yet another basic demo for openai's node.js client (#2)

Co-authored-by: uyriq <uyriq@local+buntu>

Author: uyriq

examples/node-js/.env.sample

@@ -1,2 +1,3 @@
-# Create a file called ".env" and add this env var:
+# Create a file called ".env" and add these env vars:
 OPENAI_API_KEY=...insertme...
+PROXY_PATH=...do `wrangler publish` and place working url of your worker there...

examples/node-js/README.md

@@ -0,0 +1,23 @@
+### Example client usage
+
+With `example.js`, you can test how openai-caching-proxy-worker works by running `wrangler dev` in one terminal
+and `cd examples/node-js && node example.js`in another, then you can see in your local environment
+how open-ai requests are executed using redis caching.
+If you want to publish your local instance of openai-caching-proxy-worker to Cloudflare after running `wrangler publish`
+you will get a working url, then add it in your local variable`.env` as PROXY_PATH and then you can try `example2.js`,
+it has command line options, note that order is important
+
+1. option model < text-ada-001 | text-davinci-003 >
+2. the 'cacheYES' option <boolean> : string|any
+3. option prompt - your question to chatGPT <string>
+
+so execute like this
+`node example2.js text-davinci-003 cacheNO 'is any difference between white dogs and black mice?'`
+
+```markdown
+R fresh:
+Q:
+is any difference between white dogs and black mice?
+R:
+Yes, there is a difference between white dogs and black mice. White...
+```

examples/node-js/example2.js

@@ -0,0 +1,78 @@
+// yet another basic demo usage for openai's node.js client with command line args and non-local worker.
+// Run with:
+// node example2.js
+// With example.js, you can test how openai-caching-proxy-worker works by running `wrangler dev` in one terminal and
+// `cd examples/node-js && node example.js` in another, then you can see in your local environment how requests
+// to open-ai are executed using redis caching.
+// If you want to publish your local instance of openai-caching-proxy-worker to Cloudflare after running `wrangler publish`
+// you will get a working url, add it in your local variable .env PROXY_PATH and then you can try `example2.js`, it has command line options, note that order is important:
+// 1. option model < text-ada-001 | text-davinci-003 | ...any suported by openAI >
+// 2. the 'cacheYES' option <boolean> : string|any
+// 3. option prompt - your question to chatGPT <string>
+
+import * as dotenv from 'dotenv';
+import { Configuration, OpenAIApi } from 'openai';
+
+dotenv.config();
+
+const apiKey = process.env.OPENAI_API_KEY;
+const proxyPath = process.env.PROXY_PATH;
+if (!apiKey) {
+  console.error('Error: OPENAI_API_KEY must be set');
+  process.exit(1);
+}
+
+// cli args are following: 1. option model < text-ada-001 | text-davinci-003 > 2. option 'cacheYES' <boolean> : string|any 3. option prompt - question to chatGPT <string>
+const args = process.argv.slice(2);
+let model = '';
+const prompt = '' || String(args.at(-1));
+const cache = String(args.at(1)) === 'cacheYES' ? true : false;
+
+const configuration = new Configuration({
+  apiKey,
+  // Set this to your local instance or Cloudflare deployment after doing `wrangler publish` and got cloudflare worker url to local .env PROXY_PATH variable
+  basePath: proxyPath || `http://localhost:8787/proxy`,
+  baseOptions: {
+    headers: {
+      // Cache responses for 3600 seconds (1 hour)
+      'X-Proxy-TTL': 3600,
+      // If you need to force refresh cache, you can uncomment below:
+      'X-Proxy-Refresh': !cache,
+    },
+  },
+});
+const openai = new OpenAIApi(configuration);
+
+switch (args[0]) {
+  case 'text-ada-001':
+    // cheap
+    model = 'text-ada-001';
+    break;
+  case 'text-davinci-003':
+    // overprice
+    model = 'text-davinci-003';
+    break;
+  default:
+    break;
+}
+
+const makeSampleRequests = async () => {
+  const completionOpts = {
+    model: model || 'text-ada-001',
+    prompt: prompt || 'HOW MUCH IS 2 + 2 * 2 ?',
+  };
+  const completion = await openai.createCompletion(completionOpts);
+  console.log(
+    `${cache ? 'R from chache' : 'R fresh'}:\nQ:\n ${
+      completionOpts.prompt
+    }\nR:\n${completion.data.choices[0].text.trim()}`,
+  );
+};
+
+const main = async () => {
+  // The first time these requests are made, they should
+  // be proxied as-is to OpenAI API:
+  await makeSampleRequests();
+};
+
+main();

examples/node-js/package.json

@@ -15,3 +15,4 @@
     "openai": "^3.1.0"
   }
 }
+