Merge branch '8784-dir-listings'
[arvados.git] / sdk / go / keepclient / collectionreader_test.go
1 // Copyright (C) The Arvados Authors. All rights reserved.
2 //
3 // SPDX-License-Identifier: Apache-2.0
4
5 package keepclient
6
7 import (
8         "crypto/md5"
9         "fmt"
10         "io"
11         "io/ioutil"
12         "math/rand"
13         "net/http"
14         "os"
15         "strconv"
16         "strings"
17
18         "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
19         "git.curoverse.com/arvados.git/sdk/go/arvadostest"
20         check "gopkg.in/check.v1"
21 )
22
23 var _ = check.Suite(&CollectionReaderUnit{})
24
25 type CollectionReaderUnit struct {
26         arv     *arvadosclient.ArvadosClient
27         kc      *KeepClient
28         handler SuccessHandler
29 }
30
31 func (s *CollectionReaderUnit) SetUpTest(c *check.C) {
32         var err error
33         s.arv, err = arvadosclient.MakeArvadosClient()
34         c.Assert(err, check.IsNil)
35         s.arv.ApiToken = arvadostest.ActiveToken
36
37         s.kc, err = MakeKeepClient(s.arv)
38         c.Assert(err, check.IsNil)
39
40         s.handler = SuccessHandler{
41                 disk: make(map[string][]byte),
42                 lock: make(chan struct{}, 1),
43                 ops:  new(int),
44         }
45         localRoots := make(map[string]string)
46         for i, k := range RunSomeFakeKeepServers(s.handler, 4) {
47                 localRoots[fmt.Sprintf("zzzzz-bi6l4-fakefakefake%03d", i)] = k.url
48         }
49         s.kc.SetServiceRoots(localRoots, localRoots, nil)
50 }
51
52 type SuccessHandler struct {
53         disk map[string][]byte
54         lock chan struct{} // channel with buffer==1: full when an operation is in progress.
55         ops  *int          // number of operations completed
56 }
57
58 func (h SuccessHandler) ServeHTTP(resp http.ResponseWriter, req *http.Request) {
59         switch req.Method {
60         case "PUT":
61                 buf, err := ioutil.ReadAll(req.Body)
62                 if err != nil {
63                         resp.WriteHeader(500)
64                         return
65                 }
66                 pdh := fmt.Sprintf("%x+%d", md5.Sum(buf), len(buf))
67                 h.lock <- struct{}{}
68                 h.disk[pdh] = buf
69                 if h.ops != nil {
70                         (*h.ops)++
71                 }
72                 <-h.lock
73                 resp.Write([]byte(pdh))
74         case "GET":
75                 pdh := req.URL.Path[1:]
76                 h.lock <- struct{}{}
77                 buf, ok := h.disk[pdh]
78                 if h.ops != nil {
79                         (*h.ops)++
80                 }
81                 <-h.lock
82                 if !ok {
83                         resp.WriteHeader(http.StatusNotFound)
84                 } else {
85                         resp.Write(buf)
86                 }
87         default:
88                 resp.WriteHeader(http.StatusMethodNotAllowed)
89         }
90 }
91
92 type rdrTest struct {
93         mt   string      // manifest text
94         f    string      // filename
95         want interface{} // error or string to expect
96 }
97
98 func (s *CollectionReaderUnit) TestCollectionReaderContent(c *check.C) {
99         s.kc.PutB([]byte("foo"))
100         s.kc.PutB([]byte("bar"))
101         s.kc.PutB([]byte("Hello world\n"))
102         s.kc.PutB([]byte(""))
103
104         mt := arvadostest.PathologicalManifest
105
106         for _, testCase := range []rdrTest{
107                 {mt: mt, f: "zzzz", want: os.ErrNotExist},
108                 {mt: mt, f: "frob", want: os.ErrNotExist},
