danvics/haiku.rag
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Merge pull request #38 from ggozad/feat/radius-param

Browse source 
Allow radius as parameter to expand_context() for custom tools.
This commit is contained in:
Yiorgis Gozadinos 2025-08-12 16:55:01 +02:00 committed by GitHub
commit 74f5cbd3c5
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 93 additions and 88 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

5
docs/python.md
View file

@ -138,9 +138,12 @@ Expand search results with adjacent chunks for more complete context:
# Get initial search results
search_results = await client.search("machine learning", limit=3)
# Expand with adjacent chunks based on CONTEXT_CHUNK_RADIUS setting
# Expand with adjacent chunks using config setting
expanded_results = await client.expand_context(search_results)
# Or specify a custom radius
expanded_results = await client.expand_context(search_results, radius=2)
# The expanded results contain chunks with combined content from adjacent chunks
for chunk, score in expanded_results:
print(f"Expanded content: {chunk.content}") # Now includes before/after chunks

10
src/haiku/rag/client.py
View file

@ -349,17 +349,21 @@ class HaikuRAG:
return reranked_results
async def expand_context(
self, search_results: list[tuple[Chunk, float]]
self,
search_results: list[tuple[Chunk, float]],
radius: int = Config.CONTEXT_CHUNK_RADIUS,
) -> list[tuple[Chunk, float]]:
"""Expand search results with adjacent chunks, merging overlapping chunks.
Args:
search_results: List of (chunk, score) tuples from search.
radius: Number of adjacent chunks to include before/after each chunk.
Defaults to CONTEXT_CHUNK_RADIUS config setting.
Returns:
List of (chunk, score) tuples with expanded and merged context chunks.
"""
if Config.CONTEXT_CHUNK_RADIUS == 0:
if radius == 0:
return search_results
# Group chunks by document_id to handle merging within documents
@ -377,7 +381,7 @@ class HaikuRAG:
expanded_ranges = []
for chunk, score in doc_chunks:
adjacent_chunks = await self.chunk_repository.get_adjacent_chunks(
chunk, Config.CONTEXT_CHUNK_RADIUS
chunk, radius
)
all_chunks = adjacent_chunks + [chunk]

