Merge #17288: Added TestShell class for interactive Python environments.

19139ee034 Add documentation for test_shell submodule (JamesC)
f5112369cf Add TestShell class (James Chiang)
5155602a63 Move argparse() to init() (JamesC)
2ab01462f4 Move assert num_nodes is set into main() (JamesC)
614c645643 Clear TestNode objects after shutdown (JamesC)
6f40820757 Add closing and flushing of logging handlers (JamesC)
6b71241291 Refactor TestFramework main() into setup/shutdown (JamesC)
ede8b7608e Remove network_event_loop instance in close() (JamesC)

Pull request description:

  This PR refactors BitcoinTestFramework to encapsulate setup and shutdown logic into dedicated methods, and adds a  ~~TestWrapper~~ TestShell child class. This wrapper allows the underlying BitcoinTestFramework to run _between user inputs_ in a REPL environment, such as a Jupyter notebook or any interactive Python3 interpreter.

  The ~~TestWrapper~~ TestShell is motivated by the opportunity to expose the test-framework as a prototyping and educational toolkit. Examples of code prototypes enabled by  ~~TestWrapper~~ TestShell can be found in the Optech [Taproot/Schnorr](https://github.com/bitcoinops/taproot-workshop) workshop repository.

  Usage example:
  ```
  >>> import sys
  >>> sys.path.insert(0, "/path/to/bitcoin/test/functional")
  ```
  ```
  >>> from test_framework.test_wrapper import TestShell
  >>> test = TestShell()
  >>> test.setup(num_nodes=2)
  20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework (INFO): Initializing test directory /path/to/bitcoin_func_test_XXXXXXX
  ```
  ```
  >>> test.nodes[0].generate(101)
  >>> test.nodes[0].getblockchaininfo()["blocks"]
  101
  ```
  ```
  >>> test.shutdown()
  20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework (INFO): Stopping nodes
  20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework (INFO): Cleaning up /path/to/bitcoin_func_test_XXXXXXX on exit
  20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework (INFO): Tests successful
  ```

  **Overview of changes to BitcoinTestFramework:**

  - Code moved to `setup()/shutdown()` methods.
  - Argument parsing logic encapsulated by `parse_args` method.
  - Success state moved to `BitcoinTestFramework.success`.

  _During Shutdown_

  - `BitcoinTestFramework` logging handlers are flushed and removed.
  - `BitcoinTestFrameowork.nodes` list is cleared.
  - `NetworkThread.network_event_loop` is reset. (NetworkThread class).

  **Behavioural changes:**
  - Test parameters can now also be set when overriding BitcoinTestFramework.setup() in addition to overriding `set_test_params` method.
  - Potential exceptions raised in BitcoinTestFramework.setup() will be handled in main().

  **Added files:**
  - ~~test_wrapper.py~~ `test_shell.py`
  - ~~test-wrapper.md~~ `test-shell.md`

ACKs for top commit:
  jamesob:
    ACK 19139ee034
  jonatack:
    ACK 19139ee034
  jnewbery:
    Rather than invalidate the three ACKs for a minor nit, can you force push back to 19139ee034 please? I think this PR was ready to merge before your last force push.
  jachiang:
    > Rather than invalidate the three ACKs for a minor nit, can you force push back to [19139ee](19139ee034) please? I think this PR was ready to merge before your last force push.
  jnewbery:
    ACK 19139ee034

Tree-SHA512: 0c24f405f295a8580a9c8f1b9e0182b5d753eb08cc331424616dd50a062fb773d3719db4d08943365b1f42ccb965cc363b4bcc5beae27ac90b3460b349ed46b2
pull/764/head
MarcoFalke 5 years ago
commit bc38bb9a60
No known key found for this signature in database
GPG Key ID: D2EA4850E7528B25

@ -99,6 +99,16 @@ P2PInterface object and override the callback methods.
Examples tests are [p2p_unrequested_blocks.py](p2p_unrequested_blocks.py),
[p2p_compactblocks.py](p2p_compactblocks.py).
#### Prototyping tests
The [`TestShell`](test-shell.md) class exposes the BitcoinTestFramework
functionality to interactive Python3 environments and can be used to prototype
tests. This may be especially useful in a REPL environment with session logging
utilities, such as
[IPython](https://ipython.readthedocs.io/en/stable/interactive/reference.html#session-logging-and-restoring).
The logs of such interactive sessions can later be adapted into permanent test
cases.
### Test framework modules
The following are useful modules for test developers. They are located in
[test/functional/test_framework/](test_framework).

