Merge origin/main into pr-2959

Resolve the config plumbing conflicts and keep disabled skill filtering consistent for subagent prompts after syncing with main.

Made-with: Cursor
This commit is contained in:
Xubin Ren
2026-04-12 02:02:39 +00:00
56 changed files with 9305 additions and 463 deletions
+423
View File
@@ -0,0 +1,423 @@
"""Tests for advanced EditFileTool enhancements inspired by claude-code:
- Delete-line newline cleanup
- Smart quote normalization (curly ↔ straight)
- Quote style preservation in replacements
- Indentation preservation when fallback match is trimmed
- Trailing whitespace stripping for new_text
- File size protection
- Stale detection with content-equality fallback
"""
import os
import time
import pytest
from nanobot.agent.tools.filesystem import EditFileTool, ReadFileTool, _find_match
from nanobot.agent.tools import file_state
@pytest.fixture(autouse=True)
def _clear_file_state():
file_state.clear()
yield
file_state.clear()
# ---------------------------------------------------------------------------
# Delete-line newline cleanup
# ---------------------------------------------------------------------------
class TestDeleteLineCleanup:
"""When new_text='' and deleting a line, trailing newline should be consumed."""
@pytest.fixture()
def tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_delete_line_consumes_trailing_newline(self, tool, tmp_path):
f = tmp_path / "a.py"
f.write_text("line1\nline2\nline3\n", encoding="utf-8")
result = await tool.execute(path=str(f), old_text="line2", new_text="")
assert "Successfully" in result
content = f.read_text()
# Should not leave a blank line where line2 was
assert content == "line1\nline3\n"
@pytest.mark.asyncio
async def test_delete_line_with_explicit_newline_in_old_text(self, tool, tmp_path):
f = tmp_path / "a.py"
f.write_text("line1\nline2\nline3\n", encoding="utf-8")
result = await tool.execute(path=str(f), old_text="line2\n", new_text="")
assert "Successfully" in result
assert f.read_text() == "line1\nline3\n"
@pytest.mark.asyncio
async def test_delete_preserves_content_when_not_trailing_newline(self, tool, tmp_path):
"""Deleting a word mid-line should not consume extra characters."""
f = tmp_path / "a.py"
f.write_text("hello world here\n", encoding="utf-8")
result = await tool.execute(path=str(f), old_text="world ", new_text="")
assert "Successfully" in result
assert f.read_text() == "hello here\n"
# ---------------------------------------------------------------------------
# Smart quote normalization
# ---------------------------------------------------------------------------
class TestSmartQuoteNormalization:
"""_find_match should handle curly ↔ straight quote fallback."""
def test_curly_double_quotes_match_straight(self):
content = 'She said \u201chello\u201d to him'
old_text = 'She said "hello" to him'
match, count = _find_match(content, old_text)
assert match is not None
assert count == 1
# Returned match should be the ORIGINAL content with curly quotes
assert "\u201c" in match
def test_curly_single_quotes_match_straight(self):
content = "it\u2019s a test"
old_text = "it's a test"
match, count = _find_match(content, old_text)
assert match is not None
assert count == 1
assert "\u2019" in match
def test_straight_matches_curly_in_old_text(self):
content = 'x = "hello"'
old_text = 'x = \u201chello\u201d'
match, count = _find_match(content, old_text)
assert match is not None
assert count == 1
def test_exact_match_still_preferred_over_quote_normalization(self):
content = 'x = "hello"'
old_text = 'x = "hello"'
match, count = _find_match(content, old_text)
assert match == old_text
assert count == 1
class TestQuoteStylePreservation:
"""When quote-normalized matching occurs, replacement should preserve actual quote style."""
@pytest.fixture()
def tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_replacement_preserves_curly_double_quotes(self, tool, tmp_path):
f = tmp_path / "quotes.txt"
f.write_text('message = “hello”\n', encoding="utf-8")
result = await tool.execute(
path=str(f),
old_text='message = "hello"',
new_text='message = "goodbye"',
)
assert "Successfully" in result
assert f.read_text(encoding="utf-8") == 'message = “goodbye”\n'
@pytest.mark.asyncio
async def test_replacement_preserves_curly_apostrophe(self, tool, tmp_path):
f = tmp_path / "apostrophe.txt"
f.write_text("its fine\n", encoding="utf-8")
result = await tool.execute(
path=str(f),
old_text="it's fine",
new_text="it's better",
)
assert "Successfully" in result
assert f.read_text(encoding="utf-8") == "its better\n"
# ---------------------------------------------------------------------------
# Indentation preservation
# ---------------------------------------------------------------------------
class TestIndentationPreservation:
"""Replacement should keep outer indentation when trim fallback matched."""
@pytest.fixture()
def tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_trim_fallback_preserves_outer_indentation(self, tool, tmp_path):
f = tmp_path / "indent.py"
f.write_text(
"if True:\n"
" def foo():\n"
" pass\n",
encoding="utf-8",
)
result = await tool.execute(
path=str(f),
old_text="def foo():\n pass",
new_text="def bar():\n return 1",
)
assert "Successfully" in result
assert f.read_text(encoding="utf-8") == (
"if True:\n"
" def bar():\n"
" return 1\n"
)
# ---------------------------------------------------------------------------
# Failure diagnostics
# ---------------------------------------------------------------------------
class TestEditDiagnostics:
"""Failure paths should offer actionable hints."""
