source: tests/test.py@ 2c2b60f

ADT arm-eh ast-experimental enum forall-pointer-decay jacob/cs343-translation new-ast-unique-expr pthread-emulation qualifiedEnum
Last change on this file since 2c2b60f was a2f2fda, checked in by Thierry Delisle <tdelisle@…>, 5 years ago

Finished support for running tests for new and old ast

  • Property mode set to 100755
File size: 14.9 KB
Line 
1#!/usr/bin/python3
2
3from pybin.tools import *
4from pybin.test_run import *
5from pybin import settings
6
7import argparse
8import itertools
9import re
10import sys
11import tempfile
12import time
13
14import os
15import psutil
16import signal
17
18################################################################################
19# help functions
20################################################################################
21
22def find_tests():
23 expected = []
24
25 def match_test(path):
26 match = re.search("^%s\/([\w\/\-_]*).expect\/([\w\-_]+)(\.nast|\.oast)?(\.[\w\-_]+)?\.txt$" % settings.SRCDIR, path)
27 if match :
28 test = Test()
29 test.name = match.group(2)
30 test.path = match.group(1)
31 test.arch = match.group(4)[1:] if match.group(4) else None
32
33 astv = match.group(3)[1:] if match.group(3) else None
34 if astv == 'oast':
35 test.astv = 'old'
36 elif astv == 'nast':
37 test.astv = 'new'
38 elif astv:
39 print('ERROR: "%s", expect file has astv but it is not "nast" or "oast"' % testname, file=sys.stderr)
40 sys.exit(1)
41
42 expected.append(test)
43
44 path_walk( match_test )
45
46 return expected
47
48# reads the directory ./.expect and indentifies the tests
49def list_tests( includes, excludes ):
50 # tests directly in the .expect folder will always be processed
51 test_list = find_tests()
52
53 # if we have a limited number of includes, filter by them
54 if includes:
55 test_list = [x for x in test_list if
56 x.target().startswith( tuple(includes) )
57 ]
58
59 # # if we have a folders to excludes, filter by them
60 if excludes:
61 test_list = [x for x in test_list if not
62 x.target().startswith( tuple(excludes) )
63 ]
64
65 # sort the test alphabetically for convenience
66 test_list.sort(key=lambda t: ('~' if t.arch else '') + t.target() + (t.arch if t.arch else ''))
67
68 return test_list
69
70# from the found tests, filter all the valid tests/desired tests
71def valid_tests( options ):
72 tests = []
73
74 # if we are regenerating the tests we need to find the information of the
75 # already existing tests and create new info for the new tests
76 if options.regenerate_expected :
77 for testname in options.tests :
78 testname = os.path.normpath( os.path.join(settings.SRCDIR, testname) )
79
80 # first check if this is a valid name to regenerate
81 if Test.valid_name(testname):
82 # this is a valid name, let's check if it already exists
83 found = [test for test in all_tests if canonical_path( test.target() ) == testname]
84 setup = itertools.product(settings.all_arch if options.arch else [None], settings.all_ast if options.ast else [None])
85 if not found:
86 # it's a new name, create it according to the name and specified architecture/ast version
87 tests.extend( [Test.new_target(testname, arch, ast) for arch, ast in setup] )
88 elif len(found) == 1 and not found[0].arch:
89 # we found a single test, the user better be wanting to create a cross platform test
90 if options.arch:
91 print('ERROR: "%s", test has no specified architecture but --arch was specified, ignoring it' % testname, file=sys.stderr)
92 elif options.ast:
93 print('ERROR: "%s", test has no specified ast version but --ast was specified, ignoring it' % testname, file=sys.stderr)
94 else:
95 tests.append( found[0] )
96 else:
97 # this test is already cross platform, just add a test for each platform the user asked
98 tests.extend( [Test.new_target(testname, arch, ast) for arch, ast in setup] )
99
100 # print a warning if it users didn't ask for a specific architecture
101 if not options.arch:
102 print('WARNING: "%s", test has architecture specific expected files but --arch was not specified, regenerating only for current host' % testname, file=sys.stderr)
