numInferenceSteps": 32,});
+const [result] = await window.ai.BETA_generate3DObject(
+  { prompt: "a glazed donut" },
+  { numInferenceSteps: 32 }
+)
 
 // base64 representation of your 3D object, in ply format
-const uri = result.uri;
+const uri = result.uri
 ```
 
-
-
 All public types, including error messages, are documented in the [window.ai library](/packages/lib/src/index.ts). Highlights below:
 
 ### CompletionOptions
@@ -242,11 +241,12 @@ export interface CompletionOptions {
 }
 ```
 
-### ThreeDOptions 
+### ThreeDOptions
+
 This options dictionary allows you to specify options for generating a three dimensional object.
-  
+
 ```ts
-export interface ThreeDOptions{
+export interface ThreeDOptions {
   // The number of inference steps to run. Defaults to 32, with specific default values for each model.
   numInferenceSteps?: number
   // How many generations to create. Defaults to 1.

apps/extension/package.json

@@ -19,6 +19,7 @@
     "@types/chrome": "^0.0.237",
     "@types/webextension-polyfill": "^0.10.0",
     "@vespaiach/axios-fetch-adapter": "github:alexanderatallah/axios-fetch-adapter",
+    "@xenova/transformers": "^2.17.2",
     "axios": "^1.4.0",
     "axios-retry": "^3.5.0",
     "object-hash": "^3.0.0",

apps/extension/src/background/ports/media.ts

@@ -1,33 +1,19 @@
-import {
-  ErrorCode,
-  ModelID,
-  type MediaOutput,
-} from "window.ai"
+import { ErrorCode, type MediaOutput, ModelID } from "window.ai"
+
 import type { PlasmoMessaging } from "@plasmohq/messaging"
 
-import {
-  type PortRequest,
-  type PortResponse,
-} from "~core/constants"
+import { promptInterrupts } from "~background/lib/helpers"
+import { type PortRequest, type PortResponse } from "~core/constants"
 import { PortName } from "~core/constants"
-import { configManager, AuthType } from "~core/managers/config"
-import {
-  transactionManager
-} from "~core/managers/transaction"
-import {
-  err,
-  isErr,
-  isOk,
-  ok,
-  unknownErr
-} from "~core/utils/result-monad"
+import { AuthType, configManager } from "~core/managers/config"
+import { originManager } from "~core/managers/origin"
+import { transactionManager } from "~core/managers/transaction"
+import { getMediaCaller } from "~core/media"
+import { NO_TXN_REFERRER } from "~core/model-router"
+import { err, isErr, isOk, ok, unknownErr } from "~core/utils/result-monad"
 import { log } from "~core/utils/utils"
 
 import { requestPermission } from "./permission"
-import { getMediaCaller } from "~core/media"
-import { originManager } from "~core/managers/origin"
-import { NO_TXN_REFERRER } from "~core/model-router"
-import { promptInterrupts } from "~background/lib/helpers"
 
 const handler: PlasmoMessaging.PortHandler<
   PortRequest[PortName.Media],
@@ -48,14 +34,17 @@ const handler: PlasmoMessaging.PortHandler<
 
   const txn = request.transaction
 
-  if ('messages' in txn.input) {
+  if ("messages" in txn.input) {
     return res.send(err(ErrorCode.InvalidRequest))
   }
 
   // temporarily, use external model config
-  const config = await configManager.forAuthAndModel(AuthType.External, ModelID.Shap_e)
+  const config = await configManager.forAuthAndModel(
+    AuthType.External,
+    ModelID.Shap_e
+  )
   // if not credentialed, present with login flow
-  if(!configManager.isCredentialed(config)){
+  if (!configManager.isCredentialed(config)) {
     promptInterrupts(id, err(ErrorCode.NotAuthenticated))
     return res.send({ response: err(ErrorCode.NotAuthenticated), id })
   }
@@ -73,19 +62,19 @@ const handler: PlasmoMessaging.PortHandler<
       model: txn.routedModel,
       origin: txn ? originManager.url(txn.origin) : NO_TXN_REFERRER,
       num_generations: txn.numOutputs,
-      num_inference_steps: txn.numInferenceSteps,
+      num_inference_steps: txn.numInferenceSteps
     })
   } catch (error) {
     result = unknownErr(error)
   }
-  
+
   if (isOk(result)) {
     const outputs = result.data
     res.send({ response: ok(outputs), id })
     // do not store URIs in the transaction, empty string to save storage
     txn.outputs = outputs.map((output: MediaOutput) => ({
       ...output,
-      uri: "",
+      uri: ""
     }))
   } else {
     res.send({ response: result, id })