1 # Shell class for a test, inherited by all individual tests 2 # 3 # Methods: 4 # __init__ initialise 5 # initialize run once for each job 6 # setup run once for each new version of the test installed 7 # run run the test (wrapped by job.run_test()) 8 # 9 # Data: 10 # job backreference to the job this test instance is part of 11 # outputdir eg. results/<job>/<testname.tag> 12 # resultsdir eg. results/<job>/<testname.tag>/results 13 # profdir eg. results/<job>/<testname.tag>/profiling 14 # debugdir eg. results/<job>/<testname.tag>/debug 15 # bindir eg. tests/<test> 16 # src eg. tests/<test>/src 17 # tmpdir eg. tmp/<tempname>_<testname.tag> 18 19 #pylint: disable-msg=C0111 20 21 import fcntl, json, os, re, sys, shutil, stat, tempfile, time, traceback 22 import logging 23 24 from autotest_lib.client.bin import utils 25 from autotest_lib.client.common_lib import error 26 27 28 class base_test(object): 29 preserve_srcdir = False 30 network_destabilizing = False 31 32 def __init__(self, job, bindir, outputdir): 33 self.job = job 34 self.pkgmgr = job.pkgmgr 35 self.autodir = job.autodir 36 self.outputdir = outputdir 37 self.tagged_testname = os.path.basename(self.outputdir) 38 self.resultsdir = os.path.join(self.outputdir, 'results') 39 os.mkdir(self.resultsdir) 40 self.profdir = os.path.join(self.outputdir, 'profiling') 41 os.mkdir(self.profdir) 42 self.debugdir = os.path.join(self.outputdir, 'debug') 43 os.mkdir(self.debugdir) 44 # TODO(ericli): figure out how autotest crash handler work with cros 45 # Once this is re-enabled import getpass. crosbug.com/31232 46 # crash handler, we should restore it in near term. 47 # if getpass.getuser() == 'root': 48 # self.configure_crash_handler() 49 # else: 50 self.crash_handling_enabled = False 51 self.bindir = bindir 52 self.srcdir = os.path.join(self.bindir, 'src') 53 self.tmpdir = tempfile.mkdtemp("_" + self.tagged_testname, 54 dir=job.tmpdir) 55 self._keyvals = [] 56 self._new_keyval = False 57 self.failed_constraints = [] 58 self.iteration = 0 59 self.before_iteration_hooks = [] 60 self.after_iteration_hooks = [] 61 62 # Flag to indicate if the test has succeeded or failed. 63 self.success = False 64 65 66 def configure_crash_handler(self): 67 pass 68 69 70 def crash_handler_report(self): 71 pass 72 73 74 def assert_(self, expr, msg='Assertion failed.'): 75 if not expr: 76 raise error.TestError(msg) 77 78 79 def write_test_keyval(self, attr_dict): 80 utils.write_keyval(self.outputdir, attr_dict, 81 tap_report=self.job._tap) 82 83 84 @staticmethod 85 def _append_type_to_keys(dictionary, typename): 86 new_dict = {} 87 for key, value in dictionary.iteritems(): 88 new_key = "%s{%s}" % (key, typename) 89 new_dict[new_key] = value 90 return new_dict 91 92 93 def output_perf_value(self, description, value, units=None, 94 higher_is_better=None, graph=None, replacement='_'): 95 """ 96 Records a measured performance value in an output file. 97 98 The output file will subsequently be parsed by the TKO parser to have 99 the information inserted into the results database. 100 101 @param description: A string describing the measured perf value. Must 102 be maximum length 256, and may only contain letters, numbers, 103 periods, dashes, and underscores. For example: 104 "page_load_time", "scrolling-frame-rate". 105 @param value: A number representing the measured perf value, or a list 106 of measured values if a test takes multiple measurements. 107 Measured perf values can be either ints or floats. 108 @param units: A string describing the units associated with the 109 measured perf value. Must be maximum length 32, and may only 110 contain letters, numbers, periods, dashes, and underscores. 111 For example: "msec", "fps", "score", "runs_per_second". 112 @param higher_is_better: A boolean indicating whether or not a "higher" 113 measured perf value is considered to be better. If False, it is 114 assumed that a "lower" measured value is considered to be 115 better. This impacts dashboard plotting and email notification. 