876 add rrf

tr_sys/conftest.py

@@ -1,13 +1,17 @@
-
 import pytest
-from pytest_factoryboy import register
 from unittest.mock import MagicMock
-from factories import AgentFactory, ActorFactory, MessageFactory, ChannelFactory
 
-register(AgentFactory)
-register(ActorFactory)
-register(ChannelFactory)
-register(MessageFactory)
+try:
+    from pytest_factoryboy import register
+    from factories import AgentFactory, ActorFactory, MessageFactory, ChannelFactory
+except ImportError:
+    register = None
+
+if register is not None:
+    register(AgentFactory)
+    register(ActorFactory)
+    register(ChannelFactory)
+    register(MessageFactory)
 
 @pytest.fixture
 def new_user(db, user_factory):

tr_sys/tests/unit/test_ranker_fusion.py

@@ -0,0 +1,220 @@
+import importlib.util
+from pathlib import Path
+
+
+def load_ranker_fusion():
+    module_path = Path(__file__).parents[2] / "tr_ars" / "ranker_fusion.py"
+    spec = importlib.util.spec_from_file_location("ranker_fusion", module_path)
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module
+
+
+ranker_fusion = load_ranker_fusion()
+
+
+def make_result(*ids):
+    return {
+        "node_bindings": {
+            "n%s" % index: [{"id": node_id}]
+            for index, node_id in enumerate(ids)
+        },
+        "analyses": [{"score": 0.1}],
+    }
+
+
+def test_parse_ranker_config(monkeypatch):
+    monkeypatch.setenv("ARS_RRF_ARAGORN_WEIGHT", "0.8")
+    monkeypatch.setenv("ARS_RRF_C", "50")
+    monkeypatch.setenv("ARS_RRF_ENABLED", "true")
+
+    rf_config = ranker_fusion.get_config()
+
+    weights = {
+        "infores:shepherd-aragorn": 0.8,
+        "infores:shepherd-arax": 0.2,
+    }
+
+    assert rf_config == {
+        "enabled": True,
+        "weights": weights,
+        "c_value": 50,
+    }
+
+
+def test_parse_ranker_config_derives_arax_weight(monkeypatch):
+    monkeypatch.setenv("ARS_RRF_ARAGORN_WEIGHT", "0.7")
+    monkeypatch.setenv("ARS_RRF_ARAX_WEIGHT", "0.9")
+
+    rf_config = ranker_fusion.get_config()
+
+    assert rf_config["weights"] == {
+        "infores:shepherd-aragorn": 0.7,
+        "infores:shepherd-arax": 0.3,
+    }
+
+
+def test_parse_ranker_config_defaults_on_invalid(monkeypatch):
+    monkeypatch.setenv("ARS_RRF_ARAGORN_WEIGHT", "1.2")
+    monkeypatch.setenv("ARS_RRF_C", "-1")
+    monkeypatch.setenv("ARS_RRF_ENABLED", "true")
+
+    rf_config = ranker_fusion.get_config()
+
+    default_weights = {
+        "infores:shepherd-aragorn": 0.9,
+        "infores:shepherd-arax": 0.1,
+    }
+    default_c = 40
+
+    assert rf_config == {
+        "enabled": True,
+        "weights": default_weights,
+        "c_value": default_c,
+    }
+
+
+def test_parse_ranker_config_defaults_on_nan_weight(monkeypatch):
+    monkeypatch.setenv("ARS_RRF_ARAGORN_WEIGHT", "nan")
+
+    rf_config = ranker_fusion.get_config()
+
+    assert rf_config["weights"] == {
+        "infores:shepherd-aragorn": 0.9,
+        "infores:shepherd-arax": 0.1,
+    }
+
+
+def test_parse_ranker_config_reads_enabled_from_env(monkeypatch):
+    monkeypatch.setenv("ARS_RRF_ENABLED", "false")
+
+    rf_config = ranker_fusion.get_config()
+
+    assert rf_config["enabled"] is False
+
+
+def test_configured_source_matches_infores_directly():
+    source = ranker_fusion.configured_source(
+        "infores:shepherd-arax",
+        "ara-shepherd-arax",
+        {"infores:shepherd-arax": 0.2},
+    )
+
+    assert source == "infores:shepherd-arax"
+
+
+def test_configured_source_accepts_incoming_ara_agent_alias():
+    source = ranker_fusion.configured_source(
+        None,
+        "ara-shepherd-aragorn",
+        {"infores:shepherd-aragorn": 0.8},
+    )
+