@ -0,0 +1,188 @@
Test Shell for Interactive Environments
=========================================
This document describes how to use the `TestShell` submodule in the functional
test suite.
The `TestShell` submodule extends the `BitcoinTestFramework` functionality to
external interactive environments for prototyping and educational purposes. Just
like `BitcoinTestFramework`, the `TestShell` allows the user to:
* Manage regtest bitcoind subprocesses.
* Access RPC interfaces of the underlying bitcoind instances.
* Log events to the functional test logging utility.
The `TestShell` can be useful in interactive environments where it is necessary
to extend the object lifetime of the underlying `BitcoinTestFramework` between
user inputs. Such environments include the Python3 command line interpreter or
[Jupyter](https://jupyter.org/) notebooks running a Python3 kernel.
## 1. Requirements
* Python3
* `bitcoind` built in the same repository as the `TestShell`.
## 2. Importing `TestShell` from the Bitcoin Core repository
We can import the `TestShell` by adding the path of the Bitcoin Core
`test_framework` module to the beginning of the PATH variable, and then
importing the `TestShell` class from the `test_shell` sub-package.
```
>>> import sys
>>> sys.path.insert(0, "/path/to/bitcoin/test/functional")
>>> from test_framework.test_shell import `TestShell`
```
The following `TestShell` methods manage the lifetime of the underlying bitcoind
processes and logging utilities.
* `TestShell.setup()`
* `TestShell.shutdown()`
The `TestShell` inherits all `BitcoinTestFramework` members and methods, such
as:
* `TestShell.nodes[index].rpc_method()`
* `TestShell.log.info("Custom log message")`
The following sections demonstrate how to initialize, run, and shut down a
`TestShell` object.
## 3. Initializing a `TestShell` object
```
>>> test = TestShell()
>>> test.setup(num_nodes=2, setup_clean_chain=True)
20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework (INFO): Initializing test directory /path/to/bitcoin_func_test_XXXXXXX
```
The `TestShell` forwards all functional test parameters of the parent
`BitcoinTestFramework` object. The full set of argument keywords which can be
used to initialize the `TestShell` can be found in [section
#6](#custom-testshell-parameters) of this document.
**Note: Running multiple instances of `TestShell` is not allowed.** Running a
single process also ensures that logging remains consolidated in the same
temporary folder. If you need more bitcoind nodes than set by default (1),
simply increase the `num_nodes` parameter during setup.
```
>>> test2 = TestShell()
>>> test2.setup()
TestShell is already running!
```
## 4. Interacting with the `TestShell`
Unlike the `BitcoinTestFramework` class, the `TestShell` keeps the underlying
Bitcoind subprocesses (nodes) and logging utilities running until the user
explicitly shuts down the `TestShell` object.
During the time between the `setup` and `shutdown` calls, all `bitcoind` node
processes and `BitcoinTestFramework` convenience methods can be accessed
interactively.
**Example: Mining a regtest chain**
By default, the `TestShell` nodes are initialized with a clean chain. This means
that each node of the `TestShell` is initialized with a block height of 0.
```
>>> test.nodes[0].getblockchaininfo()["blocks"]
0
```
We now let the first node generate 101 regtest blocks, and direct the coinbase
rewards to a wallet address owned by the mining node.
```
>>> address = test.nodes[0].getnewaddress()
>>> test.nodes[0].generatetoaddress(101, address)
['2b98dd0044aae6f1cca7f88a0acf366a4bfe053c7f7b00da3c0d115f03d67efb', ...
```
Since the two nodes are both initialized by default to establish an outbound
connection to each other during `setup`, the second node's chain will include
the mined blocks as soon as they propagate.
```
>>> test.nodes[1].getblockchaininfo()["blocks"]
101
```
The block rewards from the first block are now spendable by the wallet of the
first node.
```
>>> test.nodes[0].getbalance()
Decimal('50.00000000')
```
We can also log custom events to the logger.
```
