Merge pull request #35 from SilasMarvin/feat/add_release_ci

feat: add release CI

.cargo/config.toml

@@ -0,0 +1,2 @@
+[alias]
+xtask = "run --package xtask --bin xtask --"

.github/actions/github-release/Dockerfile

@@ -0,0 +1,8 @@
+FROM node:slim
+
+COPY . /action
+WORKDIR /action
+
+RUN npm install --production
+
+ENTRYPOINT ["node", "/action/main.js"]

.github/actions/github-release/README.md

@@ -0,0 +1,21 @@
+# github-release
+
+Copy-pasted from
+https://github.com/rust-lang/rust-analyzer/tree/2df30e1e07eafc1de0359566423f471920693a34/.github/actions/github-release
+
+An action used to publish GitHub releases for `wasmtime`.
+
+As of the time of this writing there's a few actions floating around which
+perform github releases but they all tend to have their set of drawbacks.
+Additionally nothing handles deleting releases which we need for our rolling
+`dev` release.
+
+To handle all this, this action rolls its own implementation using the
+actions/toolkit repository and packages published there. These run in a Docker
+container and take various inputs to orchestrate the release from the build.
+
+More comments can be found in `main.js`.
+
+Testing this is really hard. If you want to try though run `npm install` and
+then `node main.js`. You'll have to configure a bunch of env vars though to get
+anything reasonably working.

.github/actions/github-release/action.yml

@@ -0,0 +1,15 @@
+name: 'wasmtime github releases'
+description: 'wasmtime github releases'
+inputs:
+  token:
+    description: ''
+    required: true
+  name:
+    description: ''
+    required: true
+  files:
+    description: ''
+    required: true
+runs:
+  using: 'docker'
+  image: 'Dockerfile'

.github/actions/github-release/main.js

@@ -0,0 +1,144 @@
+const core = require('@actions/core');
+const path = require("path");
+const fs = require("fs");
+const github = require('@actions/github');
+const glob = require('glob');
+
+function sleep(milliseconds) {
+  return new Promise(resolve => setTimeout(resolve, milliseconds));
+}
+
+async function runOnce() {
+  // Load all our inputs and env vars. Note that `getInput` reads from `INPUT_*`
+  const files = core.getInput('files');
+  const name = core.getInput('name');
+  const token = core.getInput('token');
+  const slug = process.env.GITHUB_REPOSITORY;
+  const owner = slug.split('/')[0];
+  const repo = slug.split('/')[1];
+  const sha = process.env.HEAD_SHA;
+
+  core.info(`files: ${files}`);
+  core.info(`name: ${name}`);
+
+  const options = {
+    request: {
+      timeout: 30000,
+    }
+  };
+  const octokit = github.getOctokit(token, options);
+
+  // Delete the previous release since we can't overwrite one. This may happen
+  // due to retrying an upload or it may happen because we're doing the dev
+  // release.
+  const releases = await octokit.paginate("GET /repos/:owner/:repo/releases", { owner, repo });
+  for (const release of releases) {
+    if (release.tag_name !== name) {
+      continue;
+    }
+    const release_id = release.id;
+    core.info(`deleting release ${release_id}`);
+    await octokit.rest.repos.deleteRelease({ owner, repo, release_id });
+  }
+
+  // We also need to update the `dev` tag while we're at it on the `dev` branch.
+  if (name == 'nightly') {
+    try {
+      core.info(`updating nightly tag`);
+      await octokit.rest.git.updateRef({
+        owner,
+        repo,
+        ref: 'tags/nightly',
+        sha,
+        force: true,
+      });
+    } catch (e) {
+      core.error(e);
+      core.info(`creating nightly tag`);
+      await octokit.rest.git.createTag({
+        owner,
+        repo,
+        tag: 'nightly',
+        message: 'nightly release',
+        object: sha,
+        type: 'commit',
+      });
+    }
+  }
+
+  // Creates an official GitHub release for this `tag`, and if this is `dev`
+  // then we know that from the previous block this should be a fresh release.
+  core.info(`creating a release`);
+  const release = await octokit.rest.repos.createRelease({
+    owner,
+    repo,
+    name,
+    tag_name: name,
+    target_commitish: sha,
+    prerelease: name === 'nightly',
+  });
+  const release_id = release.data.id;
+
+  // Upload all the relevant assets for this release as just general blobs.
+  for (const file of glob.sync(files)) {
+    const size = fs.statSync(file).size;
+    const name = path.basename(file);
+
+    await runWithRetry(async function() {
+      // We can't overwrite assets, so remove existing ones from a previous try.
+      let assets = await octokit.rest.repos.listReleaseAssets({
+        owner,
+        repo,
+        release_id
+      });
+      for (const asset of assets.data) {
+        if (asset.name === name) {
+          core.info(`delete asset ${name}`);
+          const asset_id = asset.id;
+          await octokit.rest.repos.deleteReleaseAsset({ owner, repo, asset_id });
+        }
+      }
+
+      core.info(`upload ${file}`);
+      const headers = { 'content-length': size, 'content-type': 'application/octet-stream' };
+      const data = fs.createReadStream(file);
+      await octokit.rest.repos.uploadReleaseAsset({
+        data,
+        headers,
+        name,
+        url: release.data.upload_url,
+      });
+    });
+  }
+}
+
+async function runWithRetry(f) {
+  const retries = 10;
+  const maxDelay = 4000;
+  let delay = 1000;
+
+  for (let i = 0; i < retries; i++) {
+    try {
+      await f();
+      break;
+    } catch (e) {
+      if (i === retries - 1)
+        throw e;
+
+      core.error(e);
+      const currentDelay = Math.round(Math.random() * delay);
+      core.info(`sleeping ${currentDelay} ms`);
+      await sleep(currentDelay);
+      delay = Math.min(delay * 2, maxDelay);
+    }
+  }
+}
+
+async function run() {
+  await runWithRetry(runOnce);
+}
+
+run().catch(err => {
+  core.error(err);
+  core.setFailed(err.message);
+});

.github/actions/github-release/package.json

@@ -0,0 +1,10 @@
+{
+  "name": "wasmtime-github-release",
+  "version": "0.0.0",
+  "main": "main.js",
+  "dependencies": {
+    "@actions/core": "^1.6",
+    "@actions/github": "^5.0",
+    "glob": "^7.1.5"
+  }
+}

.github/workflows/README.md

@@ -0,0 +1,8 @@
+# CI pipelines
+
+## Release
+
+To create a new release for `lsp-ai`, all you'll need to do is create a new branch with the following format: `release/{release_name}`. `release_name` is usually the version of the release package in SemVer format `x.x.x{-rcx}`.
+
+This has the advantage of being able to fix issues for a specific release while continuing developping on the `main` branch by cherry-picking patches. It's inspired by trunk-based development.
+