@pytest.fixture()
def tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_ambiguous_match_reports_candidate_lines(self, tool, tmp_path):
f = tmp_path / "dup.py"
f.write_text("aaa\nbbb\naaa\nbbb\n", encoding="utf-8")
result = await tool.execute(path=str(f), old_text="aaa\nbbb", new_text="xxx")
assert "appears 2 times" in result.lower()
assert "line 1" in result.lower()
assert "line 3" in result.lower()
assert "replace_all=true" in result
@pytest.mark.asyncio
async def test_not_found_reports_whitespace_hint(self, tool, tmp_path):
f = tmp_path / "space.py"
f.write_text("value = 1\n", encoding="utf-8")
result = await tool.execute(path=str(f), old_text="value = 1", new_text="value = 2")
assert "Error" in result
assert "whitespace" in result.lower()
@pytest.mark.asyncio
async def test_not_found_reports_case_hint(self, tool, tmp_path):
f = tmp_path / "case.py"
f.write_text("HelloWorld\n", encoding="utf-8")
result = await tool.execute(path=str(f), old_text="helloworld", new_text="goodbye")
assert "Error" in result
assert "letter case differs" in result.lower()
# ---------------------------------------------------------------------------
# Advanced fallback replacement behavior
# ---------------------------------------------------------------------------
class TestAdvancedReplaceAll:
"""replace_all should work correctly for fallback-based matches too."""
@pytest.fixture()
def tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_replace_all_preserves_each_match_indentation(self, tool, tmp_path):
f = tmp_path / "indent_multi.py"
f.write_text(
"if a:\n"
" def foo():\n"
" pass\n"
"if b:\n"
" def foo():\n"
" pass\n",
encoding="utf-8",
)
result = await tool.execute(
path=str(f),
old_text="def foo():\n pass",
new_text="def bar():\n return 1",
replace_all=True,
)
assert "Successfully" in result
assert f.read_text(encoding="utf-8") == (
"if a:\n"
" def bar():\n"
" return 1\n"
"if b:\n"
" def bar():\n"
" return 1\n"
)
@pytest.mark.asyncio
async def test_trim_and_quote_fallback_match_succeeds(self, tool, tmp_path):
f = tmp_path / "quote_indent.py"
f.write_text(" message = “hello”\n", encoding="utf-8")
result = await tool.execute(
path=str(f),
old_text='message = "hello"',
new_text='message = "goodbye"',
)
assert "Successfully" in result
assert f.read_text(encoding="utf-8") == " message = “goodbye”\n"
# ---------------------------------------------------------------------------
# Advanced fallback replacement behavior
# ---------------------------------------------------------------------------
class TestAdvancedReplaceAll:
"""replace_all should work correctly for fallback-based matches too."""
@pytest.fixture()
def tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_replace_all_preserves_each_match_indentation(self, tool, tmp_path):
f = tmp_path / "indent_multi.py"
f.write_text(
"if a:\n"
" def foo():\n"
" pass\n"
"if b:\n"
" def foo():\n"
" pass\n",
encoding="utf-8",
)
result = await tool.execute(
path=str(f),
old_text="def foo():\n pass",
new_text="def bar():\n return 1",
replace_all=True,
)
assert "Successfully" in result
assert f.read_text(encoding="utf-8") == (
"if a:\n"
" def bar():\n"
" return 1\n"
"if b:\n"
" def bar():\n"
" return 1\n"
)
@pytest.mark.asyncio
async def test_trim_and_quote_fallback_match_succeeds(self, tool, tmp_path):
f = tmp_path / "quote_indent.py"
f.write_text(" message = “hello”\n", encoding="utf-8")
result = await tool.execute(
path=str(f),
old_text='message = "hello"',
new_text='message = "goodbye"',
)
assert "Successfully" in result
assert f.read_text(encoding="utf-8") == " message = “goodbye”\n"
# ---------------------------------------------------------------------------
# Trailing whitespace stripping on new_text
# ---------------------------------------------------------------------------
class TestTrailingWhitespaceStrip:
"""new_text trailing whitespace should be stripped (except .md files)."""
@pytest.fixture()
def tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_strips_trailing_whitespace_from_new_text(self, tool, tmp_path):
f = tmp_path / "a.py"
f.write_text("x = 1\n", encoding="utf-8")
result = await tool.execute(
path=str(f), old_text="x = 1", new_text="x = 2 \ny = 3 ",
)
assert "Successfully" in result
content = f.read_text()
assert "x = 2\ny = 3\n" == content
@pytest.mark.asyncio
async def test_preserves_trailing_whitespace_in_markdown(self, tool, tmp_path):
f = tmp_path / "doc.md"
f.write_text("# Title\n", encoding="utf-8")
# Markdown uses trailing double-space for line breaks
result = await tool.execute(
path=str(f), old_text="# Title", new_text="# Title \nSubtitle ",
)
assert "Successfully" in result
content = f.read_text()
# Trailing spaces should be preserved for markdown
assert "Title " in content
assert "Subtitle " in content
# ---------------------------------------------------------------------------
# File size protection
# ---------------------------------------------------------------------------
class TestFileSizeProtection:
"""Editing extremely large files should be rejected."""
@pytest.fixture()
def tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_rejects_file_over_size_limit(self, tool, tmp_path):
f = tmp_path / "huge.txt"
f.write_text("x", encoding="utf-8")
# Monkey-patch the file size check by creating a stat mock
original_stat = f.stat
class FakeStat:
def __init__(self, real_stat):
self._real = real_stat
def __getattr__(self, name):
return getattr(self._real, name)
@property
def st_size(self):
return 2 * 1024 * 1024 * 1024 # 2 GiB
import unittest.mock
with unittest.mock.patch.object(type(f), 'stat', return_value=FakeStat(f.stat())):
result = await tool.execute(path=str(f), old_text="x", new_text="y")
assert "Error" in result
assert "too large" in result.lower() or "size" in result.lower()
# ---------------------------------------------------------------------------
# Stale detection with content-equality fallback
# ---------------------------------------------------------------------------
class TestStaleDetectionContentFallback:
"""When mtime changed but file content is unchanged, edit should proceed without warning."""
