init:auto completion and add documentation code action

.github/workflows/publish.yaml

@@ -0,0 +1,61 @@
+name: Release
+
+# trigger on git push for release branch
+on:
+  workflow_dispatch: # allow manual trigger
+  push: 
+    branches:
+      - release    
+
+env:
+  GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+jobs:
+  run:
+    name: Build and publish vsix package
+    strategy:
+      matrix:
+        java-version: [17]
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout current branch
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0    
+      # cache sbt dependencies          
+      - uses: coursier/cache-action@v6 
+
+      # install java, sbt and node
+      - name: Setup Java and sbt
+        uses: actions/setup-java@v4
+        with:
+          distribution: temurin
+          java-version: ${{ matrix.java-version }}
+          cache: sbt
+      - uses: sbt/setup-sbt@v1
+
+      - name: setup node
+        uses: actions/setup-node@v4
+
+      # install dependencies and build package
+      - name: compile Scala and build js to out/ folder
+        run: 
+          sbt buildDebug
+
+      - name: package extension inside out/ folder
+        run: |
+          cd out/ 
+          ls -la 
+          npx vsce package
+
+      - name: Get upload url
+        id: get_upload_url
+        run: |
+          URL=$(curl --silent "https://api.github.com/repos/doofin/functorcoder/releases/latest" | jq -r '.upload_url')
+          echo ::set-output name=UPLOAD_URL::$URL
+
+      - name: Upload VSIX package to github release
+        uses: softprops/action-gh-release@v2
+        # if: startsWith(github.ref, 'refs/tags/')
+        with:
+          files: out/functorcoder-0.0.1.vsix

.github/workflows/scala.yml

@@ -1,4 +1,4 @@
-name: Continuous Integration
+name: CI
 
 on:
   push:
@@ -8,7 +8,7 @@ env:
 
 jobs:
   run:
-    name: Build and Run
+    name: Compile
     strategy:
       matrix:
         java-version: [17]
@@ -26,6 +26,9 @@ jobs:
           java-version: ${{ matrix.java-version }}
           cache: sbt
 
+          # cache sbt dependencies          
+      - uses: coursier/cache-action@v6 
+
       - uses: sbt/setup-sbt@v1
 
       - name: compile

.gitignore

@@ -38,3 +38,5 @@ out
 /project/target/
 /.bsp/
 
+*.worksheet.sc
+.sbtopts

README.md

@@ -1,109 +1,112 @@
-# VSCode Extension in Scala.js
-Write vscode extensions in Scala.js! This is a collection of examples and templates to get you started, with convenient sbt tasks to build and run your extension.
+# functorcoder
+**functorcoder** is an open source AI coding assistant utilizing LLM (Large Language Model) with algebraic and modular design in Scala.js. It aims at providing a clean and extensible architecture for AI coding assistants, which is helpful for understanding basic mechanics if you want to build your own AI coding assistant.
 
-contains:
-- commands from the vscode command palette
-- inline completion like github copilot
-- language server protocol client
-- code actions (when pressing Alt+Enter at a code location)
+current features implemented:
+- auto completion as you type
+- add documentation quick fix action
 
-### Setup
-Requirements:
- - [Sbt](https://www.scala-sbt.org/download.html)
+next important features to be implemented:
+- generate multiple files and folders
+- disable/enable auto completion
 
+features aiming to implement in long term:
+- code generation: completion, documentation
+- code modification: refactoring, optimization, bug fixing
+- code analysis: code understanding, code review, code quality
 
-Run the vscode extension:
-* Clone this project
-* Open the project in VSCode, run the `import build` task with Metals (it should display a popup automatically).
+## 中文简介
+作为一个copilot的用户，最近受到国产开源模型deepseek的鼓励，希望能在开源社区中贡献一些自己的力量。目前已经有一些ai插件，比如copilot, tabnine，cursor等，还有一些开源的插件，比如continue。我看了下continue的代码，发现它的设计很复杂，代码并不简洁。目前，copilot的体验还可以，但是非常封闭，无法自定义很多地方，比如代码补全的长度，模型的选择等。开源的插件则有很多稳定性问题，bug不少。
 
-* run below command, which will open a new VSCode window with the extension loaded(first time it will take some time for scalable typed to convert typescript to scala.js):
-```bash
-sbt open
-```
+所以，作为一个scala的爱好者，也希望加深对llm应用的理解，我决定自己用scala.js来实现一个简单的ai助手。
 
-After the new VSCode (extension development host) window opens:
-* Run the Hello World command from the Command Palette (`⇧⌘P`) in the new VSCode window.
-* Type `hello` and select `Hello World`.
-  * You should see a Notification _Hello World!_.
+## Getting Started
+Visit [vscode-scalajs-hello](https://github.com/doofin/vscode-scalajs-hello) to understand how to play with scala.js for VSCode extension development. Basically, sbt is used to build the project and run the extension. There you will learn:
+- setting up the development environment
+- building the project and running the extension
+- packaging the extension
 
 
-### Use it as a template
-click on the `Use this template` button to create a new repository with the same structure in github.
+Before loading the extension, you need to add options to vscode user settings, and provide your OpenAI compatible API key and URL. Here is an example:
 