116 Pure autotests are expected to specify either True or False! 117 This value can be set to "None" to indicate that the perf 118 dashboard should apply the rules encoded via Chromium 119 unit-info.json. This is only used for tracking Chromium based 120 tests (in particular telemetry). 121 @param graph: A string indicating the name of the graph on which 122 the perf value will be subsequently displayed on the chrome perf 123 dashboard. This allows multiple metrics be grouped together on 124 the same graphs. Defaults to None, indicating that the perf 125 value should be displayed individually on a separate graph. 126 @param replacement: string to replace illegal characters in 127 |description| and |units| with. 128 """ 129 if len(description) > 256: 130 raise ValueError('The description must be at most 256 characters.') 131 if units and len(units) > 32: 132 raise ValueError('The units must be at most 32 characters.') 133 134 # If |replacement| is legal replace illegal characters with it. 135 string_regex = re.compile(r'[^-\.\w]') 136 if replacement is None or re.search(string_regex, replacement): 137 raise ValueError('Invalid replacement string to mask illegal ' 138 'characters. May only contain letters, numbers, ' 139 'periods, dashes, and underscores. ' 140 'replacement: %s' % replacement) 141 description = re.sub(string_regex, replacement, description) 142 units = re.sub(string_regex, replacement, units) if units else None 143 144 entry = { 145 'description': description, 146 'value': value, 147 'units': units, 148 'higher_is_better': higher_is_better, 149 'graph': graph 150 } 151 152 output_path = os.path.join(self.resultsdir, 'perf_measurements') 153 with open(output_path, 'a') as fp: 154 fp.write(json.dumps(entry, sort_keys=True) + '\n') 155 156 157 def write_perf_keyval(self, perf_dict): 158 self.write_iteration_keyval({}, perf_dict, 159 tap_report=self.job._tap) 160 161 162 def write_attr_keyval(self, attr_dict): 163 self.write_iteration_keyval(attr_dict, {}, 164 tap_report=self.job._tap) 165 166 167 def write_iteration_keyval(self, attr_dict, perf_dict, tap_report=None): 168 # append the dictionaries before they have the {perf} and {attr} added 169 self._keyvals.append({'attr':attr_dict, 'perf':perf_dict}) 170 self._new_keyval = True 171 172 if attr_dict: 173 attr_dict = self._append_type_to_keys(attr_dict, "attr") 174 utils.write_keyval(self.resultsdir, attr_dict, type_tag="attr", 175 tap_report=tap_report) 176 177 if perf_dict: 178 perf_dict = self._append_type_to_keys(perf_dict, "perf") 179 utils.write_keyval(self.resultsdir, perf_dict, type_tag="perf", 180 tap_report=tap_report) 181 182 keyval_path = os.path.join(self.resultsdir, "keyval") 183 print >> open(keyval_path, "a"), "" 184 185 186 def analyze_perf_constraints(self, constraints): 187 if not self._new_keyval: 188 return 189 190 # create a dict from the keyvals suitable as an environment for eval 191 keyval_env = self._keyvals[-1]['perf'].copy() 192 keyval_env['__builtins__'] = None 193 self._new_keyval = False 194 failures = [] 195 196 # evaluate each constraint using the current keyvals 197 for constraint in constraints: 198 logging.info('___________________ constraint = %s', constraint) 199 logging.info('___________________ keyvals = %s', keyval_env) 200 201 try: 202 if not eval(constraint, keyval_env): 203 failures.append('%s: constraint was not met' % constraint) 204 except: 205 failures.append('could not evaluate constraint: %s' 206 % constraint) 207 208 # keep track of the errors for each iteration 209 self.failed_constraints.append(failures) 210 211 212 def process_failed_constraints(self): 213 msg = '' 214 for i, failures in enumerate(self.failed_constraints): 215 if failures: 216 msg += 'iteration %d:%s ' % (i, ','.join(failures)) 217 218 if msg: 219 raise error.TestFail(msg) 220 221 222 def register_before_iteration_hook(self, iteration_hook): 223 """ 224 This is how we expect test writers to register a before_iteration_hook. 225 This adds the method to the list of hooks which are executed 226 before each iteration. 