Source code for cocotb.regression

# Copyright (c) 2013, 2018 Potential Ventures Ltd
# Copyright (c) 2013 SolarFlare Communications Inc
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#     * Redistributions of source code must retain the above copyright
#       notice, this list of conditions and the following disclaimer.
#     * Redistributions in binary form must reproduce the above copyright
#       notice, this list of conditions and the following disclaimer in the
#       documentation and/or other materials provided with the distribution.
#     * Neither the name of Potential Ventures Ltd,
#       SolarFlare Communications Inc nor the
#       names of its contributors may be used to endorse or promote products
#       derived from this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
# ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL POTENTIAL VENTURES LTD BE LIABLE FOR ANY
# DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
# (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
# LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
# ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
# SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

"""All things relating to regression capabilities."""

import hashlib
import inspect
import math
import os
import pdb
import random
import sys
import time
import traceback
from itertools import product
from typing import Any, Iterable, Optional, Tuple

import cocotb
import cocotb.ANSI as ANSI
from cocotb import simulator
from cocotb._deprecation import deprecated
from cocotb.decorators import Task
from cocotb.decorators import test as Test
from cocotb.handle import SimHandle
from cocotb.log import SimLog
from cocotb.outcomes import Error, Outcome
from cocotb.result import SimFailure, TestSuccess
from cocotb.utils import get_sim_time, remove_traceback_frames, want_color_output
from cocotb.xunit_reporter import XUnitReporter

_pdb_on_exception = "COCOTB_PDB_ON_EXCEPTION" in os.environ

# Optional support for coverage collection of testbench files
coverage = None
if "COVERAGE" in os.environ:
    try:
        import coverage
    except ImportError as e:
        msg = (
            "Coverage collection requested but coverage module not available"
            "\n"
            "Import error was: %s\n" % repr(e)
        )
        sys.stderr.write(msg)


def _my_import(name: str) -> Any:
    mod = __import__(name)
    components = name.split(".")
    for comp in components[1:]:
        mod = getattr(mod, comp)
    return mod


_logger = SimLog(__name__)


[docs]class RegressionManager: """Encapsulates all regression capability into a single place""" def __init__(self, dut: SimHandle, tests: Iterable[Test]): """ Args: dut (SimHandle): The root handle to pass into test functions. tests (Iterable[Test]): tests to run """ self._dut = dut self._test = None self._test_task = None self._test_start_time = None self._test_start_sim_time = None self._cov = None self.log = _logger self.start_time = time.time() self.test_results = [] self.count = 0 self.passed = 0 self.skipped = 0 self.failures = 0 self._tearing_down = False # Setup XUnit ################### results_filename = os.getenv("COCOTB_RESULTS_FILE", "results.xml") suite_name = os.getenv("RESULT_TESTSUITE", "all") package_name = os.getenv("RESULT_TESTPACKAGE", "all") self.xunit = XUnitReporter(filename=results_filename) self.xunit.add_testsuite(name=suite_name, package=package_name) self.xunit.add_property(name="random_seed", value=str(cocotb.RANDOM_SEED)) # Setup Coverage #################### if coverage is not None: self.log.info("Enabling coverage collection of Python code") config_filepath = os.getenv("COVERAGE_RCFILE") if config_filepath is None: # Exclude cocotb itself from coverage collection. cocotb_package_dir = os.path.dirname(__file__) self._cov = coverage.coverage( branch=True, omit=[f"{cocotb_package_dir}/*"] ) else: # Allow the config file to handle all configuration self._cov = coverage.coverage() self._cov.start() # Test Discovery #################### self._queue = [] for test in tests: self.log.info(f"Found test {test.__module__}.{test.__qualname__}") self._queue.append(test) self.ntests = len(self._queue) if not self._queue: self.log.warning("No tests were discovered") self._queue.sort(key=lambda test: (test.stage, test._id))