-### Use it as sbt dependency
-In your `build.sbt` add the following:
-```scala
-lazy val vsc = RootProject(uri("https://github.com/doofin/vscode-scalajs-hello.git")) 
-lazy val root = Project("root", file(".")) dependsOn(vsc)
+```json
+"functorcoder": {
+        "apiKey": "somekey",
+        "apiUrl": "https://api.openai.com/v1/chat/completions",
+        "maxTokens": 512,
+        "model": "gpt-4o-mini",
+    }
 ```
 
-### Use it as a library
-**Currently not working** due to jitpack missing npm! Welcome to contribute to fix it.
+## Project Structure
+The project is divided into two main parts: the core module and the VSCode extension module under /src/main/scala/functorcoder and /src/main/scala/vscextension respectively.
 
-You can use this project as a library in your project by adding the following to your `build.sbt`:
-```scala
-resolvers += Resolver.bintrayRepo("jitpack", "https://jitpack.io")
-libraryDependencies += "com.github.doofin" % "vscode-scalajs-hello" % "master-SNAPSHOT" // might be wrong
-```
+**To get started**, read the file `extensionMain.scala` in the VSCode extension module. It is the main entry point for the extension. 
 
-You can find the latest version on
-[jitpack.](https://jitpack.io/#doofin/vscode-scalajs-hello)
+The first part is the core module, we aim keeping it concise. It contains the main logic of the ai coding assistant:
+- Large Language Model (LLM) integration
+- sending propmt to LLM and getting the response
 
-Note: 
- - I recommend using the Metals extension for Scala in VSCode.
- - If you have any issues, please open an issue on this repository.
+The second part is the VSCode extension module, which integrates the core module with the VSCode editor. It contains:
+- commands: commands to be executed in the editor
+- code actions: quick fix actions
+- code completion: auto completion
+- editor ui: status bar, notifications, etc.
+
+It's adopted from the [vscode-scalajs-hello](https://github.com/doofin/vscode-scalajs-hello) project. Refer to it for getting started with the VSCode extension development in Scala.js.
 
-## Project structure
-The project file structure in src/main/scala is as follows:
+
+project file structure for the core module:
 ```bash
-src/main/scala
-├── extensionMain.scala // main entry point for the extension
-├── commands.scala, codeActions.scala,etc // files for different extension features
-│   ├── facade // facade for vscode api
-│   ├── io // file and network io functions
+/functorcoder
+├── /src/main/scala/functorcoder
+│   ├── /llm                 # Integration with LLM (e.g., OpenAI API)
+│   ├── /actions
+│   │   ├── CodeGen.scala    # Code completion, generation, and documentation
+│   │   ├── Commands.scala   # Commands from functorcoder
+│   │   └── Debug.scala      # Debugging module
+│   ├── /types               # Types for code, context, and user actions
+│   ├── /editorUI            # Integration with the editor (e.g., VSCode)
+│   └── /tests               # Unit tests for core modules
+└── /docs                    # Documentation
 ```
 
+The project file structure for the VSCode extension module:
+```bash
+/vscextension
+├── /src/main/scala/vscextension
+│   ├── extensionMain.scala       # Main entry point for the extension
+│   ├── commands.scala            # Command definitions
+│   ├── codeActions.scala         # Code action definitions
+...
+```
 
-The project uses the following tools:
-* [SBT] build tool for building the project
-* [Scala.js] for general coding
-* [Scalably Typed] for JavaScript facades
-* [scalajs-bundler] for bundling the JavaScript dependencies
-
-SBT is configured with the `build.sbt` file. Scala.js, ScalablyTyped and the bundler are SBT plugins. With these, SBT manages your JavaScript `npm` dependencies. You should never have to run `npm` directly, simply edit the `npmDependencies` settings in `build.sbt`.
-
-[accessible-scala]: https://marketplace.visualstudio.com/items?itemName=scala-center.accessible-scala
-[helloworld-minimal-sample]: https://github.com/Microsoft/vscode-extension-samples/tree/master/helloworld-minimal-sample
-[Scalably Typed]: https://github.com/ScalablyTyped/Converter
-[SBT]: https://www.scala-sbt.org
-[ScalaJS]: http://www.scala-js.org
-[scalajs-bundler]: https://github.com/scalacenter/scalajs-bundler
 
-## How to code in Scala js?
+## design principles
+I am to design the system with mathematics, algebra and functional programming principles in mind. The system is designed to be modular and extensible, allowing for easy addition of new features and components.
 
-In general, javascript functions and classes can be used in the same way as in JS/TS!
-If the typechecker disagrees, you can insert casts with `.asInstanceOf[Type]`.
+    Input = {Query, CodeSnippet, Spec}: The set of all possible input types (queries, code snippets, or requirements/specifications).
 
-The JS types (like `js.Array`) are available from
-```scala
-import scala.scalajs.js
-```
+    Output = {Code, Explanation, Transformation, DebugSuggestion}: The set of all possible outputs.
 
-The VSCode classes and functions are available from
-```scala
-import typings.vscode.mod as vscode
+The types and objects for Input:
+- code snippet or code file: a piece of code
+- code context: a code snippet with its surrounding code
+- query: natural language query 
+- specification: natural language specification
 
-vscode.window.showInformationMessage("Hello World!")
-```
+The Output:
+- code snippet or code file: a piece of code, including completion, refactoring, optimization, bug fixing
+- explanation: a natural language explanation
+- transformation: the transformation of the input code
+- suggestion: a suggestion for debugging or improvement or refactoring
 
-Some additional types are available in the `anon` subpackage, for example:
-```scala
-import typings.vscode.anon.Dispose
-// register a command. The cast is necessary due to typescript conversion limitations.
-vscode.commands.registerCommand(name, fun).asInstanceOf[Dispose]
-```
+# feedback
+features to be implemented:
+- refactoring
+- specify which LLM to use
+- RAG(retrieval-augmented generation) to understand the whole code base
+- MCP(model context protocol) to interact with the environment, like external tools, etc.
 
-You can find more information and tutorials on the [Scala.js website](https://www.scala-js.org/).
 
 # references:
  - updated from [vscode-scalajs-hello](https://github.com/pme123/vscode-scalajs-hello) with Scala 3.3.3 and sbt.version=1.9.7.