227 228 @param iteration_hook: Method to run before each iteration. A valid 229 hook accepts a single argument which is the 230 test object. 231 """ 232 self.before_iteration_hooks.append(iteration_hook) 233 234 235 def register_after_iteration_hook(self, iteration_hook): 236 """ 237 This is how we expect test writers to register an after_iteration_hook. 238 This adds the method to the list of hooks which are executed 239 after each iteration. Hooks are executed starting with the most- 240 recently registered, in stack fashion. 241 242 @param iteration_hook: Method to run after each iteration. A valid 243 hook accepts a single argument which is the 244 test object. 245 """ 246 self.after_iteration_hooks.append(iteration_hook) 247 248 249 def initialize(self): 250 pass 251 252 253 def setup(self): 254 pass 255 256 257 def warmup(self, *args, **dargs): 258 pass 259 260 261 def drop_caches_between_iterations(self): 262 if self.job.drop_caches_between_iterations: 263 utils.drop_caches() 264 265 266 def _call_run_once_with_retry(self, constraints, profile_only, 267 postprocess_profiled_run, args, dargs): 268 """Thin wrapper around _call_run_once that retries unsuccessful tests. 269 270 If the job object's attribute test_retry is > 0 retry any tests that 271 ran unsuccessfully X times. 272 *Note this does not competely re-initialize the test, it only 273 re-executes code once all the initial job set up (packages, 274 sysinfo, etc) is complete. 275 """ 276 if self.job.test_retry != 0: 277 logging.info('Test will be retried a maximum of %d times', 278 self.job.test_retry) 279 280 max_runs = self.job.test_retry 281 for retry_run in xrange(0, max_runs+1): 282 try: 283 self._call_run_once(constraints, profile_only, 284 postprocess_profiled_run, args, dargs) 285 break 286 except error.TestFailRetry as err: 287 if retry_run == max_runs: 288 raise 289 self.job.record('INFO', None, None, 'Run %s failed with %s' % ( 290 retry_run, err)) 291 if retry_run > 0: 292 self.write_test_keyval({'test_retries_before_success': retry_run}) 293 294 295 def _call_run_once(self, constraints, profile_only, 296 postprocess_profiled_run, args, dargs): 297 self.drop_caches_between_iterations() 298 # execute iteration hooks 299 for hook in self.before_iteration_hooks: 300 hook(self) 301 302 try: 303 if profile_only: 304 if not self.job.profilers.present(): 305 self.job.record('WARN', None, None, 306 'No profilers have been added but ' 307 'profile_only is set - nothing ' 308 'will be run') 309 self.run_once_profiling(postprocess_profiled_run, 310 *args, **dargs) 311 else: 312 self.before_run_once() 313 self.run_once(*args, **dargs) 314 self.after_run_once() 315 316 self.postprocess_iteration() 317 self.analyze_perf_constraints(constraints) 318 # Catch and re-raise to let after_iteration_hooks see the exception. 319 except: 320 raise 321 finally: 322 for hook in reversed(self.after_iteration_hooks): 323 hook(self) 324 325 326 def execute(self, iterations=None, test_length=None, profile_only=None, 327 _get_time=time.time, postprocess_profiled_run=None, 328 constraints=(), *args, **dargs): 329 """ 330 This is the basic execute method for the tests inherited from base_test. 331 If you want to implement a benchmark test, it's better to implement 332 the run_once function, to cope with the profiling infrastructure. For 333 other tests, you can just override the default implementation. 334 335 @param test_length: The minimum test length in seconds. We'll run the 336 run_once function for a number of times large enough to cover the 337 minimum test length. 338 339 @param iterations: A number of iterations that we'll run the run_once 340 function. This parameter is incompatible with test_length and will 341 be silently ignored if you specify both. 342 343 @param profile_only: If true run X iterations with profilers enabled. 344 If false run X iterations and one with profiling if profiles are 345 enabled. If None, default to the value of job.default_profile_only. 346 347 @param _get_time: [time.time] Used for unit test time injection. 