Mvr/#15/conversation handler

src/app.py

@@ -8,8 +8,8 @@
 from src.modules.events import get_events
 from src.modules.factory import bind_deployment_handles
 from src.modules.modules import get_modules
-from src.modules.rag.docker_services import OllamaService, QdrantService
-
+from src.services.ollama import build_ollama
+from src.services.qdrant import build_qdrant
 
 def load_services_config() -> Any:
     config_path = Path(__file__).resolve().parents[1] / "config" / "huri.yaml"
@@ -18,24 +18,6 @@ def load_services_config() -> Any:
     return config.get("services", {})
 
 
-def build_qdrant(config: dict) -> Any:
-    return QdrantService.bind(  # type: ignore[attr-defined]
-        port=config.get("port", 6333),
-        image=config.get("image", "qdrant/qdrant:latest"),
-        storage_volume=config.get("storage_volume", "qdrant_data"),
-    )
-
-
-def build_ollama(config: dict) -> Any:
-    return OllamaService.options(  # type: ignore[attr-defined]
-        num_replicas=config.get("num_replicas", 1),
-    ).bind(
-        model=config.get("model", "mistral:7b"),
-        image=config.get("image", "ollama/ollama:latest"),
-        gpu_devices=config.get("gpu_devices", False),
-    )
-
-
 def build_app() -> Application:
     modules = get_modules()
     events = get_events()

src/core/client.py

@@ -1,12 +1,12 @@
 import asyncio
 import json
-import os
 from dataclasses import asdict
-from typing import Dict, List, Optional, Type
+from typing import Dict, List, Type
 
 import websockets
 
 from src.core.dataclasses.config import ClientConfig
+from src.core.user_config import get_or_create_user_id
 
 from .client_senders import ClientSender, get_senders
 
@@ -17,23 +17,13 @@ class Client:
     def __init__(
         self,
         config: ClientConfig,
-        user_id_file: str = os.path.expanduser("~/.huri_user_id"),
+        user_id_file: str | None = None,
         senders_dict: Dict[str, Type[ClientSender]] = get_senders(),
     ):
         self.config = config
         self.user_id_file = user_id_file
         self.senders_dict = senders_dict
 
-    def _load_user_id(self) -> Optional[str]:
-        if os.path.exists(self.user_id_file):
-            with open(self.user_id_file) as f:
-                return f.read().strip()
-        return None
-
-    def _save_user_id(self, _user_id: str):
-        with open(self.user_id_file, "w") as f:
-            f.write(_user_id)
-
     async def _receive_loop(self, ws: websockets.ClientConnection):
         try:
             while True:
@@ -48,7 +38,7 @@ async def run(self):
         async with websockets.connect(self.config.huri_url) as ws:
             print("Connected to server")
 
-            self.config.user_id = self._load_user_id()
+            self.config.user_id = get_or_create_user_id(self.user_id_file)
 
             senders: List[ClientSender] = [
                 self.senders_dict[config.name](ws=ws, **config.args)
@@ -60,7 +50,6 @@ async def run(self):
             init_msg = json.loads(await ws.recv())
             if init_msg.get("type") == "session_init":
                 user_id = init_msg["user_id"]
-                self._save_user_id(user_id)
                 print(f"Session started with _user_id: {user_id}")
 
             receive_task = asyncio.create_task(self._receive_loop(ws))

src/core/module.py

@@ -7,6 +7,7 @@ class Module:
     input_type: str
     output_type: Optional[str]
 
+
     async def process(self, _) -> Optional[Any]:
         raise NotImplementedError
 

src/core/user_config.py

@@ -0,0 +1,60 @@
+import os
+import platform
+import uuid
+from pathlib import Path
+
+
+def get_config_dir() -> Path:
+    """Cross-platform config directory."""
+    system = platform.system()
+
+    if system == "Windows":
+        # TODO: To be tested -> also consider language-specific if needed
+        base = os.environ.get("APPDATA", os.path.expanduser("~/AppData/Roaming"))
+    elif system == "Darwin":
+        # TODO: To be tested -> also consider language-specific if needed
+        base = os.path.expanduser("~/Library/Application Support")
+    else:
+        base = os.environ.get("XDG_CONFIG_HOME", os.path.expanduser("~/.config"))
+
+    config_dir = Path(base) / "huri"
+    config_dir.mkdir(parents=True, exist_ok=True)
+    return config_dir
+
+
+def load_user_id(path: str | None = None) -> str | None:
+    """Load existing _user_id, or return None if new user."""
+    id_file: Path
+
+    if path is None:
+        id_file = get_config_dir() / "_user_id"
+    else:
+        id_file = Path(path)
+    if id_file.exists():
+        uid = id_file.read_text().strip()
+        if uid:
+            return uid
+    return None
+
+
+def save_user_id(_user_id: str, path: str | None = None):
+    id_file: Path
+
+    if path is None:
+        id_file = get_config_dir() / "_user_id"
+    else:
+        id_file = Path(path)
+
+    id_file.write_text(_user_id)
+    if platform.system() != "Windows":
+        id_file.chmod(0o600)
+
+
+def get_or_create_user_id(path: str | None = None) -> str:
+    """Load existing or generate new _user_id."""
+    uid = load_user_id(path)
+    if uid:
+        return uid
+    uid = str(uuid.uuid4())
+    save_user_id(uid, path)
+    return uid

src/modules/rag/ingestion.py

@@ -1,5 +1,4 @@
 import argparse
-import os
 import re
 import sys
 import uuid
@@ -17,10 +16,10 @@
     PointStruct,
     VectorParams,
 )
-from semantic_chunker import SemanticChunker
 from sentence_transformers import SentenceTransformer
 
-USER_ID_FILE = os.path.expanduser("~/.huri_user_id")
+from src.core.user_config import get_or_create_user_id
+from src.modules.rag.semantic_chunker import SemanticChunker
 
 
 def _split_sentences(text: str) -> list[str]:
@@ -89,21 +88,6 @@ def extract_text_from_pdf(pdf_path: str) -> str:
     sys.exit(1)
 
 
-def get_user_id(provided_id: str | None = None) -> str:
-    if provided_id:
-        return provided_id
-    if os.path.exists(USER_ID_FILE):
-        with open(USER_ID_FILE) as f:
-            uid = f.read().strip()
-            if uid:
-                return uid
-    new_id = str(uuid.uuid4())
-    with open(USER_ID_FILE, "w") as f:
-        f.write(new_id)
-    print(f"Generated new user_id: {new_id}")
-    return new_id
-
-
 def ensure_collection(client: QdrantClient, collection: str, vector_size: int):
     collections = [c.name for c in client.get_collections().collections]
     if collection not in collections:
@@ -145,7 +129,6 @@ def ingest_chunks(
         )
 
     if points:
-        # Upsert in batches of 100
         batch_size = 100
         for i in range(0, len(points), batch_size):
             batch = points[i : i + batch_size]
@@ -403,7 +386,7 @@ def main():
 
     args = parser.parse_args()
 
-    _user_id = get_user_id(args._user_id)
+    _user_id = get_or_create_user_id()
     print(f"User: {_user_id}")
 
     client = QdrantClient(url=args.qdrant_url)
@@ -440,7 +423,7 @@ def main():
         # Ingest a text file
         python ingestion.py text notes.txt story.md
 
-        # Specify a user ID (otherwise reads from ~/.huri_user_id)
+        # Specify a user ID (otherwise it will be auto-generated and saved)
         python ingestion.py --user-id "abc-123" pdf report.pdf
 
         # Use a different collection