Add DocstoreFn - lookup doc via arbitrary function (#3760)

This **partially** addresses
#1524, but it's also useful
for some of our use cases.

This `DocstoreFn` allows to lookup a document given a function that
accepts the `search` string without the need to implement a custom
`Docstore`.

This could be useful when:
* you don't want to implement a `Docstore` just to provide a custom
`search`
 * it's expensive to construct an `InMemoryDocstore`/dict
 * you retrieve documents from remote sources
 * you just want to reuse existing objects

langchain/docstore/arbitrary_fn.py

@@ -0,0 +1,30 @@
+from typing import Callable, Union
+
+from langchain.docstore.base import Docstore
+from langchain.schema import Document
+
+
+class DocstoreFn(Docstore):
+    """
+    Langchain Docstore via arbitrary lookup function.
+
+    This is useful when:
+     * it's expensive to construct an InMemoryDocstore/dict
+     * you retrieve documents from remote sources
+     * you just want to reuse existing objects
+    """
+
+    def __init__(
+        self,
+        lookup_fn: Callable[[str], Union[Document, str]],
+    ):
+        self._lookup_fn = lookup_fn
+
+    def search(self, search: str) -> Document:
+        r = self._lookup_fn(search)
+        if isinstance(r, str):
+            # NOTE: assume the search string is the source ID
+            return Document(page_content=r, metadata={"source": search})
+        elif isinstance(r, Document):
+            return r
+        raise ValueError(f"Unexpected type of document {type(r)}")

tests/unit_tests/docstore/test_arbitrary_fn.py

@@ -0,0 +1,12 @@
+from langchain.docstore.arbitrary_fn import DocstoreFn
+from langchain.schema import Document
+
+
+def test_document_found() -> None:
+    # we use a dict here for simiplicity, but this could be any function
+    # including a remote lookup
+    dummy_dict = {"foo": Document(page_content="bar")}
+    docstore = DocstoreFn(lambda x: dummy_dict[x])
+    output = docstore.search("foo")
+    assert isinstance(output, Document)
+    assert output.page_content == "bar"