sdk/cwl/tests/test_pathmapper.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 import functools
   6 import mock
   7 import sys
   8 import unittest
   9 import json
  10 import logging
  11 import os
  12
  13 import arvados
  14 import arvados.keep
  15 import arvados.collection
  16 import arvados_cwl
  17 import arvados_cwl.executor
  18
  19 from cwltool.pathmapper import MapperEnt
  20 from .mock_discovery import get_rootDesc
  21
  22 from arvados_cwl.pathmapper import ArvPathMapper
  23
  24 def upload_mock(files, api, dry_run=False, num_retries=0, project=None, fnPattern="$(file %s/%s)", name=None, collection=None, packed=None):
  25     pdh = "99999999999999999999999999999991+99"
  26     for c in files:
  27         c.keepref = "%s/%s" % (pdh, os.path.basename(c.fn))
  28         c.fn = fnPattern % (pdh, os.path.basename(c.fn))
  29
  30 class TestPathmap(unittest.TestCase):
  31     def setUp(self):
  32         self.api = mock.MagicMock()
  33         self.api._rootDesc = get_rootDesc()
  34
  35     def test_keepref(self):
  36         """Test direct keep references."""
  37
  38         arvrunner = arvados_cwl.executor.ArvCwlExecutor(self.api)
  39
  40         p = ArvPathMapper(arvrunner, [{
  41             "class": "File",
  42             "location": "keep:99999999999999999999999999999991+99/hw.py"
  43         }], "", "/test/%s", "/test/%s/%s")
  44
  45         self.assertEqual({'keep:99999999999999999999999999999991+99/hw.py': MapperEnt(resolved='keep:99999999999999999999999999999991+99/hw.py', target='/test/99999999999999999999999999999991+99/hw.py', type='File', staged=True)},
  46                          p._pathmap)
  47
  48     @mock.patch("arvados.commands.run.uploadfiles")
  49     @mock.patch("arvados.commands.run.statfile")
  50     def test_upload(self, statfile, upl):
  51         """Test pathmapper uploading files."""
  52
  53         arvrunner = arvados_cwl.executor.ArvCwlExecutor(self.api)
  54
  55         def statfile_mock(prefix, fn, fnPattern="$(file %s/%s)", dirPattern="$(dir %s/%s/)", raiseOSError=False):
  56             st = arvados.commands.run.UploadFile("", "tests/hw.py")
  57             return st
  58
  59         upl.side_effect = upload_mock
  60         statfile.side_effect = statfile_mock
  61
  62         p = ArvPathMapper(arvrunner, [{
  63             "class": "File",
  64             "location": "file:tests/hw.py"
  65         }], "", "/test/%s", "/test/%s/%s")
  66
  67         self.assertEqual({'file:tests/hw.py': MapperEnt(resolved='keep:99999999999999999999999999999991+99/hw.py', target='/test/99999999999999999999999999999991+99/hw.py', type='File', staged=True)},
  68                          p._pathmap)
  69
  70     @mock.patch("arvados.commands.run.uploadfiles")
  71     @mock.patch("arvados.commands.run.statfile")
  72     def test_statfile(self, statfile, upl):
  73         """Test pathmapper handling ArvFile references."""
  74         arvrunner = arvados_cwl.executor.ArvCwlExecutor(self.api)
  75
  76         # An ArvFile object returned from arvados.commands.run.statfile means the file is located on a
  77         # keep mount, so we can construct a direct reference directly without upload.
  78         def statfile_mock(prefix, fn, fnPattern="$(file %s/%s)", dirPattern="$(dir %s/%s/)", raiseOSError=False):
  79             st = arvados.commands.run.ArvFile("", fnPattern % ("99999999999999999999999999999991+99", "hw.py"))
  80             return st
  81
  82         upl.side_effect = upload_mock
  83         statfile.side_effect = statfile_mock
  84
  85         p = ArvPathMapper(arvrunner, [{
  86             "class": "File",
  87             "location": "file:tests/hw.py"
  88         }], "", "/test/%s", "/test/%s/%s")
  89
  90         self.assertEqual({'file:tests/hw.py': MapperEnt(resolved='keep:99999999999999999999999999999991+99/hw.py', target='/test/99999999999999999999999999999991+99/hw.py', type='File', staged=True)},
  91                          p._pathmap)
  92
  93     @mock.patch("os.stat")
  94     def test_missing_file(self, stat):
  95         """Test pathmapper handling missing references."""
