tools/crunchstat-summary/tests/test_examples.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: AGPL-3.0
   4
   5 import arvados
   6 import collections
   7 import crunchstat_summary.command
   8 import difflib
   9 import glob
  10 import gzip
  11 from io import open
  12 import mock
  13 import os
  14 import unittest
  15
  16 from crunchstat_summary.command import UTF8Decode
  17
  18 TESTS_DIR = os.path.dirname(os.path.abspath(__file__))
  19
  20
  21 class ReportDiff(unittest.TestCase):
  22     def diff_known_report(self, logfile, cmd):
  23         expectfile = logfile+'.report'
  24         expect = open(expectfile, encoding='utf-8').readlines()
  25         self.diff_report(cmd, expect, expectfile=expectfile)
  26
  27     def diff_report(self, cmd, expect, expectfile='(expected)'):
  28         got = [x+"\n" for x in cmd.report().strip("\n").split("\n")]
  29         self.assertEqual(got, expect, "\n"+"".join(difflib.context_diff(
  30             expect, got, fromfile=expectfile, tofile="(generated)")))
  31
  32
  33 class SummarizeFile(ReportDiff):
  34     def test_example_files(self):
  35         for fnm in glob.glob(os.path.join(TESTS_DIR, '*.txt.gz')):
  36             logfile = os.path.join(TESTS_DIR, fnm)
  37             args = crunchstat_summary.command.ArgumentParser().parse_args(
  38                 ['--log-file', logfile])
  39             cmd = crunchstat_summary.command.Command(args)
  40             cmd.run()
  41             self.diff_known_report(logfile, cmd)
  42
  43
  44 class HTMLFromFile(ReportDiff):
  45     def test_example_files(self):
  46         # Note we don't test the output content at all yet; we're
  47         # mainly just verifying the --format=html option isn't ignored
  48         # and the HTML code path doesn't crash.
  49         for fnm in glob.glob(os.path.join(TESTS_DIR, '*.txt.gz')):
  50             logfile = os.path.join(TESTS_DIR, fnm)
  51             args = crunchstat_summary.command.ArgumentParser().parse_args(
  52                 ['--format=html', '--log-file', logfile])
  53             cmd = crunchstat_summary.command.Command(args)
  54             cmd.run()
  55             self.assertRegexpMatches(cmd.report(), r'(?is)<html>.*</html>\s*$')
  56
  57 class SummarizeEdgeCases(unittest.TestCase):
  58     def test_error_messages(self):
  59         logfile = open(os.path.join(TESTS_DIR, 'crunchstat_error_messages.txt'), encoding='utf-8')
  60         s = crunchstat_summary.summarizer.Summarizer(logfile)
  61         s.run()
  62
  63
  64 class SummarizeContainer(ReportDiff):
  65     fake_container = {
  66         'uuid': '9tee4-dz642-lymtndkpy39eibk',
  67         'created_at': '2017-08-18T14:27:25.371388141',
  68         'log': '9tee4-4zz18-ihyzym9tcwjwg4r',
  69     }
  70     fake_request = {
  71         'uuid': '9tee4-xvhdp-uper95jktm10d3w',
  72         'name': 'container',
  73         'created_at': '2017-08-18T14:27:25.242339223Z',
  74         'container_uuid': fake_container['uuid'],
  75     }
  76     reportfile = os.path.join(
  77         TESTS_DIR, 'container_9tee4-dz642-lymtndkpy39eibk.txt.gz')
  78     logfile = os.path.join(
  79         TESTS_DIR, 'container_9tee4-dz642-lymtndkpy39eibk-crunchstat.txt.gz')
  80     arvmountlog = os.path.join(
  81         TESTS_DIR, 'container_9tee4-dz642-lymtndkpy39eibk-arv-mount.txt.gz')
  82
  83     @mock.patch('arvados.collection.CollectionReader')
  84     @mock.patch('arvados.api')
  85     def test_container(self, mock_api, mock_cr):
  86         mock_api().container_requests().index().execute.return_value = {'items':[]}
  87         mock_api().container_requests().get().execute.return_value = self.fake_request
  88         mock_api().containers().get().execute.return_value = self.fake_container
  89         mock_cr().__iter__.return_value = [
  90             'crunch-run.txt', 'stderr.txt', 'node-info.txt',
  91             'container.json', 'crunchstat.txt', 'arv-mount.txt']
  92         def _open(n):
  93             if n == "crunchstat.txt":
  94                 return UTF8Decode(gzip.open(self.logfile))
  95             elif n == "arv-mount.txt":
  96                 return UTF8Decode(gzip.open(self.arvmountlog))