[docs] @classmethod def from_discovery(cls, dut: SimHandle): """ Obtains the test list by discovery. See :envvar:`MODULE` and :envvar:`TESTCASE` for details on how tests are discovered. Args: dut (SimHandle): The root handle to pass into test functions. """ tests = cls._discover_tests() return cls(dut, tests)
@classmethod def _discover_tests(cls) -> Iterable[Test]: """ Discovers tests in files automatically. See :envvar:`MODULE` and :envvar:`TESTCASE` for details on how tests are discovered. """ module_str = os.getenv("MODULE") test_str = os.getenv("TESTCASE") if module_str is None: raise ValueError( "Environment variable MODULE, which defines the module(s) to execute, is not defined." ) modules = [s.strip() for s in module_str.split(",") if s.strip()] cls._setup_pytest_assertion_rewriting(modules) tests = None if test_str: tests = [s.strip() for s in test_str.split(",") if s.strip()] for module_name in modules: try: _logger.debug("Python Path: " + ",".join(sys.path)) _logger.debug("PWD: " + os.getcwd()) module = _my_import(module_name) except Exception as E: _logger.critical("Failed to import module %s: %s", module_name, E) _logger.info('MODULE variable was "%s"', ".".join(modules)) _logger.info("Traceback: ") _logger.info(traceback.format_exc()) raise if tests is not None: not_found_tests = [] # Specific functions specified, don't auto-discover for test_name in tests: try: test = getattr(module, test_name) except AttributeError: not_found_tests.append(test_name) continue if not isinstance(test, Test): _logger.error( "Requested %s from module %s isn't a cocotb.test decorated coroutine", test_name, module_name, ) raise ImportError( "Failed to find requested test %s" % test_name ) # If we request a test manually, it should be run even if skip=True is set. test.skip = False yield test # Use the non-matching test names in the next module search tests = not_found_tests else: # auto-discover for thing in vars(module).values(): if isinstance(thing, Test): yield thing # If any test were not found in any module, raise an error if tests: _logger.error( "Requested test(s) %s wasn't found in module(s) %s", tests, modules ) raise AttributeError("Test(s) %s doesn't exist in %s" % (tests, modules)) @classmethod def _setup_pytest_assertion_rewriting(cls, test_modules: Iterable[str]) -> None: try: import pytest except ImportError: _logger.info( "pytest not found, install it to enable better AssertionError messages" ) return try: # Install the assertion rewriting hook, which must be done before we # import the test modules. from _pytest.assertion import install_importhook from _pytest.config import Config # Pass python_files to Config to support assertion rewriting only on test modules. # See https://github.com/pytest-dev/pytest/discussions/10052. test_modules_str = " ".join(f"{name}.py" for name in test_modules) pytest_conf = Config.fromdictargs( {}, ["--capture=no", "-o", f"python_files={test_modules_str}"] ) install_importhook(pytest_conf) except Exception: _logger.exception( "Configuring the assertion rewrite hook using pytest {} failed. " "Please file a bug report!".format(pytest.__version__) ) @deprecated("This method is now private.") def tear_down(self) -> None: self._tear_down() def _tear_down(self) -> None: # prevent re-entering the tear down procedure if not self._tearing_down: self._tearing_down = True else: return # fail remaining tests while True: test = self._next_test() if test is None: break self._record_result( test=test, outcome=Error(SimFailure), wall_time_s=0, sim_time_ns=0 ) # Write out final log messages self._log_test_summary() # Generate output reports self.xunit.write() if self._cov: self._cov.stop() self.log.info("Writing coverage data") self._cov.save() self._cov.html_report() if cocotb._library_coverage is not None: # TODO: move this once we have normal shutdown