@pytest.fixture()
def read_tool(self, tmp_path):
return ReadFileTool(workspace=tmp_path)
@pytest.fixture()
def edit_tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_mtime_bump_same_content_no_warning(self, read_tool, edit_tool, tmp_path):
f = tmp_path / "a.py"
f.write_text("hello world", encoding="utf-8")
await read_tool.execute(path=str(f))
# Touch the file to bump mtime without changing content
time.sleep(0.05)
original_content = f.read_text()
f.write_text(original_content, encoding="utf-8")
result = await edit_tool.execute(path=str(f), old_text="world", new_text="earth")
assert "Successfully" in result
# Should NOT warn about modification since content is the same
assert "modified" not in result.lower()
+152
View File
@@ -0,0 +1,152 @@
"""Tests for EditFileTool enhancements: read-before-edit tracking, path suggestions,
.ipynb detection, and create-file semantics."""
import pytest
from nanobot.agent.tools.filesystem import EditFileTool, ReadFileTool, WriteFileTool
from nanobot.agent.tools import file_state
# ---------------------------------------------------------------------------
# Helpers
# ---------------------------------------------------------------------------
@pytest.fixture(autouse=True)
def _clear_file_state():
"""Reset global read-state between tests."""
file_state.clear()
yield
file_state.clear()
# ---------------------------------------------------------------------------
# Read-before-edit tracking
# ---------------------------------------------------------------------------
class TestEditReadTracking:
"""edit_file should warn when file hasn't been read first."""
@pytest.fixture()
def read_tool(self, tmp_path):
return ReadFileTool(workspace=tmp_path)
@pytest.fixture()
def edit_tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_edit_warns_if_file_not_read_first(self, edit_tool, tmp_path):
f = tmp_path / "a.py"
f.write_text("hello world", encoding="utf-8")
result = await edit_tool.execute(path=str(f), old_text="world", new_text="earth")
# Should still succeed but include a warning
assert "Successfully" in result
assert "not been read" in result.lower() or "warning" in result.lower()
@pytest.mark.asyncio
async def test_edit_succeeds_cleanly_after_read(self, read_tool, edit_tool, tmp_path):
f = tmp_path / "a.py"
f.write_text("hello world", encoding="utf-8")
await read_tool.execute(path=str(f))
result = await edit_tool.execute(path=str(f), old_text="world", new_text="earth")
assert "Successfully" in result
# No warning when file was read first
assert "not been read" not in result.lower()
assert f.read_text() == "hello earth"
@pytest.mark.asyncio
async def test_edit_warns_if_file_modified_since_read(self, read_tool, edit_tool, tmp_path):
f = tmp_path / "a.py"
f.write_text("hello world", encoding="utf-8")
await read_tool.execute(path=str(f))
# External modification
f.write_text("hello universe", encoding="utf-8")
result = await edit_tool.execute(path=str(f), old_text="universe", new_text="earth")
assert "Successfully" in result
assert "modified" in result.lower() or "warning" in result.lower()
# ---------------------------------------------------------------------------
# Create-file semantics
# ---------------------------------------------------------------------------
class TestEditCreateFile:
"""edit_file with old_text='' creates new file if not exists."""
@pytest.fixture()
def tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_create_new_file_with_empty_old_text(self, tool, tmp_path):
f = tmp_path / "subdir" / "new.py"
result = await tool.execute(path=str(f), old_text="", new_text="print('hi')")
assert "created" in result.lower() or "Successfully" in result
assert f.exists()
assert f.read_text() == "print('hi')"
@pytest.mark.asyncio
async def test_create_fails_if_file_already_exists_and_not_empty(self, tool, tmp_path):
f = tmp_path / "existing.py"
f.write_text("existing content", encoding="utf-8")
result = await tool.execute(path=str(f), old_text="", new_text="new content")
assert "Error" in result or "already exists" in result.lower()
# File should be unchanged
assert f.read_text() == "existing content"
@pytest.mark.asyncio
async def test_create_succeeds_if_file_exists_but_empty(self, tool, tmp_path):
f = tmp_path / "empty.py"
f.write_text("", encoding="utf-8")
result = await tool.execute(path=str(f), old_text="", new_text="print('hi')")
assert "Successfully" in result
assert f.read_text() == "print('hi')"
# ---------------------------------------------------------------------------
# .ipynb detection
# ---------------------------------------------------------------------------
class TestEditIpynbDetection:
"""edit_file should refuse .ipynb and suggest notebook_edit."""
@pytest.fixture()
def tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_ipynb_rejected_with_suggestion(self, tool, tmp_path):
f = tmp_path / "analysis.ipynb"
f.write_text('{"cells": []}', encoding="utf-8")
result = await tool.execute(path=str(f), old_text="x", new_text="y")
assert "notebook" in result.lower()
# ---------------------------------------------------------------------------
# Path suggestion on not-found
# ---------------------------------------------------------------------------
class TestEditPathSuggestion:
"""edit_file should suggest similar paths on not-found."""
