mirror of
https://github.com/deepset-ai/haystack.git
synced 2025-07-25 01:40:22 +00:00
408d8e6ff5
* Enable the `JoinDocuments` node to work with documents with `score=None` This fixes #2983 As of now, the `JoinDocuments` node will error out if any of the documents has `score=None` - which is possible, as some retriever are not able to provide a score, like the `TfidfRetriever` on Elasticsearch or the `BM25Retriever` on Weaviate. THe reason for the error is that the `JoinDocuments` always sorts the documents by score and cannot sort when `score=None`. There was a very similar issue for `JoinAnswers` too, which was addressed by this PR: https://github.com/deepset-ai/haystack/pull/2436 This solution applies the same solution to `JoinDocuments` - so both the `JoinAnswers` and `JoinDocuments` now will have the same additional argument to disable sorting when that is requried. The solution is to add an argument to `JoinDocuments` called `sort_by_score: bool`, which allows the user to turn off the sorting of documents by score, but keeps the current functionality of sorting being performed as the default. * Fixing test bug * Addressing PR review comments - Extending unit tests - Simplifying logic * Making the sorting work even with no scores By making the no score being sorted as -Inf * Forgot to commit the change in `join_docs.py` * [EMPTY] Re-trigger CI * Added am INFO log if the `JoinDocuments` is sorting while some of the docs have `score=None` * Adjusting the arguments of `any()` * [EMPTY] Re-trigger CI
102 lines
3.5 KiB
Python
102 lines
3.5 KiB
Python
import pytest
|
|
import pandas as pd
|
|
|
|
from haystack import Document, Answer
|
|
from haystack.nodes.other.route_documents import RouteDocuments
|
|
from haystack.nodes.other.join_answers import JoinAnswers
|
|
from haystack.nodes.other.join_docs import JoinDocuments
|
|
|
|
|
|
#
|
|
# RouteDocuments tests
|
|
#
|
|
|
|
|
|
def test_routedocuments_by_content_type():
|
|
docs = [
|
|
Document(content="text document", content_type="text"),
|
|
Document(
|
|
content=pd.DataFrame(columns=["col 1", "col 2"], data=[["row 1", "row 1"], ["row 2", "row 2"]]),
|
|
content_type="table",
|
|
),
|
|
]
|
|
route_documents = RouteDocuments()
|
|
result, _ = route_documents.run(documents=docs)
|
|
assert len(result["output_1"]) == 1
|
|
assert len(result["output_2"]) == 1
|
|
assert result["output_1"][0].content_type == "text"
|
|
assert result["output_2"][0].content_type == "table"
|
|
|
|
|
|
def test_routedocuments_by_metafield(docs):
|
|
route_documents = RouteDocuments(split_by="meta_field", metadata_values=["test1", "test3", "test5"])
|
|
assert route_documents.outgoing_edges == 3
|
|
result, _ = route_documents.run(docs)
|
|
assert len(result["output_1"]) == 1
|
|
assert len(result["output_2"]) == 1
|
|
assert len(result["output_3"]) == 1
|
|
assert result["output_1"][0].meta["meta_field"] == "test1"
|
|
assert result["output_2"][0].meta["meta_field"] == "test3"
|
|
assert result["output_3"][0].meta["meta_field"] == "test5"
|
|
|
|
|
|
#
|
|
# JoinAnswers tests
|
|
#
|
|
|
|
|
|
@pytest.mark.parametrize("join_mode", ["concatenate", "merge"])
|
|
def test_joinanswers(join_mode):
|
|
inputs = [{"answers": [Answer(answer="answer 1", score=0.7)]}, {"answers": [Answer(answer="answer 2", score=0.8)]}]
|
|
|
|
join_answers = JoinAnswers(join_mode=join_mode)
|
|
result, _ = join_answers.run(inputs)
|
|
assert len(result["answers"]) == 2
|
|
assert result["answers"] == sorted(result["answers"], reverse=True)
|
|
|
|
result, _ = join_answers.run(inputs, top_k_join=1)
|
|
assert len(result["answers"]) == 1
|
|
assert result["answers"][0].answer == "answer 2"
|
|
|
|
|
|
#
|
|
# JoinDocuments tests
|
|
#
|
|
|
|
|
|
@pytest.mark.parametrize("join_mode", ["concatenate", "merge", "reciprocal_rank_fusion"])
|
|
def test_joindocuments(join_mode):
|
|
inputs = [
|
|
{"documents": [Document(content="text document 1", content_type="text", score=0.2)]},
|
|
{"documents": [Document(content="text document 2", content_type="text", score=0.7)]},
|
|
]
|
|
|
|
join_docs = JoinDocuments(join_mode=join_mode)
|
|
result, _ = join_docs.run(inputs)
|
|
assert len(result["documents"]) == 2
|
|
assert result["documents"] == sorted(result["documents"], reverse=True)
|
|
|
|
result, _ = join_docs.run(inputs, top_k_join=1)
|
|
assert len(result["documents"]) == 1
|
|
if join_mode == "reciprocal_rank_fusion":
|
|
assert result["documents"][0].content == "text document 1"
|
|
else:
|
|
assert result["documents"][0].content == "text document 2"
|
|
|
|
|
|
@pytest.mark.parametrize("join_mode", ["concatenate", "merge", "reciprocal_rank_fusion"])
|
|
@pytest.mark.parametrize("sort_by_score", [True, False])
|
|
def test_joindocuments_score_none(join_mode, sort_by_score):
|
|
"""Testing JoinDocuments() node when some of the documents have `score=None`"""
|
|
inputs = [
|
|
{"documents": [Document(content="text document 1", content_type="text", score=0.2)]},
|
|
{"documents": [Document(content="text document 2", content_type="text", score=None)]},
|
|
]
|
|
|
|
join_docs = JoinDocuments(join_mode=join_mode, sort_by_score=sort_by_score)
|
|
result, _ = join_docs.run(inputs)
|
|
assert len(result["documents"]) == 2
|
|
|
|
result, _ = join_docs.run(inputs, top_k_join=1)
|
|
assert len(result["documents"]) == 1
|