behavior to _sim_event cocotb._library_coverage.stop() cocotb._library_coverage.save() # Setup simulator finalization simulator.stop_simulator() @deprecated("This method is now private.") def next_test(self) -> Optional[Test]: return self._next_test() def _next_test(self) -> Optional[Test]: """Get the next test to run""" if not self._queue: return None self.count += 1 return self._queue.pop(0) @deprecated("This method is now private.") def handle_result(self, test: Task) -> None: self._handle_result(test) def _handle_result(self, test: Task) -> None: """Handle a test completing. Dump result to XML and schedule the next test (if any). Entered by the scheduler. Args: test: The test that completed """ assert test is self._test_task real_time = time.time() - self._test_start_time sim_time_ns = get_sim_time("ns") - self._test_start_sim_time self._record_result( test=self._test, outcome=self._test_task._outcome, wall_time_s=real_time, sim_time_ns=sim_time_ns, ) self._execute() def _init_test(self, test: Test) -> Optional[Task]: """Initialize a test. Record outcome if the initialization fails. Record skip if the test is skipped. Save the initialized test if it successfully initializes. """ if test.skip: hilight_start = ANSI.COLOR_SKIPPED if want_color_output() else "" hilight_end = ANSI.COLOR_DEFAULT if want_color_output() else "" # Want this to stand out a little bit self.log.info( "{start}skipping{end} {name} ({i}/{total})".format( start=hilight_start, i=self.count, total=self.ntests, end=hilight_end, name=test.__qualname__, ) ) self._record_result(test, None, 0, 0) return None test_init_outcome = cocotb.outcomes.capture(test, self._dut) if isinstance(test_init_outcome, cocotb.outcomes.Error): self.log.error( "Failed to initialize test %s" % test.__qualname__, exc_info=test_init_outcome.error, ) self._record_result(test, test_init_outcome, 0, 0) return None running_test = test_init_outcome.get() # seed random number generator based on test module, name, and RANDOM_SEED hasher = hashlib.sha1() hasher.update(test.__qualname__.encode()) hasher.update(test.__module__.encode()) seed = cocotb.RANDOM_SEED + int(hasher.hexdigest(), 16) random.seed(seed) return running_test def _score_test(self, test: Test, outcome: Outcome) -> Tuple[bool, bool]: """ Given a test and the test's outcome, determine if the test met expectations and log pertinent information """ # scoring outcomes result_pass = True sim_failed = False try: outcome.get() except Exception as e: result = remove_traceback_frames(e, ["_score_test", "get"]) else: result = TestSuccess() if ( isinstance(result, TestSuccess) and not test.expect_fail and not test.expect_error ): self._log_test_passed(test, None, None) elif isinstance(result, AssertionError) and test.expect_fail: self._log_test_passed(test, result, "failed as expected") elif isinstance(result, TestSuccess) and test.expect_error: self._log_test_failed(test, None, "passed but we expected an error") result_pass = False elif isinstance(result, TestSuccess): self._log_test_failed(test, None, "passed but we expected a failure") result_pass = False elif isinstance(result, SimFailure): if isinstance(result, test.expect_error): self._log_test_passed(test, result, "errored as expected") else: self.log.error("Test error has lead to simulator shutting us down") result_pass = False # whether we expected it or not, the simulation has failed unrecoverably sim_failed = True elif test.expect_error: if isinstance(result, test.expect_error): self._log_test_passed(test, result, "errored as expected") else: self._log_test_failed(test, result, "errored with unexpected type ") result_pass = False else: self._log_test_failed(test, result, None) result_pass = False if _pdb_on_exception: pdb.post_mortem(result.__traceback__) return result_pass, sim_failed def _log_test_passed( self, test: Test, result: Optional[Exception] = None, msg: Optional[str] = None ) -> None: start_hilight = ANSI.COLOR_PASSED if want_color_output() else "" stop_hilight = ANSI.COLOR_DEFAULT if want_color_output() else "" if msg is None: rest = "" else: rest = f": {msg}" if result is None: result_was = "" else: result_was = f" (result was {type(result).