RFC: BaseRetriever.return_str()

libs/core/langchain_core/prompts/base.py

@@ -210,7 +210,7 @@ class BasePromptTemplate(RunnableSerializable[Dict, PromptValue], ABC):
             raise ValueError(f"{save_path} must be json or yaml")
 
 
-def format_document(doc: Document, prompt: BasePromptTemplate) -> str:
+def format_document(doc: Document, prompt: Optional[BasePromptTemplate] = None) -> str:
     """Format a document into a string based on a prompt template.
 
     First, this pulls information from the document from two sources:
@@ -244,16 +244,17 @@ def format_document(doc: Document, prompt: BasePromptTemplate) -> str:
             format_document(doc, prompt)
             >>> "Page 1: This is a joke"
     """
+    _prompt = prompt or PromptTemplate.from_template("{page_content}")
     base_info = {"page_content": doc.page_content, **doc.metadata}
     missing_metadata = set(prompt.input_variables).difference(base_info)
     if len(missing_metadata) > 0:
         required_metadata = [
-            iv for iv in prompt.input_variables if iv != "page_content"
+            iv for iv in _prompt.input_variables if iv != "page_content"
         ]
         raise ValueError(
             f"Document prompt requires documents to have metadata variables: "
             f"{required_metadata}. Received document with missing metadata: "
             f"{list(missing_metadata)}."
         )
-    document_info = {k: base_info[k] for k in prompt.input_variables}
-    return prompt.format(**document_info)
+    document_info = {k: base_info[k] for k in _prompt.input_variables}
+    return _prompt.format(**document_info)

libs/core/langchain_core/retrievers.py

@@ -7,6 +7,7 @@ from typing import TYPE_CHECKING, Any, Dict, List, Optional
 
 from langchain_core.documents import Document
 from langchain_core.load.dump import dumpd
+from langchain_core.prompts import format_document
 from langchain_core.runnables import (
     Runnable,
     RunnableConfig,
@@ -285,3 +286,29 @@ class BaseRetriever(RunnableSerializable[RetrieverInput, RetrieverOutput], ABC):
                 **kwargs,
             )
             return result
+
+    def return_str(
+        self, prompt: Optional[BasePromptTemplate] = None, separator: str = "\n\n"
+    ) -> RunnableSerializable[RetrieverInput, str]:
+        """
+
+        Example:
+            .. code-block:: python
+
+            from langchain_community.retrievers import FAISS
+
+            retriever = FAISS.from_texts(["hi", "bye"])
+            chain = {"context": retriever.return_str()} | prompt | llm | StrOutputParser()
+        """
+        return self | partial(_transform_docs, prompt=prompt, separator=separator)
+
+
+def _transform_docs(
+    doc_stream: Iterator[Document],
+    prompt: Optional[BasePromptTemplate],
+    separator: str,
+) -> Iterator[str]:
+    doc = next(doc_stream)
+    yield format_document(doc, prompt=prompt)
+    for doc in doc_stream:
+        yield separator + format_document(doc, prompt=prompt)