.github/workflows/release.yml

@@ -0,0 +1,188 @@
+name: release
+on:
+  workflow_dispatch:
+
+  push:
+    branches:
+      - 'release/**'
+
+env:
+  CARGO_INCREMENTAL: 0
+  CARGO_NET_RETRY: 10
+  RUSTFLAGS: "-D warnings -W unreachable-pub"
+  RUSTUP_MAX_RETRIES: 10
+  FETCH_DEPTH: 0 # pull in the tags for the version string
+  CARGO_TARGET_AARCH64_UNKNOWN_LINUX_GNU_LINKER: aarch64-linux-gnu-gcc
+  CARGO_TARGET_ARM_UNKNOWN_LINUX_GNUEABIHF_LINKER: arm-linux-gnueabihf-gcc
+
+jobs:
+  dist:
+    strategy:
+      matrix:
+        include:
+          - os: windows-latest
+            target: x86_64-pc-windows-msvc
+            code-target: win32-x64
+          - os: windows-latest
+            target: i686-pc-windows-msvc
+            code-target: win32-ia32
+          - os: windows-latest
+            target: aarch64-pc-windows-msvc
+            code-target: win32-arm64
+          - os: ubuntu-22.04
+            target: x86_64-unknown-linux-gnu
+            code-target: linux-x64
+          - os: ubuntu-22.04
+            target: aarch64-unknown-linux-gnu
+            code-target: linux-arm64
+          - os: ubuntu-22.04
+            target: arm-unknown-linux-gnueabihf
+            code-target: linux-armhf
+          - os: macos-12
+            target: x86_64-apple-darwin
+            code-target: darwin-x64
+          - os: macos-12
+            target: aarch64-apple-darwin
+            code-target: darwin-arm64
+
+    env:
+      LSP_AI_TARGET: ${{ matrix.target }}
+
+    name: dist (${{ matrix.target }})
+    runs-on: ${{ matrix.os }}
+    container: ${{ matrix.container }}
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: ${{ env.FETCH_DEPTH }}
+
+      - name: Install Rust toolchain
+        run: |
+          rustup update --no-self-update stable
+          rustup target add ${{ matrix.target }}
+          rustup component add rust-src
+
+      - name: Update apt repositories
+        if: contains(matrix.os, 'ubuntu')
+        run: sudo apt-get update -y
+
+      - name: Install AArch64 target toolchain
+        if: matrix.target == 'aarch64-unknown-linux-gnu'
+        run: sudo apt-get install gcc-aarch64-linux-gnu libc6-dev-arm64-cross  g++-aarch64-linux-gnu
+
+      - name: Install ARM target toolchain
+        if: matrix.target == 'arm-unknown-linux-gnueabihf'
+        run: sudo apt-get install gcc-arm-linux-gnueabihf g++-arm-linux-gnueabihf
+
+      - name: Dist
+        run: cargo xtask dist
+
+      - name: Upload artifacts
+        uses: actions/upload-artifact@v4
+        with:
+          name: dist-${{ matrix.target }}
+          path: ./dist
+
+  dist-x86_64-unknown-linux-musl:
+    name: dist (x86_64-unknown-linux-musl)
+    runs-on: ubuntu-latest
+    env:
+      LSP_AI_TARGET: x86_64-unknown-linux-musl
+      # For some reason `-crt-static` is not working for clang without lld
+      RUSTFLAGS: "-C link-arg=-fuse-ld=lld -C target-feature=-crt-static"
+    container:
+      image: rust:alpine
+      volumes:
+        - /usr/local/cargo/registry:/usr/local/cargo/registry
+
+    steps:
+      - name: Install dependencies
+        run: apk add --no-cache git clang lld musl-dev nodejs npm openssl-dev pkgconfig g++ openssl-dev perl make linux-headers
+
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: ${{ env.FETCH_DEPTH }}
+
+      - name: Install Rust toolchain
+        run: |
+          rustup update --no-self-update stable
+          rustup target add x86_64-unknown-linux-gnu
+          rustup component add rust-src
+
+      - name: Dist
+        run: cargo xtask dist
+
+      - name: Upload artifacts
+        uses: actions/upload-artifact@v4
+        with:
+          name: dist-x86_64-unknown-linux-musl
+          path: ./dist
+
+  publish:
+    name: publish
+    runs-on: ubuntu-latest
+    needs: ["dist", "dist-x86_64-unknown-linux-musl"]
+    permissions:
+      contents: write
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: ${{ env.FETCH_DEPTH }}
+
+      - run: echo "HEAD_SHA=$(git rev-parse HEAD)" >> $GITHUB_ENV
+      - run: 'echo "HEAD_SHA: $HEAD_SHA"'
+
+      - name: Split branch name
+        env:
+          BRANCH: ${{ github.ref_name }}
+        id: split
+        run: echo "tag=${BRANCH##*/}" >> $GITHUB_OUTPUT
+
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist-aarch64-apple-darwin
+          path: dist
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist-x86_64-apple-darwin
+          path: dist
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist-x86_64-unknown-linux-gnu
+          path: dist
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist-x86_64-unknown-linux-musl
+          path: dist
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist-aarch64-unknown-linux-gnu
+          path: dist
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist-arm-unknown-linux-gnueabihf
+          path: dist
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist-x86_64-pc-windows-msvc
+          path: dist
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist-i686-pc-windows-msvc
+          path: dist
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist-aarch64-pc-windows-msvc
+          path: dist
+      - run: ls -al ./dist
+
+      - name: Publish Release
+        uses: ./.github/actions/github-release
+        with:
+          files: "dist/*"
+          name: ${{ steps.split.outputs.tag }}
+          token: ${{ secrets.GITHUB_TOKEN }}

Cargo.lock

@@ -266,9 +266,9 @@ dependencies = [
 
 [[package]]
 name = "bumpalo"
-version = "3.15.4"
+version = "3.16.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7ff69b9dd49fd426c69a0db9fc04dd934cdb6645ff000864d98f7e2af8830eaa"
+checksum = "79296716171880943b8470b5f8d03aa55eb2e645a4874bdbb28adb49162e012c"
 
 [[package]]
 name = "byteorder"
@@ -1658,9 +1658,9 @@ checksum = "490cc448043f947bae3cbee9c203358d62dbee0db12107a74be5c30ccfd09771"
 
 [[package]]
 name = "memchr"
-version = "2.7.1"
+version = "2.7.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "523dc4f511e55ab87b694dc30d0f820d60906ef06413f93d4d7a1385599cc149"
+checksum = "78ca9ab1a0babb1e7d5695e3530886289c18cf2f87ec19a575a0abdce112e3a3"
 
 [[package]]
 name = "memo-map"
@@ -3218,9 +3218,9 @@ dependencies = [
 
 [[package]]
 name = "time"
-version = "0.3.34"
+version = "0.3.36"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c8248b6521bb14bc45b4067159b9b6ad792e2d6d754d6c41fb50e29fefe38749"
+checksum = "5dfd88e563464686c916c7e46e623e520ddc6d79fa6641390f2e3fa86e83e885"
 dependencies = [
  "deranged",
  "itoa",
@@ -3239,9 +3239,9 @@ checksum = "ef927ca75afb808a4d64dd374f00a2adf8d0fcff8e7b184af886c3c87ec4a3f3"
 
 [[package]]
 name = "time-macros"
-version = "0.2.17"
+version = "0.2.18"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7ba3a3ef41e6672a2f0f001392bb5dcd3ff0a9992d618ca761a11c3121547774"
+checksum = "3f252a68540fde3a3877aeea552b832b40ab9a69e318efd078774a01ddee1ccf"
 dependencies = [
  "num-conv",
  "time-core",
@@ -4134,6 +4134,55 @@ dependencies = [
  "windows-sys 0.48.0",
 ]
 
+[[package]]
+name = "write-json"
+version = "0.1.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "23f6174b2566cc4a74f95e1367ec343e7fa80c93cc8087f5c4a3d6a1088b2118"
+
+[[package]]
+name = "xflags"
+version = "0.3.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7d9e15fbb3de55454b0106e314b28e671279009b363e6f1d8e39fdc3bf048944"
+dependencies = [
+ "xflags-macros",
+]
+
+[[package]]
+name = "xflags-macros"
+version = "0.3.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "672423d4fea7ffa2f6c25ba60031ea13dc6258070556f125cc4d790007d4a155"
+
+[[package]]
+name = "xshell"
+version = "0.2.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6db0ab86eae739efd1b054a8d3d16041914030ac4e01cd1dca0cf252fd8b6437"
+dependencies = [
+ "xshell-macros",
+]
+
+[[package]]
+name = "xshell-macros"
+version = "0.2.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9d422e8e38ec76e2f06ee439ccc765e9c6a9638b9e7c9f2e8255e4d41e8bd852"
+
+[[package]]
+name = "xtask"
+version = "0.1.0"
+dependencies = [
+ "anyhow",
+ "flate2",
+ "time",
+ "write-json",
+ "xflags",
+ "xshell",
+ "zip",
+]
+
 [[package]]
 name = "xxhash-rust"
 version = "0.8.10"
@@ -4165,3 +4214,16 @@ name = "zeroize"
 version = "1.7.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "525b4ec142c6b68a2d10f01f7bbf6755599ca3f81ea53b8431b7dd348f5fdb2d"
+
+[[package]]
+name = "zip"
+version = "0.6.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "760394e246e4c28189f19d488c058bf16f564016aefac5d32bb1f3b51d5e9261"
+dependencies = [
+ "byteorder",
+ "crc32fast",
+ "crossbeam-utils",
+ "flate2",
+ "time",
+]

Cargo.toml

@@ -1,7 +1,5 @@
 [workspace]
-members = [
+members = ["crates/*", "xtask/"]
-    "crates/*",
-]
 resolver = "2"
 