+    assert source == "infores:shepherd-aragorn"
+
+
+def test_weighted_rrf_sorts_results_and_adds_metadata():
+    result_a = make_result("CHEBI:1", "MONDO:1")
+    result_b = make_result("CHEBI:2", "MONDO:1")
+    result_c = make_result("CHEBI:3", "MONDO:1")
+    data = {"message": {"results": [result_a, result_b, result_c]}}
+
+    summary = ranker_fusion.apply_weighted_rrf(
+        data,
+        {
+            "infores:shepherd-aragorn": [result_a, result_b],
+            "infores:shepherd-arax": [result_b, result_c],
+        },
+        {"infores:shepherd-aragorn": 0.5, "infores:shepherd-arax": 0.5},
+        60,
+    )
+
+    assert summary["applied"] is True
+    assert [r["node_bindings"]["n0"][0]["id"] for r in data["message"]["results"]] == [
+        "CHEBI:2",
+        "CHEBI:1",
+        "CHEBI:3",
+    ]
+    assert data["message"]["results"][0]["rrf_rank"] == 1
+    assert data["message"]["results"][0]["rrf_ranker_ranks"] == {
+        "infores:shepherd-aragorn": 2,
+        "infores:shepherd-arax": 1,
+    }
+    assert data["message"]["results"][0]["rrf_score"] > data["message"]["results"][1]["rrf_score"]
+
+
+def test_weighted_rrf_keeps_unmatched_results_at_end_stably():
+    result_a = make_result("CHEBI:1")
+    result_b = make_result("CHEBI:2")
+    result_c = make_result("CHEBI:3")
+    data = {"message": {"results": [result_c, result_b, result_a]}}
+
+    summary = ranker_fusion.apply_weighted_rrf(
+        data,
+        {"infores:shepherd-aragorn": [result_a]},
+        {"infores:shepherd-aragorn": 1.0},
+        60,
+    )
+
+    assert summary["applied"] is True
+    assert [r["node_bindings"]["n0"][0]["id"] for r in data["message"]["results"]] == [
+        "CHEBI:1",
+        "CHEBI:3",
+        "CHEBI:2",
+    ]
+    assert data["message"]["results"][1]["rrf_score"] == 0.0
+    assert data["message"]["results"][2]["rrf_score"] == 0.0
+
+
+def test_weighted_rrf_tied_scores_share_rank():
+    result_a = make_result("CHEBI:1")
+    result_b = make_result("CHEBI:2")
+    data = {"message": {"results": [result_b, result_a]}}
+
+    summary = ranker_fusion.apply_weighted_rrf(
+        data,
+        {
+            "infores:shepherd-aragorn": [result_a, result_b],
+            "infores:shepherd-arax": [result_b, result_a],
+        },
+        {"infores:shepherd-aragorn": 0.5, "infores:shepherd-arax": 0.5},
+        60,
+    )
+
+    assert summary["applied"] is True
+    assert data["message"]["results"] == [result_b, result_a]
+    assert data["message"]["results"][0]["rrf_score"] == data["message"]["results"][1]["rrf_score"]
+    assert data["message"]["results"][0]["rrf_rank"] == 1
+    assert data["message"]["results"][1]["rrf_rank"] == 1
+
+
+def test_weighted_rrf_noops_without_matching_ranker_results():
+    result_a = make_result("CHEBI:1")
+    data = {"message": {"results": [result_a]}}
+
+    summary = ranker_fusion.apply_weighted_rrf(
+        data,
+        {"infores:other": [result_a]},
+        {"infores:shepherd-aragorn": 1.0},
+        60,
+    )
+
+    assert summary == {"applied": False, "reason": "missing_ranker_results"}
+    assert data["message"]["results"] == [result_a]
+
+
+def test_weighted_rrf_noops_without_matching_answers():
+    result_a = make_result("CHEBI:1")
+    result_b = make_result("CHEBI:2")
+    data = {"message": {"results": [result_a]}}
+
+    summary = ranker_fusion.apply_weighted_rrf(
+        data,
+        {"infores:shepherd-aragorn": [result_b]},
+        {"infores:shepherd-aragorn": 1.0},
+        60,
+    )
+
+    assert summary == {"applied": False, "reason": "no_matching_results"}
+    assert data["message"]["results"] == [result_a]
+    assert "rrf_score" not in result_a