Spaces:
Running on Zero
Running on Zero
File size: 7,233 Bytes
a964978 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 | from unittest.mock import AsyncMock, patch
import pytest
from app.llm.provider import ResolvedProvider
from app.pipeline.chunker import Chunk
from app.pipeline.graph import (
KnowledgeGraph,
expand_graph_knowledge,
extract_entities_and_graph,
extract_heuristic_entities,
)
from app.pipeline.indexes import DocumentIndex
from app.pipeline.refmap import ReferenceMap
from app.pipeline.twohop import TwoHopChunk
from app.sessions import DocumentSession
from app.trace import get_trace_manager
def test_knowledge_graph_neighbor_lookup_and_cap():
"""Verify neighbor fact retrieval and strict <= 4 cap enforcement."""
kg = KnowledgeGraph()
kg.add_node("Resignation Policy", entity_type="POLICY")
kg.add_node("Notice Period", entity_type="REQUIREMENT")
kg.add_node("Severance Pay", entity_type="BENEFIT")
kg.add_node("Exit Interview", entity_type="PROCESS")
kg.add_node("Company Laptop", entity_type="EQUIPMENT")
kg.add_node("Security Clearance", entity_type="SECURITY")
kg.add_edge("Resignation Policy", "Notice Period", "requires", "Voluntary resignation requires 60 days advance written notice.")
kg.add_edge("Resignation Policy", "Severance Pay", "provides", "Standard resignation does not provide severance pay.")
kg.add_edge("Resignation Policy", "Exit Interview", "mandates", "All departing personnel must attend an Exit Interview.")
kg.add_edge("Resignation Policy", "Company Laptop", "requires", "Departing personnel must return Company Laptop within 48 hours.")
kg.add_edge("Resignation Policy", "Security Clearance", "governs", "Security Clearance is revoked on the final working day.")
# There are 5 connected edges, but cap must strictly enforce max 4
neighbors = kg.get_neighbor_facts(["Resignation Policy"], max_facts=4)
assert len(neighbors) == 4, f"Expected strictly <= 4 facts, got {len(neighbors)}"
facts = [e.fact for e in neighbors]
assert "Voluntary resignation requires 60 days advance written notice." in facts
def test_heuristic_entity_extraction():
"""Verify pattern-based entity extraction on chunk text."""
chunks = [
Chunk(
id="c1",
doc_id="d1",
text="Voluntary Resignation requires 60 days notice to the Human Resources department.",
token_count=12,
content_hash="h1",
),
Chunk(
id="c2",
doc_id="d1",
text="Health Insurance provides comprehensive dental and optical coverage.",
token_count=10,
content_hash="h2",
),
]
kg = extract_heuristic_entities(chunks)
assert len(kg.nodes) >= 2
assert len(kg.edges) >= 1
edge_relations = [e.relation for e in kg.edges]
assert "requires" in edge_relations or "provides" in edge_relations
@pytest.mark.anyio
async def test_llm_graph_extraction_mock():
"""Verify LLM extraction populates graph nodes and directed edges."""
provider = ResolvedProvider(
provider_type="groq",
api_key="gsk_test",
base_url="https://api.groq.com/openai/v1",
model="llama-3.3-70b-versatile",
)
mock_json = """
{
"nodes": [
{"name": "Leave Policy", "entity_type": "POLICY", "description": "Annual vacation and sick leave rules"},
{"name": "Manager Approval", "entity_type": "REQUIREMENT", "description": "Supervisor sign-off"}
],
"edges": [
{"source_node": "Leave Policy", "target_node": "Manager Approval", "relation": "requires", "fact": "All leave requests require Manager Approval 2 weeks in advance."}
]
}
"""
with patch("app.pipeline.graph.get_brain_client") as mock_get_brain:
mock_brain = AsyncMock()
mock_brain.generate.return_value = mock_json
mock_get_brain.return_value = mock_brain
chunks = [
Chunk(id="c1", doc_id="d1", text="Leave Policy text", token_count=4, content_hash="h1")
]
kg = await extract_entities_and_graph(chunks, provider=provider)
assert "leave policy" in kg.nodes
assert len(kg.edges) == 1
assert kg.edges[0].relation == "requires"
assert "Manager Approval" in kg.edges[0].fact
def test_expand_graph_knowledge_cap_and_labels():
"""Verify entity-adjacent query pulls in related facts formatted as (related: ...)."""
kg = KnowledgeGraph()
kg.add_node("Resignation Policy")
kg.add_node("Handover Document")
kg.add_edge(
"Resignation Policy",
"Handover Document",
"requires",
"Employees resigning must submit a Handover Document to their team lead.",
)
candidate_chunk = Chunk(
id="c1",
doc_id="d1",
text="Section 9: Resignation Policy details for full-time staff.",
token_count=10,
content_hash="h1",
)
initial_candidates = [TwoHopChunk(chunk=candidate_chunk)]
expanded, added_count = expand_graph_knowledge(
query="What are the resignation rules?",
candidates=initial_candidates,
graph=kg,
max_expansion=4,
)
assert added_count == 1
assert len(expanded) == 2
related_chunk = expanded[1]
assert related_chunk.is_two_hop is True
assert "related:" in (related_chunk.added_reason or "")
assert "(related: Employees resigning must submit a Handover Document" in related_chunk.chunk.text
@pytest.mark.anyio
async def test_stream_answer_pipeline_graph_expansion_integration():
"""
End-to-end gate: an entity-adjacent question pulls in a related fact with the label;
cap respected in trace.
"""
from app.pipeline.answer import stream_answer_pipeline
kg = KnowledgeGraph()
kg.add_node("Notice Period")
kg.add_node("Transition Plan")
kg.add_edge(
"Notice Period",
"Transition Plan",
"requires",
"Voluntary resignation notice requires an approved written Transition Plan.",
)
chunk = Chunk(
id="c1",
doc_id="d1",
text="Section 5.1: Voluntary resignation requires a Notice Period of 60 days.",
token_count=12,
content_hash="h1",
section_path=["5. Termination", "5.1 Notice Period"],
)
doc_index = DocumentIndex()
doc_index.build([chunk])
session = DocumentSession(
session_id="test_graph_stream_sess",
filename="policy.txt",
file_size_bytes=100,
page_count=1,
blocks=[],
chunks=[chunk],
ref_map=ReferenceMap(),
doc_index=doc_index,
knowledge_graph=kg,
)
events = []
async for sse in stream_answer_pipeline(
session, "What does the Notice Period require?", question_id="q_graph_test"
):
events.append(sse)
# Check expand event contains graph_expansion_count
expand_events = [e for e in events if "event: expand" in e]
assert len(expand_events) == 1
assert '"graph_expansion_count"' in expand_events[0]
# Check trace
trace = get_trace_manager().get_trace("q_graph_test")
assert trace is not None
assert "graph_expansion_count" in trace
assert trace["graph_expansion_count"] <= 4
assert trace["graph_expansion_count"] >= 1
|