 [workspace.package]
@@ -10,3 +8,13 @@ license = "MIT"
 description = "LSP-AI is an open-source language server that serves as a backend for AI-powered functionality, designed to assist and empower software engineers, not replace them."
 repository = "https://github.com/SilasMarvin/lsp-ai"
 readme = "README.md"
+authors = ["Silas Marvin <silas.marvin@dmarvin.net>"]
+
+[profile.dev.package]
+# This speeds up `cargo xtask dist`.
+miniz_oxide.opt-level = 3
+
+[profile.release]
+incremental = true
+# Set this to 1 or 2 to get more useful backtraces in debugger.
+debug = 0

crates/lsp-ai/src/config.rs

@@ -3,7 +3,7 @@ use serde::{Deserialize, Serialize};
 use serde_json::Value;
 use std::collections::HashMap;
 
-pub type Kwargs = HashMap<String, Value>;
+pub(crate) type Kwargs = HashMap<String, Value>;
 
 const fn max_requests_per_second_default() -> f32 {
     1.
@@ -70,7 +70,7 @@ pub struct TextSplitter {
 }
 
 #[derive(Debug, Clone, Deserialize)]
-pub enum ValidMemoryBackend {
+pub(crate) enum ValidMemoryBackend {
     #[serde(rename = "file_store")]
     FileStore(FileStore),
     #[serde(rename = "postgresml")]
@@ -79,7 +79,7 @@ pub enum ValidMemoryBackend {
 
 #[derive(Debug, Clone, Deserialize)]
 #[serde(tag = "type")]
-pub enum ValidModel {
+pub(crate) enum ValidModel {
     #[cfg(feature = "llama_cpp")]
     #[serde(rename = "llama_cpp")]
     LLaMACPP(LLaMACPP),
@@ -97,13 +97,13 @@ pub enum ValidModel {
 
 #[derive(Debug, Clone, Deserialize, Serialize)]
 #[serde(deny_unknown_fields)]
-pub struct ChatMessage {
+pub(crate) struct ChatMessage {
-    pub role: String,
+    pub(crate) role: String,
-    pub content: String,
+    pub(crate) content: String,
 }
 
 impl ChatMessage {
-    pub fn new(role: String, content: String) -> Self {
+    pub(crate) fn new(role: String, content: String) -> Self {
         Self {
             role,
             content,
@@ -115,10 +115,10 @@ impl ChatMessage {
 #[derive(Clone, Debug, Deserialize)]
 #[allow(clippy::upper_case_acronyms)]
 #[serde(deny_unknown_fields)]
-pub struct FIM {
+pub(crate) struct FIM {
-    pub start: String,
+    pub(crate) start: String,
-    pub middle: String,
+    pub(crate) middle: String,
-    pub end: String,
+    pub(crate) end: String,
 }
 
 const fn max_crawl_memory_default() -> u64 {
@@ -131,71 +131,71 @@ const fn max_crawl_file_size_default() -> u64 {
 
 #[derive(Clone, Debug, Deserialize)]
 #[serde(deny_unknown_fields)]
-pub struct Crawl {
+pub(crate) struct Crawl {
     #[serde(default = "max_crawl_file_size_default")]
-    pub max_file_size: u64,
+    pub(crate) max_file_size: u64,
     #[serde(default = "max_crawl_memory_default")]
-    pub max_crawl_memory: u64,
+    pub(crate) max_crawl_memory: u64,
     #[serde(default)]
-    pub all_files: bool,
+    pub(crate) all_files: bool,
 }
 
 #[derive(Clone, Debug, Deserialize)]
-pub struct PostgresMLEmbeddingModel {
+pub(crate) struct PostgresMLEmbeddingModel {
-    pub model: String,
+    pub(crate) model: String,
-    pub embed_parameters: Option<Value>,
+    pub(crate) embed_parameters: Option<Value>,
-    pub query_parameters: Option<Value>,
+    pub(crate) query_parameters: Option<Value>,
 }
 
 #[derive(Clone, Debug, Deserialize)]
 #[serde(deny_unknown_fields)]
-pub struct PostgresML {
+pub(crate) struct PostgresML {
-    pub database_url: Option<String>,
+    pub(crate) database_url: Option<String>,
-    pub crawl: Option<Crawl>,
+    pub(crate) crawl: Option<Crawl>,
     #[serde(default)]
-    pub splitter: ValidSplitter,
+    pub(crate) splitter: ValidSplitter,
-    pub embedding_model: Option<PostgresMLEmbeddingModel>,
+    pub(crate) embedding_model: Option<PostgresMLEmbeddingModel>,
 }
 
 #[derive(Clone, Debug, Deserialize, Default)]
 #[serde(deny_unknown_fields)]
-pub struct FileStore {
+pub(crate) struct FileStore {
-    pub crawl: Option<Crawl>,
+    pub(crate) crawl: Option<Crawl>,
 }
 
 impl FileStore {
-    pub fn new_without_crawl() -> Self {
+    pub(crate) fn new_without_crawl() -> Self {
         Self { crawl: None }
     }
 }
 
 #[derive(Clone, Debug, Deserialize)]
 #[serde(deny_unknown_fields)]
-pub struct Ollama {
+pub(crate) struct Ollama {
     // The generate endpoint, default: 'http://localhost:11434/api/generate'
-    pub generate_endpoint: Option<String>,
+    pub(crate) generate_endpoint: Option<String>,
     // The chat endpoint, default: 'http://localhost:11434/api/chat'
-    pub chat_endpoint: Option<String>,
+    pub(crate) chat_endpoint: Option<String>,
     // The model name
-    pub model: String,
+    pub(crate) model: String,
     // The maximum requests per second
     #[serde(default = "max_requests_per_second_default")]
-    pub max_requests_per_second: f32,
+    pub(crate) max_requests_per_second: f32,
 }
 
 #[derive(Clone, Debug, Deserialize)]
 #[serde(deny_unknown_fields)]
-pub struct MistralFIM {
+pub(crate) struct MistralFIM {
     // The auth token env var name
-    pub auth_token_env_var_name: Option<String>,
+    pub(crate) auth_token_env_var_name: Option<String>,
-    pub auth_token: Option<String>,
+    pub(crate) auth_token: Option<String>,
     // The fim endpoint
-    pub fim_endpoint: Option<String>,
+    pub(crate) fim_endpoint: Option<String>,
     // The model name
-    pub model: String,
+    pub(crate) model: String,
     // The maximum requests per second
     #[serde(default = "max_requests_per_second_default")]
-    pub max_requests_per_second: f32,
+    pub(crate) max_requests_per_second: f32,
 }
 
 #[cfg(feature = "llama_cpp")]
@@ -229,87 +229,89 @@ pub struct LLaMACPP {
 
 #[derive(Clone, Debug, Deserialize)]
 #[serde(deny_unknown_fields)]
-pub struct OpenAI {
+pub(crate) struct OpenAI {
     // The auth token env var name
-    pub auth_token_env_var_name: Option<String>,
+    pub(crate) auth_token_env_var_name: Option<String>,
     // The auth token
-    pub auth_token: Option<String>,
+    pub(crate) auth_token: Option<String>,
     // The completions endpoint
-    pub completions_endpoint: Option<String>,
+    pub(crate) completions_endpoint: Option<String>,
     // The chat endpoint
-    pub chat_endpoint: Option<String>,
+    pub(crate) chat_endpoint: Option<String>,
     // The maximum requests per second
     #[serde(default = "max_requests_per_second_default")]
-    pub max_requests_per_second: f32,
+    pub(crate) max_requests_per_second: f32,
     // The model name
-    pub model: String,
+    pub(crate) model: String,
 }
 
 #[derive(Clone, Debug, Deserialize)]
 #[serde(deny_unknown_fields)]
-pub struct Gemini {
+pub(crate) struct Gemini {
     // The auth token env var name
-    pub auth_token_env_var_name: Option<String>,
+    pub(crate) auth_token_env_var_name: Option<String>,
     // The auth token
-    pub auth_token: Option<String>,
+    pub(crate) auth_token: Option<String>,
     // The completions endpoint
-    pub completions_endpoint: Option<String>,
+    #[allow(dead_code)]
+    pub(crate) completions_endpoint: Option<String>,
     // The chat endpoint
-    pub chat_endpoint: Option<String>,
+    pub(crate) chat_endpoint: Option<String>,
     // The maximum requests per second
     #[serde(default = "max_requests_per_second_default")]
-    pub max_requests_per_second: f32,
+    pub(crate) max_requests_per_second: f32,
     // The model name
-    pub model: String,
+    pub(crate) model: String,
 }
 