>>> test.nodes[0].log.info("Successfully mined regtest chain!")
20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework.node0 (INFO): Successfully mined regtest chain!
```
**Note: Please also consider the functional test
[readme](../test/functional/README.md), which provides an overview of the
test-framework**. Modules such as
[key.py](../test/functional/test_framework/key.py),
[script.py](../test/functional/test_framework/script.py) and
[messages.py](../test/functional/test_framework/messages.py) are particularly
useful in constructing objects which can be passed to the bitcoind nodes managed
by a running `TestShell` object.
## 5. Shutting the `TestShell` down
Shutting down the `TestShell` will safely tear down all running bitcoind
instances and remove all temporary data and logging directories.
```
>>> test.shutdown()
20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework (INFO): Stopping nodes
20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework (INFO): Cleaning up /path/to/bitcoin_func_test_XXXXXXX on exit
20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework (INFO): Tests successful
```
To prevent the logs from being removed after a shutdown, simply set the
`TestShell.options.nocleanup` member to `True`.
```
>>> test.options.nocleanup = True
>>> test.shutdown()
20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework (INFO): Stopping nodes
20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework (INFO): Not cleaning up dir /path/to/bitcoin_func_test_XXXXXXX on exit
20XX-XX-XXTXX:XX:XX.XXXXXXX TestFramework (INFO): Tests successful
```
The following utility consolidates logs from the bitcoind nodes and the
underlying `BitcoinTestFramework`:
* `/path/to/bitcoin/test/functional/combine_logs.py
'/path/to/bitcoin_func_test_XXXXXXX'`
## 6. Custom `TestShell` parameters
The `TestShell` object initializes with the default settings inherited from the
`BitcoinTestFramework` class. The user can override these in
`TestShell.setup(key=value)`.
**Note:** `TestShell.reset()` will reset test parameters to default values and
can be called after the TestShell is shut down.
| Test parameter key | Default Value | Description |
|---|---|---|
| `bind_to_localhost_only` | `True` | Binds bitcoind RPC services to `127.0.0.1` if set to `True`.|
| `cachedir` | `"/path/to/bitcoin/test/cache"` | Sets the bitcoind datadir directory. |
| `chain` | `"regtest"` | Sets the chain-type for the underlying test bitcoind processes. |
| `configfile` | `"/path/to/bitcoin/test/config.ini"` | Sets the location of the test framework config file. |
| `coveragedir` | `None` | Records bitcoind RPC test coverage into this directory if set. |
| `loglevel` | `INFO` | Logs events at this level and higher. Can be set to `DEBUG`, `INFO`, `WARNING`, `ERROR` or `CRITICAL`. |
| `nocleanup` | `False` | Cleans up temporary test directory if set to `True` during `shutdown`. |
| `noshutdown` | `False` | Does not stop bitcoind instances after `shutdown` if set to `True`. |
| `num_nodes` | `1` | Sets the number of initialized bitcoind processes. |
| `perf` | False | Profiles running nodes with `perf` for the duration of the test if set to `True`. |
| `rpc_timeout` | `60` | Sets the RPC server timeout for the underlying bitcoind processes. |
| `setup_clean_chain` | `False` | Initializes an empty blockchain by default. A 199-block-long chain is initialized if set to `True`. |
| `randomseed` | Random Integer | `TestShell.options.randomseed` is a member of `TestShell` which can be accessed during a test to seed a random generator. User can override default with a constant value for reproducible test runs. |
| `supports_cli` | `False` | Whether the bitcoin-cli utility is compiled and available for the test. |
| `tmpdir` | `"/var/folders/.../"` | Sets directory for test logs. Will be deleted upon a successful test run unless `nocleanup` is set to `True` |
| `trace_rpc` | `False` | Logs all RPC calls if set to `True`. |
| `usecli` | `False` | Uses the bitcoin-cli interface for all bitcoind commands instead of directly calling the RPC server. Requires `supports_cli`. |