348 349 @param postprocess_profiled_run: Run the postprocessing for the 350 profiled run. 351 """ 352 353 # For our special class of tests, the benchmarks, we don't want 354 # profilers to run during the test iterations. Let's reserve only 355 # the last iteration for profiling, if needed. So let's stop 356 # all profilers if they are present and active. 357 profilers = self.job.profilers 358 if profilers.active(): 359 profilers.stop(self) 360 if profile_only is None: 361 profile_only = self.job.default_profile_only 362 # If the user called this test in an odd way (specified both iterations 363 # and test_length), let's warn them. 364 if iterations and test_length: 365 logging.debug('Iterations parameter ignored (timed execution)') 366 if test_length: 367 test_start = _get_time() 368 time_elapsed = 0 369 timed_counter = 0 370 logging.debug('Test started. Specified %d s as the minimum test ' 371 'length', test_length) 372 while time_elapsed < test_length: 373 timed_counter = timed_counter + 1 374 if time_elapsed == 0: 375 logging.debug('Executing iteration %d', timed_counter) 376 elif time_elapsed > 0: 377 logging.debug('Executing iteration %d, time_elapsed %d s', 378 timed_counter, time_elapsed) 379 self._call_run_once_with_retry(constraints, profile_only, 380 postprocess_profiled_run, args, 381 dargs) 382 test_iteration_finish = _get_time() 383 time_elapsed = test_iteration_finish - test_start 384 logging.debug('Test finished after %d iterations, ' 385 'time elapsed: %d s', timed_counter, time_elapsed) 386 else: 387 if iterations is None: 388 iterations = 1 389 if iterations > 1: 390 logging.debug('Test started. Specified %d iterations', 391 iterations) 392 for self.iteration in xrange(1, iterations + 1): 393 if iterations > 1: 394 logging.debug('Executing iteration %d of %d', 395 self.iteration, iterations) 396 self._call_run_once_with_retry(constraints, profile_only, 397 postprocess_profiled_run, args, 398 dargs) 399 400 if not profile_only: 401 self.iteration += 1 402 self.run_once_profiling(postprocess_profiled_run, *args, **dargs) 403 404 # Do any postprocessing, normally extracting performance keyvals, etc 405 self.postprocess() 406 self.process_failed_constraints() 407 408 409 def run_once_profiling(self, postprocess_profiled_run, *args, **dargs): 410 profilers = self.job.profilers 411 # Do a profiling run if necessary 412 if profilers.present(): 413 self.drop_caches_between_iterations() 414 profilers.before_start(self) 415 416 self.before_run_once() 417 profilers.start(self) 418 logging.debug('Profilers present. Profiling run started') 419 420 try: 421 self.run_once(*args, **dargs) 422 423 # Priority to the run_once() argument over the attribute. 424 postprocess_attribute = getattr(self, 425 'postprocess_profiled_run', 426 False) 427 428 if (postprocess_profiled_run or 429 (postprocess_profiled_run is None and 430 postprocess_attribute)): 431 self.postprocess_iteration() 432 433 finally: 434 profilers.stop(self) 435 profilers.report(self) 436 437 self.after_run_once() 438 439 440 def postprocess(self): 441 pass 442 443 444 def postprocess_iteration(self): 445 pass 446 447 448 def cleanup(self): 449 pass 450 451 452 def before_run_once(self): 453 """ 454 Override in tests that need it, will be called before any run_once() 455 call including the profiling run (when it's called before starting 456 the profilers). 457 """ 458 pass 459 460 461 def after_run_once(self): 462 """ 463 Called after every run_once (including from a profiled run when it's 464 called after stopping the profilers). 