 #[derive(Clone, Debug, Deserialize)]
 #[serde(deny_unknown_fields)]
-pub struct Anthropic {
+pub(crate) struct Anthropic {
     // The auth token env var name
-    pub auth_token_env_var_name: Option<String>,
+    pub(crate) auth_token_env_var_name: Option<String>,
-    pub auth_token: Option<String>,
+    pub(crate) auth_token: Option<String>,
     // The completions endpoint
-    pub completions_endpoint: Option<String>,
+    #[allow(dead_code)]
+    pub(crate) completions_endpoint: Option<String>,
     // The chat endpoint
-    pub chat_endpoint: Option<String>,
+    pub(crate) chat_endpoint: Option<String>,
     // The maximum requests per second
     #[serde(default = "max_requests_per_second_default")]
-    pub max_requests_per_second: f32,
+    pub(crate) max_requests_per_second: f32,
     // The model name
-    pub model: String,
+    pub(crate) model: String,
 }
 
 #[derive(Clone, Debug, Deserialize)]
-pub struct Completion {
+pub(crate) struct Completion {
     // The model key to use
-    pub model: String,
+    pub(crate) model: String,
     // Args are deserialized by the backend using them
     #[serde(default)]
-    pub parameters: Kwargs,
+    pub(crate) parameters: Kwargs,
     // Parameters for post processing
     #[serde(default)]
-    pub post_process: PostProcess,
+    pub(crate) post_process: PostProcess,
 }
 
 #[derive(Clone, Debug, Deserialize)]
 #[serde(deny_unknown_fields)]
-pub struct ValidConfig {
+pub(crate) struct ValidConfig {
-    pub memory: ValidMemoryBackend,
+    pub(crate) memory: ValidMemoryBackend,
-    pub models: HashMap<String, ValidModel>,
+    pub(crate) models: HashMap<String, ValidModel>,
-    pub completion: Option<Completion>,
+    pub(crate) completion: Option<Completion>,
 }
 
 #[derive(Clone, Debug, Deserialize, Default)]
-pub struct ValidClientParams {
+pub(crate) struct ValidClientParams {
     #[serde(alias = "rootUri")]
-    pub root_uri: Option<String>,
+    pub(crate) root_uri: Option<String>,
 }
 
 #[derive(Clone, Debug)]
 pub struct Config {
-    pub config: ValidConfig,
+    pub(crate) config: ValidConfig,
-    pub client_params: ValidClientParams,
+    pub(crate) client_params: ValidClientParams,
 }
 
 impl Config {

crates/lsp-ai/src/crawl.rs

@@ -4,7 +4,7 @@ use tracing::{error, instrument};
 
 use crate::config::{self, Config};
 
-pub struct Crawl {
+pub(crate) struct Crawl {
     crawl_config: config::Crawl,
     config: Config,
     crawled_file_types: HashSet<String>,
@@ -12,7 +12,7 @@ pub struct Crawl {
 }
 
 impl Crawl {
-    pub fn new(crawl_config: config::Crawl, config: Config) -> Self {
+    pub(crate) fn new(crawl_config: config::Crawl, config: Config) -> Self {
         Self {
             crawl_config,
             config,
@@ -37,11 +37,10 @@ impl Crawl {
             }
 
             let extension_to_match = triggered_file
-                .map(|tf| {
+                .and_then(|tf| {
                     let path = std::path::Path::new(&tf);
                     path.extension().map(|f| f.to_str().map(|f| f.to_owned()))
                 })
-                .flatten()
                 .flatten();
 
             if let Some(extension_to_match) = &extension_to_match {
@@ -70,7 +69,7 @@ impl Crawl {
                             }
                         } else {
                             match (
-                                path.extension().map(|pe| pe.to_str()).flatten(),
+                                path.extension().and_then(|pe| pe.to_str()),
                                 &extension_to_match,
                             ) {
                                 (Some(path_extension), Some(extension_to_match)) => {

crates/lsp-ai/src/custom_requests/generation.rs

@@ -4,28 +4,28 @@ use serde_json::Value;
 
 use crate::config;
 
-pub enum Generation {}
+pub(crate) enum Generation {}
 
 #[derive(Debug, Clone, Deserialize, Serialize)]
 #[serde(rename_all = "camelCase")]
-pub struct GenerationParams {
+pub(crate) struct GenerationParams {
     // This field was "mixed-in" from TextDocumentPositionParams
     #[serde(flatten)]
-    pub text_document_position: TextDocumentPositionParams,
+    pub(crate) text_document_position: TextDocumentPositionParams,
     // The model key to use
-    pub model: String,
+    pub(crate) model: String,
     #[serde(default)]
     // Args are deserialized by the backend using them
-    pub parameters: Value,
+    pub(crate) parameters: Value,
     // Parameters for post processing
     #[serde(default)]
-    pub post_process: config::PostProcess,
+    pub(crate) post_process: config::PostProcess,
 }
 
 #[derive(Debug, Clone, Deserialize, Serialize)]
 #[serde(rename_all = "camelCase")]
-pub struct GenerateResult {
+pub(crate) struct GenerateResult {
-    pub generated_text: String,
+    pub(crate) generated_text: String,
 }
 
 impl lsp_types::request::Request for Generation {

crates/lsp-ai/src/custom_requests/generation_stream.rs

@@ -1,7 +1,7 @@
 use lsp_types::{ProgressToken, TextDocumentPositionParams};
 use serde::{Deserialize, Serialize};
 
-pub enum GenerationStream {}
+pub(crate) enum GenerationStream {}
 
 #[derive(Debug, PartialEq, Clone, Deserialize, Serialize)]
 #[serde(rename_all = "camelCase")]
@@ -15,9 +15,9 @@ pub struct GenerationStreamParams {
 
 #[derive(Debug, PartialEq, Clone, Deserialize, Serialize)]
 #[serde(rename_all = "camelCase")]
-pub struct GenerationStreamResult {
+pub(crate) struct GenerationStreamResult {
-    pub generated_text: String,
+    pub(crate) generated_text: String,
-    pub partial_result_token: ProgressToken,
+    pub(crate) partial_result_token: ProgressToken,
 }
 
 impl lsp_types::request::Request for GenerationStream {

crates/lsp-ai/src/custom_requests/mod.rs

@@ -1,2 +1,2 @@
-pub mod generation;
+pub(crate) mod generation;
-pub mod generation_stream;
+pub(crate) mod generation_stream;

crates/lsp-ai/src/memory_backends/file_store.rs

@@ -17,12 +17,12 @@ use crate::{
 use super::{ContextAndCodePrompt, FIMPrompt, MemoryBackend, MemoryRunParams, Prompt, PromptType};
 
 #[derive(Default)]
-pub struct AdditionalFileStoreParams {
+pub(crate) struct AdditionalFileStoreParams {
     build_tree: bool,
 }
 
 impl AdditionalFileStoreParams {
-    pub fn new(build_tree: bool) -> Self {
+    pub(crate) fn new(build_tree: bool) -> Self {
         Self { build_tree }
     }
 }
@@ -47,7 +47,7 @@ impl File {
     }
 }
 
-pub struct FileStore {
+pub(crate) struct FileStore {
     params: AdditionalFileStoreParams,
     file_map: Mutex<HashMap<String, File>>,
     accessed_files: Mutex<IndexSet<String>>,
@@ -55,7 +55,10 @@ pub struct FileStore {
 }
 
 impl FileStore {
-    pub fn new(mut file_store_config: config::FileStore, config: Config) -> anyhow::Result<Self> {
+    pub(crate) fn new(
+        mut file_store_config: config::FileStore,
+        config: Config,
+    ) -> anyhow::Result<Self> {
         let crawl = file_store_config
             .crawl
             .take()
@@ -72,7 +75,7 @@ impl FileStore {
         Ok(s)
     }
 
-    pub fn new_with_params(
+    pub(crate) fn new_with_params(
         mut file_store_config: config::FileStore,
         config: Config,
         params: AdditionalFileStoreParams,
@@ -192,7 +195,7 @@ impl FileStore {
         Ok((rope, cursor_index))
     }
 