@ -478,7 +478,8 @@ class NetworkThread(threading.Thread):
wait_until(lambda: not self.network_event_loop.is_running(), timeout=timeout)
self.network_event_loop.close()
self.join(timeout)
# Safe to remove event loop.
NetworkThread.network_event_loop = None
class P2PDataStore(P2PInterface):
"""A P2P data store class.

@ -99,12 +99,39 @@ class BitcoinTestFramework(metaclass=BitcoinTestMetaClass):
self.supports_cli = False
self.bind_to_localhost_only = True
self.set_test_params()
assert hasattr(self, "num_nodes"), "Test must set self.num_nodes in set_test_params()"
self.parse_args()
def main(self):
"""Main function. This should not be overridden by the subclass test scripts."""
assert hasattr(self, "num_nodes"), "Test must set self.num_nodes in set_test_params()"
try:
self.setup()
self.run_test()
except JSONRPCException:
self.log.exception("JSONRPC error")
self.success = TestStatus.FAILED
except SkipTest as e:
self.log.warning("Test Skipped: %s" % e.message)
self.success = TestStatus.SKIPPED
except AssertionError:
self.log.exception("Assertion failed")
self.success = TestStatus.FAILED
except KeyError:
self.log.exception("Key error")
self.success = TestStatus.FAILED
except Exception:
self.log.exception("Unexpected exception caught during testing")
self.success = TestStatus.FAILED
except KeyboardInterrupt:
self.log.warning("Exiting after keyboard interrupt")
self.success = TestStatus.FAILED
finally:
exit_code = self.shutdown()
sys.exit(exit_code)
def parse_args(self):
parser = argparse.ArgumentParser(usage="%(prog)s [options]")
parser.add_argument("--nocleanup", dest="nocleanup", default=False, action="store_true",
help="Leave bitcoinds and test.* datadir on exit or error")
@ -135,6 +162,9 @@ class BitcoinTestFramework(metaclass=BitcoinTestMetaClass):
self.add_options(parser)
self.options = parser.parse_args()
def setup(self):
"""Call this method to start up the test framework object with options set."""
PortSeed.n = self.options.port_seed
check_json_precision()
@ -181,33 +211,20 @@ class BitcoinTestFramework(metaclass=BitcoinTestMetaClass):
self.network_thread = NetworkThread()
self.network_thread.start()
success = TestStatus.FAILED
if self.options.usecli:
if not self.supports_cli:
raise SkipTest("--usecli specified but test does not support using CLI")
self.skip_if_no_cli()
self.skip_test_if_missing_module()
self.setup_chain()
self.setup_network()
try:
if self.options.usecli:
if not self.supports_cli:
raise SkipTest("--usecli specified but test does not support using CLI")
self.skip_if_no_cli()
self.skip_test_if_missing_module()
self.setup_chain()
self.setup_network()
self.run_test()
success = TestStatus.PASSED
except JSONRPCException:
self.log.exception("JSONRPC error")
except SkipTest as e:
self.log.warning("Test Skipped: %s" % e.message)
success = TestStatus.SKIPPED
except AssertionError:
self.log.exception("Assertion failed")
except KeyError:
self.log.exception("Key error")
except Exception:
self.log.exception("Unexpected exception caught during testing")
except KeyboardInterrupt:
self.log.warning("Exiting after keyboard interrupt")
self.success = TestStatus.PASSED
if success == TestStatus.FAILED and self.options.pdbonfailure:
def shutdown(self):
"""Call this method to shut down the test framework object."""
if self.success == TestStatus.FAILED and self.options.pdbonfailure:
print("Testcase failed. Attaching python debugger. Enter ? for help")
pdb.set_trace()
@ -225,7 +242,7 @@ class BitcoinTestFramework(metaclass=BitcoinTestMetaClass):
should_clean_up = (
not self.options.nocleanup and
not self.options.noshutdown and
success != TestStatus.FAILED and
self.success != TestStatus.FAILED and
not self.options.perf
)
if should_clean_up:
@ -238,20 +255,33 @@ class BitcoinTestFramework(metaclass=BitcoinTestMetaClass):
self.log.warning("Not cleaning up dir {}".format(self.options.tmpdir))
cleanup_tree_on_exit = False
if success == TestStatus.PASSED:
if self.success == TestStatus.PASSED:
self.log.info("Tests successful")
exit_code = TEST_EXIT_PASSED
elif success == TestStatus.SKIPPED:
elif self.success == TestStatus.SKIPPED:
self.log.info("Test skipped")
exit_code = TEST_EXIT_SKIPPED
else:
self.log.error("Test failed. Test logging available at %s/test_framework.log", self.options.tmpdir)
self.log.error("Hint: Call {} '{}' to consolidate all logs".format(os.path.normpath(os.path.dirname(os.path.realpath(__file__)) + "/../combine_logs.py"), self.options.tmpdir))
exit_code = TEST_EXIT_FAILED
logging.shutdown()
# Logging.shutdown will not remove stream- and filehandlers, so we must
# do it explicitly. Handlers are removed so the next test run can apply
# different log handler settings.
# See: https://docs.python.org/3/library/logging.html#logging.shutdown
for h in list(self.log.handlers):
h.flush()
h.close()
self.log.removeHandler(h)
rpc_logger = logging.getLogger("BitcoinRPC")
for h in list(rpc_logger.handlers):
h.flush()
rpc_logger.removeHandler(h)
if cleanup_tree_on_exit:
shutil.rmtree(self.options.tmpdir)
sys.exit(exit_code)
self.nodes.clear()
return exit_code
# Methods to override in subclass test scripts.
def set_test_params(self):

