[bugfix] corrected declaration of top-k parameter for chromaDB invoca…

README.md

@@ -52,6 +52,7 @@ graph TD
 ## Autor
 
 **[@CuriousGu](https://www.github.com/CuriousGu) 🇧🇷**
+**[@PrimeRibs2501](https://www.github.com/PrimeRibs2501) 🇧🇷**
 
 ## Docs
 1. [Estrutura](docs/pt_br/ESTRUTURA.md)
@@ -70,3 +71,6 @@ Veja o arquivo [LICENSE](LICENSE) para mais detalhes.
 - Email: gustavo_ortega@usp.br
 - Linkedin: [Gustavo M. Ortega](https://www.linkedin.com/in/gustavomendoncaortega/)
 
+- Email: pedro.hbueno03@usp.br
+- Linkedin: [Pedro H. B. Ribeiro](https://www.linkedin.com/in//pedrohenriquebuenoribeiro)
+

docker/docker-compose.yml

@@ -22,7 +22,7 @@ services:
     attach: false
 
   chroma:
-    image: chromadb/chroma:latest
+    image: chromadb/chroma:1.0.12
     env_file:
       - ../.env
     ports:

src/api/controllers/files.py

@@ -1,5 +1,7 @@
 from src.services.document_reader import DocumentReader
 from src.infrastructure.database import ChromaDB
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain_core.documents import Document
 from src.infrastructure.config import settings
 from src.api.models import FileMetadata
 
@@ -17,10 +19,20 @@ async def controller_upload_file(
         metadata["extension"] = content["extension"]
         metadata["file_name"] = content["name"]
 
+        langchain_docs = []
+
+        for j, text_content in enumerate(content["content"]):
+                    doc = Document(
+                        page_content=text_content,
+                        metadata={
+                            'extension' : content["extension"],
+                            'file_name' : content["name"]
+
+                            }
+                    )
+                    langchain_docs.append(doc)
         await vector_store.add_documents(
-            documents=content["content"],
-            collection_name=settings.INDEX_NAME,
-            metadatas=[metadata for _ in content["content"]],
+            documents=langchain_docs
         )
         return True
     except Exception as e:

src/infrastructure/database/chromadb/connector.py

@@ -21,6 +21,11 @@ def __init__(self):
         )
         try:
             self.client = self._connect()
+            self.vector_store = Chroma(
+                client=self.client,
+                collection_name=self.collection_name,
+                embedding_function=self.embedding_function,
+            )
             self.retriever = self._as_retriever()
         except Exception as e:
             raise ConnectionError(f"Failed to connect to ChromaDB: {e}")
@@ -56,35 +61,22 @@ def _as_retriever(self, collection_name: str = None, k: int = 10) -> dict:
         """
         Cria um retriever adaptado à dimensão da coleção
         """
-        vector_store = Chroma(
-            client=self.client,
-            collection_name=collection_name or self.collection_name,
-            embedding_function=self.embedding_function,
-        )
-        retriever = vector_store.as_retriever(k=k)
+
+        retriever = self.vector_store.as_retriever(k=k)
 
         return retriever
 
     async def add_documents(
         self,
         documents: List[str],
-        collection_name: str,
-        metadatas: Optional[List[dict]] = None,
     ):
         """
         Adiciona documentos à coleção com dimensão configurada no .env
         """
-        if not self.collection or self.collection.name != collection_name:
-            self.collection = self._create_collection(collection_name)
-
-        embeddings = self.embedding_function.embed_documents(documents)
-
-        return self.collection.add(
-            documents=documents,
-            embeddings=embeddings,
-            metadatas=metadatas,
-            ids=[str(uuid.uuid4()) for _ in enumerate(documents)],
-        )
+        return self.vector_store.add_documents(
+                documents=documents,
+                ids=[str(uuid.uuid4()) for _ in enumerate(documents)],
+            )
 
     async def list_collections(self):
         return self.client.list_collections()
@@ -94,7 +86,7 @@ async def list_documents(
         collection_name: str
     ):
         if collection := self.client.get_collection(collection_name):
-            return collection.get()
+            return collection.get(include=["metadatas", "documents", "embeddings"])
         return []
 
     async def query_documents(
@@ -164,7 +156,7 @@ def retrieve(query: str, n: int = 10) -> List:
             # Limitando a quantidade de documentos retornados para n+10
             # Isso garante que tenhamos documentos suficientes para filtrar
             # posteriormente
-            response = retriever.invoke(query)
+            response = retriever.get_relevant_documents(query)
 
             created_at = [
                 (index, x.metadata.get("created_at"))