TriCacheLLM-MMA 0.1.6__tar.gz → 0.1.8__tar.gz

{tricachellm_mma-0.1.6 → tricachellm_mma-0.1.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: TriCacheLLM_MMA
-Version: 0.1.6
+Version: 0.1.8
 Summary: A robust three-tier LLM-aware caching SDK using Redis, ChromaDB, and automatic SQLite bootstrap.
 Author-email: Mohib Ashfaq Butt <inboxmohib@gmail.com>
 License-Expression: LGPL-3.0-only
@@ -424,7 +424,7 @@ Or install the development version directly from the repository:
 pip install .
 ```
 
-PyPI link: [https://pypi.org/project/TriCacheLLM-MMA/0.1.6/](https://pypi.org/project/TriCacheLLM-MMA/0.1.6/)​
+PyPI link: [https://pypi.org/project/TriCacheLLM-MMA/0.1.8/](https://pypi.org/project/TriCacheLLM-MMA/0.1.8/)​
 
 ---
 
@@ -478,6 +478,25 @@ Keep the Celery worker running while using the cache.
 
 ---
 
+### 📂 Model Caching Locations
+
+To ensure heavy Hugging Face embedding models don't re-download or clutter your project directory, `TriCacheLLM_MMA` automatically stores them in a stable, user-level system cache directory depending on your operating system:
+
+* **Windows:** 
+  `%LOCALAPPDATA%\TriCacheLLM_MMA\embedding_model\` 
+
+  *(Falls back to `C:\Users\<Username>\AppData\Local\TriCacheLLM_MMA\embedding_model\`)*
+
+* **macOS:** 
+  `~/Library/Caches/TriCacheLLM_MMA/embedding_model/`
+
+* **Linux / Unix:** 
+  `$XDG_CACHE_HOME/TriCacheLLM_MMA/embedding_model/` 
+  
+  *(Falls back to `~/.cache/TriCacheLLM_MMA/embedding_model/`)*
+
+---
+
 # Cohere
 
 The persistent cache tier uses Cohere reranking.
@@ -521,6 +540,10 @@ app = FastAPI(
 )
 
 
+# NOTE: Run celery first it will download the embedding model for you then initialize create_cache_system(user_id=0)
+# In upcoming versions, this process will be automated, so you won't need to manually run the Celery command.
+
+
 # 1. ADMIN SETUP (Run once on startup / first deploy with user_id=0)
 @app.post("/api/consumer/init")
 async def initialize_consumer():

{tricachellm_mma-0.1.6 → tricachellm_mma-0.1.8}/README.md

@@ -382,7 +382,7 @@ Or install the development version directly from the repository:
 pip install .
 ```
 
-PyPI link: [https://pypi.org/project/TriCacheLLM-MMA/0.1.6/](https://pypi.org/project/TriCacheLLM-MMA/0.1.6/)​
+PyPI link: [https://pypi.org/project/TriCacheLLM-MMA/0.1.8/](https://pypi.org/project/TriCacheLLM-MMA/0.1.8/)​
 
 ---
 
@@ -436,6 +436,25 @@ Keep the Celery worker running while using the cache.
 
 ---
 
+### 📂 Model Caching Locations
+
+To ensure heavy Hugging Face embedding models don't re-download or clutter your project directory, `TriCacheLLM_MMA` automatically stores them in a stable, user-level system cache directory depending on your operating system:
+
+* **Windows:** 
+  `%LOCALAPPDATA%\TriCacheLLM_MMA\embedding_model\` 
+
+  *(Falls back to `C:\Users\<Username>\AppData\Local\TriCacheLLM_MMA\embedding_model\`)*
+
+* **macOS:** 
+  `~/Library/Caches/TriCacheLLM_MMA/embedding_model/`
+
+* **Linux / Unix:** 
+  `$XDG_CACHE_HOME/TriCacheLLM_MMA/embedding_model/` 
+  
+  *(Falls back to `~/.cache/TriCacheLLM_MMA/embedding_model/`)*
+
+---
+
 # Cohere
 
 The persistent cache tier uses Cohere reranking.
@@ -479,6 +498,10 @@ app = FastAPI(
 )
 
 
+# NOTE: Run celery first it will download the embedding model for you then initialize create_cache_system(user_id=0)
+# In upcoming versions, this process will be automated, so you won't need to manually run the Celery command.
+
+
 # 1. ADMIN SETUP (Run once on startup / first deploy with user_id=0)
 @app.post("/api/consumer/init")
 async def initialize_consumer():

{tricachellm_mma-0.1.6 → tricachellm_mma-0.1.8}/TriCacheLLM_MMA/portable_cache_main.py

@@ -1,7 +1,9 @@
 # File: portable_cache_main.py
 import asyncio
+from asyncio import subprocess
 import hashlib
 import json
+import os
 from pathlib import Path
 from typing import Any
 from .portable_cache_schemas.portable_cache_dbConf import init_cache_database, init_db_tables, db_manager
@@ -29,6 +31,8 @@ from .portable_cache_redis import get_redis
 from .portable_cache_redis import redis_manager
 from .portable_cache_utils.protable_cache_DynamicEnv_maker import system_key
 from .portable_cache_dbSchema import Paths
+import sys
+
 
 def create_vector_index_schema(dim: int, distance_metric: str = "COSINE", m: int = 16, ef_construction: int = 200, ef_runtime: int = 10) -> list:
     return [
@@ -439,10 +443,49 @@ async def create_cache_system(
             if not key_res["success"]:
                 raise ValueError("Failed to initialize system keys. Verify if the keys are correct.")
             