  97         mock_cr().open.side_effect = _open
  98         args = crunchstat_summary.command.ArgumentParser().parse_args(
  99             ['--job', self.fake_request['uuid']])
 100         cmd = crunchstat_summary.command.Command(args)
 101         cmd.run()
 102         self.diff_known_report(self.reportfile, cmd)
 103
 104
 105 class SummarizeJob(ReportDiff):
 106     fake_job_uuid = '4xphq-8i9sb-jq0ekny1xou3zoh'
 107     fake_log_id = 'fake-log-collection-id'
 108     fake_job = {
 109         'uuid': fake_job_uuid,
 110         'log': fake_log_id,
 111     }
 112     logfile = os.path.join(TESTS_DIR, 'logfile_20151204190335.txt.gz')
 113
 114     @mock.patch('arvados.collection.CollectionReader')
 115     @mock.patch('arvados.api')
 116     def test_job_report(self, mock_api, mock_cr):
 117         mock_api().jobs().get().execute.return_value = self.fake_job
 118         mock_cr().__iter__.return_value = ['fake-logfile.txt']
 119         mock_cr().open.return_value = UTF8Decode(gzip.open(self.logfile))
 120         args = crunchstat_summary.command.ArgumentParser().parse_args(
 121             ['--job', self.fake_job_uuid])
 122         cmd = crunchstat_summary.command.Command(args)
 123         cmd.run()
 124         self.diff_known_report(self.logfile, cmd)
 125         mock_api().jobs().get.assert_called_with(uuid=self.fake_job_uuid)
 126         mock_cr.assert_called_with(self.fake_log_id)
 127         mock_cr().open.assert_called_with('fake-logfile.txt')
 128
 129
 130 class SummarizePipeline(ReportDiff):
 131     fake_instance = {
 132         'uuid': 'zzzzz-d1hrv-i3e77t9z5y8j9cc',
 133         'owner_uuid': 'zzzzz-tpzed-xurymjxw79nv3jz',
 134         'components': collections.OrderedDict([
 135             ['foo', {
 136                 'job': {
 137                     'uuid': 'zzzzz-8i9sb-000000000000000',
 138                     'log': 'fake-log-pdh-0',
 139                     'runtime_constraints': {
 140                         'min_ram_mb_per_node': 900,
 141                         'min_cores_per_node': 1,
 142                     },
 143                 },
 144             }],
 145             ['bar', {
 146                 'job': {
 147                     'uuid': 'zzzzz-8i9sb-000000000000001',
 148                     'log': 'fake-log-pdh-1',
 149                     'runtime_constraints': {
 150                         'min_ram_mb_per_node': 900,
 151                         'min_cores_per_node': 1,
 152                     },
 153                 },
 154             }],
 155             ['no-job-assigned', {}],
 156             ['unfinished-job', {
 157                 'job': {
 158                     'uuid': 'zzzzz-8i9sb-xxxxxxxxxxxxxxx',
 159                 },
 160             }],
 161             ['baz', {
 162                 'job': {
 163                     'uuid': 'zzzzz-8i9sb-000000000000002',
 164                     'log': 'fake-log-pdh-2',
 165                     'runtime_constraints': {
 166                         'min_ram_mb_per_node': 900,
 167                         'min_cores_per_node': 1,
 168                     },
 169                 },
 170             }]]),
 171     }
 172
 173     @mock.patch('arvados.collection.CollectionReader')
 174     @mock.patch('arvados.api')
 175     def test_pipeline(self, mock_api, mock_cr):
 176         logfile = os.path.join(TESTS_DIR, 'logfile_20151204190335.txt.gz')
 177         mock_api().pipeline_instances().get().execute. \
 178             return_value = self.fake_instance
 179         mock_cr().__iter__.return_value = ['fake-logfile.txt']
 180         mock_cr().open.side_effect = [UTF8Decode(gzip.open(logfile)) for _ in range(3)]
 181         args = crunchstat_summary.command.ArgumentParser().parse_args(
 182             ['--pipeline-instance', self.fake_instance['uuid']])
 183         cmd = crunchstat_summary.command.Command(args)
 184         cmd.run()
 185
 186         job_report = [
 187             line for line in open(logfile+'.report', encoding='utf-8').readlines()
 188             if not line.startswith('#!! ')]
 189         expect = (
 190             ['### Summary for foo (zzzzz-8i9sb-000000000000000)\n'] +
