up

reme4/application.py

@@ -166,4 +166,7 @@ class Application(BaseComponent):
 
     def run_app(self):
         """Start the service and serve the application."""
+        if self.context.service is None:
+            raise RuntimeError("Service not configured")
+
         self.context.service.run_app(app=self)

reme4/components/embedding/base_embedding_model.py

@@ -62,22 +62,22 @@ class BaseEmbeddingModel(BaseComponent):
 
     # -- Public API --
 
-    async def get_embedding(self, input_text: str, **kwargs) -> list[float] | None:
+    async def get_embedding(self, input_text: str, **kwargs) -> np.ndarray | None:
         """Get embedding for a single text."""
         results = await self.get_embeddings([input_text], **kwargs)
         return results[0] if results else None
 
-    async def get_embeddings(self, input_text: list[str], **kwargs) -> list[list[float] | None]:
+    async def get_embeddings(self, input_text: list[str], **kwargs) -> list[np.ndarray | None]:
         """Get embeddings for a list of texts, with caching and batching."""
         truncated = [t[: self.max_input_length] for t in input_text]
-        results: list[list[float] | None] = [None] * len(truncated)
+        results: list[np.ndarray | None] = [None] * len(truncated)
         to_compute: list[tuple[int, str]] = []
 
         # Split into cache hits and misses
         for idx, text in enumerate(truncated):
             cached = self._get_from_cache(text)
             if cached is not None:
-                results[idx] = cached.tolist()
+                results[idx] = cached
             else:
                 to_compute.append((idx, text))
 
@@ -114,7 +114,7 @@ class BaseEmbeddingModel(BaseComponent):
                             emb_array = np.pad(emb_array, (0, self.dimensions - len(emb_array)))
                         else:
                             emb_array = emb_array[: self.dimensions]
-                    results[orig_idx] = emb_array.tolist()
+                    results[orig_idx] = emb_array
                     self._put_to_cache(text, emb_array)
 
         return results

reme4/components/file_store/local_file_store.py

@@ -68,12 +68,12 @@ class LocalFileStore(BaseFileStore):
             if old_node and self.embedding_model:
                 for cid in old_node.chunk_ids:
                     old = self.file_chunks.pop(cid, None)
-                    if old and old.embedding:
+                    if old and old.embedding is not None:
                         cached[cid] = old.embedding
 
             node.chunk_ids = []
             for c in chunks:
-                if self.embedding_model and not c.embedding:
+                if self.embedding_model and c.embedding is None:
                     if c.id in cached:
                         c.embedding = cached[c.id]
                     elif c.text:
@@ -120,7 +120,7 @@ class LocalFileStore(BaseFileStore):
             return []
 
         query_embedding = await self.embedding_model.get_embedding(query)
-        if not query_embedding:
+        if query_embedding is None:
             return []
 
         candidates = [c for c in self.file_chunks.values() if c.embedding is not None]

reme4/reme.py

@@ -15,7 +15,7 @@ class ReMe(Application):
 
 def main():
     """Parse CLI arguments and launch the appropriate mode."""
-    action, config = parse_args(sys.argv[1:])
+    action, config = parse_args(*sys.argv[1:])
     if action == "start":
         reme = ReMe(**config)
         reme.run_app()