+            
+            # This bit is for 0.1.9 (at import time embedding model will be loaded anyways so run this anytime u want) 
+            """
+            is_in_venv = sys.prefix != sys.base_prefix
+            celery_name = "celery.exe" if os.name == "nt" else "celery"
+            celery_executable = Path(sys.executable).parent / celery_name
+            
+            if is_in_venv and celery_executable.exists():
+                celery_cmd = [
+                    str(celery_executable),
+                    "-A",
+                    "TriCacheLLM_MMA.portable_cache_bgWorkers.portable_cache_celery_conf.celery_app",
+                    "worker",
+                    "--loglevel=info",
+                    "-Q",
+                    "ai"
+                ]
+                
+                # Check OS and launch accordingly
+                if os.name == "nt":
+                    subprocess.Popen(celery_cmd, creationflags=subprocess.CREATE_NEW_CONSOLE)
+                    print("Don't worry, TriCacheLLM_MMA is running celery background workers automatically")
+                    
+                elif sys.platform == "darwin":
+                    try:
+                        cmd_str = " ".join(celery_cmd)
+                        subprocess.Popen(["osascript", "-e", f'tell application "Terminal" to do script "{cmd_str}"'])
+                        print("Don't worry, TriCacheLLM_MMA is running celery background workers automatically")
+                    except Exception:
+                        subprocess.Popen(celery_cmd)    
+                        print("Don't worry, TriCacheLLM_MMA is running celery background workers automatically")
+                else:
+                    # Linux / Unix fallback (runs as a background child process)
+                    subprocess.Popen(celery_cmd)
+                    print("Don't worry, TriCacheLLM_MMA is running celery background workers automatically")
+            else:
+                print("---WARNING--- auto celery start failed manually run this command:\ncelery -A TriCacheLLM_MMA.portable_cache_bgWorkers.portable_cache_celery_conf.celery_app worker --loglevel=info -Q ai")
+            
+        """
+        
             return #user_id=0 doesnt need its own vdb! if you want to be it user-self be user_id=1 -> single user
                     #if you want multi-tanent then keep sending in user_ids lol
         
-        
         async with db_manager.async_session() as db:
             ans: None | dict = await create_cache_vdb_inishiator(user_id=user_id, db=db) 
             

tricachellm_mma-0.1.8/TriCacheLLM_MMA/portable_cache_utils/portable_cache_embedding_model.py

@@ -0,0 +1,63 @@
+# File: portable_cache_utils/portable_cache_embedding_model.py
+
+import os
+import sys
+from pathlib import Path
+from langchain_huggingface import HuggingFaceEmbeddings
+from ..portable_cache_utils.protable_cache_DynamicEnv_maker import get_settings
+
+
+def get_embedding_cache_dir() -> Path:
+    """
+    Return a stable per-user cache directory for the Hugging Face
+    embedding model, independent of the consumer application's location,
+    respecting OS conventions and XDG standards.
+    """
+    if os.name == "nt":
+        local_app_data = os.environ.get("LOCALAPPDATA")
+        if local_app_data:
+            return Path(local_app_data) / "TriCacheLLM_MMA" / "embedding_model"
+        return Path.home() / "AppData" / "Local" / "TriCacheLLM_MMA" / "embedding_model"
+
+    if sys.platform == "darwin":
+        return Path.home() / "Library" / "Caches" / "TriCacheLLM_MMA" / "embedding_model"
+
+    # Linux / Unix: Respect XDG_CACHE_HOME if explicitly configured
+    xdg_cache = os.environ.get("XDG_CACHE_HOME")
+    if xdg_cache:
+        return Path(xdg_cache) / "TriCacheLLM_MMA" / "embedding_model"
+    
+    return Path.home() / ".cache" / "TriCacheLLM_MMA" / "embedding_model"
+
+
+# Ensure the stable cache path exists prior to initialization
+embedding_cache_dir = get_embedding_cache_dir()
+embedding_cache_dir.mkdir(parents=True, exist_ok=True)
+
+
+#on each --reload this reads weights
+embedding_model = HuggingFaceEmbeddings(
+    model_name=get_settings().portable_cache_cache_proj_embedding_model,
+    cache_folder=str(embedding_cache_dir),
+)
+
+
+
+#lazy load, on --reload it will read weights but not at the start but when needed (un-comment me if u need lazyload)
+"""
+from langchain_huggingface import HuggingFaceEmbeddings
+from ..portable_cache_utils.protable_cache_DynamicEnv_maker import get_settings
+
+class EmbeddingModel:
+    def __init__(self):
+        self.model = None
+
+    def get_model(self):
+        if self.model is None:
+            self.model = HuggingFaceEmbeddings(
+                model_name=get_settings().portable_cache_cache_proj_embedding_model,
+                cache_folder=get_settings().portable_cache_chroma_db_dir,
+            )
+        return self.model
+embedding_manager = EmbeddingModel()
+"""