 191             job_report + ['\n'] +
 192             ['### Summary for bar (zzzzz-8i9sb-000000000000001)\n'] +
 193             job_report + ['\n'] +
 194             ['### Summary for unfinished-job (partial) (zzzzz-8i9sb-xxxxxxxxxxxxxxx)\n',
 195              '(no report generated)\n',
 196              '\n'] +
 197             ['### Summary for baz (zzzzz-8i9sb-000000000000002)\n'] +
 198             job_report)
 199         self.diff_report(cmd, expect)
 200         mock_cr.assert_has_calls(
 201             [
 202                 mock.call('fake-log-pdh-0'),
 203                 mock.call('fake-log-pdh-1'),
 204                 mock.call('fake-log-pdh-2'),
 205             ], any_order=True)
 206         mock_cr().open.assert_called_with('fake-logfile.txt')
 207
 208
 209 class SummarizeACRJob(ReportDiff):
 210     fake_job = {
 211         'uuid': 'zzzzz-8i9sb-i3e77t9z5y8j9cc',
 212         'owner_uuid': 'zzzzz-tpzed-xurymjxw79nv3jz',
 213         'components': {
 214             'foo': 'zzzzz-8i9sb-000000000000000',
 215             'bar': 'zzzzz-8i9sb-000000000000001',
 216             'unfinished-job': 'zzzzz-8i9sb-xxxxxxxxxxxxxxx',
 217             'baz': 'zzzzz-8i9sb-000000000000002',
 218         }
 219     }
 220     fake_jobs_index = { 'items': [
 221         {
 222             'uuid': 'zzzzz-8i9sb-000000000000000',
 223             'log': 'fake-log-pdh-0',
 224             'runtime_constraints': {
 225                 'min_ram_mb_per_node': 900,
 226                 'min_cores_per_node': 1,
 227             },
 228         },
 229         {
 230             'uuid': 'zzzzz-8i9sb-000000000000001',
 231             'log': 'fake-log-pdh-1',
 232             'runtime_constraints': {
 233                 'min_ram_mb_per_node': 900,
 234                 'min_cores_per_node': 1,
 235             },
 236         },
 237         {
 238             'uuid': 'zzzzz-8i9sb-xxxxxxxxxxxxxxx',
 239         },
 240         {
 241             'uuid': 'zzzzz-8i9sb-000000000000002',
 242             'log': 'fake-log-pdh-2',
 243             'runtime_constraints': {
 244                 'min_ram_mb_per_node': 900,
 245                 'min_cores_per_node': 1,
 246             },
 247         },
 248     ]}
 249     @mock.patch('arvados.collection.CollectionReader')
 250     @mock.patch('arvados.api')
 251     def test_acr_job(self, mock_api, mock_cr):
 252         logfile = os.path.join(TESTS_DIR, 'logfile_20151204190335.txt.gz')
 253         mock_api().jobs().index().execute.return_value = self.fake_jobs_index
 254         mock_api().jobs().get().execute.return_value = self.fake_job
 255         mock_cr().__iter__.return_value = ['fake-logfile.txt']
 256         mock_cr().open.side_effect = [UTF8Decode(gzip.open(logfile)) for _ in range(3)]
 257         args = crunchstat_summary.command.ArgumentParser().parse_args(
 258             ['--job', self.fake_job['uuid']])
 259         cmd = crunchstat_summary.command.Command(args)
 260         cmd.run()
 261
 262         job_report = [
 263             line for line in open(logfile+'.report', encoding='utf-8').readlines()
 264             if not line.startswith('#!! ')]
 265         expect = (
 266             ['### Summary for zzzzz-8i9sb-i3e77t9z5y8j9cc (partial) (zzzzz-8i9sb-i3e77t9z5y8j9cc)\n',
 267              '(no report generated)\n',
 268              '\n'] +
 269             ['### Summary for bar (zzzzz-8i9sb-000000000000001)\n'] +
 270             job_report + ['\n'] +
 271             ['### Summary for baz (zzzzz-8i9sb-000000000000002)\n'] +
 272             job_report + ['\n'] +
 273             ['### Summary for foo (zzzzz-8i9sb-000000000000000)\n'] +
 274             job_report + ['\n'] +
 275             ['### Summary for unfinished-job (partial) (zzzzz-8i9sb-xxxxxxxxxxxxxxx)\n',
 276              '(no report generated)\n']
 277         )
 278         self.diff_report(cmd, expect)
 279         mock_cr.assert_has_calls(
 280             [
 281                 mock.call('fake-log-pdh-0'),
 282                 mock.call('fake-log-pdh-1'),
 283                 mock.call('fake-log-pdh-2'),
 284             ], any_order=True)
 285         mock_cr().open.assert_called_with('fake-logfile.txt')