requirements.txt updated

Dockerfile

@@ -2,33 +2,53 @@ FROM python:3.10-slim-bullseye
 
 ENV PYTHONUNBUFFERED=1
 ENV PIP_NO_CACHE_DIR=1
+ENV DOCKER_CONTAINER=true  
+ENV SPACE_APP_DATA=/data
+ENV HF_HOME=/data/huggingface  # For HF cache
+
+# Optimize llama-cpp-python build for CPU only
+ENV CMAKE_ARGS="-DLLAMA_BLAS=0 -DLLAMA_CUBLAS=0"
+ENV FORCE_CMAKE=1
 
 WORKDIR /app
 
-# System deps (lightweight) - ADD pthread development headers
+# System deps - minimal for HuggingFace Spaces
 RUN apt-get update && apt-get install -y --no-install-recommends \
     build-essential \
     libglib2.0-0 \
     libjpeg62-turbo \
     poppler-utils \
-    libpthread-stubs0-dev \
-    && rm -rf /var/lib/apt/lists/*
-
-# Optimize llama-cpp-python build for CPU only to reduce build time/memory usage
-ENV CMAKE_ARGS="-DLLAMA_BLAS=0 -DLLAMA_CUBLAS=0"
-ENV FORCE_CMAKE=1
+    libmagic1 \
+    curl \
+    git \               # ← ADD for potential git operations
+    && rm -rf /var/lib/apt/lists/* \
+    && apt-get clean
 
+# Copy requirements first for better layer caching
 COPY requirements.txt /app/requirements.txt
 
-# Install Python dependencies
+# Install Python dependencies with specific versions
 RUN pip install --upgrade pip && \
-    pip install -r requirements.txt
+    pip install -r requirements.txt --no-cache-dir
 
-# Download spaCy model
+# Download spaCy model (after dependencies)
 RUN python -m spacy download en_core_web_sm
 
+# Create directories that your app expects
+RUN mkdir -p /data/models /data/uploads /data/cache /data/logs /data/huggingface
+
+# Copy app code
 COPY . .
 
+# Set proper permissions
+RUN chmod -R 755 /app && \
+    chmod -R 755 /data
+
+# Health check (remove if you don't have /health endpoint)
+HEALTHCHECK --interval=30s --timeout=10s --start-period=30s --retries=3 \
+    CMD curl -f http://localhost:7860/docs || exit 1  # Changed to /docs endpoint
+
 EXPOSE 7860
 
-CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]
+# Use multiple workers for better performance
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860", "--workers", "2"]

config/settings.py

@@ -74,7 +74,7 @@ class Settings(BaseSettings):
     ANTHROPIC_MAX_TOKENS   : int                                                = 1024
 
     # Priority order for LLM providers
-    LLM_PROVIDER_PRIORITY  : list                                               = ["ollama", "openai", "anthropic", "llama_cpp"]
+    LLM_PROVIDER_PRIORITY  : list                                               = ["llama_cpp", "ollama", "openai", "anthropic", ]
     
     # Which providers are available
     ENABLE_OLLAMA          : bool                                               = True
@@ -84,7 +84,7 @@ class Settings(BaseSettings):
     ENABLE_HF_INFERENCE    : bool                                               = False  # HuggingFace Inference API
 
     # Default provider (auto-selected based on environment)
-    LLM_DEFAULT_PROVIDER   : str                                                = "ollama"
+    LLM_DEFAULT_PROVIDER   : str                                                = "llama_cpp"
     
     # Huggingface Inference Settings (Optional)
     HF_MODEL_ID            : Optional[str]                                      = None   # e.g. "meta-llama/Llama-2-7b-chat-hf"
@@ -234,12 +234,12 @@ class Settings(BaseSettings):
             return priority if priority else ["ollama"]
     
 
-    @field_validator('LLM_DEFAULT_PROVIDER', mode='after')
+    @field_validator('LLM_DEFAULT_PROVIDER', mode = 'after')
     def set_default_provider(cls, v, info):
         """
         Set default provider based on availability
         """
-        values = info.data
+        values   = info.data
         
         # Get the priority list (after adjustments)
         priority = values.get('LLM_PROVIDER_PRIORITY', [])

services/llm_interpreter.py

@@ -29,7 +29,7 @@ class LLMClauseInterpreter:
     """
     Uses LLM to generate plain-English explanations for legal clauses and integrated with RiskAnalyzer results and RiskRules framework
     """
-    def __init__(self, llm_manager: LLMManager, default_provider: LLMProvider = LLMProvider.OLLAMA):
+    def __init__(self, llm_manager: LLMManager, default_provider: LLMProvider = None):
         """
         Initialize LLM interpreter