__qualname__})" self.log.info( f"{test.__qualname__} {start_hilight}passed{stop_hilight}{rest}{result_was}" ) def _log_test_failed( self, test: Test, result: Optional[Exception] = None, msg: Optional[str] = None ) -> None: start_hilight = ANSI.COLOR_FAILED if want_color_output() else "" stop_hilight = ANSI.COLOR_DEFAULT if want_color_output() else "" if msg is None: rest = "" else: rest = f": {msg}" self.log.info( f"{test.__qualname__} {start_hilight}failed{stop_hilight}{rest}", exc_info=result, ) def _record_result( self, test: Test, outcome: Optional[Outcome], wall_time_s: float, sim_time_ns: float, ) -> None: ratio_time = self._safe_divide(sim_time_ns, wall_time_s) try: lineno = inspect.getsourcelines(test._func)[1] except OSError: lineno = 1 self.xunit.add_testcase( name=test.__qualname__, classname=test.__module__, file=inspect.getfile(test._func), lineno=repr(lineno), time=repr(wall_time_s), sim_time_ns=repr(sim_time_ns), ratio_time=repr(ratio_time), ) if outcome is None: # skipped test_pass, sim_failed = None, False self.xunit.add_skipped() self.skipped += 1 else: test_pass, sim_failed = self._score_test(test, outcome) if not test_pass: self.xunit.add_failure( message=f"Test failed with RANDOM_SEED={cocotb.RANDOM_SEED}" ) self.failures += 1 else: self.passed += 1 self.test_results.append( { "test": ".".join([test.__module__, test.__qualname__]), "pass": test_pass, "sim": sim_time_ns, "real": wall_time_s, "ratio": ratio_time, } ) if sim_failed: self._tear_down() return @deprecated("This method is now private.") def execute(self) -> None: self._execute() def _execute(self) -> None: while True: self._test = self._next_test() if self._test is None: return self._tear_down() self._test_task = self._init_test(self._test) if self._test_task is not None: return self._start_test() def _start_test(self) -> None: # Want this to stand out a little bit start = "" end = "" if want_color_output(): start = ANSI.COLOR_TEST end = ANSI.COLOR_DEFAULT self.log.info( "{start}running{end} {name} ({i}/{total}){description}".format( start=start, i=self.count, total=self.ntests, end=end, name=self._test.__qualname__, description=_trim(self._test.__doc__), ) ) self._test_start_time = time.time() self._test_start_sim_time = get_sim_time("ns") cocotb.scheduler._add_test(self._test_task) def _log_test_summary(self) -> None: real_time = time.time() - self.start_time sim_time_ns = get_sim_time("ns") ratio_time = self._safe_divide(sim_time_ns, real_time) if len(self.test_results) == 0: return TEST_FIELD = "TEST" RESULT_FIELD = "STATUS" SIM_FIELD = "SIM TIME (ns)" REAL_FIELD = "REAL TIME (s)" RATIO_FIELD = "RATIO (ns/s)" TOTAL_NAME = f"TESTS={self.ntests} PASS={self.passed} FAIL={self.failures} SKIP={self.skipped}" TEST_FIELD_LEN = max( len(TEST_FIELD), len(TOTAL_NAME), len(max([x["test"] for x in self.test_results], key=len)), ) RESULT_FIELD_LEN = len(RESULT_FIELD) SIM_FIELD_LEN = len(SIM_FIELD) REAL_FIELD_LEN = len(REAL_FIELD) RATIO_FIELD_LEN = len(RATIO_FIELD) header_dict = dict( a=TEST_FIELD, b=RESULT_FIELD, c=SIM_FIELD, d=REAL_FIELD, e=RATIO_FIELD, a_len=TEST_FIELD_LEN, b_len=RESULT_FIELD_LEN, c_len=SIM_FIELD_LEN, d_len=REAL_FIELD_LEN, e_len=RATIO_FIELD_LEN, ) LINE_LEN = ( 3 + TEST_FIELD_LEN + 2 + RESULT_FIELD_LEN + 2 + SIM_FIELD_LEN + 2 + REAL_FIELD_LEN + 2 + RATIO_FIELD_LEN + 3 ) LINE_SEP = "*" * LINE_LEN + "\n" summary = "" summary += LINE_SEP summary += "** {a:<{a_len}} {b:^{b_len}} {c:>{c_len}} {d:>{d_len}} {e:>{e_len}} **\n".format( **header_dict ) summary += LINE_SEP test_line = "** {a:<{a_len}} {start}{b:^{b_len}}{end} {c:>{c_len}.2f} {d:>{d_len}.2f} {e:>{e_len}} **\n" for result in self.test_results: hilite = "" lolite = "" if result["pass"] is None: ratio = "-.