-    pub fn get_characters_around_position(
+    pub(crate) fn get_characters_around_position(
         &self,
         position: &TextDocumentPositionParams,
         characters: usize,
@@ -216,7 +219,7 @@ impl FileStore {
         Ok(rope_slice.to_string())
     }
 
-    pub fn build_code(
+    pub(crate) fn build_code(
         &self,
         position: &TextDocumentPositionParams,
         prompt_type: PromptType,
@@ -272,15 +275,18 @@ impl FileStore {
         })
     }
 
-    pub fn file_map(&self) -> &Mutex<HashMap<String, File>> {
+    pub(crate) fn file_map(&self) -> &Mutex<HashMap<String, File>> {
         &self.file_map
     }
 
-    pub fn contains_file(&self, uri: &str) -> bool {
+    pub(crate) fn contains_file(&self, uri: &str) -> bool {
         self.file_map.lock().contains_key(uri)
     }
 
-    pub fn position_to_byte(&self, position: &TextDocumentPositionParams) -> anyhow::Result<usize> {
+    pub(crate) fn position_to_byte(
+        &self,
+        position: &TextDocumentPositionParams,
+    ) -> anyhow::Result<usize> {
         let file_map = self.file_map.lock();
         let uri = position.text_document.uri.to_string();
         let file = file_map
@@ -322,7 +328,7 @@ impl MemoryBackend for FileStore {
         prompt_type: PromptType,
         params: &Value,
     ) -> anyhow::Result<Prompt> {
-        let params: MemoryRunParams = params.try_into()?;
+        let params: MemoryRunParams = params.into();
         self.build_code(position, prompt_type, params, true)
     }
 

crates/lsp-ai/src/memory_backends/mod.rs

@@ -6,7 +6,7 @@ use serde_json::Value;
 
 use crate::config::{Config, ValidMemoryBackend};
 
-pub mod file_store;
+pub(crate) mod file_store;
 mod postgresml;
 
 #[derive(Clone, Debug)]
@@ -16,9 +16,9 @@ pub enum PromptType {
 }
 
 #[derive(Clone)]
-pub struct MemoryRunParams {
+pub(crate) struct MemoryRunParams {
-    pub is_for_chat: bool,
+    pub(crate) is_for_chat: bool,
-    pub max_context: usize,
+    pub(crate) max_context: usize,
 }
 
 impl From<&Value> for MemoryRunParams {

crates/lsp-ai/src/memory_backends/postgresml/mod.rs

@@ -84,7 +84,7 @@ async fn split_and_upsert_file(
 }
 
 #[derive(Clone)]
-pub struct PostgresML {
+pub(crate) struct PostgresML {
     config: Config,
     postgresml_config: config::PostgresML,
     file_store: Arc<FileStore>,
@@ -240,12 +240,7 @@ impl PostgresML {
                         })
                         .collect();
                     if let Err(e) = task_collection
-                        .delete_documents(
+                        .delete_documents(json!({ "$or": delete_or_statements }).into())
-                            json!({
-                                "$or": delete_or_statements
-                            })
-                            .into(),
-                        )
                         .await
                         .context("PGML - error deleting documents")
                     {
@@ -255,15 +250,14 @@ impl PostgresML {
                     let documents: Vec<pgml::types::Json> = chunks
                         .into_iter()
                         .zip(&file_uris)
-                        .map(|(chunks, uri)| {
+                        .flat_map(|(chunks, uri)| {
                             chunks
                                 .into_iter()
                                 .map(|chunk| {
-                                    chunk_to_document(&uri, chunk, task_root_uri.as_deref())
+                                    chunk_to_document(uri, chunk, task_root_uri.as_deref())
                                 })
                                 .collect::<Vec<Value>>()
                         })
-                        .flatten()
                         .map(|f: Value| f.into())
                         .collect();
                     if let Err(e) = task_collection
@@ -365,15 +359,11 @@ impl PostgresML {
                 current_chunks_bytes += contents.len();
                 let chunks: Vec<pgml::types::Json> = self
                     .splitter
-                    .split_file_contents(&uri, &contents)
+                    .split_file_contents(uri, &contents)
                     .into_iter()
                     .map(|chunk| {
-                        chunk_to_document(
+                        chunk_to_document(uri, chunk, self.config.client_params.root_uri.as_deref())
-                            &uri,
+                            .into()
-                            chunk,
-                            self.config.client_params.root_uri.as_deref(),
-                        )
-                        .into()
                     })
                     .collect();
                 chunks_to_upsert.extend(chunks);
@@ -389,7 +379,7 @@ impl PostgresML {
             }
         }
         // Upsert any remaining chunks
-        if chunks_to_upsert.len() > 0 {
+        if chunks_to_upsert.is_empty() {
             collection
                 .upsert_documents(chunks_to_upsert, None)
                 .await
@@ -479,7 +469,7 @@ impl PostgresML {
                     Ok(true)
                 })?;
             // Upsert any remaining documents
-            if documents.len() > 0 {
+            if documents.is_empty() {
                 let mut collection = self.collection.clone();
                 TOKIO_RUNTIME.spawn(async move {
                     if let Err(e) = collection
@@ -510,7 +500,7 @@ impl MemoryBackend for PostgresML {
         prompt_type: PromptType,
         params: &Value,
     ) -> anyhow::Result<Prompt> {
-        let params: MemoryRunParams = params.try_into()?;
+        let params: MemoryRunParams = params.into();
         let chunk_size = self.splitter.chunk_size();
         let total_allowed_characters = tokens_to_estimated_characters(params.max_context);
 
@@ -535,8 +525,7 @@ impl MemoryBackend for PostgresML {
             .postgresml_config
             .embedding_model
             .as_ref()
-            .map(|m| m.query_parameters.clone())
+            .and_then(|m| m.query_parameters.clone())
-            .flatten()
         {
             Some(query_parameters) => query_parameters,
             None => json!({
@@ -602,7 +591,7 @@ impl MemoryBackend for PostgresML {
                 Prompt::ContextAndCode(ContextAndCodePrompt::new(
                     context.to_owned(),
                     format_file_excerpt(
-                        &position.text_document.uri.to_string(),
+                        position.text_document.uri.as_str(),
                         &context_and_code.code,
                         self.config.client_params.root_uri.as_deref(),
                     ),

crates/lsp-ai/src/memory_worker.rs

@@ -13,7 +13,7 @@ use crate::{
 };
 
 #[derive(Debug)]
-pub struct PromptRequest {
+pub(crate) struct PromptRequest {
     position: TextDocumentPositionParams,
     prompt_type: PromptType,
     params: Value,
@@ -21,7 +21,7 @@ pub struct PromptRequest {
 }
 
 impl PromptRequest {
-    pub fn new(
+    pub(crate) fn new(
         position: TextDocumentPositionParams,
         prompt_type: PromptType,
         params: Value,
@@ -37,13 +37,13 @@ impl PromptRequest {
 }
 
 #[derive(Debug)]
-pub struct FilterRequest {
+pub(crate) struct FilterRequest {
     position: TextDocumentPositionParams,
     tx: tokio::sync::oneshot::Sender<String>,
 }
 
 impl FilterRequest {
-    pub fn new(
+    pub(crate) fn new(
         position: TextDocumentPositionParams,
         tx: tokio::sync::oneshot::Sender<String>,
     ) -> Self {
@@ -51,7 +51,7 @@ impl FilterRequest {
     }
 }
 
-pub enum WorkerRequest {
+pub(crate) enum WorkerRequest {
     FilterText(FilterRequest),
     Prompt(PromptRequest),
     DidOpenTextDocument(DidOpenTextDocumentParams),
@@ -115,7 +115,7 @@ fn do_run(
     }
 }
 
-pub fn run(
+pub(crate) fn run(
     memory_backend: Box<dyn MemoryBackend + Send + Sync>,
     rx: std::sync::mpsc::Receiver<WorkerRequest>,
 ) {

crates/lsp-ai/src/splitters/text_splitter.rs

@@ -2,20 +2,20 @@ use crate::{config, memory_backends::file_store::File};
 
 use super::{ByteRange, Chunk, Splitter};
 
