refactor: remove SKIP_LLAMA_CPP_BUILD argument and related logic from workflows and configuration

.github/workflows/deploy-prod.yml

@@ -49,7 +49,6 @@ jobs:
             APP_BASE=/chat
             PUBLIC_APP_COLOR=yellow
             PUBLIC_COMMIT_SHA=${{ env.GITHUB_SHA_SHORT }}
-            SKIP_LLAMA_CPP_BUILD=true
   deploy:
     name: Deploy on prod
     runs-on: ubuntu-latest

.github/workflows/lint-and-test.yml

@@ -53,7 +53,6 @@ jobs:
         run: |
           docker build \
             --build-arg INCLUDE_DB=true \
-            --build-arg SKIP_LLAMA_CPP_BUILD=true \
             -t chat-ui-test:latest .
 
       - name: Run Docker container

Dockerfile

@@ -52,9 +52,7 @@ COPY --link --chown=1000 package-lock.json package.json ./
 
 ARG APP_BASE=
 ARG PUBLIC_APP_COLOR=blue
-ARG SKIP_LLAMA_CPP_BUILD
 ENV BODY_SIZE_LIMIT=15728640
-ENV SKIP_LLAMA_CPP_BUILD=$SKIP_LLAMA_CPP_BUILD
 
 RUN --mount=type=cache,target=/app/.npm \
     npm set cache /app/.npm && \

vite.config.ts

@@ -2,10 +2,6 @@ import { sveltekit } from "@sveltejs/kit/vite";
 import Icons from "unplugin-icons/vite";
 import { promises } from "fs";
 import { defineConfig } from "vitest/config";
-import { resolve } from "path";
-import fs from "fs-extra";
-import { spawn } from "child_process";
-import type { Plugin } from "vite";
 
 // used to load fonts server side for thumbnail generation
 function loadTTFAsArrayBuffer() {
@@ -20,68 +16,6 @@ function loadTTFAsArrayBuffer() {
 		},
 	};
 }
-const isViteNode = process.argv.some((arg) => arg.includes("vite-node")) || !!process.env.VITE_NODE;
-const skipLlamaCppBuild = process.env.SKIP_LLAMA_CPP_BUILD === "true";
-const shouldCopyLlama =
-	process.env.npm_lifecycle_event === "build" && !isViteNode && !skipLlamaCppBuild; // Copy node-llama-cpp/llama files to build output
-
-function copyLlamaFiles() {
-	return {
-		name: "copy-llama-files",
-		apply: "build" as const,
-		closeBundle: async () => {
-			try {
-				// Run npx command first and pipe IO
-				console.log("Running node-llama-cpp source download...");
-
-				await new Promise((resolve, reject) => {
-					const npxProcess = spawn("npx", ["--no", "node-llama-cpp", "source", "download"], {
-						stdio: "inherit", // Pipe all IO to parent process
-						shell: true,
-					});
-
-					npxProcess.on("close", (code) => {
-						if (code === 0) {
-							console.log("✓ Successfully downloaded llama source files");
-							resolve(code);
-						} else {
-							reject(new Error(`npx command failed with code ${code}`));
-						}
-					});
-
-					npxProcess.on("error", (err) => {
-						reject(err);
-					});
-				});
-
-				const sourcePath = resolve("node_modules/node-llama-cpp/llama");
-				const destPath = resolve("build/server/llama");
-
-				// Ensure destination directory exists
-				await fs.ensureDir(destPath);
-
-				// Copy files - using a filter to prevent copying files to subdirectories of themselves
-				await fs.copy(sourcePath, destPath, {
-					filter: (src, dest) => {
-						// Skip if source path is inside destination path or vice versa
-						if (src.includes(destPath) || dest.includes(sourcePath)) {
-							console.log(`Skipping problematic copy: ${src} -> ${dest}`);
-							return false;
-						}
-						return true;
-					},
-					overwrite: true,
-					dereference: true,
-				});
-
-				console.log("✓ Successfully copied llama files to build output");
-			} catch (error) {
-				console.error("Error in llama files process:", error);
-			}
-		},
-	} satisfies Plugin;
-}
-
 export default defineConfig({
 	plugins: [
 		sveltekit(),
@@ -89,7 +23,6 @@ export default defineConfig({
 			compiler: "svelte",
 		}),
 		loadTTFAsArrayBuffer(),
-		...(shouldCopyLlama ? [copyLlamaFiles()] : []),
 	],
 	// Allow external access via ngrok tunnel host
 	server: {