"""
Tests for context pack generation
Tests GET /context/pack endpoint
"""
import pytest
from src.codebase_rag.services.pipeline import PackBuilder
class TestPackBuilder:
"""Test pack builder service"""
@pytest.mark.unit
def test_build_context_pack_basic(self):
"""Test basic context pack building"""
pack_builder = PackBuilder()
nodes = [
{
"type": "file",
"path": "src/auth/token.py",
"lang": "python",
"score": 0.9,
"summary": "Python file token.py in auth/ directory",
"ref": "ref://file/src/auth/token.py#L1-L100"
},
{
"type": "file",
"path": "src/auth/user.py",
"lang": "python",
"score": 0.8,
"summary": "Python file user.py in auth/ directory",
"ref": "ref://file/src/auth/user.py#L1-L200"
}
]
pack = pack_builder.build_context_pack(
nodes=nodes,
budget=1500,
stage="plan",
repo_id="test-repo",
keywords=["auth"],
focus_paths=[]
)
assert "items" in pack
assert "budget_used" in pack
assert "budget_limit" in pack
assert pack["budget_limit"] == 1500
assert pack["budget_used"] <= 1500
assert len(pack["items"]) > 0
@pytest.mark.unit
def test_pack_respects_budget(self):
"""Test that pack builder respects token budget"""
pack_builder = PackBuilder()
# Create many nodes
nodes = []
for i in range(50):
nodes.append({
"type": "file",
"path": f"src/module_{i}/file.py",
"lang": "python",
"score": 1.0 - (i * 0.01),
"summary": f"Python file for module {i} with some description",
"ref": f"ref://file/src/module_{i}/file.py#L1-L100"
})
# Small budget
pack = pack_builder.build_context_pack(
nodes=nodes,
budget=500,
stage="plan",
repo_id="test-repo"
)
# Should fit within budget
assert pack["budget_used"] <= 500
# Should have selected some but not all items
assert 0 < len(pack["items"]) < len(nodes)
@pytest.mark.unit
def test_pack_prioritizes_high_scores(self):
"""Test that higher scored items are prioritized"""
pack_builder = PackBuilder()
nodes = [
{
"type": "file",
"path": "low_score.py",
"lang": "python",
"score": 0.1,
"summary": "Low score file",
"ref": "ref://file/low_score.py#L1-L100"
},
{
"type": "file",
"path": "high_score.py",
"lang": "python",
"score": 0.9,
"summary": "High score file",
"ref": "ref://file/high_score.py#L1-L100"
}
]
pack = pack_builder.build_context_pack(
nodes=nodes,
budget=300, # Only room for one
stage="plan",
repo_id="test-repo"
)
# Should select the high score file
if len(pack["items"]) > 0:
first_item = pack["items"][0]
assert "high_score" in first_item["ref"]
@pytest.mark.unit
def test_pack_focus_paths_priority(self):
"""Test that focus paths get priority"""
pack_builder = PackBuilder()
nodes = [
{
"type": "file",
"path": "src/important/critical.py",
"lang": "python",
"score": 0.5,
"summary": "Critical file",
"ref": "ref://file/src/important/critical.py#L1-L100"
},
{
"type": "file",
"path": "src/other/regular.py",
"lang": "python",
"score": 0.9,
"summary": "Regular file",
"ref": "ref://file/src/other/regular.py#L1-L100"
}
]
pack = pack_builder.build_context_pack(
nodes=nodes,
budget=300, # Only room for one
stage="plan",
repo_id="test-repo",
focus_paths=["src/important"]
)
# Focus path should be prioritized even with lower score
if len(pack["items"]) > 0:
first_item = pack["items"][0]
assert "important" in first_item["ref"]
@pytest.mark.unit
def test_title_extraction_via_build_context_pack(self):
"""Test that titles are properly extracted from paths via build_context_pack"""
pack_builder = PackBuilder()
nodes = [
{
"type": "file",
"path": "src/auth/token.py",
"lang": "python",
"score": 0.9,
"summary": "Test file",
"ref": "ref://file/src/auth/token.py"
},
{
"type": "file",
"path": "main.py",
"lang": "python",
"score": 0.8,
"summary": "Test file",
"ref": "ref://file/main.py"
}
]
pack = pack_builder.build_context_pack(
nodes=nodes,
budget=1500,
stage="plan",
repo_id="test-repo"
)
# Verify titles are extracted correctly (last 2 path segments)
assert len(pack["items"]) == 2
# Multi-level path should show last 2 segments
multi_level_item = next(item for item in pack["items"] if "auth" in item["ref"])
assert multi_level_item["title"] == "auth/token.py"
# Single level path should show as-is
single_level_item = next(item for item in pack["items"] if item["ref"] == "ref://file/main.py")
assert single_level_item["title"] == "main.py"
class TestContextPackAPI:
"""Test context pack API endpoint"""
@pytest.mark.integration