{tricachellm_mma-0.1.6 → tricachellm_mma-0.1.8}/TriCacheLLM_MMA.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: TriCacheLLM_MMA
-Version: 0.1.6
+Version: 0.1.8
 Summary: A robust three-tier LLM-aware caching SDK using Redis, ChromaDB, and automatic SQLite bootstrap.
 Author-email: Mohib Ashfaq Butt <inboxmohib@gmail.com>
 License-Expression: LGPL-3.0-only
@@ -424,7 +424,7 @@ Or install the development version directly from the repository:
 pip install .
 ```
 
-PyPI link: [https://pypi.org/project/TriCacheLLM-MMA/0.1.6/](https://pypi.org/project/TriCacheLLM-MMA/0.1.6/)​
+PyPI link: [https://pypi.org/project/TriCacheLLM-MMA/0.1.8/](https://pypi.org/project/TriCacheLLM-MMA/0.1.8/)​
 
 ---
 
@@ -478,6 +478,25 @@ Keep the Celery worker running while using the cache.
 
 ---
 
+### 📂 Model Caching Locations
+
+To ensure heavy Hugging Face embedding models don't re-download or clutter your project directory, `TriCacheLLM_MMA` automatically stores them in a stable, user-level system cache directory depending on your operating system:
+
+* **Windows:** 
+  `%LOCALAPPDATA%\TriCacheLLM_MMA\embedding_model\` 
+
+  *(Falls back to `C:\Users\<Username>\AppData\Local\TriCacheLLM_MMA\embedding_model\`)*
+
+* **macOS:** 
+  `~/Library/Caches/TriCacheLLM_MMA/embedding_model/`
+
+* **Linux / Unix:** 
+  `$XDG_CACHE_HOME/TriCacheLLM_MMA/embedding_model/` 
+  
+  *(Falls back to `~/.cache/TriCacheLLM_MMA/embedding_model/`)*
+
+---
+
 # Cohere
 
 The persistent cache tier uses Cohere reranking.
@@ -521,6 +540,10 @@ app = FastAPI(
 )
 
 
+# NOTE: Run celery first it will download the embedding model for you then initialize create_cache_system(user_id=0)
+# In upcoming versions, this process will be automated, so you won't need to manually run the Celery command.
+
+
 # 1. ADMIN SETUP (Run once on startup / first deploy with user_id=0)
 @app.post("/api/consumer/init")
 async def initialize_consumer():

{tricachellm_mma-0.1.6 → tricachellm_mma-0.1.8}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "TriCacheLLM_MMA"
-version = "0.1.6"
+version = "0.1.8"
 description = "A robust three-tier LLM-aware caching SDK using Redis, ChromaDB, and automatic SQLite bootstrap."
 readme = "README.md"
 requires-python = ">=3.11"

tricachellm_mma-0.1.6/TriCacheLLM_MMA/portable_cache_utils/portable_cache_embedding_model.py

@@ -1,31 +0,0 @@
-# File: portable_cache_utils/portable_cache_embedding_model.py
-
-#on each --reload this reads weights
-from langchain_huggingface import HuggingFaceEmbeddings
-from ..portable_cache_utils.protable_cache_DynamicEnv_maker import get_settings
-
-embedding_model = HuggingFaceEmbeddings(
-    model_name=get_settings().portable_cache_cache_proj_embedding_model,
-    cache_folder=get_settings().portable_cache_chroma_db_dir
-)
-
-
-
-#lazy load, on --reload it will read weights but not at the start but when needed (un-comment me if u need lazyload)
-"""
-from langchain_huggingface import HuggingFaceEmbeddings
-from ..portable_cache_utils.protable_cache_DynamicEnv_maker import get_settings
-
-class EmbeddingModel:
-    def __init__(self):
-        self.model = None
-
-    def get_model(self):
-        if self.model is None:
-            self.model = HuggingFaceEmbeddings(
-                model_name=get_settings().portable_cache_cache_proj_embedding_model,
-                cache_folder=get_settings().portable_cache_chroma_db_dir,
-            )
-        return self.model
-embedding_manager = EmbeddingModel()
-"""