465 """ 466 pass 467 468 469 @staticmethod 470 def _make_writable_to_others(directory): 471 mode = os.stat(directory).st_mode 472 mode = mode | stat.S_IROTH | stat.S_IWOTH | stat.S_IXOTH 473 os.chmod(directory, mode) 474 475 476 def _exec(self, args, dargs): 477 self.job.logging.tee_redirect_debug_dir(self.debugdir, 478 log_name=self.tagged_testname) 479 try: 480 if self.network_destabilizing: 481 self.job.disable_warnings("NETWORK") 482 483 # write out the test attributes into a keyval 484 dargs = dargs.copy() 485 run_cleanup = dargs.pop('run_cleanup', self.job.run_test_cleanup) 486 keyvals = dargs.pop('test_attributes', {}).copy() 487 keyvals['version'] = self.version 488 for i, arg in enumerate(args): 489 keyvals['param-%d' % i] = repr(arg) 490 for name, arg in dargs.iteritems(): 491 keyvals['param-%s' % name] = repr(arg) 492 self.write_test_keyval(keyvals) 493 494 _validate_args(args, dargs, self.initialize, self.setup, 495 self.execute, self.cleanup) 496 497 try: 498 # Make resultsdir and tmpdir accessible to everyone. We may 499 # output data to these directories as others, e.g., chronos. 500 self._make_writable_to_others(self.tmpdir) 501 self._make_writable_to_others(self.resultsdir) 502 503 # Initialize: 504 _cherry_pick_call(self.initialize, *args, **dargs) 505 506 lockfile = open(os.path.join(self.job.tmpdir, '.testlock'), 'w') 507 try: 508 fcntl.flock(lockfile, fcntl.LOCK_EX) 509 # Setup: (compile and install the test, if needed) 510 p_args, p_dargs = _cherry_pick_args(self.setup, args, dargs) 511 utils.update_version(self.srcdir, self.preserve_srcdir, 512 self.version, self.setup, 513 *p_args, **p_dargs) 514 finally: 515 fcntl.flock(lockfile, fcntl.LOCK_UN) 516 lockfile.close() 517 518 # Execute: 519 os.chdir(self.outputdir) 520 521 # call self.warmup cherry picking the arguments it accepts and 522 # translate exceptions if needed 523 _call_test_function(_cherry_pick_call, self.warmup, 524 *args, **dargs) 525 526 if hasattr(self, 'run_once'): 527 p_args, p_dargs = _cherry_pick_args(self.run_once, 528 args, dargs) 529 # pull in any non-* and non-** args from self.execute 530 for param in _get_nonstar_args(self.execute): 531 if param in dargs: 532 p_dargs[param] = dargs[param] 533 else: 534 p_args, p_dargs = _cherry_pick_args(self.execute, 535 args, dargs) 536 537 _call_test_function(self.execute, *p_args, **p_dargs) 538 except Exception: 539 # Save the exception while we run our cleanup() before 540 # reraising it, but log it to so actual time of error is known. 541 exc_info = sys.exc_info() 542 logging.warning('Autotest caught exception when running test:', 543 exc_info=True) 544 545 try: 546 try: 547 if run_cleanup: 548 _cherry_pick_call(self.cleanup, *args, **dargs) 549 except Exception: 550 logging.error('Ignoring exception during cleanup() ' 551 'phase:') 552 traceback.print_exc() 553 logging.error('Now raising the earlier %s error', 554 exc_info[0]) 555 self.crash_handler_report() 556 finally: 557 self.job.logging.restore() 558 try: 559 raise exc_info[0], exc_info[1], exc_info[2] 560 finally: 561 # http://docs.python.org/library/sys.html#sys.exc_info 562 # Be nice and prevent a circular reference. 563 del exc_info 564 else: 565 try: 566 if run_cleanup: 567 _cherry_pick_call(self.cleanup, *args, **dargs) 568 self.crash_handler_report() 569 finally: 570 self.job.logging.restore() 571 except error.AutotestError: 572 if self.network_destabilizing: 573 self.job.enable_warnings("NETWORK") 574 # Pass already-categorized errors on up. 575 raise 576 except Exception, e: 577 if self.network_destabilizing: 578 self.job.enable_warnings("NETWORK") 579 # Anything else is an ERROR in our own code, not execute(). 580 raise error.UnhandledTestError(e) 581 else: 582 if self.network_destabilizing: 583 self.job.enable_warnings("NETWORK") 584 585 586 def runsubtest(self, url, *args, **dargs): 587 """ 588 Execute another autotest test from inside the current test's scope. 589 590 @param test: Parent test. 591 @param url: Url of new test. 592 @param tag: Tag added to test name. 593 @param args: Args for subtest. 594 @param dargs: Dictionary with args for subtest. 595 @iterations: Number of subtest iterations. 596 @profile_only: If true execute one profiled run. 597 """ 598 dargs["profile_only"] = dargs.get("profile_only", False) 599 test_basepath = self.outputdir[len(self.job.resultdir + "/"):] 600 return self.job.run_test(url, master_testpath=test_basepath, 601 *args, **dargs) 602 603 604 def _get_nonstar_args(func): 605 """Extract all the (normal) function parameter names. 