From 95bdc59af0bad389fca636b92f30621cbe2961b3 Mon Sep 17 00:00:00 2001
From: Lucas Oliveira <lucas.edu.oli@hotmail.com>
Date: Mon, 6 Oct 2025 09:45:33 -0300
Subject: [PATCH] Use opensearch queries on documents and processors

---
 src/api/documents.py     | 32 ++++++--------------------------
 src/models/processors.py | 20 ++++----------------
 2 files changed, 10 insertions(+), 42 deletions(-)

diff --git a/src/api/documents.py b/src/api/documents.py
index a367c9fe..048f746a 100644
--- a/src/api/documents.py
+++ b/src/api/documents.py
@@ -23,20 +23,9 @@ async def check_filename_exists(request: Request, document_service, session_mana
         )
 
         # Search for any document with this exact filename
-        # Try both .keyword (exact match) and regular field (analyzed match)
-        search_body = {
-            "query": {
-                "bool": {
-                    "should": [
-                        {"term": {"filename.keyword": filename}},
-                        {"term": {"filename": filename}}
-                    ],
-                    "minimum_should_match": 1
-                }
-            },
-            "size": 1,
-            "_source": ["filename"]
-        }
+        from utils.opensearch_queries import build_filename_search_body
+
+        search_body = build_filename_search_body(filename, size=1, source=["filename"])
 
         logger.debug(f"Checking filename existence: {filename}")
 
@@ -83,18 +72,9 @@ async def delete_documents_by_filename(request: Request, document_service, sessi
         )
 
         # Delete by query to remove all chunks of this document
-        # Use both .keyword and regular field to ensure we catch all variations
-        delete_query = {
-            "query": {
-                "bool": {
-                    "should": [
-                        {"term": {"filename.keyword": filename}},
-                        {"term": {"filename": filename}}
-                    ],
-                    "minimum_should_match": 1
-                }
-            }
-        }
+        from utils.opensearch_queries import build_filename_delete_body
+
+        delete_query = build_filename_delete_body(filename)
 
         logger.debug(f"Deleting documents with filename: {filename}")
 
diff --git a/src/models/processors.py b/src/models/processors.py
index 6d7b74b4..4a5d96b5 100644
--- a/src/models/processors.py
+++ b/src/models/processors.py
@@ -65,6 +65,7 @@ class TaskProcessor:
         Returns True if any chunks with this filename exist.
         """
         from config.settings import INDEX_NAME
+        from utils.opensearch_queries import build_filename_search_body
         import asyncio
 
         max_retries = 3
@@ -73,15 +74,7 @@ class TaskProcessor:
         for attempt in range(max_retries):
             try:
                 # Search for any document with this exact filename
-                search_body = {
-                    "query": {
-                        "term": {
-                            "filename.keyword": filename
-                        }
-                    },
-                    "size": 1,
-                    "_source": False
-                }
+                search_body = build_filename_search_body(filename, size=1, source=False)
 
                 response = await opensearch_client.search(
                     index=INDEX_NAME,
@@ -126,16 +119,11 @@ class TaskProcessor:
         Delete all chunks of a document with the given filename from OpenSearch.
         """
         from config.settings import INDEX_NAME
+        from utils.opensearch_queries import build_filename_delete_body
 
         try:
             # Delete all documents with this filename
-            delete_body = {
-                "query": {
-                    "term": {
-                        "filename.keyword": filename
-                    }
-                }
-            }
+            delete_body = build_filename_delete_body(filename)
 
             response = await opensearch_client.delete_by_query(
                 index=INDEX_NAME,