109                 {mt: mt, f: "/segmented/frob", want: "frob"},
110                 {mt: mt, f: "./segmented/frob", want: "frob"},
111                 {mt: mt, f: "/f", want: "f"},
112                 {mt: mt, f: "./f", want: "f"},
113                 {mt: mt, f: "foo bar//baz", want: "foo"},
114                 {mt: mt, f: "foo/zero", want: ""},
115                 {mt: mt, f: "zero@0", want: ""},
116                 {mt: mt, f: "zero@1", want: ""},
117                 {mt: mt, f: "zero@4", want: ""},
118                 {mt: mt, f: "zero@9", want: ""},
119                 {mt: mt, f: "f", want: "f"},
120                 {mt: mt, f: "ooba", want: "ooba"},
121                 {mt: mt, f: "overlapReverse/o", want: "o"},
122                 {mt: mt, f: "overlapReverse/oo", want: "oo"},
123                 {mt: mt, f: "overlapReverse/ofoo", want: "ofoo"},
124                 {mt: mt, f: "foo bar/baz", want: "foo"},
125                 {mt: mt, f: "segmented/frob", want: "frob"},
126                 {mt: mt, f: "segmented/oof", want: "oof"},
127         } {
128                 c.Logf("%#v", testCase)
129                 rdr, err := s.kc.CollectionFileReader(map[string]interface{}{"manifest_text": testCase.mt}, testCase.f)
130                 switch want := testCase.want.(type) {
131                 case error:
132                         c.Check(rdr, check.IsNil)
133                         c.Check(err, check.Equals, want)
134                 case string:
135                         buf := make([]byte, len(want))
136                         n, err := io.ReadFull(rdr, buf)
137                         c.Check(err, check.IsNil)
138                         for i := 0; i < 4; i++ {
139                                 c.Check(string(buf), check.Equals, want)
140                                 n, err = rdr.Read(buf)
141                                 c.Check(n, check.Equals, 0)
142                                 c.Check(err, check.Equals, io.EOF)
143                         }
144
145                         for a := len(want) - 2; a >= 0; a-- {
146                                 for b := a + 1; b <= len(want); b++ {
147                                         offset, err := rdr.Seek(int64(a), io.SeekStart)
148                                         c.Logf("...a=%d, b=%d", a, b)
149                                         c.Check(err, check.IsNil)
150                                         c.Check(offset, check.Equals, int64(a))
151                                         buf := make([]byte, b-a)
152                                         n, err := io.ReadFull(rdr, buf)
153                                         c.Check(n, check.Equals, b-a)
154                                         c.Check(string(buf), check.Equals, want[a:b])
155                                 }
156                         }
157                         offset, err := rdr.Seek(-1, io.SeekStart)
158                         c.Check(err, check.NotNil)
159                         c.Check(offset, check.Equals, int64(len(want)))
160
161                         c.Check(rdr.Close(), check.Equals, nil)
162                 }
163         }
164 }
165
166 func (s *CollectionReaderUnit) TestCollectionReaderManyBlocks(c *check.C) {
167         h := md5.New()
168         var testdata []byte
169         buf := make([]byte, 4096)
170         locs := make([]string, len(buf))
171         filesize := 0
172         for i := 0; i < len(locs); i++ {
173                 _, err := rand.Read(buf[:i])
174                 h.Write(buf[:i])
175                 locs[i], _, err = s.kc.PutB(buf[:i])
176                 c.Assert(err, check.IsNil)
177                 filesize += i
178                 testdata = append(testdata, buf[:i]...)