@pytest.fixture()
def tool(self, tmp_path):
return EditFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_suggests_similar_filename(self, tool, tmp_path):
f = tmp_path / "config.py"
f.write_text("x = 1", encoding="utf-8")
# Typo: conifg.py
result = await tool.execute(
path=str(tmp_path / "conifg.py"), old_text="x = 1", new_text="x = 2",
)
assert "Error" in result
assert "config.py" in result
@pytest.mark.asyncio
async def test_shows_cwd_in_error(self, tool, tmp_path):
result = await tool.execute(
path=str(tmp_path / "nonexistent.py"), old_text="a", new_text="b",
)
assert "Error" in result
+31
View File
@@ -43,3 +43,34 @@ async def test_exec_path_append_preserves_system_path():
tool = ExecTool(path_append="/opt/custom/bin")
result = await tool.execute(command="ls /")
assert "Exit code: 0" in result
@_UNIX_ONLY
@pytest.mark.asyncio
async def test_exec_allowed_env_keys_passthrough(monkeypatch):
"""Env vars listed in allowed_env_keys should be visible to commands."""
monkeypatch.setenv("MY_CUSTOM_VAR", "hello-from-config")
tool = ExecTool(allowed_env_keys=["MY_CUSTOM_VAR"])
result = await tool.execute(command="printenv MY_CUSTOM_VAR")
assert "hello-from-config" in result
@_UNIX_ONLY
@pytest.mark.asyncio
async def test_exec_allowed_env_keys_does_not_leak_others(monkeypatch):
"""Env vars NOT in allowed_env_keys should still be blocked."""
monkeypatch.setenv("MY_CUSTOM_VAR", "hello-from-config")
monkeypatch.setenv("MY_SECRET_VAR", "secret-value")
tool = ExecTool(allowed_env_keys=["MY_CUSTOM_VAR"])
result = await tool.execute(command="printenv MY_SECRET_VAR")
assert "secret-value" not in result
@_UNIX_ONLY
@pytest.mark.asyncio
async def test_exec_allowed_env_keys_missing_var_ignored(monkeypatch):
"""If an allowed key is not set in the parent process, it should be silently skipped."""
monkeypatch.delenv("NONEXISTENT_VAR_12345", raising=False)
tool = ExecTool(allowed_env_keys=["NONEXISTENT_VAR_12345"])
result = await tool.execute(command="printenv NONEXISTENT_VAR_12345")
assert "Exit code: 1" in result
+76 -72
View File
@@ -271,15 +271,11 @@ async def test_connect_mcp_servers_enabled_tools_supports_raw_names(
) -> None:
fake_mcp_runtime["session"] = _make_fake_session(["demo", "other"])
registry = ToolRegistry()
stack = AsyncExitStack()
await stack.__aenter__()
try:
await connect_mcp_servers(
{"test": MCPServerConfig(command="fake", enabled_tools=["demo"])},
registry,
stack,
)
finally:
stacks = await connect_mcp_servers(
{"test": MCPServerConfig(command="fake", enabled_tools=["demo"])},
registry,
)
for stack in stacks.values():
await stack.aclose()
assert registry.tool_names == ["mcp_test_demo"]
@@ -291,15 +287,11 @@ async def test_connect_mcp_servers_enabled_tools_defaults_to_all(
) -> None:
fake_mcp_runtime["session"] = _make_fake_session(["demo", "other"])
registry = ToolRegistry()
stack = AsyncExitStack()
await stack.__aenter__()
try:
await connect_mcp_servers(
{"test": MCPServerConfig(command="fake")},
registry,
stack,
)
finally:
stacks = await connect_mcp_servers(
{"test": MCPServerConfig(command="fake")},
registry,
)
for stack in stacks.values():
await stack.aclose()
assert registry.tool_names == ["mcp_test_demo", "mcp_test_other"]
@@ -311,15 +303,11 @@ async def test_connect_mcp_servers_enabled_tools_supports_wrapped_names(
) -> None:
fake_mcp_runtime["session"] = _make_fake_session(["demo", "other"])
registry = ToolRegistry()
stack = AsyncExitStack()
await stack.__aenter__()
try:
await connect_mcp_servers(
{"test": MCPServerConfig(command="fake", enabled_tools=["mcp_test_demo"])},
registry,
stack,
)
finally:
stacks = await connect_mcp_servers(
{"test": MCPServerConfig(command="fake", enabled_tools=["mcp_test_demo"])},
registry,
)
for stack in stacks.values():
await stack.aclose()
assert registry.tool_names == ["mcp_test_demo"]
@@ -331,15 +319,11 @@ async def test_connect_mcp_servers_enabled_tools_empty_list_registers_none(
) -> None:
fake_mcp_runtime["session"] = _make_fake_session(["demo", "other"])
registry = ToolRegistry()
stack = AsyncExitStack()
await stack.__aenter__()
try:
await connect_mcp_servers(
{"test": MCPServerConfig(command="fake", enabled_tools=[])},
registry,
stack,
)
finally:
stacks = await connect_mcp_servers(
{"test": MCPServerConfig(command="fake", enabled_tools=[])},
registry,
)
for stack in stacks.values():
await stack.aclose()
assert registry.tool_names == []
@@ -358,15 +342,11 @@ async def test_connect_mcp_servers_enabled_tools_warns_on_unknown_entries(
monkeypatch.setattr("nanobot.agent.tools.mcp.logger.warning", _warning)