def test_context_pack_basic(self, test_client, test_repo_id):
"""Test basic context pack endpoint"""
response = test_client.get(
f"/api/v1/context/pack?repoId={test_repo_id}&stage=plan&budget=1500"
)
assert response.status_code == 200
data = response.json()
assert "items" in data
assert "budget_used" in data
assert "budget_limit" in data
assert "stage" in data
assert "repo_id" in data
assert data["stage"] == "plan"
assert data["repo_id"] == test_repo_id
assert data["budget_limit"] == 1500
@pytest.mark.integration
def test_context_pack_with_keywords(self, test_client, test_repo_id):
"""Test context pack with keyword filtering"""
response = test_client.get(
f"/api/v1/context/pack?repoId={test_repo_id}&stage=plan&budget=1500&keywords=auth,token"
)
assert response.status_code == 200
data = response.json()
# Items should be relevant to keywords
assert len(data["items"]) >= 0
@pytest.mark.integration
def test_context_pack_with_focus(self, test_client, test_repo_id):
"""Test context pack with focus paths"""
response = test_client.get(
f"/api/v1/context/pack?repoId={test_repo_id}&stage=plan&budget=1500&focus=src/auth"
)
assert response.status_code == 200
data = response.json()
assert "items" in data
@pytest.mark.integration
def test_context_pack_different_stages(self, test_client, test_repo_id):
"""Test context pack with different stages"""
stages = ["plan", "review", "implement"]
for stage in stages:
response = test_client.get(
f"/api/v1/context/pack?repoId={test_repo_id}&stage={stage}&budget=1000"
)
assert response.status_code == 200
data = response.json()
assert data["stage"] == stage
@pytest.mark.integration
def test_context_pack_budget_limits(self, test_client, test_repo_id):
"""Test that different budgets produce different sized packs"""
# Small budget
response_small = test_client.get(
f"/api/v1/context/pack?repoId={test_repo_id}&stage=plan&budget=500"
)
# Large budget
response_large = test_client.get(
f"/api/v1/context/pack?repoId={test_repo_id}&stage=plan&budget=5000"
)
assert response_small.status_code == 200
assert response_large.status_code == 200
small_data = response_small.json()
large_data = response_large.json()
# Large budget should allow more items (if data available)
assert small_data["budget_used"] <= 500
assert large_data["budget_used"] <= 5000
@pytest.mark.integration
def test_context_pack_item_format(self, test_client, test_repo_path, test_repo_id):
"""Test that context pack items have correct format"""
# First ingest data
ingest_response = test_client.post("/api/v1/ingest/repo", json={
"local_path": test_repo_path,
"include_globs": ["**/*.py"],
"exclude_globs": []
})
assert ingest_response.status_code == 200
# Get context pack
response = test_client.get(
f"/api/v1/context/pack?repoId={test_repo_id}&stage=plan&budget=2000"
)
assert response.status_code == 200
data = response.json()
# Check item format
for item in data["items"]:
assert "kind" in item
assert "title" in item
assert "summary" in item
assert "ref" in item
assert item["kind"] in ["file", "symbol", "guideline"]
assert item["ref"].startswith("ref://")
class TestContextPackIntegration:
"""Integration tests for context pack workflow"""
@pytest.mark.integration
def test_full_context_workflow(self, test_client, test_repo_path, test_repo_id):
"""Test full workflow: ingest -> related -> context pack"""
# 1. Ingest repository
ingest_response = test_client.post("/api/v1/ingest/repo", json={
"local_path": test_repo_path,
"include_globs": ["**/*.py", "**/*.ts"],
"exclude_globs": []
})
assert ingest_response.status_code == 200
# 2. Find related files
related_response = test_client.get(
f"/api/v1/graph/related?query=helper&repoId={test_repo_id}&limit=10"
)
assert related_response.status_code == 200
# 3. Build context pack
pack_response = test_client.get(
f"/api/v1/context/pack?repoId={test_repo_id}&stage=plan&budget=1500&keywords=helper"
)
assert pack_response.status_code == 200
pack_data = pack_response.json()
# Verify context pack is usable for prompts
assert pack_data["budget_used"] <= 1500
assert len(pack_data["items"]) > 0
# Each item should have enough info for MCP integration
first_item = pack_data["items"][0]
assert "ref" in first_item # Handle for MCP
assert "summary" in first_item # Brief description
assert "title" in first_item # Display name