-pub struct TextSplitter {
+pub(crate) struct TextSplitter {
     chunk_size: usize,
     splitter: text_splitter::TextSplitter<text_splitter::Characters>,
 }
 
 impl TextSplitter {
-    pub fn new(config: config::TextSplitter) -> Self {
+    pub(crate) fn new(config: config::TextSplitter) -> Self {
         Self {
             chunk_size: config.chunk_size,
             splitter: text_splitter::TextSplitter::new(config.chunk_size),
         }
     }
 
-    pub fn new_with_chunk_size(chunk_size: usize) -> Self {
+    pub(crate) fn new_with_chunk_size(chunk_size: usize) -> Self {
         Self {
             chunk_size,
             splitter: text_splitter::TextSplitter::new(chunk_size),

crates/lsp-ai/src/splitters/tree_sitter.rs

@@ -6,14 +6,14 @@ use crate::{config, memory_backends::file_store::File, utils::parse_tree};
 
 use super::{text_splitter::TextSplitter, ByteRange, Chunk, Splitter};
 
-pub struct TreeSitter {
+pub(crate) struct TreeSitter {
     chunk_size: usize,
     splitter: TreeSitterCodeSplitter,
     text_splitter: TextSplitter,
 }
 
 impl TreeSitter {
-    pub fn new(config: config::TreeSitter) -> anyhow::Result<Self> {
+    pub(crate) fn new(config: config::TreeSitter) -> anyhow::Result<Self> {
         let text_splitter = TextSplitter::new_with_chunk_size(config.chunk_size);
         Ok(Self {
             chunk_size: config.chunk_size,

crates/lsp-ai/src/transformer_backends/anthropic.rs

@@ -30,18 +30,18 @@ const fn temperature_default() -> f32 {
 
 // NOTE: We cannot deny unknown fields as the provided parameters may contain other fields relevant to other processes
 #[derive(Debug, Deserialize)]
-pub struct AnthropicRunParams {
+pub(crate) struct AnthropicRunParams {
     system: String,
     messages: Vec<ChatMessage>,
     #[serde(default = "max_tokens_default")]
-    pub max_tokens: usize,
+    pub(crate) max_tokens: usize,
     #[serde(default = "top_p_default")]
-    pub top_p: f32,
+    pub(crate) top_p: f32,
     #[serde(default = "temperature_default")]
-    pub temperature: f32,
+    pub(crate) temperature: f32,
 }
 
-pub struct Anthropic {
+pub(crate) struct Anthropic {
     config: config::Anthropic,
 }
 
@@ -56,11 +56,11 @@ struct AnthropicChatResponse {
     error: Option<Value>,
     #[serde(default)]
     #[serde(flatten)]
-    pub other: HashMap<String, Value>,
+    pub(crate) other: HashMap<String, Value>,
 }
 
 impl Anthropic {
-    pub fn new(config: config::Anthropic) -> Self {
+    pub(crate) fn new(config: config::Anthropic) -> Self {
         Self { config }
     }
 

crates/lsp-ai/src/transformer_backends/gemini.rs

@@ -39,7 +39,7 @@ const fn max_tokens_default() -> usize {
 
 #[derive(Debug, Serialize, Deserialize, Clone)]
 struct Part {
-    pub text: String,
+    pub(crate) text: String,
 }
 
 #[derive(Debug, Serialize, Deserialize, Clone)]
@@ -55,37 +55,33 @@ impl GeminiContent {
 }
 
 #[derive(Debug, Deserialize, Serialize, Clone)]
+#[serde(rename = "camelCase")]
 #[serde(deny_unknown_fields)]
-pub struct GeminiGenerationConfig {
+pub(crate) struct GeminiGenerationConfig {
-    #[serde(rename = "stopSequences")]
     #[serde(default)]
-    pub stop_sequences: Vec<String>,
+    pub(crate) stop_sequences: Vec<String>,
-    #[serde(rename = "maxOutputTokens")]
     #[serde(default = "max_tokens_default")]
-    pub max_output_tokens: usize,
+    pub(crate) max_output_tokens: usize,
-    pub temperature: Option<f32>,
+    pub(crate) temperature: Option<f32>,
-    #[serde(rename = "topP")]
+    pub(crate) top_p: Option<f32>,
-    pub top_p: Option<f32>,
+    pub(crate) top_k: Option<f32>,
-    #[serde(rename = "topK")]
-    pub top_k: Option<f32>,
 }
 
 // NOTE: We cannot deny unknown fields as the provided parameters may contain other fields relevant to other processes
 #[derive(Debug, Deserialize, Serialize, Clone)]
-pub struct GeminiRunParams {
+#[serde(rename = "camelCase")]
+pub(crate) struct GeminiRunParams {
     contents: Vec<GeminiContent>,
-    #[serde(rename = "systemInstruction")]
     system_instruction: GeminiContent,
-    #[serde(rename = "generationConfig")]
     generation_config: Option<GeminiGenerationConfig>,
 }
 
-pub struct Gemini {
+pub(crate) struct Gemini {
     configuration: config::Gemini,
 }
 
 impl Gemini {
-    pub fn new(configuration: config::Gemini) -> Self {
+    pub(crate) fn new(configuration: config::Gemini) -> Self {
         Self { configuration }
     }
 

crates/lsp-ai/src/transformer_backends/mistral_fim.rs

@@ -26,25 +26,25 @@ const fn temperature_default() -> f32 {
 
 // NOTE: We cannot deny unknown fields as the provided parameters may contain other fields relevant to other processes
 #[derive(Debug, Deserialize)]
-pub struct MistralFIMRunParams {
+pub(crate) struct MistralFIMRunParams {
     #[serde(default = "max_tokens_default")]
-    pub max_tokens: usize,
+    pub(crate) max_tokens: usize,
     #[serde(default = "top_p_default")]
-    pub top_p: f32,
+    pub(crate) top_p: f32,
     #[serde(default = "temperature_default")]
-    pub temperature: f32,
+    pub(crate) temperature: f32,
-    pub min_tokens: Option<u64>,
+    pub(crate) min_tokens: Option<u64>,
-    pub random_seed: Option<u64>,
+    pub(crate) random_seed: Option<u64>,
     #[serde(default)]
-    pub stop: Vec<String>,
+    pub(crate) stop: Vec<String>,
 }
 
-pub struct MistralFIM {
+pub(crate) struct MistralFIM {
     config: config::MistralFIM,
 }
 
 impl MistralFIM {
-    pub fn new(config: config::MistralFIM) -> Self {
+    pub(crate) fn new(config: config::MistralFIM) -> Self {
         Self { config }
     }
 

crates/lsp-ai/src/transformer_backends/ollama.rs

@@ -16,8 +16,8 @@ use super::TransformerBackend;
 
 // NOTE: We cannot deny unknown fields as the provided parameters may contain other fields relevant to other processes
 #[derive(Debug, Deserialize)]
-pub struct OllamaRunParams {
+pub(crate) struct OllamaRunParams {
-    pub fim: Option<FIM>,
+    pub(crate) fim: Option<FIM>,
     messages: Option<Vec<ChatMessage>>,
     #[serde(default)]
     options: HashMap<String, Value>,
@@ -26,7 +26,7 @@ pub struct OllamaRunParams {
     keep_alive: Option<String>,
 }
 
-pub struct Ollama {
+pub(crate) struct Ollama {
     configuration: config::Ollama,
 }
 

crates/lsp-ai/src/transformer_backends/open_ai/mod.rs

@@ -38,22 +38,22 @@ const fn temperature_default() -> f32 {
 
 // NOTE: We cannot deny unknown fields as the provided parameters may contain other fields relevant to other processes
 #[derive(Debug, Deserialize)]
-pub struct OpenAIRunParams {
+pub(crate) struct OpenAIRunParams {
-    pub fim: Option<FIM>,
+    pub(crate) fim: Option<FIM>,
     messages: Option<Vec<ChatMessage>>,
     #[serde(default = "max_tokens_default")]
-    pub max_tokens: usize,
+    pub(crate) max_tokens: usize,
     #[serde(default = "top_p_default")]
-    pub top_p: f32,
+    pub(crate) top_p: f32,
     #[serde(default = "presence_penalty_default")]
-    pub presence_penalty: f32,
+    pub(crate) presence_penalty: f32,
     #[serde(default = "frequency_penalty_default")]
-    pub frequency_penalty: f32,
+    pub(crate) frequency_penalty: f32,
     #[serde(default = "temperature_default")]
-    pub temperature: f32,
+    pub(crate) temperature: f32,
 }
 