stack = AsyncExitStack()
await stack.__aenter__()
try:
await connect_mcp_servers(
{"test": MCPServerConfig(command="fake", enabled_tools=["unknown"])},
registry,
stack,
)
finally:
stacks = await connect_mcp_servers(
{"test": MCPServerConfig(command="fake", enabled_tools=["unknown"])},
registry,
)
for stack in stacks.values():
await stack.aclose()
assert registry.tool_names == []
@@ -376,6 +356,46 @@ async def test_connect_mcp_servers_enabled_tools_warns_on_unknown_entries(
assert "Available wrapped names: mcp_test_demo" in warnings[-1]
@pytest.mark.asyncio
async def test_connect_mcp_servers_one_failure_does_not_block_others(
monkeypatch: pytest.MonkeyPatch,
) -> None:
sessions = {"good": _make_fake_session(["demo"])}
class _SelectiveClientSession:
def __init__(self, read: object, _write: object) -> None:
self._session = sessions[read]
async def __aenter__(self) -> object:
return self._session
async def __aexit__(self, exc_type, exc, tb) -> bool:
return False
@asynccontextmanager
async def _selective_stdio_client(params: object):
if params.command == "bad":
raise RuntimeError("boom")
yield params.command, object()
monkeypatch.setattr(sys.modules["mcp"], "ClientSession", _SelectiveClientSession)
monkeypatch.setattr(sys.modules["mcp.client.stdio"], "stdio_client", _selective_stdio_client)
registry = ToolRegistry()
stacks = await connect_mcp_servers(
{
"good": MCPServerConfig(command="good"),
"bad": MCPServerConfig(command="bad"),
},
registry,
)
for stack in stacks.values():
await stack.aclose()
assert registry.tool_names == ["mcp_good_demo"]
assert set(stacks) == {"good"}
# ---------------------------------------------------------------------------
# MCPResourceWrapper tests
# ---------------------------------------------------------------------------
@@ -389,9 +409,7 @@ def _make_resource_def(
return SimpleNamespace(name=name, uri=uri, description=description)
def _make_resource_wrapper(
session: object, *, timeout: float = 0.1
) -> MCPResourceWrapper:
def _make_resource_wrapper(session: object, *, timeout: float = 0.1) -> MCPResourceWrapper:
return MCPResourceWrapper(session, "srv", _make_resource_def(), resource_timeout=timeout)
@@ -434,9 +452,7 @@ async def test_resource_wrapper_execute_handles_timeout() -> None:
await asyncio.sleep(1)
return SimpleNamespace(contents=[])
wrapper = _make_resource_wrapper(
SimpleNamespace(read_resource=read_resource), timeout=0.01
)
wrapper = _make_resource_wrapper(SimpleNamespace(read_resource=read_resource), timeout=0.01)
result = await wrapper.execute()
assert result == "(MCP resource read timed out after 0.01s)"
@@ -464,20 +480,14 @@ def _make_prompt_def(
return SimpleNamespace(name=name, description=description, arguments=arguments)
def _make_prompt_wrapper(
session: object, *, timeout: float = 0.1
) -> MCPPromptWrapper:
return MCPPromptWrapper(
session, "srv", _make_prompt_def(), prompt_timeout=timeout
)
def _make_prompt_wrapper(session: object, *, timeout: float = 0.1) -> MCPPromptWrapper:
return MCPPromptWrapper(session, "srv", _make_prompt_def(), prompt_timeout=timeout)
def test_prompt_wrapper_properties() -> None:
arg1 = SimpleNamespace(name="topic", required=True)
arg2 = SimpleNamespace(name="style", required=False)
wrapper = MCPPromptWrapper(
None, "myserver", _make_prompt_def(arguments=[arg1, arg2])
)
wrapper = MCPPromptWrapper(None, "myserver", _make_prompt_def(arguments=[arg1, arg2]))
assert wrapper.name == "mcp_myserver_prompt_myprompt"
assert "[MCP Prompt]" in wrapper.description
assert "A test prompt" in wrapper.description
@@ -528,9 +538,7 @@ async def test_prompt_wrapper_execute_handles_timeout() -> None:
await asyncio.sleep(1)
return SimpleNamespace(messages=[])
wrapper = _make_prompt_wrapper(
SimpleNamespace(get_prompt=get_prompt), timeout=0.01
)
wrapper = _make_prompt_wrapper(SimpleNamespace(get_prompt=get_prompt), timeout=0.01)
result = await wrapper.execute()
assert result == "(MCP prompt call timed out after 0.01s)"
@@ -616,15 +624,11 @@ async def test_connect_registers_resources_and_prompts(
prompt_names=["prompt_c"],
)
registry = ToolRegistry()
stack = AsyncExitStack()
await stack.__aenter__()
try:
await connect_mcp_servers(
{"test": MCPServerConfig(command="fake")},
registry,
stack,
)
finally:
stacks = await connect_mcp_servers(
{"test": MCPServerConfig(command="fake")},
registry,
)
for stack in stacks.values():
await stack.aclose()
assert "mcp_test_tool_a" in registry.tool_names
+38 -1
View File
@@ -1,5 +1,6 @@
"""Test message tool suppress logic for final replies."""
