apunkt/SurfSense
1
0
Fork 0
mirror of https://github.com/MODSetter/SurfSense.git synced 2026-06-26 21:39:43 +02:00
Code Issues Projects Releases Packages Wiki Activity

search-kb: on-demand KB tool on the [n] spine; drop kb_matched_chunk_ids

Browse source 
The main agent's search_knowledge_base tool runs the hybrid spine, renders
a <retrieved_context> of numbered [n] passages, and persists the registry.
KB subagent prompts teach citing [n] from <document view="full"> reads
(evidence.chunk_ids -> evidence.citations). Delete the now-unused
search->read highlighting hand-off: the kb_matched_chunk_ids state field,
its reducer default, the tool's _matched_chunk_ids writer, and the dead
KnowledgePriorityMiddleware writes.
This commit is contained in:
CREDO23 2026-06-25 15:26:39 +02:00
parent 04a76b163b
commit c98bdea5cf
16 changed files with 518 additions and 325 deletions
Download patch file Download diff file Expand all files Collapse all files

85
surfsense_backend/tests/unit/middleware/test_knowledge_search.py
View file

@ -6,9 +6,6 @@ import pytest
from langchain_core.messages import AIMessage, HumanMessage
from app.agents.chat.multi_agent_chat.shared.middleware import knowledge_search as ks
from app.agents.chat.multi_agent_chat.shared.middleware.filesystem.backends.document_xml import (
build_document_xml as _build_document_xml,
)
from app.agents.chat.multi_agent_chat.shared.middleware.knowledge_search import (
KBSearchPlan,
KnowledgePriorityMiddleware,
@ -59,88 +56,6 @@ class TestResolveSearchTypes:
assert result.count("FILE") == 1
# ── _build_document_xml ────────────────────────────────────────────────
class TestBuildDocumentXml:
@pytest.fixture
def sample_document(self):
return {
"document_id": 42,
"document": {
"id": 42,
"document_type": "FILE",
"title": "Test Doc",
"metadata": {"url": "https://example.com"},
},
"chunks": [
{"chunk_id": 101, "content": "First chunk content"},
{"chunk_id": 102, "content": "Second chunk content"},
{"chunk_id": 103, "content": "Third chunk content"},
],
}
def test_contains_document_metadata(self, sample_document):
xml = _build_document_xml(sample_document)
assert "<document_id>42</document_id>" in xml
assert "<document_type>FILE</document_type>" in xml
assert "Test Doc" in xml
def test_contains_chunk_index(self, sample_document):
xml = _build_document_xml(sample_document)
assert "<chunk_index>" in xml
assert "</chunk_index>" in xml
assert 'chunk_id="101"' in xml
assert 'chunk_id="102"' in xml
assert 'chunk_id="103"' in xml
def test_matched_chunks_flagged_in_index(self, sample_document):
xml = _build_document_xml(sample_document, matched_chunk_ids={101, 103})
lines = xml.split("\n")
for line in lines:
if 'chunk_id="101"' in line:
assert 'matched="true"' in line
if 'chunk_id="102"' in line:
assert 'matched="true"' not in line
if 'chunk_id="103"' in line:
assert 'matched="true"' in line
def test_chunk_content_in_document_content_section(self, sample_document):
xml = _build_document_xml(sample_document)
assert "<document_content>" in xml
assert "First chunk content" in xml
assert "Second chunk content" in xml
assert "Third chunk content" in xml
def test_line_numbers_in_chunk_index_are_accurate(self, sample_document):
"""Verify that the line ranges in chunk_index actually point to the right content."""
xml = _build_document_xml(sample_document, matched_chunk_ids={101})
xml_lines = xml.split("\n")
for line in xml_lines:
if 'chunk_id="101"' in line and "lines=" in line:
import re
m = re.search(r'lines="(\d+)-(\d+)"', line)
assert m, f"No lines= attribute found in: {line}"
start, _end = int(m.group(1)), int(m.group(2))
target_line = xml_lines[start - 1]
assert "101" in target_line
assert "First chunk content" in target_line
break
else:
pytest.fail("chunk_id=101 entry not found in chunk_index")
def test_splits_into_lines_correctly(self, sample_document):
"""Each chunk occupies exactly one line (no embedded newlines)."""
xml = _build_document_xml(sample_document)
lines = xml.split("\n")
chunk_lines = [
line for line in lines if "<![CDATA[" in line and "<chunk" in line
]
assert len(chunk_lines) == 3
# ── planner parsing / date normalization ───────────────────────────────