103
104
105 # print a warning if it users didn't ask for a specific ast version
106 if not options.ast:
107 print('WARNING: "%s", test has ast version specific expected files but --ast was not specified, regenerating only for current ast' % testname, file=sys.stderr)
108
109 else :
110 print('ERROR: "%s", tests are not allowed to end with a C/C++/CFA extension, ignoring it' % testname, file=sys.stderr)
111
112 else :
113 # otherwise we only need to validate that all tests are present in the complete list
114 for testname in options.tests:
115 test = [t for t in all_tests if path_cmp( t.target(), testname )]
116
117 if test :
118 tests.extend( test )
119 else :
120 print('ERROR: No expected file for test %s, ignoring it' % testname, file=sys.stderr)
121
122 return tests
123
124# parses the option
125def parse_args():
126 # create a parser with the arguments for the tests script
127 parser = argparse.ArgumentParser(description='Script which runs cforall tests')
128 parser.add_argument('--ast', help='Test for specific ast', type=comma_separated(str), default=None)
129 parser.add_argument('--arch', help='Test for specific architecture', type=comma_separated(str), default=None)
130 parser.add_argument('--debug', help='Run all tests in debug or release', type=comma_separated(yes_no), default='yes')
131 parser.add_argument('--install', help='Run all tests based on installed binaries or tree binaries', type=comma_separated(yes_no), default='no')
132 parser.add_argument('--continue', help='When multiple specifications are passed (debug/install/arch), sets whether or not to continue if the last specification failed', type=yes_no, default='yes', dest='continue_')
133 parser.add_argument('--timeout', help='Maximum duration in seconds after a single test is considered to have timed out', type=int, default=120)
134 parser.add_argument('--global-timeout', help='Maximum cumulative duration in seconds after the ALL tests are considered to have timed out', type=int, default=7200)
135 parser.add_argument('--timeout-with-gdb', help='Instead of killing the command when it times out, orphan it and print process id to allow gdb to attach', type=yes_no, default="no")
136 parser.add_argument('--dry-run', help='Don\'t run the tests, only output the commands', action='store_true')
137 parser.add_argument('--list', help='List all test available', action='store_true')
138 parser.add_argument('--all', help='Run all test available', action='store_true')
139 parser.add_argument('--regenerate-expected', help='Regenerate the .expect by running the specified tets, can be used with --all option', action='store_true')
140 parser.add_argument('--archive-errors', help='If called with a valid path, on test crashes the test script will copy the core dump and the executable to the specified path.', type=str, default='')
141 parser.add_argument('-j', '--jobs', help='Number of tests to run simultaneously', type=int)
142 parser.add_argument('--list-comp', help='List all valide arguments', action='store_true')
143 parser.add_argument('-I','--include', help='Directory of test to include, can be used multiple time, All if omitted', action='append')
144 parser.add_argument('-E','--exclude', help='Directory of test to exclude, can be used multiple time, None if omitted', action='append')
145 parser.add_argument('tests', metavar='test', type=str, nargs='*', help='a list of tests to run')
146
147 try:
148 options = parser.parse_args()
149 except:
150 print('ERROR: invalid arguments', file=sys.stderr)
151 parser.print_help(sys.stderr)
152 sys.exit(1)
153
154 # script must have at least some tests to run or be listing
155 listing = options.list or options.list_comp
156 all_tests = options.all
157 some_tests = len(options.tests) > 0
158 some_dirs = len(options.include) > 0 if options.include else 0
159
160 # check that exactly one of the booleans is set to true
161 if not sum( (listing, all_tests, some_tests, some_dirs) ) > 0 :
162 print('''ERROR: must have option '--all', '--list', '--include', '-I' or non-empty test list''', file=sys.stderr)
163 parser.print_help()