606 607 Given a function, returns a tuple of parameter names, specifically 608 excluding the * and ** parameters, if the function accepts them. 609 610 @param func: A callable that we want to chose arguments for. 611 612 @return: A tuple of parameters accepted by the function. 613 """ 614 return func.func_code.co_varnames[:func.func_code.co_argcount] 615 616 617 def _cherry_pick_args(func, args, dargs): 618 """Sanitize positional and keyword arguments before calling a function. 619 620 Given a callable (func), an argument tuple and a dictionary of keyword 621 arguments, pick only those arguments which the function is prepared to 622 accept and return a new argument tuple and keyword argument dictionary. 623 624 Args: 625 func: A callable that we want to choose arguments for. 626 args: A tuple of positional arguments to consider passing to func. 627 dargs: A dictionary of keyword arguments to consider passing to func. 628 Returns: 629 A tuple of: (args tuple, keyword arguments dictionary) 630 """ 631 # Cherry pick args: 632 if func.func_code.co_flags & 0x04: 633 # func accepts *args, so return the entire args. 634 p_args = args 635 else: 636 p_args = () 637 638 # Cherry pick dargs: 639 if func.func_code.co_flags & 0x08: 640 # func accepts **dargs, so return the entire dargs. 641 p_dargs = dargs 642 else: 643 # Only return the keyword arguments that func accepts. 644 p_dargs = {} 645 for param in _get_nonstar_args(func): 646 if param in dargs: 647 p_dargs[param] = dargs[param] 648 649 return p_args, p_dargs 650 651 652 def _cherry_pick_call(func, *args, **dargs): 653 """Cherry picks arguments from args/dargs based on what "func" accepts 654 and calls the function with the picked arguments.""" 655 p_args, p_dargs = _cherry_pick_args(func, args, dargs) 656 return func(*p_args, **p_dargs) 657 658 659 def _validate_args(args, dargs, *funcs): 660 """Verify that arguments are appropriate for at least one callable. 661 662 Given a list of callables as additional parameters, verify that 663 the proposed keyword arguments in dargs will each be accepted by at least 664 one of the callables. 665 666 NOTE: args is currently not supported and must be empty. 667 668 Args: 669 args: A tuple of proposed positional arguments. 670 dargs: A dictionary of proposed keyword arguments. 671 *funcs: Callables to be searched for acceptance of args and dargs. 672 Raises: 673 error.AutotestError: if an arg won't be accepted by any of *funcs. 674 """ 675 all_co_flags = 0 676 all_varnames = () 677 for func in funcs: 678 all_co_flags |= func.func_code.co_flags 679 all_varnames += func.func_code.co_varnames[:func.func_code.co_argcount] 680 681 # Check if given args belongs to at least one of the methods below. 682 if len(args) > 0: 683 # Current implementation doesn't allow the use of args. 684 raise error.TestError('Unnamed arguments not accepted. Please ' 685 'call job.run_test with named args only') 686 687 # Check if given dargs belongs to at least one of the methods below. 688 if len(dargs) > 0: 689 if not all_co_flags & 0x08: 690 # no func accepts *dargs, so: 691 for param in dargs: 692 if not param in all_varnames: 693 raise error.AutotestError('Unknown parameter: %s' % param) 694 695 696 def _installtest(job, url): 697 (group, name) = job.pkgmgr.get_package_name(url, 'test') 698 699 # Bail if the test is already installed 700 group_dir = os.path.join(job.testdir, "download", group) 701 if os.path.exists(os.path.join(group_dir, name)): 702 return (group, name) 703 704 # If the group directory is missing create it and add 705 # an empty __init__.py so that sub-directories are 706 # considered for import. 