chore: improved OpenAI + EDOT Node.js examples

.gitignore

@@ -1,3 +1,4 @@
+.env
 node_modules
 npm-debug.log*
 build

examples/openai/.npmrc

@@ -0,0 +1 @@
+package-lock=false

examples/openai/README.md

@@ -0,0 +1,64 @@
+# OpenAI Zero-Code Instrumentation Examples
+
+This is an example of how to instrument OpenAI calls with zero code changes,
+using `@elastic/opentelemetry-node` included in the Elastic Distribution of
+OpenTelemetry Node.js ([EDOT Node.js][edot-node]).
+
+When OpenAI examples run, they export traces, metrics and logs to an OTLP
+compatible endpoint. Traces and metrics include details such as the model used
+and the duration of the LLM request. In the case of chat, Logs capture the
+request and the generated response. The combination of these provide a
+comprehensive view of the performance and behavior of your OpenAI usage.
+
+## Install
+
+First, set up a Node.js environment for the examples like this:
+
+```bash
+nvm use --lts  # or similar to setup Node.js v20 or later
+npm install
+```
+
+## Configure
+
+Copy [env.example](env.example) to `.env` and update its `OPENAI_API_KEY`.
+
+An OTLP compatible endpoint should be listening for traces, metrics and logs on
+`http://localhost:4318`. If not, update `OTEL_EXPORTER_OTLP_ENDPOINT` as well.
+For example, if Elastic APM server is running locally, edit `.env` like this:
+```
+OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:8200
+```
+
+## Run
+
+There are two examples, and they run the same way:
+
+### Chat
+
+[chat.js](chat.js) asks the LLM a geography question and prints the response.
+
+Run it like this:
+```bash
+node --env-file .env --require @elastic/opentelemetry-node chat.js
+```
+
+You should see something like "Atlantic Ocean" unless your LLM hallucinates!
+
+### Embeddings
+
+
+[embeddings.js](embeddings.js) creates in-memory VectorDB embeddings about
+Elastic products. Then, it searches for one similar to a question.
+
+Run it like this:
+```bash
+node --env-file .env --require @elastic/opentelemetry-node embeddings.js
+```
+
+You should see something like "Connectors can help you connect to a database",
+unless your LLM hallucinates!
+
+---
+
+[edot-node]: https://github.com/elastic/elastic-otel-node/blob/main/packages/opentelemetry-node/README.md#install

examples/openai-chat.js → examples/openai/chat.js

@@ -17,20 +17,26 @@
  * under the License.
  */
 
-// Usage:
-//  OPENAI_API_KEY=sk-... \
-//      node -r @elastic/opentelemetry-node openai-chat.js
-
 const {OpenAI} = require('openai');
 
+let chatModel = process.env.CHAT_MODEL ?? 'gpt-4o-mini';
+
 async function main() {
-    const openai = new OpenAI();
-    const result = await openai.chat.completions.create({
-        model: 'gpt-4o-mini',
-        messages: [
-            {role: 'user', content: 'Why is the sky blue? Answer briefly.'},
-        ],
+    const client = new OpenAI();
+
+    const messages = [
+        {
+            role: 'user',
+            content:
+                'Answer in up to 3 words: Which ocean contains Bouvet Island?',
+        },
+    ];
+
+    const chatCompletion = await client.chat.completions.create({
+        model: chatModel,
+        messages: messages,
     });
-    console.log(result.choices[0]?.message?.content);
+    console.log(chatCompletion.choices[0].message.content);
 }
+
 main();