import asyncio
from pathlib import Path
from unittest.mock import AsyncMock, MagicMock
@@ -86,6 +87,42 @@ class TestMessageToolSuppressLogic:
assert result is not None
assert "Hello" in result.content
@pytest.mark.asyncio
async def test_injected_followup_with_message_tool_does_not_emit_empty_fallback(
self, tmp_path: Path
) -> None:
loop = _make_loop(tmp_path)
tool_call = ToolCallRequest(
id="call1", name="message",
arguments={"content": "Tool reply", "channel": "feishu", "chat_id": "chat123"},
)
calls = iter([
LLMResponse(content="First answer", tool_calls=[]),
LLMResponse(content="", tool_calls=[tool_call]),
LLMResponse(content="", tool_calls=[]),
LLMResponse(content="", tool_calls=[]),
LLMResponse(content="", tool_calls=[]),
])
loop.provider.chat_with_retry = AsyncMock(side_effect=lambda *a, **kw: next(calls))
loop.tools.get_definitions = MagicMock(return_value=[])
sent: list[OutboundMessage] = []
mt = loop.tools.get("message")
if isinstance(mt, MessageTool):
mt.set_send_callback(AsyncMock(side_effect=lambda m: sent.append(m)))
pending_queue = asyncio.Queue()
await pending_queue.put(
InboundMessage(channel="feishu", sender_id="user1", chat_id="chat123", content="follow-up")
)
msg = InboundMessage(channel="feishu", sender_id="user1", chat_id="chat123", content="Start")
result = await loop._process_message(msg, pending_queue=pending_queue)
assert len(sent) == 1
assert sent[0].content == "Tool reply"
assert result is None
async def test_progress_hides_internal_reasoning(self, tmp_path: Path) -> None:
loop = _make_loop(tmp_path)
tool_call = ToolCallRequest(id="call1", name="read_file", arguments={"path": "foo.txt"})
@@ -107,7 +144,7 @@ class TestMessageToolSuppressLogic:
async def on_progress(content: str, *, tool_hint: bool = False) -> None:
progress.append((content, tool_hint))
final_content, _, _ = await loop._run_agent_loop([], on_progress=on_progress)
final_content, _, _, _, _ = await loop._run_agent_loop([], on_progress=on_progress)
assert final_content == "Done"
assert progress == [
+147
View File
@@ -0,0 +1,147 @@
"""Tests for NotebookEditTool — Jupyter .ipynb editing."""
import json
import pytest
from nanobot.agent.tools.notebook import NotebookEditTool
def _make_notebook(cells: list[dict] | None = None, nbformat: int = 4, nbformat_minor: int = 5) -> dict:
"""Build a minimal valid .ipynb structure."""
return {
"nbformat": nbformat,
"nbformat_minor": nbformat_minor,
"metadata": {"kernelspec": {"display_name": "Python 3", "language": "python", "name": "python3"}},
"cells": cells or [],
}
def _code_cell(source: str, cell_id: str | None = None) -> dict:
cell = {"cell_type": "code", "source": source, "metadata": {}, "outputs": [], "execution_count": None}
if cell_id:
cell["id"] = cell_id
return cell
def _md_cell(source: str, cell_id: str | None = None) -> dict:
cell = {"cell_type": "markdown", "source": source, "metadata": {}}
if cell_id:
cell["id"] = cell_id
return cell
def _write_nb(tmp_path, name: str, nb: dict) -> str:
p = tmp_path / name
p.write_text(json.dumps(nb), encoding="utf-8")
return str(p)
class TestNotebookEdit:
@pytest.fixture()
def tool(self, tmp_path):
return NotebookEditTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_replace_cell_content(self, tool, tmp_path):
nb = _make_notebook([_code_cell("print('hello')"), _code_cell("x = 1")])
path = _write_nb(tmp_path, "test.ipynb", nb)
result = await tool.execute(path=path, cell_index=0, new_source="print('world')")
assert "Successfully" in result
saved = json.loads((tmp_path / "test.ipynb").read_text())
assert saved["cells"][0]["source"] == "print('world')"
assert saved["cells"][1]["source"] == "x = 1"
@pytest.mark.asyncio
async def test_insert_cell_after_target(self, tool, tmp_path):
nb = _make_notebook([_code_cell("cell 0"), _code_cell("cell 1")])
path = _write_nb(tmp_path, "test.ipynb", nb)
result = await tool.execute(path=path, cell_index=0, new_source="inserted", edit_mode="insert")
assert "Successfully" in result
saved = json.loads((tmp_path / "test.ipynb").read_text())
assert len(saved["cells"]) == 3
assert saved["cells"][0]["source"] == "cell 0"
assert saved["cells"][1]["source"] == "inserted"
assert saved["cells"][2]["source"] == "cell 1"
@pytest.mark.asyncio
async def test_delete_cell(self, tool, tmp_path):
nb = _make_notebook([_code_cell("A"), _code_cell("B"), _code_cell("C")])
path = _write_nb(tmp_path, "test.ipynb", nb)
result = await tool.execute(path=path, cell_index=1, edit_mode="delete")
assert "Successfully" in result
saved = json.loads((tmp_path / "test.ipynb").read_text())
assert len(saved["cells"]) == 2
assert saved["cells"][0]["source"] == "A"
assert saved["cells"][1]["source"] == "C"
@pytest.mark.asyncio
async def test_create_new_notebook_from_scratch(self, tool, tmp_path):
path = str(tmp_path / "new.ipynb")
result = await tool.execute(path=path, cell_index=0, new_source="# Hello", edit_mode="insert", cell_type="markdown")
assert "Successfully" in result or "created" in result.lower()
saved = json.loads((tmp_path / "new.ipynb").read_text())
assert saved["nbformat"] == 4
assert len(saved["cells"]) == 1
assert saved["cells"][0]["cell_type"] == "markdown"
assert saved["cells"][0]["source"] == "# Hello"
@pytest.mark.asyncio