--" pass_fail_str = "SKIP" if want_color_output(): hilite = ANSI.COLOR_SKIPPED lolite = ANSI.COLOR_DEFAULT elif result["pass"]: ratio = format(result["ratio"], "0.2f") pass_fail_str = "PASS" if want_color_output(): hilite = ANSI.COLOR_PASSED lolite = ANSI.COLOR_DEFAULT else: ratio = format(result["ratio"], "0.2f") pass_fail_str = "FAIL" if want_color_output(): hilite = ANSI.COLOR_FAILED lolite = ANSI.COLOR_DEFAULT test_dict = dict( a=result["test"], b=pass_fail_str, c=result["sim"], d=result["real"], e=ratio, a_len=TEST_FIELD_LEN, b_len=RESULT_FIELD_LEN, c_len=SIM_FIELD_LEN - 1, d_len=REAL_FIELD_LEN - 1, e_len=RATIO_FIELD_LEN - 1, start=hilite, end=lolite, ) summary += test_line.format(**test_dict) summary += LINE_SEP summary += test_line.format( a=TOTAL_NAME, b="", c=sim_time_ns, d=real_time, e=format(ratio_time, "0.2f"), a_len=TEST_FIELD_LEN, b_len=RESULT_FIELD_LEN, c_len=SIM_FIELD_LEN - 1, d_len=REAL_FIELD_LEN - 1, e_len=RATIO_FIELD_LEN - 1, start="", end="", ) summary += LINE_SEP self.log.info(summary) @staticmethod def _safe_divide(a: float, b: float) -> float: try: return a / b except ZeroDivisionError: if a == 0: return float("nan") else: return float("inf")
def _create_test(function, name, documentation, mod, *args, **kwargs): """Factory function to create tests, avoids late binding. Creates a test dynamically. The test will call the supplied function with the supplied arguments. Args: function (function): The test function to run. name (str): The name of the test. documentation (str): The docstring for the test. mod (module): The module this function belongs to. *args: Remaining args to pass to test function. **kwargs: Passed to the test function. Returns: Decorated test function """ async def _my_test(dut): await function(dut, *args, **kwargs) _my_test.__name__ = name _my_test.__qualname__ = name _my_test.__doc__ = documentation _my_test.__module__ = mod.__name__ return cocotb.test()(_my_test)
[docs]class TestFactory: """Factory to automatically generate tests. Args: test_function: A Callable that returns the test Coroutine. Must take *dut* as the first argument. *args: Remaining arguments are passed directly to the test function. Note that these arguments are not varied. An argument that varies with each test must be a keyword argument to the test function. **kwargs: Remaining keyword arguments are passed directly to the test function. Note that these arguments are not varied. An argument that varies with each test must be a keyword argument to the test function. Assuming we have a common test function that will run a test. This test function will take keyword arguments (for example generators for each of the input interfaces) and generate tests that call the supplied function. This Factory allows us to generate sets of tests based on the different permutations of the possible arguments to the test function. For example, if we have a module that takes backpressure, has two configurable features where enabling ``feature_b`` requires ``feature_a`` to be active, and need to test against data generation routines ``gen_a`` and ``gen_b``: >>> tf = TestFactory(test_function=run_test) >>> tf.add_option(name='data_in', optionlist=[gen_a, gen_b]) >>> tf.add_option('backpressure', [None, random_backpressure]) >>> tf.add_option(('feature_a', 'feature_b'), [(False, False), (True, False), (True, True)]) >>> tf.generate_tests() We would get the following tests: * ``gen_a`` with no backpressure and both features disabled * ``gen_a`` with no backpressure and only ``feature_a`` enabled * ``gen_a`` with no backpressure and both features enabled * ``gen_a`` with ``random_backpressure`` and both features disabled * ``gen_a`` with ``random_backpressure`` and only ``feature_a`` enabled * ``gen_a`` with ``random_backpressure`` and both features enabled * ``gen_b`` with no backpressure and both features disabled * ``gen_b`` with no backpressure and only ``feature_a`` enabled * ``gen_b`` with no backpressure and both features enabled * ``gen_b`` with ``random_backpressure`` and both features disabled * ``gen_b`` with ``random_backpressure`` and only ``feature_a`` enabled * ``gen_b`` with ``random_backpressure`` and both features enabled The tests are appended to the calling module for auto-discovery. Tests are simply named ``test_function_N``. The docstring for the test (hence the test description) includes the name and description of each generator. .. versionchanged:: 1.5 Groups of options are now supported """ # Prevent warnings from collection of TestFactories by unit testing frameworks. __test__ = False def __init__(self, test_function, *args, **kwargs): self.test_function = test_function self.name = self.test_function.__qualname__ self.args = args self.kwargs_constant = kwargs self.kwargs = {} self.log = _logger