examples/openai/embeddings.js

@@ -0,0 +1,73 @@
+/*
+ * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
+ * or more contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright
+ * ownership. Elasticsearch B.V. licenses this file to you under
+ * the Apache License, Version 2.0 (the "License"); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *	http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+const {OpenAI} = require('openai');
+const {dot, norm} = require('mathjs');
+
+let embeddingsModel = process.env.EMBEDDINGS_MODEL ?? 'text-embedding-3-small';
+
+async function main() {
+    const client = new OpenAI();
+
+    const products = [
+        "Search: Ingest your data, and explore Elastic's machine learning and retrieval augmented generation (RAG) capabilities.",
+        'Observability: Unify your logs, metrics, traces, and profiling at scale in a single platform.',
+        'Security: Protect, investigate, and respond to cyber threats with AI-driven security analytics.',
+        'Elasticsearch: Distributed, RESTful search and analytics.',
+        'Kibana: Visualize your data. Navigate the Stack.',
+        'Beats: Collect, parse, and ship in a lightweight fashion.',
+        'Connectors: Connect popular databases, file systems, collaboration tools, and more.',
+        'Logstash: Ingest, transform, enrich, and output.',
+    ];
+
+    // Generate embeddings for each product. Keep them in an array instead of a vector DB.
+    const productEmbeddings = [];
+    for (const product of products) {
+        productEmbeddings.push(await createEmbedding(client, product));
+    }
+
+    const queryEmbedding = await createEmbedding(
+        client,
+        'What can help me connect to a database?'
+    );
+
+    // Calculate cosine similarity between the query and document embeddings
+    const similarities = productEmbeddings.map((productEmbedding) => {
+        return (
+            dot(queryEmbedding, productEmbedding) /
+            (norm(queryEmbedding) * norm(productEmbedding))
+        );
+    });
+
+    // Get the index of the most similar document
+    const mostSimilarIndex = similarities.indexOf(Math.max(...similarities));
+
+    console.log(products[mostSimilarIndex]);
+}
+
+async function createEmbedding(client, text) {
+    const response = await client.embeddings.create({
+        input: [text],
+        model: embeddingsModel,
+        encoding_format: 'float',
+    });
+    return response.data[0].embedding;
+}
+
+main();

examples/openai/env.example

@@ -0,0 +1,22 @@
+# Update this with your real OpenAI API key
+OPENAI_API_KEY=sk-YOUR_API_KEY
+
+# Uncomment to use Ollama instead of OpenAI
+# OPENAI_BASE_URL=http://localhost:11434/v1
+# OPENAI_API_KEY=unused
+# CHAT_MODEL=qwen2.5:0.5b
+# EMBEDDINGS_MODEL=all-minilm:33m
+
+# OTEL_EXPORTER_* variables are not required. If you would like to change your
+# OTLP endpoint to Elastic APM server using HTTP, uncomment the following:
+# OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:8200
+# OTEL_EXPORTER_OTLP_PROTOCOL=http/protobuf
+
+OTEL_SERVICE_NAME=openai-example
+OTEL_LOG_LEVEL=warn
+
+# Change to 'false' to hide prompt and completion content
+OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT=true
+# Change to affect behavior of which resources are detected. Note: these
+# choices are specific to the runtime, in this case Node.js.
+OTEL_NODE_RESOURCE_DETECTORS=container,env,host,os,serviceinstance,process,alibaba,aws,azure

examples/openai/package.json

@@ -0,0 +1,24 @@
+{
+  "name": "edot-node-openai-example",
+  "version": "1.0.0",
+  "private": true,
+  "type": "commonjs",
+  "engines": {
+    "node": ">=20"
+  },
+  "scripts": {
+    "chat": "node --env-file .env --require @elastic/opentelemetry-node chat.js",
+    "embeddings": "node --env-file .env --require @elastic/opentelemetry-node embeddings.js"
+  },
+  "dependencies": {
+    "@elastic/opentelemetry-node": "*",
+    "mathjs": "^14.0.1",
+    "openai": "^4.77.0"
+  },
+  "// overrides comment": "Override to avoid punycode warnings in recent versions of Node.JS",
+  "overrides": {
+    "[email protected]": {
+      "whatwg-url": "14.x"
+    }
+  }
+}