async def test_invalid_cell_index_error(self, tool, tmp_path):
nb = _make_notebook([_code_cell("only cell")])
path = _write_nb(tmp_path, "test.ipynb", nb)
result = await tool.execute(path=path, cell_index=5, new_source="x")
assert "Error" in result
@pytest.mark.asyncio
async def test_non_ipynb_rejected(self, tool, tmp_path):
f = tmp_path / "script.py"
f.write_text("pass")
result = await tool.execute(path=str(f), cell_index=0, new_source="x")
assert "Error" in result
assert ".ipynb" in result
@pytest.mark.asyncio
async def test_preserves_metadata_and_outputs(self, tool, tmp_path):
cell = _code_cell("old")
cell["outputs"] = [{"output_type": "stream", "text": "hello\n"}]
cell["execution_count"] = 42
nb = _make_notebook([cell])
path = _write_nb(tmp_path, "test.ipynb", nb)
await tool.execute(path=path, cell_index=0, new_source="new")
saved = json.loads((tmp_path / "test.ipynb").read_text())
assert saved["metadata"]["kernelspec"]["language"] == "python"
@pytest.mark.asyncio
async def test_nbformat_45_generates_cell_id(self, tool, tmp_path):
nb = _make_notebook([], nbformat_minor=5)
path = _write_nb(tmp_path, "test.ipynb", nb)
await tool.execute(path=path, cell_index=0, new_source="x = 1", edit_mode="insert")
saved = json.loads((tmp_path / "test.ipynb").read_text())
assert "id" in saved["cells"][0]
assert len(saved["cells"][0]["id"]) > 0
@pytest.mark.asyncio
async def test_insert_with_cell_type_markdown(self, tool, tmp_path):
nb = _make_notebook([_code_cell("code")])
path = _write_nb(tmp_path, "test.ipynb", nb)
await tool.execute(path=path, cell_index=0, new_source="# Title", edit_mode="insert", cell_type="markdown")
saved = json.loads((tmp_path / "test.ipynb").read_text())
assert saved["cells"][1]["cell_type"] == "markdown"
@pytest.mark.asyncio
async def test_invalid_edit_mode_rejected(self, tool, tmp_path):
nb = _make_notebook([_code_cell("code")])
path = _write_nb(tmp_path, "test.ipynb", nb)
result = await tool.execute(path=path, cell_index=0, new_source="x", edit_mode="replcae")
assert "Error" in result
assert "edit_mode" in result
@pytest.mark.asyncio
async def test_invalid_cell_type_rejected(self, tool, tmp_path):
nb = _make_notebook([_code_cell("code")])
path = _write_nb(tmp_path, "test.ipynb", nb)
result = await tool.execute(path=path, cell_index=0, new_source="x", cell_type="raw")
assert "Error" in result
assert "cell_type" in result
+180
View File
@@ -0,0 +1,180 @@
"""Tests for ReadFileTool enhancements: description fix, read dedup, PDF support, device blacklist."""
import pytest
from nanobot.agent.tools.filesystem import ReadFileTool, WriteFileTool
from nanobot.agent.tools import file_state
@pytest.fixture(autouse=True)
def _clear_file_state():
file_state.clear()
yield
file_state.clear()
# ---------------------------------------------------------------------------
# Description fix
# ---------------------------------------------------------------------------
class TestReadDescriptionFix:
def test_description_mentions_image_support(self):
tool = ReadFileTool()
assert "image" in tool.description.lower()
def test_description_no_longer_says_cannot_read_images(self):
tool = ReadFileTool()
assert "cannot read binary files or images" not in tool.description.lower()
# ---------------------------------------------------------------------------
# Read deduplication
# ---------------------------------------------------------------------------
class TestReadDedup:
"""Same file + same offset/limit + unchanged mtime -> short stub."""
@pytest.fixture()
def tool(self, tmp_path):
return ReadFileTool(workspace=tmp_path)
@pytest.fixture()
def write_tool(self, tmp_path):
return WriteFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_second_read_returns_unchanged_stub(self, tool, tmp_path):
f = tmp_path / "data.txt"
f.write_text("\n".join(f"line {i}" for i in range(100)), encoding="utf-8")
first = await tool.execute(path=str(f))
assert "line 0" in first
second = await tool.execute(path=str(f))
assert "unchanged" in second.lower()
# Stub should not contain file content
assert "line 0" not in second
@pytest.mark.asyncio
async def test_read_after_external_modification_returns_full(self, tool, tmp_path):
f = tmp_path / "data.txt"
f.write_text("original", encoding="utf-8")
await tool.execute(path=str(f))
# Modify the file externally
f.write_text("modified content", encoding="utf-8")
second = await tool.execute(path=str(f))
assert "modified content" in second
@pytest.mark.asyncio
async def test_different_offset_returns_full(self, tool, tmp_path):
f = tmp_path / "data.txt"
f.write_text("\n".join(f"line {i}" for i in range(1, 21)), encoding="utf-8")
await tool.execute(path=str(f), offset=1, limit=5)
second = await tool.execute(path=str(f), offset=6, limit=5)
# Different offset → full read, not stub
assert "line 6" in second
@pytest.mark.asyncio
async def test_first_read_after_write_returns_full_content(self, tool, write_tool, tmp_path):
f = tmp_path / "fresh.txt"
result = await write_tool.execute(path=str(f), content="hello")
assert "Successfully" in result
read_result = await tool.execute(path=str(f))
assert "hello" in read_result
assert "unchanged" not in read_result.lower()
@pytest.mark.asyncio
async def test_dedup_does_not_apply_to_images(self, tool, tmp_path):