164 sys.exit(1)
165
166 return options
167
168################################################################################
169# running test functions
170################################################################################
171def success(val):
172 return val == 0 or settings.dry_run
173
174def no_rule(file, target):
175 return not settings.dry_run and file_contains_only(file, "make: *** No rule to make target `%s'. Stop." % target)
176
177# logic to run a single test and return the result (No handling of printing or other test framework logic)
178def run_single_test(test):
179
180 # find the output file based on the test name and options flag
181 exe_file = test.target_executable();
182 out_file = test.target_output()
183 err_file = test.error_log()
184 cmp_file = test.expect()
185 in_file = test.input()
186
187 # prepare the proper directories
188 test.prepare()
189
190 # ----------
191 # MAKE
192 # ----------
193 # build, skipping to next test on error
194 with Timed() as comp_dur:
195 make_ret, _ = make( test.target(), output_file=subprocess.DEVNULL, error=out_file, error_file = err_file )
196
197 # ----------
198 # RUN
199 # ----------
200 # run everything in a temp directory to make sure core file are handled properly
201 run_dur = None
202 with tempdir():
203 # if the make command succeeds continue otherwise skip to diff
204 if success(make_ret):
205 with Timed() as run_dur:
206 if settings.dry_run or is_exe(exe_file):
207 # run test
208 retcode, _ = sh(exe_file, output_file=out_file, input_file=in_file, timeout=True)
209 else :
210 # simply cat the result into the output
211 retcode = cat(exe_file, out_file)
212 else:
213 retcode = mv(err_file, out_file)
214
215 if success(retcode):
216 if settings.generating :
217 # if we are only generating the output we still need to check that the test actually exists
218 if no_rule(out_file, test.target()) :
219 retcode = 1
220 error = "\t\tNo make target for test %s!" % test.target()
221 rm(out_file)
222 else:
223 error = None
224 else :
225 # fetch return code and error from the diff command
226 retcode, error = diff(cmp_file, out_file)
227
228 else:
229 if os.stat(out_file).st_size < 1048576:
230 with open (out_file, "r", encoding='latin-1') as myfile: # use latin-1 so all chars mean something.
231 error = myfile.read()
232 else:
233 error = "Output log can't be read, file is bigger than 1MB, see {} for actual error\n".format(out_file)
234
235 ret, info = core_info(exe_file)
236 error = error + info if error else info
237
238 if settings.archive:
239 error = error + '\n' + core_archive(settings.archive, test.target(), exe_file)
240
241
242
243 # clean the executable
244 rm(exe_file)
245
246 return retcode, error, [comp_dur.duration, run_dur.duration if run_dur else None]
247
248# run a single test and handle the errors, outputs, printing, exception handling, etc.
249def run_test_worker(t) :
250 try :
251 # print formated name
252 name_txt = '{0:{width}} '.format(t.target(), width=settings.output_width)
253
254 retcode, error, duration = run_single_test(t)
255
256 # update output based on current action
257 result_txt = TestResult.toString( retcode, duration )
258
259 #print result with error if needed
260 text = '\t' + name_txt + result_txt
261 out = sys.stdout
262 if error :
263 text = text + '\n' + error
264
265 return retcode == TestResult.SUCCESS, text
266 except KeyboardInterrupt:
267 return False, ""
268 # except Exception as ex:
269 # print("Unexpected error in worker thread running {}: {}".format(t.target(), ex), file=sys.stderr)
270 # sys.stderr.flush()
271 # return False, ""
272
273
274# run the given list of tests with the given parameters
275def run_tests(tests, jobs) :
276 # clean the sandbox from previous commands
277 make('clean', output_file=subprocess.DEVNULL, error=subprocess.DEVNULL)
278
279 # create the executor for our jobs
280 pool = multiprocessing.Pool(jobs)
281
282 failed = False
283
284 # for each test to run
285 try :
286 num = len(tests)
287 fancy = sys.stdout.isatty()
288 results = pool.imap_unordered(
289 run_test_worker,
