Add Files

Dockerfile

@@ -4,7 +4,12 @@ FROM python:3.10-slim
 ENV PYTHONDONTWRITEBYTECODE=1 \
     PYTHONUNBUFFERED=1 \
     PIP_NO_CACHE_DIR=1 \
-    CHROMA_DB_PATH=/data/chroma
+    CHROMA_DB_PATH=/data/chroma \
+    CACHE_ROOT=/data/cache \
+    HF_HOME=/data/cache/hf \
+    TRANSFORMERS_CACHE=/data/cache/transformers \
+    SENTENCE_TRANSFORMERS_HOME=/data/cache/sentence-transformers \
+    XDG_CACHE_HOME=/data/cache
 
 # System deps
 RUN apt-get update && apt-get install -y --no-install-recommends \
@@ -14,8 +19,9 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
 
 WORKDIR /app
 
-# Ensure writable path for Chroma
-RUN mkdir -p "$CHROMA_DB_PATH" && chmod -R 777 "$CHROMA_DB_PATH"
+# Ensure writable paths for Chroma and model caches
+RUN mkdir -p "$CHROMA_DB_PATH" "$CACHE_ROOT" "$HF_HOME" "$TRANSFORMERS_CACHE" "$SENTENCE_TRANSFORMERS_HOME" \
+    && chmod -R 777 "$CHROMA_DB_PATH" "$CACHE_ROOT"
 
 # Copy requirements first for better caching
 COPY requirements.txt ./

process_aware_rag.py

@@ -1,5 +1,7 @@
 # process_aware_rag.py
 import chromadb
+from chromadb.config import Settings
+from chromadb.utils import embedding_functions
 import google.generativeai as genai
 from query_classifier import QueryClassifier
 from graph_builder import LegalProcessGraph
@@ -16,13 +18,36 @@ class ProcessAwareRAG:
         # Initialize vector store (use writable path by default)
         chroma_path = os.getenv('CHROMA_DB_PATH', '/tmp/legal_vector_db')
         os.makedirs(chroma_path, exist_ok=True)
-        client = chromadb.PersistentClient(path=chroma_path)
+
+        # Redirect model caches to writable directories
+        default_cache_root = os.getenv('CACHE_ROOT', '/data/cache')
+        os.makedirs(default_cache_root, exist_ok=True)
+        os.environ.setdefault('HF_HOME', os.path.join(default_cache_root, 'hf'))
+        os.environ.setdefault('TRANSFORMERS_CACHE', os.path.join(default_cache_root, 'transformers'))
+        os.environ.setdefault('SENTENCE_TRANSFORMERS_HOME', os.path.join(default_cache_root, 'sentence-transformers'))
+        os.environ.setdefault('XDG_CACHE_HOME', default_cache_root)
+        for env_key in ['HF_HOME', 'TRANSFORMERS_CACHE', 'SENTENCE_TRANSFORMERS_HOME', 'XDG_CACHE_HOME']:
+            os.makedirs(os.environ[env_key], exist_ok=True)
+
+        # Disable Chroma anonymized telemetry and initialize client
+        client = chromadb.PersistentClient(
+            path=chroma_path,
+            settings=Settings(anonymized_telemetry=False)
+        )
         
         # Ensure collection exists
+        # Use explicit embedding function to ensure queries can compute embeddings
+        embedding_function = embedding_functions.DefaultEmbeddingFunction()
         try:
-            self.vector_collection = client.get_collection("legal_context")
+            self.vector_collection = client.get_collection(
+                "legal_context",
+                embedding_function=embedding_function
+            )
         except Exception:
-            self.vector_collection = client.create_collection("legal_context")
+            self.vector_collection = client.create_collection(
+                "legal_context",
+                embedding_function=embedding_function
+            )
         
         # Initialize LLM
         genai.configure(api_key=os.getenv('GOOGLE_API_KEY'))

vector_store_builder.py

@@ -1,5 +1,7 @@
 # vector_store_builder.py
+import os
 import chromadb
+from chromadb.config import Settings
 from chromadb.utils import embedding_functions
 from legal_aid_context import ALL_CONTEXT_DOCS
 import uuid
@@ -7,16 +9,25 @@ import uuid
 def build_vector_store():
     """Build ChromaDB vector store with legal context"""
     
-    # Initialize ChromaDB client
-    client = chromadb.PersistentClient(path="./legal_vector_db")
+    # Initialize ChromaDB client with telemetry disabled and writable paths
+    chroma_path = os.getenv('CHROMA_DB_PATH', './legal_vector_db')
+    os.makedirs(chroma_path, exist_ok=True)
+
+    default_cache_root = os.getenv('CACHE_ROOT', './cache')
+    os.makedirs(default_cache_root, exist_ok=True)
+    os.environ.setdefault('HF_HOME', os.path.join(default_cache_root, 'hf'))
+    os.environ.setdefault('TRANSFORMERS_CACHE', os.path.join(default_cache_root, 'transformers'))
+    os.environ.setdefault('SENTENCE_TRANSFORMERS_HOME', os.path.join(default_cache_root, 'sentence-transformers'))
+    os.environ.setdefault('XDG_CACHE_HOME', default_cache_root)
+    for env_key in ['HF_HOME', 'TRANSFORMERS_CACHE', 'SENTENCE_TRANSFORMERS_HOME', 'XDG_CACHE_HOME']:
+        os.makedirs(os.environ[env_key], exist_ok=True)
+
+    client = chromadb.PersistentClient(path=chroma_path, settings=Settings(anonymized_telemetry=False))
     
     # Create collection with default embedding function
     embedding_function = embedding_functions.DefaultEmbeddingFunction()
     
-    collection = client.get_or_create_collection(
-        name="legal_context",
-        embedding_function=embedding_function
-    )
+    collection = client.get_or_create_collection(name="legal_context", embedding_function=embedding_function)
     
     # Prepare documents for vector store
     documents = []