f = tmp_path / "img.png"
f.write_bytes(b"\x89PNG\r\n\x1a\nfake-png-data")
first = await tool.execute(path=str(f))
assert isinstance(first, list)
second = await tool.execute(path=str(f))
# Images should always return full content blocks, not a stub
assert isinstance(second, list)
# ---------------------------------------------------------------------------
# PDF support
# ---------------------------------------------------------------------------
class TestReadPdf:
@pytest.fixture()
def tool(self, tmp_path):
return ReadFileTool(workspace=tmp_path)
@pytest.mark.asyncio
async def test_pdf_returns_text_content(self, tool, tmp_path):
fitz = pytest.importorskip("fitz")
pdf_path = tmp_path / "test.pdf"
doc = fitz.open()
page = doc.new_page()
page.insert_text((72, 72), "Hello PDF World")
doc.save(str(pdf_path))
doc.close()
result = await tool.execute(path=str(pdf_path))
assert "Hello PDF World" in result
@pytest.mark.asyncio
async def test_pdf_pages_parameter(self, tool, tmp_path):
fitz = pytest.importorskip("fitz")
pdf_path = tmp_path / "multi.pdf"
doc = fitz.open()
for i in range(5):
page = doc.new_page()
page.insert_text((72, 72), f"Page {i + 1} content")
doc.save(str(pdf_path))
doc.close()
result = await tool.execute(path=str(pdf_path), pages="2-3")
assert "Page 2 content" in result
assert "Page 3 content" in result
assert "Page 1 content" not in result
@pytest.mark.asyncio
async def test_pdf_file_not_found_error(self, tool, tmp_path):
result = await tool.execute(path=str(tmp_path / "nope.pdf"))
assert "Error" in result
assert "not found" in result
# ---------------------------------------------------------------------------
# Device path blacklist
# ---------------------------------------------------------------------------
class TestReadDeviceBlacklist:
@pytest.fixture()
def tool(self):
return ReadFileTool()
@pytest.mark.asyncio
async def test_dev_random_blocked(self, tool):
result = await tool.execute(path="/dev/random")
assert "Error" in result
assert "blocked" in result.lower() or "device" in result.lower()
@pytest.mark.asyncio
async def test_dev_urandom_blocked(self, tool):
result = await tool.execute(path="/dev/urandom")
assert "Error" in result
@pytest.mark.asyncio
async def test_dev_zero_blocked(self, tool):
result = await tool.execute(path="/dev/zero")
assert "Error" in result
@pytest.mark.asyncio
async def test_proc_fd_blocked(self, tool):
result = await tool.execute(path="/proc/self/fd/0")
assert "Error" in result
@pytest.mark.asyncio
async def test_symlink_to_dev_zero_blocked(self, tmp_path):
tool = ReadFileTool(workspace=tmp_path)
link = tmp_path / "zero-link"
link.symlink_to("/dev/zero")
result = await tool.execute(path=str(link))
assert "Error" in result
assert "blocked" in result.lower() or "device" in result.lower()
+24
View File
@@ -323,3 +323,27 @@ async def test_subagent_registers_grep_and_glob(tmp_path: Path) -> None:
assert "grep" in captured["tool_names"]
assert "glob" in captured["tool_names"]
def test_subagent_prompt_respects_disabled_skills(tmp_path: Path) -> None:
bus = MessageBus()
provider = MagicMock()
provider.get_default_model.return_value = "test-model"
skills_dir = tmp_path / "skills"
(skills_dir / "alpha").mkdir(parents=True)
(skills_dir / "alpha" / "SKILL.md").write_text("# Alpha\n\nhidden\n", encoding="utf-8")
(skills_dir / "beta").mkdir(parents=True)
(skills_dir / "beta" / "SKILL.md").write_text("# Beta\n\nshown\n", encoding="utf-8")
mgr = SubagentManager(
provider=provider,
workspace=tmp_path,
bus=bus,
max_tool_result_chars=4096,
disabled_skills=["alpha"],
)
prompt = mgr._build_subagent_prompt()
assert "alpha" not in prompt
assert "beta" in prompt
+38
View File
@@ -120,6 +120,27 @@ async def test_jina_search(monkeypatch):
assert "https://jina.ai" in result
@pytest.mark.asyncio
async def test_kagi_search(monkeypatch):
async def mock_get(self, url, **kw):
assert "kagi.com/api/v0/search" in url
assert kw["headers"]["Authorization"] == "Bot kagi-key"
assert kw["params"] == {"q": "test", "limit": 2}
return _response(json={
"data": [
{"t": 0, "title": "Kagi Result", "url": "https://kagi.com", "snippet": "Premium search"},
{"t": 1, "list": ["ignored related search"]},
]
})
monkeypatch.setattr(httpx.AsyncClient, "get", mock_get)
tool = _tool(provider="kagi", api_key="kagi-key")
result = await tool.execute(query="test", count=2)
assert "Kagi Result" in result
assert "https://kagi.com" in result
assert "ignored related search" not in result
@pytest.mark.asyncio
async def test_unknown_provider():
tool = _tool(provider="unknown")
@@ -189,6 +210,23 @@ async def test_jina_422_falls_back_to_duckduckgo(monkeypatch):
assert "DuckDuckGo fallback" in result
@pytest.mark.asyncio
async def test_kagi_fallback_to_duckduckgo_when_no_key(monkeypatch):
class MockDDGS:
def __init__(self, **kw):
pass
def text(self, query, max_results=5):
return [{"title": "Fallback", "href": "https://ddg.example", "body": "DuckDuckGo fallback"}]
monkeypatch.setattr("ddgs.DDGS", MockDDGS)
monkeypatch.delenv("KAGI_API_KEY", raising=False)
tool = _tool(provider="kagi", api_key="")
result = await tool.execute(query="test")
assert "Fallback" in result
@pytest.mark.asyncio
async def test_jina_search_uses_path_encoded_query(monkeypatch):
calls = {}