[docs] def add_option(self, name, optionlist): """Add a named option to the test. Args: name (str or iterable of str): An option name, or an iterable of several option names. Passed to test as keyword arguments. optionlist (list): A list of possible options for this test knob. If N names were specified, this must be a list of N-tuples or lists, where each element specifies a value for its respective option. .. versionchanged:: 1.5 Groups of options are now supported """ if not isinstance(name, str): name = tuple(name) for opt in optionlist: if len(name) != len(opt): raise ValueError( "Mismatch between number of options and number of option values in group" ) self.kwargs[name] = optionlist
[docs] def generate_tests(self, prefix="", postfix=""): """ Generate an exhaustive set of tests using the cartesian product of the possible keyword arguments. The generated tests are appended to the namespace of the calling module. Args: prefix (str): Text string to append to start of ``test_function`` name when naming generated test cases. This allows reuse of a single ``test_function`` with multiple :class:`TestFactories <.TestFactory>` without name clashes. postfix (str): Text string to append to end of ``test_function`` name when naming generated test cases. This allows reuse of a single ``test_function`` with multiple :class:`TestFactories <.TestFactory>` without name clashes. """ frm = inspect.stack()[1] mod = inspect.getmodule(frm[0]) d = self.kwargs for index, testoptions in enumerate( dict(zip(d, v)) for v in product(*d.values()) ): name = "%s%s%s_%03d" % (prefix, self.name, postfix, index + 1) doc = "Automatically generated test\n\n" # preprocess testoptions to split tuples testoptions_split = {} for optname, optvalue in testoptions.items(): if isinstance(optname, str): testoptions_split[optname] = optvalue else: # previously checked in add_option; ensure nothing has changed assert len(optname) == len(optvalue) for n, v in zip(optname, optvalue): testoptions_split[n] = v for optname, optvalue in testoptions_split.items(): if callable(optvalue): if not optvalue.__doc__: desc = "No docstring supplied" else: desc = optvalue.__doc__.split("\n")[0] doc += "\t{}: {} ({})\n".format( optname, optvalue.__qualname__, desc ) else: doc += "\t{}: {}\n".format(optname, repr(optvalue)) self.log.debug( 'Adding generated test "%s" to module "%s"' % (name, mod.__name__) ) kwargs = {} kwargs.update(self.kwargs_constant) kwargs.update(testoptions_split) if hasattr(mod, name): self.log.error( "Overwriting %s in module %s. " "This causes a previously defined testcase " "not to be run. Consider setting/changing " "name_postfix" % (name, mod) ) setattr( mod, name, _create_test(self.test_function, name, doc, mod, *self.args, **kwargs), )
def _trim(docstring: Optional[str]) -> str: """Normalizes test docstrings Based on https://www.python.org/dev/peps/pep-0257/#handling-docstring-indentation. """ if docstring is None: return "" # Convert tabs to spaces (following the normal Python rules) # and split into a list of lines: lines = docstring.expandtabs().splitlines() # Determine minimum indentation (first line doesn't count): indent = math.inf for line in lines[1:]: stripped = line.lstrip() if stripped: indent = min(indent, len(line) - len(stripped)) # Remove indentation (first line is special): trimmed = [lines[0].strip()] if indent < math.inf: for line in lines[1:]: trimmed.append(line[indent:].rstrip()) # Strip off trailing and leading blank lines: while trimmed and not trimmed[-1]: trimmed.pop() while trimmed and not trimmed[0]: trimmed.pop(0) # Add one newline back trimmed.insert(0, "") # Return a single string: return "\n ".join(trimmed)