707 if not os.path.exists(group_dir): 708 os.makedirs(group_dir) 709 f = file(os.path.join(group_dir, '__init__.py'), 'w+') 710 f.close() 711 712 logging.debug("%s: installing test url=%s", name, url) 713 tarball = os.path.basename(url) 714 tarball_path = os.path.join(group_dir, tarball) 715 test_dir = os.path.join(group_dir, name) 716 job.pkgmgr.fetch_pkg(tarball, tarball_path, 717 repo_url = os.path.dirname(url)) 718 719 # Create the directory for the test 720 if not os.path.exists(test_dir): 721 os.mkdir(os.path.join(group_dir, name)) 722 723 job.pkgmgr.untar_pkg(tarball_path, test_dir) 724 725 os.remove(tarball_path) 726 727 # For this 'sub-object' to be importable via the name 728 # 'group.name' we need to provide an __init__.py, 729 # so link the main entry point to this. 730 os.symlink(name + '.py', os.path.join(group_dir, name, 731 '__init__.py')) 732 733 # The test is now installed. 734 return (group, name) 735 736 737 def _call_test_function(func, *args, **dargs): 738 """Calls a test function and translates exceptions so that errors 739 inside test code are considered test failures.""" 740 try: 741 return func(*args, **dargs) 742 except error.AutotestError: 743 raise 744 except Exception, e: 745 # Other exceptions must be treated as a FAIL when 746 # raised during the test functions 747 raise error.UnhandledTestFail(e) 748 749 750 def runtest(job, url, tag, args, dargs, 751 local_namespace={}, global_namespace={}, 752 before_test_hook=None, after_test_hook=None, 753 before_iteration_hook=None, after_iteration_hook=None): 754 local_namespace = local_namespace.copy() 755 global_namespace = global_namespace.copy() 756 # if this is not a plain test name then download and install the 757 # specified test 758 if url.endswith('.tar.bz2'): 759 (testgroup, testname) = _installtest(job, url) 760 bindir = os.path.join(job.testdir, 'download', testgroup, testname) 761 importdir = os.path.join(job.testdir, 'download') 762 modulename = '%s.%s' % (re.sub('/', '.', testgroup), testname) 763 classname = '%s.%s' % (modulename, testname) 764 path = testname 765 else: 766 # If the test is local, it may be under either testdir or site_testdir. 767 # Tests in site_testdir override tests defined in testdir 768 testname = path = url 769 testgroup = '' 770 path = re.sub(':', '/', testname) 771 modulename = os.path.basename(path) 772 classname = '%s.%s' % (modulename, modulename) 773 774 # Try installing the test package 775 # The job object may be either a server side job or a client side job. 776 # 'install_pkg' method will be present only if it's a client side job. 777 if hasattr(job, 'install_pkg'): 778 try: 779 bindir = os.path.join(job.testdir, testname) 780 job.install_pkg(testname, 'test', bindir) 781 except error.PackageInstallError: 782 # continue as a fall back mechanism and see if the test code 783 # already exists on the machine 784 pass 785 786 bindir = None 787 for dir in [job.testdir, getattr(job, 'site_testdir', None)]: 788 if dir is not None and os.path.exists(os.path.join(dir, path)): 789 importdir = bindir = os.path.join(dir, path) 790 if not bindir: 791 raise error.TestError(testname + ': test does not exist') 792 793 subdir = os.path.join(dargs.pop('master_testpath', ""), testname) 794 outputdir = os.path.join(job.resultdir, subdir) 795 if tag: 796 outputdir += '.' + tag 797 798 local_namespace['job'] = job 799 local_namespace['bindir'] = bindir 800 local_namespace['outputdir'] = outputdir 801 802 sys.path.insert(0, importdir) 803 try: 804 exec ('import %s' % modulename, local_namespace, global_namespace) 805 exec ("mytest = %s(job, bindir, outputdir)" % classname, 806 local_namespace, global_namespace) 807 finally: 808 sys.path.pop(0) 809 810 pwd = os.getcwd() 811 os.chdir(outputdir) 812 813 try: 814 mytest = global_namespace['mytest'] 815 mytest.success = False 816 if before_test_hook: 817 before_test_hook(mytest) 818 819 # we use the register iteration hooks methods to register the passed 820 # in hooks 821 if before_iteration_hook: 822 mytest.register_before_iteration_hook(before_iteration_hook) 823 if after_iteration_hook: 824 mytest.register_after_iteration_hook(after_iteration_hook) 825 mytest._exec(args, dargs) 826 mytest.success = True 827 finally: 828 os.chdir(pwd) 829 if after_test_hook: 830 after_test_hook(mytest) 831 shutil.rmtree(mytest.tmpdir, ignore_errors=True) 832