lib/deduplicationreport/report_test.go

   1 // Copyright (C) The Arvados Authors. All rights reserved.
   2 //
   3 // SPDX-License-Identifier: AGPL-3.0
   4
   5 package deduplicationreport
   6
   7 import (
   8         "bytes"
   9         "testing"
  10
  11         "git.arvados.org/arvados.git/sdk/go/arvados"
  12         "git.arvados.org/arvados.git/sdk/go/arvadostest"
  13         "gopkg.in/check.v1"
  14 )
  15
  16 func Test(t *testing.T) {
  17         check.TestingT(t)
  18 }
  19
  20 var _ = check.Suite(&Suite{})
  21
  22 type Suite struct{}
  23
  24 func (s *Suite) TearDownSuite(c *check.C) {
  25         // Undo any changes/additions to the database so they don't affect subsequent tests.
  26         arvadostest.ResetEnv()
  27 }
  28
  29 func (*Suite) TestUsage(c *check.C) {
  30         var stdout, stderr bytes.Buffer
  31         exitcode := Command.RunCommand("deduplicationreport.test", []string{"-h", "-log-level=debug"}, &bytes.Buffer{}, &stdout, &stderr)
  32         c.Check(exitcode, check.Equals, 0)
  33         c.Check(stdout.String(), check.Equals, "")
  34         c.Log(stderr.String())
  35         c.Check(stderr.String(), check.Matches, `(?ms).*Usage:.*`)
  36 }
  37
  38 func (*Suite) TestTwoIdenticalUUIDs(c *check.C) {
  39         var stdout, stderr bytes.Buffer
  40         // Run dedupreport with 2 identical uuids
  41         exitcode := Command.RunCommand("deduplicationreport.test", []string{arvadostest.FooCollection, arvadostest.FooCollection}, &bytes.Buffer{}, &stdout, &stderr)
  42         c.Check(exitcode, check.Equals, 0)
  43         c.Check(stdout.String(), check.Matches, "(?ms).*Collections:[[:space:]]+1.*")
  44         c.Check(stdout.String(), check.Matches, "(?ms).*Nominal size of stored data:[[:space:]]+3 bytes \\(3 B\\).*")
  45         c.Check(stdout.String(), check.Matches, "(?ms).*Actual size of stored data:[[:space:]]+3 bytes \\(3 B\\).*")
  46         c.Check(stdout.String(), check.Matches, "(?ms).*Saved by Keep deduplication:[[:space:]]+0 bytes \\(0 B\\).*")
  47         c.Log(stderr.String())
  48 }
  49
  50 func (*Suite) TestTwoUUIDsInvalidPDH(c *check.C) {
  51         var stdout, stderr bytes.Buffer
  52         // Run dedupreport with pdh,uuid where pdh does not match
  53         exitcode := Command.RunCommand("deduplicationreport.test", []string{arvadostest.FooAndBarFilesInDirPDH + "," + arvadostest.FooCollection, arvadostest.FooCollection}, &bytes.Buffer{}, &stdout, &stderr)
  54         c.Check(exitcode, check.Equals, 1)
  55         c.Check(stdout.String(), check.Equals, "")
  56         c.Log(stderr.String())
  57         c.Check(stderr.String(), check.Matches, `(?ms).*Error: the collection with UUID zzzzz-4zz18-fy296fx3hot09f7 has PDH 1f4b0bc7583c2a7f9102c395f4ffc5e3\+45, but a different PDH was provided in the arguments: 870369fc72738603c2fad16664e50e2d\+58.*`)
  58 }
  59
  60 func (*Suite) TestNonExistentCollection(c *check.C) {
  61         var stdout, stderr bytes.Buffer
  62         // Run dedupreport with many UUIDs
  63         exitcode := Command.RunCommand("deduplicationreport.test", []string{arvadostest.FooCollection, arvadostest.NonexistentCollection}, &bytes.Buffer{}, &stdout, &stderr)
  64         c.Check(exitcode, check.Equals, 1)
  65         c.Check(stdout.String(), check.Equals, "Collection zzzzz-4zz18-fy296fx3hot09f7: pdh 1f4b0bc7583c2a7f9102c395f4ffc5e3+45; nominal size 3 (3 B)\n")
