Add node support for local sidecar server and request types (#165)

* Add support for local sidecar server * Remove un-needed types * Update README.MD * Update README.MD API Constructor Docs * Update README.MD examples * Fix default port in tests * Update README types * Update README typo
vikhyat · Dec 5, 2024 · fc49f31 · fc49f31
1 parent fc8a50e
commit fc49f31
Show file tree

Hide file tree

Showing 5 changed files with 430 additions and 67 deletions.
diff --git a/clients/node/README.MD b/clients/node/README.MD
@@ -21,8 +21,9 @@ yarn add moondream
 ## Quick Start
 
 Before using this client library, you'll need an API key to access Moondream's hosted service.
-You can get a free API key from [console.moondream.ai](https://console.moondream.ai). Currently
-local inference is only available in Python, but Node.js support is coming very soon.
+You can get a free API key from [console.moondream.ai](https://console.moondream.ai).
+
+### Cloud
 
 ```javascript
 import { vl } from "moondream";
@@ -39,15 +40,72 @@ const image = fs.readFileSync("path/to/image.jpg");
 // Basic usage examples
 async function main() {
   // Generate a caption for the image
-  const caption = await model.caption(image);
+  const caption = await model.caption({
+    image: image,
+    length: "normal",
+    stream: false
+  });
   console.log("Caption:", caption);
 
   // Ask a question about the image
-  const answer = await model.query(image, "What's in this image?");
+  const answer = await model.query({
+    image: image,
+    question: "What's in this image?",
+    stream: false
+  });
   console.log("Answer:", answer);
 
   // Stream the response
-  const stream = await model.caption(image, "normal", true);
+  const stream = await model.caption({
+    image: image,
+    length: "normal",
+    stream: true
+  });
+  for await (const chunk of stream.caption) {
+    process.stdout.write(chunk);
+  }
+}
+
+main();
+```
+
+### Local Inference
+
+- Install the `moondream` CLI: `pip install moondream`
+- Run the local server: `moondream serve --model <path-to-model>`
+- Set the `apiUrl` parameter to the URL of the local server (the default is `http://localhost:3475`)
+
+```javascript
+const model = new vl({
+  apiUrl: "http://localhost:3475",
+});
+
+const image = fs.readFileSync("path/to/image.jpg");
+
+// Basic usage examples
+async function main() {
+  // Generate a caption for the image
+  const caption = await model.caption({
+    image: image,
+    length: "normal",
+    stream: false
+  });
+  console.log("Caption:", caption);
+
+  // Ask a question about the image
+  const answer = await model.query({
+    image: image,
+    question: "What's in this image?",
+    stream: false
+  });
+  console.log("Answer:", answer);
+
+  // Stream the response
+  const stream = await model.caption({
+    image: image,
+    length: "normal",
+    stream: true
+  });
   for await (const chunk of stream.caption) {
     process.stdout.write(chunk);
   }
@@ -68,47 +126,77 @@ main();
 ### Constructor
 
 ```javascript
+// for cloud inference
 const model = new vl({
   apiKey: "your-api-key",
 });
+
+// or for local inference
+const model = new vl({
+  apiUrl: "http://localhost:3475",
+});
 ```
 
 ### Methods
 
-#### caption(image, length?, stream?, settings?)
+#### caption({ image: string, length: string, stream?: boolean })
 
 Generate a caption for an image.
 
 ```javascript
-const result = await model.caption(image, "normal", false);
+const result = await model.caption({
+  image: image,
+  length: "normal",
+  stream: false
+});
+
 // or with streaming
-const stream = await model.caption(image, "normal", true);
+const stream = await model.caption({
+  image: image,
+  length: "normal",
+  stream: true
+});
 ```
 
-#### query(image, question, stream?, settings?)
+#### query({ image: string, question: string, stream?: boolean })
 
 Ask a question about an image.
 
 ```javascript
-const result = await model.query(image, "What's in this image?", false);
+const result = await model.query({
+  image: image,
+  question: "What's in this image?",
+  stream: false
+});
+
 // or with streaming
-const stream = await model.query(image, "What's in this image?", true);
+const stream = await model.query({
+  image: image,
+  question: "What's in this image?",
+  stream: true
+});
 ```
 
-#### detect(image, object)
+#### detect({ image: string, object: string })
 
 Detect specific objects in an image.
 
 ```javascript
-const result = await model.detect(image, "car");
+const result = await model.detect({
+  image: image,
+  object: "car"
+});
 ```
 
-#### point(image, object)
+#### point({ image: string, object: string })
 
 Get coordinates of specific objects in an image.
 
 ```javascript
-const result = await model.point(image, "person");
+const result = await model.point({
+  image: image,
+  object: "person"
+});
 ```
 
 ### Input Types