feat(tests): use Qwen2.5 (Ollama) for invoice extraction tests and add helpers for model management; normalize dates and coerce numeric fields

test/helpers/docker.ts

@@ -295,3 +295,66 @@ export async function ensurePaddleOcrVlFull(): Promise<boolean> {
   }
   return ensureService(IMAGES.paddleocrVlFull);
 }
+
+/**
+ * Ensure an Ollama model is pulled and available
+ * Uses the MiniCPM container (which runs Ollama) to pull the model
+ */
+export async function ensureOllamaModel(modelName: string): Promise<boolean> {
+  const OLLAMA_URL = 'http://localhost:11434';
+
+  console.log(`\n[Ollama] Ensuring model: ${modelName}`);
+
+  // Check if model exists
+  try {
+    const response = await fetch(`${OLLAMA_URL}/api/tags`);
+    if (response.ok) {
+      const data = await response.json();
+      const models = data.models || [];
+      const exists = models.some((m: { name: string }) =>
+        m.name === modelName || m.name.startsWith(modelName.split(':')[0])
+      );
+
+      if (exists) {
+        console.log(`[Ollama] Model already available: ${modelName}`);
+        return true;
+      }
+    }
+  } catch {
+    console.log(`[Ollama] Cannot check models, Ollama may not be running`);
+    return false;
+  }
+
+  // Pull the model
+  console.log(`[Ollama] Pulling model: ${modelName} (this may take a while)...`);
+  try {
+    const response = await fetch(`${OLLAMA_URL}/api/pull`, {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify({ name: modelName, stream: false }),
+    });
+
+    if (response.ok) {
+      console.log(`[Ollama] Model pulled successfully: ${modelName}`);
+      return true;
+    } else {
+      console.log(`[Ollama] Failed to pull model: ${response.status}`);
+      return false;
+    }
+  } catch (err) {
+    console.log(`[Ollama] Error pulling model: ${err}`);
+    return false;
+  }
+}
+
+/**
+ * Ensure Qwen2.5 7B model is available (for text-only JSON extraction)
+ */
+export async function ensureQwen25(): Promise<boolean> {
+  // First ensure the Ollama service (MiniCPM container) is running
+  const ollamaOk = await ensureMiniCpm();
+  if (!ollamaOk) return false;
+
+  // Then ensure the Qwen2.5 model is pulled
+  return ensureOllamaModel('qwen2.5:7b');
+}