test_selector.py

"""
Manual verification tests for the ToolSelector.
Run with: python3 test_selector.py
"""

from tool_selector import ToolRegistry, ToolSelector, ToolEntry

CASES = [
    # (task description, expected top tool name)
    ("convert 100 USD to Japanese yen", "currency_converter"),
    ("calculate 17 percent of 340", "calculator"),
    ("what is the latest news about AI?", "web_search"),
    ("run this python snippet and show output", "code_executor"),
    ("summarize this long article", "summarizer"),
    ("convert 5 miles to kilometers", "unit_converter"),
    ("how many days between Jan 1 and March 15?", "calendar_tool"),
    ("read the CSV file and show column names", "file_reader"),
    ("send an email to john@example.com", "email_sender"),
    ("count users who signed up last month from the database", "database_query"),
]

PASS = 0
FAIL = 0

registry = ToolRegistry.default()
selector = ToolSelector(registry, top_n=3)

print("=" * 60)
print("ToolSelector — test suite")
print("=" * 60)

for task, expected_top in CASES:
    recs = selector.recommend(task)
    top_name = recs[0].tool.name if recs else "(none)"
    passed = top_name == expected_top
    status = "PASS" if passed else "FAIL"
    if passed:
        PASS += 1
    else:
        FAIL += 1
    score = recs[0].score if recs else 0.0
    print(f"[{status}] task='{task[:45]}...' | expected={expected_top} | got={top_name} (score={score:.3f})")

print()
print(f"Results: {PASS}/{PASS+FAIL} passed")

# Custom registry test
print()
print("--- Custom registry test ---")
custom = ToolRegistry()
custom.register(ToolEntry(
    name="my_custom_tool",
    description="Handles invoice parsing and extraction.",
    capabilities=["invoice", "parsing", "extraction", "ocr"],
    use_cases=["extract invoice", "parse invoice", "invoice data"],
    tags=["finance", "document"],
))
sel2 = ToolSelector(custom, top_n=1)
recs2 = sel2.recommend("extract data from this invoice PDF")
assert recs2 and recs2[0].tool.name == "my_custom_tool", "Custom tool test FAILED"
print("[PASS] Custom tool registry test")

# JSON round-trip test
import tempfile, json, pathlib
print()
print("--- Persistence round-trip test ---")
with tempfile.NamedTemporaryFile(suffix=".json", delete=False) as f:
    tmp_path = f.name
registry.save(tmp_path)
reloaded = ToolRegistry.load(tmp_path)
assert len(reloaded) == len(registry), "Persistence test FAILED: length mismatch"
print(f"[PASS] Saved and reloaded {len(reloaded)} tools from JSON")

# CLI JSON output test
print()
print("--- CLI JSON output test ---")
from tool_selector.cli import main as cli_main
import io, contextlib

buf = io.StringIO()
with contextlib.redirect_stdout(buf):
    cli_main(["--json", "convert 50 EUR to dollars"])
output = json.loads(buf.getvalue())
assert output["recommendations"][0]["name"] == "currency_converter", "CLI test FAILED"
print(f"[PASS] CLI JSON output correct: top={output['recommendations'][0]['name']}")

print()
if FAIL == 0:
    print("All tests passed.")
else:
    print(f"WARNING: {FAIL} test(s) failed.")