@ -0,0 +1,75 @@
#!/usr/bin/env python3
# Copyright (c) 2019 The Bitcoin Core developers
# Distributed under the MIT software license, see the accompanying
# file COPYING or http://www.opensource.org/licenses/mit-license.php.
from test_framework.test_framework import BitcoinTestFramework
class TestShell:
"""Wrapper Class for BitcoinTestFramework.
The TestShell class extends the BitcoinTestFramework
rpc & daemon process management functionality to external
python environments.
It is a singleton class, which ensures that users only
start a single TestShell at a time."""
class __TestShell(BitcoinTestFramework):
def set_test_params(self):
pass
def run_test(self):
pass
def setup(self, **kwargs):
if self.running:
print("TestShell is already running!")
return
# Num_nodes parameter must be set
# by BitcoinTestFramework child class.
self.num_nodes = kwargs.get('num_nodes', 1)
kwargs.pop('num_nodes', None)
# User parameters override default values.
for key, value in kwargs.items():
if hasattr(self, key):
setattr(self, key, value)
elif hasattr(self.options, key):
setattr(self.options, key, value)
else:
raise KeyError(key + " not a valid parameter key!")
super().setup()
self.running = True
def shutdown(self):
if not self.running:
print("TestShell is not running!")
else:
super().shutdown()
self.running = False
def reset(self):
if self.running:
print("Shutdown TestWrapper before resetting!")
else:
self.num_nodes = None
super().__init__()
instance = None
def __new__(cls):
# This implementation enforces singleton pattern, and will return the
# previously initialized instance if available
if not TestShell.instance:
TestShell.instance = TestShell.__TestShell()
TestShell.instance.running = False
return TestShell.instance
def __getattr__(self, name):
return getattr(self.instance, name)
def __setattr__(self, name, value):
return setattr(self.instance, name, value)
Loading…
Cancel
Save