retriever adapter

.gitignore

@@ -1,2 +1,3 @@
 .env
-*.DS_Store
+*.DS_Store
+__pycache__/

app/models.py

@@ -18,6 +18,8 @@ class GraphState(TypedDict):
     metadata: Optional[Dict[str, Any]]
     file_type: Optional[str]
     workflow_type: Optional[str]  # 'standard' or 'geojson_direct'
+    metadata_filters: Optional[Dict[str, Any]]
+    metadata: Dict[str, Any]
 
 class ChatUIInput(BaseModel):
     """Input model for text-only ChatUI requests"""
@@ -27,4 +29,3 @@ class ChatUIFileInput(BaseModel):
     """Input model for ChatUI requests with file attachments"""
     text: str
     files: Optional[List[Dict[str, Any]]] = None
-

app/nodes.py

@@ -1,8 +1,7 @@
-from utils import detect_file_type, convert_context_to_list, merge_state, getconfig
-from models import GraphState
-from datetime import datetime
 import tempfile
 import os
+from models import GraphState
+from datetime import datetime
 from gradio_client import Client, file
 import logging
 import dotenv
@@ -10,6 +9,9 @@ import httpx
 import json
 from typing import Generator, Optional
 
+from utils import detect_file_type, convert_context_to_list, merge_state, getconfig
+from retriever_adapter import RetrieverAdapter
+
 dotenv.load_dotenv()
 
 logger = logging.getLogger(__name__)
@@ -22,6 +24,7 @@ INGESTOR = config.get("ingestor", "INGESTOR", fallback="https://mtyrrell-chatfed
 GEOJSON_INGESTOR = config.get("ingestor", "GEOJSON_INGESTOR", fallback="https://giz-eudr-chatfed-ingestor.hf.space")
 MAX_CONTEXT_CHARS = int(config.get("general", "MAX_CONTEXT_CHARS"))
 
+retriever_adapter = RetrieverAdapter("params.cfg")
 
 #----------------------------------------
 # LANGGRAPH NODE FUNCTIONS
@@ -122,19 +125,18 @@ def geojson_direct_result_node(state: GraphState) -> GraphState:
 
 
 def retrieve_node(state: GraphState) -> GraphState:
-    """Retrieve relevant context from vector store"""
+    """Retrieve relevant context using adapter"""
     start_time = datetime.now()
     logger.info(f"Retrieval: {state['query'][:50]}...")
     
     try:
-        client = Client(RETRIEVER, hf_token=os.getenv("HF_TOKEN"))
-        context = client.predict(
+        # Get filters from state (provided by ChatUI or LLM agent)
+        filters = state.get("metadata_filters")
+        
+        context = retriever_adapter.retrieve(
             query=state["query"],
-            reports_filter=state.get("reports_filter", ""),
-            sources_filter=state.get("sources_filter", ""),
-            subtype_filter=state.get("subtype_filter", ""),
-            year_filter=state.get("year_filter", ""),
-            api_name="/retrieve"
+            filters=filters,
+            hf_token=os.getenv("HF_TOKEN")
         )
         
         duration = (datetime.now() - start_time).total_seconds()
@@ -142,7 +144,9 @@ def retrieve_node(state: GraphState) -> GraphState:
         metadata.update({
             "retrieval_duration": duration,
             "context_length": len(context) if context else 0,
-            "retrieval_success": True
+            "retrieval_success": True,
+            "filters_applied": filters,
+            "retriever_config": retriever_adapter.get_metadata()
         })
         
         return {"context": context, "metadata": metadata}

app/retriever_adapter.py

@@ -0,0 +1,73 @@
+import logging
+from typing import Dict, Any, Optional
+from gradio_client import Client
+from utils import getconfig
+
+logger = logging.getLogger(__name__)
+
+class RetrieverAdapter:
+    """
+    Simple adapter that passes query and filters to retriever.
+    Filters are provided by ChatUI frontend or LLM agent, not config.
+    """
+    
+    def __init__(self, config_path: str = "params.cfg"):
+        self.config = getconfig(config_path)
+        self.retriever_url = self.config.get("retriever", "RETRIEVER")
+        self.collection_name = self.config.get("retriever", "COLLECTION_NAME", fallback=None)
+        
+        logger.info(f"RetrieverAdapter initialized: {self.retriever_url}")
+        if self.collection_name:
+            logger.info(f"Collection: {self.collection_name}")
+    
+    def retrieve(
+        self, 
+        query: str, 
+        filters: Optional[Dict[str, Any]] = None,
+        hf_token: Optional[str] = None
+    ) -> str:
+        """
+        Execute retrieval with query and optional filters.
+        
+        Args:
+            query: Search query
+            filters: Metadata filters dict (from ChatUI or LLM agent)
+            hf_token: HuggingFace token for authentication
+            
+        Returns:
+            Retrieved context string
+        """
+        try:
+            client = Client(self.retriever_url, hf_token=hf_token)
+            
+            # Build parameters - always include query
+            params = {"query": query}
+            
+            # Add collection name if configured
+            if self.collection_name:
+                params["collection_name"] = self.collection_name
+            
+            # Add filters if provided
+            if filters:
+                params["filters"] = filters
+            
+            params["api_name"] = "/retrieve"
+            
+            # Log request
+            log_params = {k: f"<dict with {len(v)} keys>" if isinstance(v, dict) else v 
+                         for k, v in params.items() if k != "api_name"}
+            logger.info(f"Retrieval request: {log_params}")
+            
+            context = client.predict(**params)
+            return context
+            
+        except Exception as e:
+            logger.error(f"Retrieval failed: {str(e)}")
+            raise
+    
+    def get_metadata(self) -> Dict[str, Any]:
+        """Return retriever configuration metadata"""
+        return {
+            "retriever_url": self.retriever_url,
+            "collection_name": self.collection_name
+        }

params.cfg

@@ -1,5 +1,6 @@
 [retriever]
-RETRIEVER = https://giz-chatfed-retriever0-3.hf.space
+RETRIEVER = https://giz-chatfed-retriever0-3.hf.space/
+COLLECTION_NAME = EUDR  # Optional, only if retriever needs it
 
 [generator]
 GENERATOR = https://giz-eudr-chabo-generator.hf.space