add basic setup for Ollama to support exercises for generating random items

Author: Caroline Ladek

backend/pom.xml

@@ -137,14 +137,18 @@
     <dependency>
       <groupId>io.quarkus</groupId>
       <artifactId>quarkus-rest-client-reactive-jackson</artifactId>
-      <scope>test</scope>
     </dependency>
     <dependency>
       <groupId>org.wiremock</groupId>
       <artifactId>wiremock</artifactId>
       <version>${wiremock.version}</version>
       <scope>test</scope>
     </dependency>
+    <dependency>
+      <groupId>io.quarkiverse.openapi.generator</groupId>
+      <artifactId>quarkus-openapi-generator</artifactId>
+      <version>${openapi.generator.version}</version>
+    </dependency>
   </dependencies>
   <build>
     <plugins>

backend/src/main/openapi/ollama_api.yaml

@@ -0,0 +1,96 @@
+openapi: 3.1.0
+info:
+  title: Ollama LLM API
+  description: API specification for interacting with a locally hosted Ollama model
+  version: 1.0.0
+servers:
+  - url: http://localhost:11434
+paths:
+  /api/generate:
+    post:
+      summary: Generate a response for a given prompt
+      description: |
+        Generates a response using the specified model. This is a streaming endpoint, meaning multiple responses may be returned.
+      operationId: queryLlm
+      requestBody:
+        required: true
+        content:
+          application/json:
+            schema:
+              type: object
+              required:
+                - model
+                - prompt
+              properties:
+                model:
+                  type: string
+                  description: The model name to use for generating responses.
+                prompt:
+                  type: string
+                  description: The input text for the model.
+                suffix:
+                  type: string
+                  description: The text to append after the model's response.
+                  nullable: true
+                images:
+                  type: array
+                  description: List of base64-encoded images (for multimodal models).
+                  items:
+                    type: string
+                format:
+                  type: object
+                  additionalProperties: true
+                  description: The format of the response. Can be `json` or a JSON schema.
+                options:
+                  type: object
+                  description: Additional model parameters from the Modelfile (e.g., temperature).
+                system:
+                  type: string
+                  description: Overrides the system message defined in the Modelfile.
+                template:
+                  type: string
+                  description: Overrides the prompt template defined in the Modelfile.
+                stream:
+                  type: boolean
+                  description: If false, returns a single response instead of a stream.
+                  default: true
+                raw:
+                  type: boolean
+                  description: If true, disables prompt formatting.
+                  default: false
+                keep_alive:
+                  type: string
+                  description: Controls how long the model stays loaded in memory (default 5m).
+                  example: "5m"
+                context:
+                  type: string
+                  description: (Deprecated) Context from a previous request to maintain short-term memory.
+                  nullable: true
+      responses:
+        "200":
+          description: Successful response with generated text.
+          content:
+            application/json:
+              schema:
+                type: object
+                properties:
+                  response:
+                    type: string
+                    description: The generated text response.
+                  model:
+                    type: string
+                    description: The model used for generation.
+                  usage:
+                    type: object
+                    description: Statistics on token usage.
+                    properties:
+                      prompt_tokens:
+                        type: integer
+                      completion_tokens:
+                        type: integer
+                      total_tokens:
+                        type: integer
+        "400":
+          description: Bad request due to invalid parameters.
+        "500":
+          description: Internal server error.

docker-compose.yaml

@@ -35,10 +35,27 @@ services:
     networks:
       - quarkus
 
+  ollama:
+    image: ollama/ollama:0.5.13
+    container_name: ollama
+    restart: unless-stopped
+    pull_policy: always
+    tty: true
+    volumes:
+      - ollama:/root/.ollama
+      - ./entrypoint.sh:/entrypoint.sh
+    ports:
+      - 11434:11434
+    networks:
+      - quarkus
+    entrypoint: [ "/usr/bin/bash", "/entrypoint.sh" ]
+    env_file: "./backend/.env"
+
 networks:
   quarkus:
     driver: bridge
 
 volumes:
+  ollama: {}
   db-data:
     driver: local

entrypoint.sh

@@ -0,0 +1,24 @@
+#!/bin/bash
+
+# Start Ollama in the background
+/bin/ollama serve &
+pid=$!
+
+# Wait for Ollama to start
+sleep 5
+
+# Check if the ollama model exists
+if ! ollama list | grep -q "^${OLLAMA_MODEL}"; then
+    echo "${OLLAMA_MODEL} model not found. Pulling now..."
+    ollama pull "${OLLAMA_MODEL}"
+    echo "Model download complete!"
+else
+    echo "${OLLAMA_MODEL} model is already available."
+fi
+
+# Run the model with the specified keep-alive time
+echo "Starting ${OLLAMA_MODEL} model with keep-alive time of ${OLLAMA_KEEP_ALIVE_TIME}..."
+ollama run "${OLLAMA_MODEL}" --keepalive "${OLLAMA_KEEP_ALIVE_TIME}"
+
+# Wait for Ollama process to finish
+wait $pid

pom.xml

@@ -24,6 +24,7 @@
     <quarkus.platform.group-id>io.quarkus.platform</quarkus.platform.group-id>
     <quarkus.platform.version>3.6.3</quarkus.platform.version>
     <querydsl.version>5.0.0</querydsl.version>
+    <openapi.generator.version>2.8.1</openapi.generator.version>
     <!-- Run integration tests by default -->
     <skipITs>false</skipITs>
     <surefire-plugin.version>3.0.0</surefire-plugin.version>