bug: JoinDocuments nodes produce incorrect results if preceded by another JoinDocuments node (#3170)

* don't send the list of inputs back as an output in the running of a node.

* updated documentation

* Update pydoc-markdown.py

* added test case for pipeline join fix

Co-authored-by: JeffRisberg <jrisberg@aol.com>

haystack/nodes/base.py

@@ -265,9 +265,9 @@ class BaseComponent(ABC):
         if all_debug:
             output["_debug"] = all_debug
 
-        # add "extra" args that were not used by the node
+        # add "extra" args that were not used by the node, but not the 'inputs' value
         for k, v in arguments.items():
-            if k not in output.keys():
+            if k not in output.keys() and k != "inputs":
                 output[k] = v
 
         output["params"] = params

test/pipelines/test_pipeline.py

@@ -18,9 +18,11 @@ import pandas as pd
 from haystack import __version__
 from haystack.document_stores.deepsetcloud import DeepsetCloudDocumentStore
 from haystack.document_stores.elasticsearch import ElasticsearchDocumentStore
+from haystack.document_stores.memory import InMemoryDocumentStore
 from haystack.nodes.other.join_docs import JoinDocuments
 from haystack.nodes.base import BaseComponent
 from haystack.nodes.retriever.sparse import BM25Retriever
+from haystack.nodes.retriever.sparse import FilterRetriever
 from haystack.pipelines import (
     Pipeline,
     RootNode,
@@ -1999,3 +2001,66 @@ def test_batch_querying_multiple_queries(document_store_with_docs):
     assert isinstance(result["answers"][0][0], Answer)
     assert len(result["answers"]) == 2  # Predictions for 2 collections of documents
     assert len(result["answers"][0]) == 5  # top-k of 5 for collection of docs
+
+
+def test_fix_to_pipeline_execution_when_join_follows_join():
+    # wire up 4 retrievers, each with one document
+    document_store_1 = InMemoryDocumentStore()
+    retriever_1 = FilterRetriever(document_store_1, scale_score=True)
+    dicts_1 = [{"content": "Alpha", "score": 0.552}]
+    document_store_1.write_documents(dicts_1)
+
+    document_store_2 = InMemoryDocumentStore()
+    retriever_2 = FilterRetriever(document_store_2, scale_score=True)
+    dicts_2 = [{"content": "Beta", "score": 0.542}]
+    document_store_2.write_documents(dicts_2)
+
+    document_store_3 = InMemoryDocumentStore()
+    retriever_3 = FilterRetriever(document_store_3, scale_score=True)
+    dicts_3 = [{"content": "Gamma", "score": 0.532}]
+    document_store_3.write_documents(dicts_3)
+
+    document_store_4 = InMemoryDocumentStore()
+    retriever_4 = FilterRetriever(document_store_4, scale_score=True)
+    dicts_4 = [{"content": "Delta", "score": 0.512}]
+    document_store_4.write_documents(dicts_4)
+
+    # wire up a pipeline of the retrievers, with 4-way join
+    pipeline = Pipeline()
+    pipeline.add_node(component=retriever_1, name="Retriever1", inputs=["Query"])
+    pipeline.add_node(component=retriever_2, name="Retriever2", inputs=["Query"])
+    pipeline.add_node(component=retriever_3, name="Retriever3", inputs=["Query"])
+    pipeline.add_node(component=retriever_4, name="Retriever4", inputs=["Query"])
+    pipeline.add_node(
+        component=JoinDocuments(weights=[0.25, 0.25, 0.25, 0.25], join_mode="merge"),
+        name="Join",
+        inputs=["Retriever1", "Retriever2", "Retriever3", "Retriever4"],
+    )
+
+    res = pipeline.run(query="Alpha Beta Gamma Delta")
+    documents = res["documents"]
+    assert len(documents) == 4  # all four documents should be found
+
+    # wire up a pipeline of the retrievers, with join following join
+    pipeline = Pipeline()
+    pipeline.add_node(component=retriever_1, name="Retriever1", inputs=["Query"])
+    pipeline.add_node(component=retriever_2, name="Retriever2", inputs=["Query"])
+    pipeline.add_node(component=retriever_3, name="Retriever3", inputs=["Query"])
+    pipeline.add_node(component=retriever_4, name="Retriever4", inputs=["Query"])
+    pipeline.add_node(
+        component=JoinDocuments(weights=[0.5, 0.5], join_mode="merge"),
+        name="Join12",
+        inputs=["Retriever1", "Retriever2"],
+    )
+    pipeline.add_node(
+        component=JoinDocuments(weights=[0.5, 0.5], join_mode="merge"),
+        name="Join34",
+        inputs=["Retriever3", "Retriever4"],
+    )
+    pipeline.add_node(
+        component=JoinDocuments(weights=[0.5, 0.5], join_mode="merge"), name="JoinFinal", inputs=["Join12", "Join34"]
+    )
+
+    res = pipeline.run(query="Alpha Beta Gamma Delta")
+    documents = res["documents"]
+    assert len(documents) == 4  # all four documents should be found