-pub struct OpenAI {
+pub(crate) struct OpenAI {
     configuration: config::OpenAI,
 }
 
@@ -68,27 +68,27 @@ struct OpenAICompletionsResponse {
     error: Option<Value>,
     #[serde(default)]
     #[serde(flatten)]
-    pub other: HashMap<String, Value>,
+    pub(crate) other: HashMap<String, Value>,
 }
 
 #[derive(Debug, Deserialize, Serialize)]
-pub struct OpenAIChatMessage {
+pub(crate) struct OpenAIChatMessage {
-    pub role: String,
+    pub(crate) role: String,
-    pub content: String,
+    pub(crate) content: String,
 }
 
 #[derive(Deserialize)]
-pub struct OpenAIChatChoices {
+pub(crate) struct OpenAIChatChoices {
-    pub message: OpenAIChatMessage,
+    pub(crate) message: OpenAIChatMessage,
 }
 
 #[derive(Deserialize)]
-pub struct OpenAIChatResponse {
+pub(crate) struct OpenAIChatResponse {
-    pub choices: Option<Vec<OpenAIChatChoices>>,
+    pub(crate) choices: Option<Vec<OpenAIChatChoices>>,
-    pub error: Option<Value>,
+    pub(crate) error: Option<Value>,
     #[serde(default)]
     #[serde(flatten)]
-    pub other: HashMap<String, Value>,
+    pub(crate) other: HashMap<String, Value>,
 }
 
 impl OpenAI {

crates/lsp-ai/src/transformer_worker.rs

@@ -20,25 +20,25 @@ use crate::transformer_backends::TransformerBackend;
 use crate::utils::{ToResponseError, TOKIO_RUNTIME};
 
 #[derive(Clone, Debug)]
-pub struct CompletionRequest {
+pub(crate) struct CompletionRequest {
     id: RequestId,
     params: CompletionParams,
 }
 
 impl CompletionRequest {
-    pub fn new(id: RequestId, params: CompletionParams) -> Self {
+    pub(crate) fn new(id: RequestId, params: CompletionParams) -> Self {
         Self { id, params }
     }
 }
 
 #[derive(Clone, Debug)]
-pub struct GenerationRequest {
+pub(crate) struct GenerationRequest {
     id: RequestId,
     params: GenerationParams,
 }
 
 impl GenerationRequest {
-    pub fn new(id: RequestId, params: GenerationParams) -> Self {
+    pub(crate) fn new(id: RequestId, params: GenerationParams) -> Self {
         Self { id, params }
     }
 }
@@ -58,7 +58,7 @@ impl GenerationStreamRequest {
 }
 
 #[derive(Clone, Debug)]
-pub enum WorkerRequest {
+pub(crate) enum WorkerRequest {
     Completion(CompletionRequest),
     Generation(GenerationRequest),
     GenerationStream(GenerationStreamRequest),
@@ -159,7 +159,7 @@ fn post_process_response(
     }
 }
 
-pub fn run(
+pub(crate) fn run(
     transformer_backends: HashMap<String, Box<dyn TransformerBackend + Send + Sync>>,
     memory_tx: std::sync::mpsc::Sender<memory_worker::WorkerRequest>,
     transformer_rx: std::sync::mpsc::Receiver<WorkerRequest>,
@@ -338,7 +338,7 @@ async fn do_completion(
     let mut response = transformer_backend.do_completion(&prompt, params).await?;
 
     if let Some(post_process) = config.get_completions_post_process() {
-        response.insert_text = post_process_response(response.insert_text, &prompt, &post_process);
+        response.insert_text = post_process_response(response.insert_text, &prompt, post_process);
     }
 
     // Build and send the response

crates/lsp-ai/src/utils.rs

@@ -6,7 +6,7 @@ use tree_sitter::Tree;
 
 use crate::{config::ChatMessage, memory_backends::ContextAndCodePrompt, splitters::Chunk};
 
-pub static TOKIO_RUNTIME: Lazy<runtime::Runtime> = Lazy::new(|| {
+pub(crate) static TOKIO_RUNTIME: Lazy<runtime::Runtime> = Lazy::new(|| {
     runtime::Builder::new_multi_thread()
         .worker_threads(4)
         .enable_all()
@@ -14,7 +14,7 @@ pub static TOKIO_RUNTIME: Lazy<runtime::Runtime> = Lazy::new(|| {
         .expect("Error building tokio runtime")
 });
 
-pub trait ToResponseError {
+pub(crate) trait ToResponseError {
     fn to_response_error(&self, code: i32) -> ResponseError;
 }
 
@@ -28,11 +28,11 @@ impl ToResponseError for anyhow::Error {
     }
 }
 
-pub fn tokens_to_estimated_characters(tokens: usize) -> usize {
+pub(crate) fn tokens_to_estimated_characters(tokens: usize) -> usize {
     tokens * 4
 }
 
-pub fn format_chat_messages(
+pub(crate) fn format_chat_messages(
     messages: &[ChatMessage],
     prompt: &ContextAndCodePrompt,
 ) -> Vec<ChatMessage> {
@@ -47,24 +47,28 @@ pub fn format_chat_messages(
         .collect()
 }
 
-pub fn format_context_code_in_str(s: &str, context: &str, code: &str) -> String {
+pub(crate) fn format_context_code_in_str(s: &str, context: &str, code: &str) -> String {
     s.replace("{CONTEXT}", context).replace("{CODE}", code)
 }
 
-pub fn format_context_code(context: &str, code: &str) -> String {
+pub(crate) fn format_context_code(context: &str, code: &str) -> String {
     format!("{context}\n\n{code}")
 }
 
-pub fn chunk_to_id(uri: &str, chunk: &Chunk) -> String {
+pub(crate) fn chunk_to_id(uri: &str, chunk: &Chunk) -> String {
     format!("{uri}#{}-{}", chunk.range.start_byte, chunk.range.end_byte)
 }
 
-pub fn parse_tree(uri: &str, contents: &str, old_tree: Option<&Tree>) -> anyhow::Result<Tree> {
+pub(crate) fn parse_tree(
+    uri: &str,
+    contents: &str,
+    old_tree: Option<&Tree>,
+) -> anyhow::Result<Tree> {
     let path = std::path::Path::new(uri);
     let extension = path.extension().map(|x| x.to_string_lossy());
     let extension = extension.as_deref().unwrap_or("");
     let mut parser = utils_tree_sitter::get_parser_for_extension(extension)?;
     parser
-        .parse(&contents, old_tree)
+        .parse(contents, old_tree)
         .with_context(|| format!("parsing tree failed for {uri}"))
 }

crates/splitter-tree-sitter/src/lib.rs

@@ -55,11 +55,7 @@ impl TreeSitterCodeSplitter {
         }
     }
 
-    pub fn split<'a, 'b, 'c>(
+    pub fn split<'c>(&self, tree: &Tree, utf8: &'c [u8]) -> Result<Vec<Chunk<'c>>, SplitError> {
-        &'a self,
-        tree: &'b Tree,
-        utf8: &'c [u8],
-    ) -> Result<Vec<Chunk<'c>>, SplitError> {
         let cursor = tree.walk();
         Ok(self
             .split_recursive(cursor, utf8)?
@@ -68,7 +64,7 @@ impl TreeSitterCodeSplitter {
             // Let's combine some of our smaller chunks together
             // We also want to do this in reverse as it (seems) to make more sense to combine code slices from bottom to top
             .try_fold(vec![], |mut acc, current| {
-                if acc.len() == 0 {
+                if acc.is_empty() {
                     acc.push(current);
                     Ok::<_, SplitError>(acc)
                 } else {
@@ -94,9 +90,9 @@ impl TreeSitterCodeSplitter {
             .collect())
     }
 