  66         c.Log(stderr.String())
  67         c.Check(stderr.String(), check.Matches, `(?ms).*Error: unable to retrieve collection:.*404 Not Found.*`)
  68 }
  69
  70 func (*Suite) TestManyUUIDsNoOverlap(c *check.C) {
  71         var stdout, stderr bytes.Buffer
  72         // Run dedupreport with 5 UUIDs
  73         exitcode := Command.RunCommand("deduplicationreport.test", []string{arvadostest.FooCollection, arvadostest.HelloWorldCollection, arvadostest.FooBarDirCollection, arvadostest.WazVersion1Collection, arvadostest.UserAgreementCollection}, &bytes.Buffer{}, &stdout, &stderr)
  74         c.Check(exitcode, check.Equals, 0)
  75         c.Check(stdout.String(), check.Matches, "(?ms).*Collections:[[:space:]]+5.*")
  76         c.Check(stdout.String(), check.Matches, "(?ms).*Nominal size of stored data:[[:space:]]+249049 bytes \\(243 KiB\\).*")
  77         c.Check(stdout.String(), check.Matches, "(?ms).*Actual size of stored data:[[:space:]]+249049 bytes \\(243 KiB\\).*")
  78         c.Check(stdout.String(), check.Matches, "(?ms).*Saved by Keep deduplication:[[:space:]]+0 bytes \\(0 B\\).*")
  79         c.Log(stderr.String())
  80         c.Check(stderr.String(), check.Equals, "")
  81 }
  82
  83 func (*Suite) TestTwoOverlappingCollections(c *check.C) {
  84         var stdout, stderr bytes.Buffer
  85         // Create two collections
  86         arv := arvados.NewClientFromEnv()
  87
  88         var c1 arvados.Collection
  89         err := arv.RequestAndDecode(&c1, "POST", "arvados/v1/collections", nil, map[string]interface{}{"collection": map[string]interface{}{"manifest_text": ". d3b07384d113edec49eaa6238ad5ff00+4 0:4:foo\n"}})
  90         c.Assert(err, check.Equals, nil)
  91
  92         var c2 arvados.Collection
  93         err = arv.RequestAndDecode(&c2, "POST", "arvados/v1/collections", nil, map[string]interface{}{"collection": map[string]interface{}{"manifest_text": ". c157a79031e1c40f85931829bc5fc552+4 d3b07384d113edec49eaa6238ad5ff00+4 0:4:bar 4:4:foo\n"}})
  94         c.Assert(err, check.Equals, nil)
  95
  96         for _, trial := range []struct {
  97                 field1 string
  98                 field2 string
  99         }{
 100                 {
 101                         // Run dedupreport with 2 arguments: uuid uuid
 102                         field1: c1.UUID,
 103                         field2: c2.UUID,
 104                 },
 105                 {
 106                         // Run dedupreport with 2 arguments: pdh,uuid uuid
 107                         field1: c1.PortableDataHash + "," + c1.UUID,
 108                         field2: c2.UUID,
 109                 },
 110         } {
 111                 exitcode := Command.RunCommand("deduplicationreport.test", []string{trial.field1, trial.field2}, &bytes.Buffer{}, &stdout, &stderr)
 112                 c.Check(exitcode, check.Equals, 0)
 113                 c.Check(stdout.String(), check.Matches, "(?ms).*Nominal size of stored data:[[:space:]]+12 bytes \\(12 B\\).*")
 114                 c.Check(stdout.String(), check.Matches, "(?ms).*Actual size of stored data:[[:space:]]+8 bytes \\(8 B\\).*")
 115                 c.Check(stdout.String(), check.Matches, "(?ms).*Saved by Keep deduplication:[[:space:]]+4 bytes \\(4 B\\).*")
 116                 c.Log(stderr.String())
 117                 c.Check(stderr.String(), check.Equals, "")
 118         }
 119 }