290 tests,
291 chunksize = 1
292 )
293
294 for i, (succ, txt) in enumerate(timed(results, timeout = settings.timeout.total), 1) :
295 if not succ :
296 failed = True
297
298 print(" " + txt)
299
300 if(fancy and i != num):
301 print("%d/%d" % (i, num), end='\r')
302 sys.stdout.flush()
303
304 except KeyboardInterrupt:
305 print("Tests interrupted by user", file=sys.stderr)
306 pool.terminate()
307 pool.join()
308 failed = True
309 except multiprocessing.TimeoutError:
310 print("ERROR: Test suite timed out", file=sys.stderr)
311 pool.terminate()
312 pool.join()
313 failed = True
314 killgroup() # needed to cleanly kill all children
315
316
317 # clean the workspace
318 make('clean', output_file=subprocess.DEVNULL, error=subprocess.DEVNULL)
319
320 return failed
321
322
323################################################################################
324# main loop
325################################################################################
326if __name__ == "__main__":
327
328 # parse the command line arguments
329 options = parse_args()
330
331 # init global settings
332 settings.init( options )
333
334 # users may want to simply list the tests
335 if options.list_comp :
336 # fetch the liest of all valid tests
337 tests = list_tests( None, None )
338
339 # print the possible options
340 print("-h --help --debug --dry-run --list --arch --all --regenerate-expected --archive-errors --install --timeout --global-timeout --timeout-with-gdb -j --jobs -I --include -E --exclude --continue ", end='')
341 print(" ".join(map(lambda t: "%s" % (t.target()), tests)))
342
343 elif options.list :
344 # fetch the liest of all valid tests
345 tests = list_tests( options.include, options.exclude )
346
347 # print the available tests
348 fancy_print("\n".join(map(lambda t: t.toString(), tests)))
349
350 else :
351 # fetch the liest of all valid tests
352 all_tests = list_tests( options.include, options.exclude )
353
354 # if user wants all tests than no other treatement of the test list is required
355 if options.all or options.include :
356 tests = all_tests
357
358 #otherwise we need to validate that the test list that was entered is valid
359 else :
360 tests = valid_tests( options )
361
362 # make sure we have at least some test to run
363 if not tests :
364 print('ERROR: No valid test to run', file=sys.stderr)
365 sys.exit(1)
366
367 # prep invariants
368 settings.prep_output(tests)
369 failed = 0
370
371 # check if the expected files aren't empty
372 if not options.regenerate_expected:
373 for t in tests:
374 if is_empty(t.expect()):
375 print('WARNING: test "{}" has empty .expect file'.format(t.target()), file=sys.stderr)
376
377 # for each build configurations, run the test
378 with Timed() as total_dur:
379 for ast, arch, debug, install in itertools.product(settings.all_ast, settings.all_arch, settings.all_debug, settings.all_install):
380 settings.ast = ast
381 settings.arch = arch
382 settings.debug = debug
383 settings.install = install
384
385 # filter out the tests for a different architecture
386 # tests are the same across debug/install
387 local_tests = settings.ast.filter( tests )
388 local_tests = settings.arch.filter( local_tests )
389 options.jobs, forceJobs = job_count( options, local_tests )
390 settings.update_make_cmd(forceJobs, options.jobs)
391
392 # check the build configuration works
393 settings.validate()
394
395 # print configuration
396 print('%s %i tests on %i cores (%s:%s - %s)' % (
397 'Regenerating' if settings.generating else 'Running',
398 len(local_tests),
399 options.jobs,
400 settings.ast.string,
401 settings.arch.string,
402 settings.debug.string
403 ))
404 if not local_tests :
405 print('WARNING: No tests for this configuration')
406 continue
407
408 # otherwise run all tests and make sure to return the correct error code
409 failed = run_tests(local_tests, options.jobs)
410 if failed:
411 result = 1
412 if not settings.continue_:
413 break
414
415 print('Tests took %s' % fmtDur( total_dur.duration ))
416 sys.exit( failed )
Note: See TracBrowser for help on using the repository browser.