  96         arvrunner = arvados_cwl.executor.ArvCwlExecutor(self.api)
  97
  98         stat.side_effect = OSError(2, "No such file or directory")
  99
 100         with self.assertRaises(OSError):
 101             p = ArvPathMapper(arvrunner, [{
 102                 "class": "File",
 103                 "location": "file:tests/hw.py"
 104             }], "", "/test/%s", "/test/%s/%s")
 105
 106     def test_needs_new_collection(self):
 107         arvrunner = arvados_cwl.executor.ArvCwlExecutor(self.api)
 108         a = {
 109             "class": "File",
 110             "location": "keep:99999999999999999999999999999991+99/hw.py",
 111             "basename": "hw.py"
 112         }
 113         p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s")
 114         p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True
 115         self.assertFalse(p.needs_new_collection(a))
 116
 117         p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s")
 118         self.assertTrue(p.needs_new_collection(a))
 119
 120         a = {
 121             "class": "File",
 122             "location": "keep:99999999999999999999999999999991+99/hw.py",
 123             "basename": "hw.py",
 124             "secondaryFiles": [{
 125                 "class": "File",
 126                 "location": "keep:99999999999999999999999999999991+99/hw.pyc",
 127                 "basename": "hw.pyc"
 128             }]
 129         }
 130         p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s")
 131         p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True
 132         p._pathmap["keep:99999999999999999999999999999991+99/hw.pyc"] = True
 133         self.assertFalse(p.needs_new_collection(a))
 134
 135         a = {
 136             "class": "File",
 137             "location": "keep:99999999999999999999999999999991+99/hw.py",
 138             "basename": "hw.py",
 139             "secondaryFiles": [{
 140                 "class": "File",
 141                 "location": "keep:99999999999999999999999999999992+99/hw.pyc",
 142                 "basename": "hw.pyc"
 143             }]
 144         }
 145         p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s")
 146         p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True
 147         p._pathmap["keep:99999999999999999999999999999992+99/hw.pyc"] = True
 148         self.assertTrue(p.needs_new_collection(a))
 149
 150         a = {
 151             "class": "File",
 152             "location": "keep:99999999999999999999999999999991+99/hw.py",
 153             "basename": "hw.py",
 154             "secondaryFiles": [{
 155                 "class": "File",
 156                 "location": "keep:99999999999999999999999999999991+99/hw.pyc",
 157                 "basename": "hw.other"
 158             }]
 159         }
 160         p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s")
 161         p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True
 162         p._pathmap["keep:99999999999999999999999999999991+99/hw.pyc"] = True
 163         self.assertTrue(p.needs_new_collection(a))
 164
 165         a = {
 166             "class": "File",
 167             "location": "keep:99999999999999999999999999999991+99/hw.py",
 168             "basename": "hw.py",
 169             "secondaryFiles": [{
 170                 "class": "Directory",
 171                 "location": "keep:99999999999999999999999999999991+99/hw",
 172                 "basename": "hw",
 173                 "listing": [{
 174                     "class": "File",
 175                     "location": "keep:99999999999999999999999999999991+99/hw/h2",
 176                     "basename": "h2"
 177                 }]
 178             }]
 179         }
 180         p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s")
 181         p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True
 182         p._pathmap["keep:99999999999999999999999999999991+99/hw"] = True
 183         p._pathmap["keep:99999999999999999999999999999991+99/hw/h2"] = True
 184         self.assertFalse(p.needs_new_collection(a))
 185
 186         a = {
 187             "class": "File",
 188             "location": "keep:99999999999999999999999999999991+99/hw.py",
 189             "basename": "hw.py",
 190             "secondaryFiles": [{
 191                 "class": "Directory",
 192                 "location": "keep:99999999999999999999999999999991+99/hw",
 193                 "basename": "wh",
 194                 "listing": [{
 195                     "class": "File",
 196                     "location": "keep:99999999999999999999999999999991+99/hw/h2",
 197                     "basename": "h2"
 198                 }]
 199             }]
 200         }
 201         p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s")
 202         p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True
 203         p._pathmap["keep:99999999999999999999999999999991+99/hw"] = True
 204         p._pathmap["keep:99999999999999999999999999999991+99/hw/h2"] = True
 205         self.assertTrue(p.needs_new_collection(a))
 206
 207         a = {
 208             "class": "File",
 209             "location": "keep:99999999999999999999999999999991+99/hw.py",
 210             "basename": "hw.py",
 211             "secondaryFiles": [{
 212                 "class": "File",
 213                 "location": "_:123",
 214                 "basename": "hw.pyc",
 215                 "contents": "123"
 216             }]
 217         }
 218         p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s")
 219         p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True
 220         p._pathmap["_:123"] = True
 221         self.assertTrue(p.needs_new_collection(a))