Add some llama.cpp-specific endpoints

src/endpoints/backends/text-completions.js

@@ -473,6 +473,76 @@ llamacpp.post('/caption-image', jsonParser, async function (request, response) {
     }
 });
 
+llamacpp.post('/props', jsonParser, async function (request, response) {
+    try {
+        if (!request.body.server_url) {
+            return response.sendStatus(400);
+        }
+
+        console.log('LlamaCpp props request:', request.body);
+        const baseUrl = trimV1(request.body.server_url);
+
+        const fetchResponse = await fetch(`${baseUrl}/props`, {
+            method: 'GET',
+            timeout: 0,
+        });
+
+        if (!fetchResponse.ok) {
+            console.log('LlamaCpp props error:', fetchResponse.status, fetchResponse.statusText);
+            return response.status(500).send({ error: true });
+        }
+
+        const data = await fetchResponse.json();
+        console.log('LlamaCpp props response:', data);
+
+        return response.send(data);
+
+    } catch (error) {
+        console.error(error);
+        return response.status(500);
+    }
+});
+
+llamacpp.post('/slots', jsonParser, async function (request, response) {
+    try {
+        if (!request.body.server_url) {
+            return response.sendStatus(400);
+        }
+        if (!/^\d+$/.test(request.body.id_slot)) {
+            return response.sendStatus(400);
+        }
+        if (!/^(erase|restore|save)$/.test(request.body.action)) {
+            return response.sendStatus(400);
+        }
+
+        console.log('LlamaCpp slots request:', request.body);
+        const baseUrl = trimV1(request.body.server_url);
+
+        const fetchResponse = await fetch(`${baseUrl}/slots/${request.body.id_slot}?action=${request.body.action}`, {
+            method: 'POST',
+            headers: { 'Content-Type': 'application/json' },
+            timeout: 0,
+            body: JSON.stringify({
+                filename: `${request.body.filename}`,
+            }),
+        });
+
+        if (!fetchResponse.ok) {
+            console.log('LlamaCpp slots error:', fetchResponse.status, fetchResponse.statusText);
+            return response.status(500).send({ error: true });
+        }
+
+        const data = await fetchResponse.json();
+        console.log('LlamaCpp slots response:', data);
+
+        return response.send(data);
+
+    } catch (error) {
+        console.error(error);
+        return response.status(500);
+    }
+});
+
 router.use('/ollama', ollama);
 router.use('/llamacpp', llamacpp);