Merge pull request #45 from phillip-kruger/main

Allow OpenAI custom server
quarkiverse · Sep 12, 2024 · d52aa7e · d52aa7e
2 parents fe9445f + af574c6
commit d52aa7e
Show file tree

Hide file tree

Showing 8 changed files with 76 additions and 28 deletions.
diff --git a/README.md b/README.md
@@ -17,7 +17,7 @@ extension to you project, example in maven:
 
 Chappie is a Dev Mode only extension, so this does not add anything to your production application.
 
-To use chappie, you need to configure it with either an OpenAI Api Key or have Ollama running locally.
+To use chappie, you need to configure it with either an OpenAI Compatible Service or have Ollama running locally.
 
 ## Using OpenAI
 To use OpenAI you need to provide an [OpenAI Api Key](https://help.openai.com/en/articles/4936850-where-do-i-find-my-openai-api-key) 
@@ -29,10 +29,35 @@ Example:
 mvn quarkus:dev -Dquarkus.assistant.openai.api-key=sk....
 ```
 
+## Using Podman Desktop AI
+You can use Podman Desktop AI by setting the openai base-url and the api key to a dummy value.
+
+Example:
+
+```
+mvn quarkus:dev -Dquarkus.assistant.openai.api-key=sk-dummy -Dquarkus.assistant.openai.base-url=http://localhost:46311/v1 -Dquarkus.assistant.openai.model-name=instructlab/granite-7b-lab-GGUF
+```
+
+Change the values to your own setup.
+
+# Using MAAS
+
+You can also use [Models as a Service on OpenShift AI](https://maas.apps.prod.rhoai.rh-aiservices-bu.com/)
+
+Example:
+
+```
+mvn quarkus:dev -Dquarkus.assistant.openai.api-key=your-key-here -Dquarkus.assistant.openai.base-url=quarkus.assistant.openai.base-url=https://granite-8b-code-instruct-maas-apicast-production.apps.prod.rhoai.rh-aiservices-bu.com:443/v1 -Dquarkus.assistant.openai.model-name=granite-8b-code-instruct-128k
+```
+
+Change the values to your own setup.
+
 ## Using Ollama
 To use Ollama you need to install and run ollama. See [ollama.com/download](https://ollama.com/download)
 
-By default, Ollama will use the `codellama` model. You can configure this with `quarkus.assistant.ollama.model-name` property, example:
+By default, Ollama will use the `codellama` model. You can configure this with `quarkus.assistant.ollama.model-name` property.
+
+Example:
 
 ```
 quarkus.assistant.ollama.model-name=instructlab/granite-7b-lab

diff --git a/deployment/pom.xml b/deployment/pom.xml
@@ -11,7 +11,7 @@
     <name>Quarkus Chappie - Deployment</name>
 
     <properties>
-        <chappie-server.version>0.0.12</chappie-server.version>
+        <chappie-server.version>0.0.13</chappie-server.version>
     </properties>
 
     <dependencies>

diff --git a/deployment/src/main/java/io/quarkiverse/chappie/deployment/ChappieDevUIProcessor.java b/deployment/src/main/java/io/quarkiverse/chappie/deployment/ChappieDevUIProcessor.java
@@ -20,7 +20,7 @@ void pages(List<ChappiePageBuildItem> chappiePageBuildItems,
             BuildProducer<CardPageBuildItem> cardPageProducer,
             ChappieConfig config) {
 
-        if (config.openai().apiKey().isPresent()) {
+        if (config.openai().apiKey().isPresent() || config.openai().baseUrl().isPresent()) {
             configuredOpenAiPage(chappiePageBuildItems, cardPageProducer, config);
         } else if (ollamaBuildItem.isPresent()) {
             configuredOllamaPage(chappiePageBuildItems, cardPageProducer, config);
@@ -32,7 +32,9 @@ void pages(List<ChappiePageBuildItem> chappiePageBuildItems,
     private void configuredOpenAiPage(List<ChappiePageBuildItem> chappiePageBuildItems,
             BuildProducer<CardPageBuildItem> cardPageProducer,
             ChappieConfig config) {
-        configuredPage(chappiePageBuildItems, cardPageProducer, "OpenAI", config.openai().modelName());
+
+        configuredPage(chappiePageBuildItems, cardPageProducer, "OpenAI ",
+                config.openai().baseUrl().orElse(config.openai().modelName()));
     }
 
     private void configuredOllamaPage(List<ChappiePageBuildItem> chappiePageBuildItems,

diff --git a/...rc/main/java/io/quarkiverse/chappie/deployment/devservice/ChappieDevServiceProcessor.java b/...rc/main/java/io/quarkiverse/chappie/deployment/devservice/ChappieDevServiceProcessor.java
@@ -44,7 +44,8 @@ public void createContainer(BuildProducer<DevServicesResultBuildItem> devService
             Optional<OllamaBuildItem> ollamaBuildItem,
             ChappieConfig config) {
 
-        if (process == null && (config.openai().apiKey().isPresent() || ollamaBuildItem.isPresent())) {
+        if (process == null && (config.openai().apiKey().isPresent() || config.openai().baseUrl().isPresent()
+                || ollamaBuildItem.isPresent())) {
 