-    fn split_recursive<'a, 'b, 'c>(
+    fn split_recursive<'c>(
-        &'a self,
+        &self,
-        mut cursor: TreeCursor<'b>,
+        mut cursor: TreeCursor<'_>,
         utf8: &'c [u8],
     ) -> Result<Vec<Chunk<'c>>, SplitError> {
         let node = cursor.node();

xtask/Cargo.toml

@@ -0,0 +1,19 @@
+[package]
+name = "xtask"
+version = "0.1.0"
+publish = false
+edition.workspace = true
+license.workspace = true
+authors.workspace = true
+
+[dependencies]
+anyhow = "1"
+flate2 = "1"
+write-json = "0.1"
+xshell = "0.2"
+xflags = "0.3"
+time = { version = "0.3", default-features = false }
+zip = { version = "0.6", default-features = false, features = [
+  "deflate",
+  "time",
+] }

xtask/src/dist.rs

@@ -0,0 +1,147 @@
+use std::{
+    env,
+    fs::File,
+    io::{self, BufWriter},
+    path::{Path, PathBuf},
+};
+
+use flate2::{write::GzEncoder, Compression};
+use time::OffsetDateTime;
+use xshell::{cmd, Shell};
+use zip::{write::FileOptions, DateTime, ZipWriter};
+
+use crate::{flags, project_root};
+
+impl flags::Dist {
+    pub(crate) fn run(self, sh: &Shell) -> anyhow::Result<()> {
+        let branch = sh.var("GITHUB_REF").unwrap_or_default();
+        let release = if branch.starts_with("refs/heads/release/") {
+            branch.replace("refs/heads/release/", "")
+        } else {
+            "0.0.0".to_owned()
+        };
+        let project_root = project_root();
+        let target = Target::get(&project_root);
+        let dist = project_root.join("dist");
+        sh.remove_path(&dist)?;
+        sh.create_dir(&dist)?;
+
+        dist_server(sh, &release, &target)?;
+        Ok(())
+    }
+}
+
+fn dist_server(sh: &Shell, release: &str, target: &Target) -> anyhow::Result<()> {
+    let _e = sh.push_env("CFG_RELEASE", release);
+    let _e = sh.push_env("CARGO_PROFILE_RELEASE_LTO", "thin");
+
+    // Uncomment to enable debug info for releases. Note that:
+    //   * debug info is split on windows and macs, so it does nothing for those platforms,
+    //   * on Linux, this blows up the binary size from 8MB to 43MB, which is unreasonable.
+    // let _e = sh.push_env("CARGO_PROFILE_RELEASE_DEBUG", "1");
+
+    if target.name.contains("-linux-") {
+        env::set_var("CC", "clang");
+    }
+
+    let target_name = &target.name;
+    cmd!(sh, "cargo build --manifest-path ./crates/lsp-ai/Cargo.toml --bin lsp-ai --target {target_name} --release").run()?;
+
+    let dst = Path::new("dist").join(&target.artifact_name);
+    gzip(&target.server_path, &dst.with_extension("gz"))?;
+    if target_name.contains("-windows-") {
+        zip(
+            &target.server_path,
+            target.symbols_path.as_ref(),
+            &dst.with_extension("zip"),
+        )?;
+    }
+
+    Ok(())
+}
+
+fn gzip(src_path: &Path, dest_path: &Path) -> anyhow::Result<()> {
+    let mut encoder = GzEncoder::new(File::create(dest_path)?, Compression::best());
+    let mut input = io::BufReader::new(File::open(src_path)?);
+    io::copy(&mut input, &mut encoder)?;
+    encoder.finish()?;
+    Ok(())
+}
+
+fn zip(src_path: &Path, symbols_path: Option<&PathBuf>, dest_path: &Path) -> anyhow::Result<()> {
+    let file = File::create(dest_path)?;
+    let mut writer = ZipWriter::new(BufWriter::new(file));
+    writer.start_file(
+        src_path.file_name().unwrap().to_str().unwrap(),
+        FileOptions::default()
+            .last_modified_time(
+                DateTime::try_from(OffsetDateTime::from(
+                    std::fs::metadata(src_path)?.modified()?,
+                ))
+                .unwrap(),
+            )
+            .unix_permissions(0o755)
+            .compression_method(zip::CompressionMethod::Deflated)
+            .compression_level(Some(9)),
+    )?;
+    let mut input = io::BufReader::new(File::open(src_path)?);
+    io::copy(&mut input, &mut writer)?;
+    if let Some(symbols_path) = symbols_path {
+        writer.start_file(
+            symbols_path.file_name().unwrap().to_str().unwrap(),
+            FileOptions::default()
+                .last_modified_time(
+                    DateTime::try_from(OffsetDateTime::from(
+                        std::fs::metadata(src_path)?.modified()?,
+                    ))
+                    .unwrap(),
+                )
+                .compression_method(zip::CompressionMethod::Deflated)
+                .compression_level(Some(9)),
+        )?;
+        let mut input = io::BufReader::new(File::open(symbols_path)?);
+        io::copy(&mut input, &mut writer)?;
+    }
+    writer.finish()?;
+    Ok(())
+}
+
+struct Target {
+    name: String,
+    server_path: PathBuf,
+    symbols_path: Option<PathBuf>,
+    artifact_name: String,
+}
+
+impl Target {
+    fn get(project_root: &Path) -> Self {
+        let name = match env::var("LSP_AI_TARGET") {
+            Ok(target) => target,
+            _ => {
+                if cfg!(target_os = "linux") {
+                    "x86_64-unknown-linux-gnu".to_string()
+                } else if cfg!(target_os = "windows") {
+                    "x86_64-pc-windows-msvc".to_string()
+                } else if cfg!(target_os = "macos") {
+                    "x86_64-apple-darwin".to_string()
+                } else {
+                    panic!("Unsupported OS, maybe try setting LSP_AI_TARGET")
+                }
+            }
+        };
+        let out_path = project_root.join("target").join(&name).join("release");
+        let (exe_suffix, symbols_path) = if name.contains("-windows-") {
+            (".exe".into(), Some(out_path.join("lsp_ai.pdb")))
+        } else {
+            (String::new(), None)
+        };
+        let server_path = out_path.join(format!("lsp-ai{exe_suffix}"));
+        let artifact_name = format!("lsp-ai-{name}{exe_suffix}");
+        Self {
+            name,
+            server_path,
+            symbols_path,
+            artifact_name,
+        }
+    }
+}

xtask/src/flags.rs

@@ -0,0 +1,43 @@
+#![allow(unreachable_pub)]
+
+xflags::xflags! {
+    src "./src/flags.rs"
+
+    /// Run custom build command.
+    cmd xtask {
+        cmd dist {}
+    }
+}
+
+// generated start
+// The following code is generated by `xflags` macro.
+// Run `env UPDATE_XFLAGS=1 cargo build` to regenerate.
+#[derive(Debug)]
+pub struct Xtask {
+    pub subcommand: XtaskCmd,
+}
+
+#[derive(Debug)]
+pub enum XtaskCmd {
+    Dist(Dist),
+}
+
+#[derive(Debug)]
+pub struct Dist;
+
+impl Xtask {
+    #[allow(dead_code)]
+    pub fn from_env_or_exit() -> Self {
+        Self::from_env_or_exit_()
+    }
+
+    #[allow(dead_code)]
+    pub fn from_env() -> xflags::Result<Self> {
+        Self::from_env_()
+    }
+
+    #[allow(dead_code)]
+    pub fn from_vec(args: Vec<std::ffi::OsString>) -> xflags::Result<Self> {
+        Self::from_vec_(args)
+    }
+}

xtask/src/main.rs

@@ -0,0 +1,44 @@
+//! See <https://github.com/matklad/cargo-xtask/>.
+//!
+//! This binary defines various auxiliary build commands, which are not
+//! expressible with just `cargo`.
+//!
+//! This binary is integrated into the `cargo` command line by using an alias in
+//! `.cargo/config`.
+
+#![warn(
+    rust_2018_idioms,
+    unused_lifetimes,
+    semicolon_in_expressions_from_macros
+)]
+
+mod flags;
+
+mod dist;
+
+use std::{
+    env,
+    path::{Path, PathBuf},
+};
+use xshell::Shell;
+
+fn main() -> anyhow::Result<()> {
+    let flags = flags::Xtask::from_env_or_exit();
+
+    let sh = &Shell::new()?;
+    sh.change_dir(project_root());
+
+    match flags.subcommand {
+        flags::XtaskCmd::Dist(cmd) => cmd.run(sh),
+    }
+}
+
+fn project_root() -> PathBuf {
+    Path::new(
+        &env::var("CARGO_MANIFEST_DIR").unwrap_or_else(|_| env!("CARGO_MANIFEST_DIR").to_owned()),
+    )
+    .ancestors()
+    .nth(1)
+    .unwrap()
+    .to_path_buf()
+}