166
tests/test_client.py
View file

@ -644,109 +644,107 @@ async def test_client_expand_context_multiple_chunks():
@pytest.mark.asyncio
async def test_client_expand_context_merges_overlapping_chunks():
"""Test that overlapping expanded chunks are merged into one."""
with patch("haiku.rag.client.Config.CONTEXT_CHUNK_RADIUS", 1):
async with HaikuRAG(":memory:") as client:
# Create document with 5 chunks
manual_chunks = [
Chunk(content="Chunk 0", metadata={"order": 0}),
Chunk(content="Chunk 1", metadata={"order": 1}),
Chunk(content="Chunk 2", metadata={"order": 2}),
Chunk(content="Chunk 3", metadata={"order": 3}),
Chunk(content="Chunk 4", metadata={"order": 4}),
]
async with HaikuRAG(":memory:") as client:
# Create document with 5 chunks
manual_chunks = [
Chunk(content="Chunk 0", metadata={"order": 0}),
Chunk(content="Chunk 1", metadata={"order": 1}),
Chunk(content="Chunk 2", metadata={"order": 2}),
Chunk(content="Chunk 3", metadata={"order": 3}),
Chunk(content="Chunk 4", metadata={"order": 4}),
]
doc = await client.create_document(
content="Full document content", chunks=manual_chunks
)
doc = await client.create_document(
content="Full document content", chunks=manual_chunks
)
assert doc.id is not None
chunks = await client.chunk_repository.get_by_document_id(doc.id)
assert doc.id is not None
chunks = await client.chunk_repository.get_by_document_id(doc.id)
# Get adjacent chunks (orders 1 and 2) - these will overlap when expanded
chunk1 = next(c for c in chunks if c.metadata.get("order") == 1)
chunk2 = next(c for c in chunks if c.metadata.get("order") == 2)
# Get adjacent chunks (orders 1 and 2) - these will overlap when expanded
chunk1 = next(c for c in chunks if c.metadata.get("order") == 1)
chunk2 = next(c for c in chunks if c.metadata.get("order") == 2)
# With radius=1:
# chunk1 expanded would be [0,1,2]
# chunk2 expanded would be [1,2,3]
# These should merge into one chunk containing [0,1,2,3]
search_results = [(chunk1, 0.8), (chunk2, 0.7)]
expanded_results = await client.expand_context(search_results)
# With radius=1:
# chunk1 expanded would be [0,1,2]
# chunk2 expanded would be [1,2,3]
# These should merge into one chunk containing [0,1,2,3]
search_results = [(chunk1, 0.8), (chunk2, 0.7)]
expanded_results = await client.expand_context(search_results, radius=1)
# Should have only 1 merged result instead of 2 overlapping ones
assert len(expanded_results) == 1
# Should have only 1 merged result instead of 2 overlapping ones
assert len(expanded_results) == 1
merged_chunk, score = expanded_results[0]
merged_chunk, score = expanded_results[0]
# Should contain all chunks from 0 to 3
assert "Chunk 0" in merged_chunk.content
assert "Chunk 1" in merged_chunk.content
assert "Chunk 2" in merged_chunk.content
assert "Chunk 3" in merged_chunk.content
assert "Chunk 4" not in merged_chunk.content # Should not include chunk 4
# Should contain all chunks from 0 to 3
assert "Chunk 0" in merged_chunk.content
assert "Chunk 1" in merged_chunk.content
assert "Chunk 2" in merged_chunk.content
assert "Chunk 3" in merged_chunk.content
assert "Chunk 4" not in merged_chunk.content # Should not include chunk 4
# Should use the higher score (0.8)
assert score == 0.8
# Should use the higher score (0.8)
assert score == 0.8
@pytest.mark.asyncio
async def test_client_expand_context_keeps_separate_non_overlapping():
"""Test that non-overlapping expanded chunks remain separate."""
with patch("haiku.rag.client.Config.CONTEXT_CHUNK_RADIUS", 1):
async with HaikuRAG(":memory:") as client:
# Create document with chunks far apart
manual_chunks = [
Chunk(content="Chunk 0", metadata={"order": 0}),
Chunk(content="Chunk 1", metadata={"order": 1}),
Chunk(content="Chunk 2", metadata={"order": 2}),
Chunk(content="Chunk 5", metadata={"order": 5}), # Gap here
Chunk(content="Chunk 6", metadata={"order": 6}),
Chunk(content="Chunk 7", metadata={"order": 7}),
]
async with HaikuRAG(":memory:") as client:
# Create document with chunks far apart
manual_chunks = [
Chunk(content="Chunk 0", metadata={"order": 0}),
Chunk(content="Chunk 1", metadata={"order": 1}),
Chunk(content="Chunk 2", metadata={"order": 2}),
Chunk(content="Chunk 5", metadata={"order": 5}), # Gap here
Chunk(content="Chunk 6", metadata={"order": 6}),
Chunk(content="Chunk 7", metadata={"order": 7}),
]
doc = await client.create_document(
content="Full document content", chunks=manual_chunks
)
doc = await client.create_document(
content="Full document content", chunks=manual_chunks
)
assert doc.id is not None
chunks = await client.chunk_repository.get_by_document_id(doc.id)
assert doc.id is not None
chunks = await client.chunk_repository.get_by_document_id(doc.id)
# Get chunks by index - they will have sequential orders 0,1,2,3,4,5
# So get chunk with order=0 and chunk with order=5 (far enough apart)
chunk0 = next(
c for c in chunks if c.metadata.get("order") == 0
) # Content: "Chunk 0"
chunk5 = next(
c for c in chunks if c.metadata.get("order") == 5
) # Content: "Chunk 7"
# Get chunks by index - they will have sequential orders 0,1,2,3,4,5
# So get chunk with order=0 and chunk with order=5 (far enough apart)
chunk0 = next(
c for c in chunks if c.metadata.get("order") == 0
) # Content: "Chunk 0"
chunk5 = next(
c for c in chunks if c.metadata.get("order") == 5
) # Content: "Chunk 7"
# chunk0 expanded: [0,1] with radius=1 (orders 0,1)
# chunk5 expanded: [4,5] with radius=1 (orders 4,5)
# These should remain separate (max_order 1 < min_order 4 - 1)
search_results = [(chunk0, 0.8), (chunk5, 0.7)]
expanded_results = await client.expand_context(search_results)
# chunk0 expanded: [0,1] with radius=1 (orders 0,1)
# chunk5 expanded: [4,5] with radius=1 (orders 4,5)
# These should remain separate (max_order 1 < min_order 4 - 1)
search_results = [(chunk0, 0.8), (chunk5, 0.7)]
expanded_results = await client.expand_context(search_results, radius=1)
# Should have 2 separate results
assert len(expanded_results) == 2
# Should have 2 separate results
assert len(expanded_results) == 2
# Sort by score to ensure predictable order
expanded_results.sort(key=lambda x: x[1], reverse=True)
# Sort by score to ensure predictable order
expanded_results.sort(key=lambda x: x[1], reverse=True)
chunk0_expanded, score1 = expanded_results[0]
chunk5_expanded, score2 = expanded_results[1]
chunk0_expanded, score1 = expanded_results[0]
chunk5_expanded, score2 = expanded_results[1]
# First chunk (order=0) expanded should contain orders [0,1]
# Content should be "Chunk 0" + "Chunk 1"
assert "Chunk 0" in chunk0_expanded.content
assert "Chunk 1" in chunk0_expanded.content
assert (
"Chunk 7" not in chunk0_expanded.content
) # Should not have chunk 7 content
assert score1 == 0.8
# First chunk (order=0) expanded should contain orders [0,1]
# Content should be "Chunk 0" + "Chunk 1"
assert "Chunk 0" in chunk0_expanded.content
assert "Chunk 1" in chunk0_expanded.content
assert (
"Chunk 7" not in chunk0_expanded.content
) # Should not have chunk 7 content
assert score1 == 0.8
# Second chunk (order=5) expanded should contain orders [4,5]
# Content should be "Chunk 6" + "Chunk 7" (orders 4 and 5)
assert "Chunk 6" in chunk5_expanded.content # Order 4 content
assert "Chunk 7" in chunk5_expanded.content # Order 5 content
assert "Chunk 0" not in chunk5_expanded.content
assert score2 == 0.7
# Second chunk (order=5) expanded should contain orders [4,5]
# Content should be "Chunk 6" + "Chunk 7" (orders 4 and 5)
assert "Chunk 6" in chunk5_expanded.content # Order 4 content
assert "Chunk 7" in chunk5_expanded.content # Order 5 content
assert "Chunk 0" not in chunk5_expanded.content
assert score2 == 0.7