             Map<String, String> properties = new HashMap<>();
             int port = findAvailablePort(4315);
@@ -54,14 +55,25 @@ public void createContainer(BuildProducer<DevServicesResultBuildItem> devService
             properties.put("quarkus.http.host", "localhost");
             properties.put("quarkus.http.port", String.valueOf(port));
 
-            if (config.openai().apiKey().isPresent()) {
-                properties.put("chappie.openai.api-key", config.openai().apiKey().get());
+            if (config.devservices().log()) {
+                properties.put("chappie.log.request", "true");
+                properties.put("chappie.log.response", "true");
+            }
+
+            properties.put("chappie.timeout", config.devservices().timeout());
+            if (config.openai().apiKey().isPresent() || config.openai().baseUrl().isPresent()) {
+                config.openai().apiKey().ifPresent((t) -> {
+                    properties.put("chappie.openai.api-key", t);
+                });
+                config.openai().baseUrl().ifPresent((t) -> {
+                    properties.put("chappie.openai.base-url", t);
+                });
+
                 properties.put("chappie.openai.model-name", config.openai().modelName());
+
             } else if (ollamaBuildItem.isPresent()) {
                 properties.put("chappie.ollama.base-url", ollamaBuildItem.get().getUrl());
                 properties.put("chappie.ollama.model-name", config.ollama().modelName());
-                properties.put("chappie.ollama.timeout", config.ollama().timeout());
-
             }
 
             String extVersion = extensionVersionBuildItem.getVersion();
@@ -96,7 +108,6 @@ public void createContainer(BuildProducer<DevServicesResultBuildItem> devService
         }
 
         chappieClientProducer.produce(new ChappieClientBuildItem(chappieClient));
-
     }
 
     @BuildStep
@@ -204,10 +215,7 @@ private void runServer(String version, Map<String, String> properties, boolean l
         for (Map.Entry<String, String> es : properties.entrySet()) {
             command.add("-D" + es.getKey() + "=" + es.getValue());
         }
-        if (log) {
-            command.add("-Dchappie.log.request=true");
-            command.add("-Dchappie.log.response=true");
-        }
+
         command.add("-jar");
         command.add(chappieServer.toString());
 

diff --git a/deployment/src/main/java/io/quarkiverse/chappie/deployment/devservice/DevServicesConfig.java b/deployment/src/main/java/io/quarkiverse/chappie/deployment/devservice/DevServicesConfig.java
@@ -20,12 +20,6 @@ public interface DevServicesConfig {
      */
     Optional<Integer> port();
 
-    /**
-     * The version to use
-     */
-    @WithDefault("0.0.12")
-    String version();
-
     /**
      * The base url for chappie
      */
@@ -36,4 +30,9 @@ public interface DevServicesConfig {
      */
     Optional<Long> processId();
 
+    /**
+     * Timeout for the request
+     */
+    @WithDefault("PT120S")
+    String timeout();
 }
diff --git a/deployment/src/main/java/io/quarkiverse/chappie/deployment/devservice/OpenAIConfig.java b/deployment/src/main/java/io/quarkiverse/chappie/deployment/devservice/OpenAIConfig.java
@@ -13,6 +13,11 @@ public interface OpenAIConfig {
      */
     Optional<String> apiKey();
 
+    /**
+     * Your OpenAI Base Url
+     */
+    Optional<String> baseUrl();
+
     /**
      * The Model to use
      */

diff --git a/...yment/src/main/java/io/quarkiverse/chappie/deployment/devservice/ollama/OllamaConfig.java b/...yment/src/main/java/io/quarkiverse/chappie/deployment/devservice/ollama/OllamaConfig.java
@@ -30,10 +30,4 @@ public interface OllamaConfig {
     @WithDefault("true")
     boolean preload();
 
-    /**
-     * Timeout for the request
-     */
-    @WithDefault("PT120S")
-    String timeout();
-
 }
diff --git a/sample/src/main/resources/application.properties b/sample/src/main/resources/application.properties
@@ -1 +1,16 @@
-#quarkus.assistant.ollama.model-name=instructlab/granite-7b-lab
+#quarkus.assistant.devservices.log=true
+
+# Local podman ai
+#quarkus.assistant.openai.base-url=http://localhost:46311/v1
+#quarkus.assistant.openai.model-name=instructlab/granite-7b-lab-GGUF
+#quarkus.assistant.openai.api-key=sk-dummy
+
+# Mike's server
+#quarkus.assistant.openai.base-url=https://sno-llama31-llama-serving.apps.sno.sandbox1771.opentlc.com/v1
+#quarkus.assistant.openai.model-name=Meta-Llama-3.1-8B-Instruct-Q8_0.gguf
+#quarkus.assistant.openai.api-key=sk-dummy
+
+# MOOS
+#quarkus.assistant.openai.base-url=https://granite-8b-code-instruct-maas-apicast-production.apps.prod.rhoai.rh-aiservices-bu.com:443/v1
+#quarkus.assistant.openai.model-name=granite-8b-code-instruct-128k
+#quarkus.assistant.openai.api-key=your-key-here