179         }
180         manifest := "./random " + strings.Join(locs, " ") + " 0:" + strconv.Itoa(filesize) + ":bytes.bin\n"
181         dataMD5 := h.Sum(nil)
182
183         checkMD5 := md5.New()
184         rdr, err := s.kc.CollectionFileReader(map[string]interface{}{"manifest_text": manifest}, "random/bytes.bin")
185         c.Assert(err, check.IsNil)
186         defer rdr.Close()
187
188         _, err = io.Copy(checkMD5, rdr)
189         c.Check(err, check.IsNil)
190         _, err = rdr.Read(make([]byte, 1))
191         c.Check(err, check.Equals, io.EOF)
192         c.Check(checkMD5.Sum(nil), check.DeepEquals, dataMD5)
193
194         size, err := rdr.Seek(0, io.SeekEnd)
195         c.Check(err, check.IsNil)
196         buf = make([]byte, len(testdata))
197         copy(buf, testdata)
198         curPos := size
199         for i := 0; i < 16; i++ {
200                 offset := rand.Intn(len(buf) - 1)
201                 count := rand.Intn(len(buf) - offset)
202                 if rand.Intn(2) == 0 {
203                         curPos, err = rdr.Seek(int64(offset)-curPos, io.SeekCurrent)
204                 } else {
205                         curPos, err = rdr.Seek(int64(offset), io.SeekStart)
206                 }
207                 c.Check(curPos, check.Equals, int64(offset))
208                 for count > 0 {
209                         n, err := rdr.Read(buf[offset : offset+count])
210                         c.Assert(err, check.IsNil)
211                         c.Assert(n > 0, check.Equals, true)
212                         offset += n
213                         count -= n
214                 }
215                 curPos, err = rdr.Seek(0, io.SeekCurrent)
216                 c.Check(curPos, check.Equals, int64(offset))
217         }
218         c.Check(md5.Sum(buf), check.DeepEquals, md5.Sum(testdata))
219         c.Check(buf[:1000], check.DeepEquals, testdata[:1000])
220
221         curPos, err = rdr.Seek(size+12345, io.SeekCurrent)
222         c.Check(err, check.IsNil)
223         c.Check(curPos, check.Equals, size)
224
225         curPos, err = rdr.Seek(8-size, io.SeekCurrent)
226         c.Check(err, check.IsNil)
227         c.Check(curPos, check.Equals, int64(8))
228
229         curPos, err = rdr.Seek(-9, io.SeekCurrent)
230         c.Check(err, check.NotNil)
231         c.Check(curPos, check.Equals, int64(8))
232 }
233
234 func (s *CollectionReaderUnit) TestCollectionReaderCloseEarly(c *check.C) {
235         s.kc.BlockCache = &BlockCache{}
236         s.kc.PutB([]byte("foo"))
237         s.kc.PutB([]byte("bar"))
238         s.kc.PutB([]byte("baz"))
239
240         mt := ". "
241         for i := 0; i < 300; i++ {
242                 mt += "acbd18db4cc2f85cedef654fccc4a4d8+3 37b51d194a7513e45b56f6524f2d51f2+3 73feffa4b7f6bb68e44cf984c85f6e88+3 "
243         }
244         mt += "0:2700:foo900.txt\n"
245
246         // Grab the stub server's lock, ensuring our cfReader doesn't
247         // get anything back from its first call to kc.Get() before we
248         // have a chance to call Close().
249         s.handler.lock <- struct{}{}
250         opsBeforeRead := *s.handler.ops
251
252         rdr, err := s.kc.CollectionFileReader(map[string]interface{}{"manifest_text": mt}, "foo900.txt")
253         c.Assert(err, check.IsNil)
254
255         firstReadDone := make(chan struct{})
256         go func() {
257                 n, err := rdr.Read(make([]byte, 3))
258                 c.Check(n, check.Equals, 3)
259                 c.Check(err, check.IsNil)
260                 close(firstReadDone)
261         }()
262
263         // Release the stub server's lock. The first GET operation will proceed.
264         <-s.handler.lock
265
266         // Make sure our first read operation consumes the data
267         // received from the first GET.
268         <-firstReadDone
269
270         err = rdr.Close()
271         c.Check(err, check.IsNil)
272
273         // Stub should have handled exactly one GET request.
274         c.Check(*s.handler.ops, check.Equals, opsBeforeRead+1)
275 }
276
277 func (s *CollectionReaderUnit) TestCollectionReaderDataError(c *check.C) {
278         manifest := ". ffffffffffffffffffffffffffffffff+1 0:1:notfound.txt\n"
279         buf := make([]byte, 1)
280         rdr, err := s.kc.CollectionFileReader(map[string]interface{}{"manifest_text": manifest}, "notfound.txt")
281         c.Check(err, check.IsNil)
282         for i := 0; i < 2; i++ {
283                 _, err = io.ReadFull(rdr, buf)
284                 c.Check(err, check.NotNil)
285                 c.Check(err, check.Not(check.Equals), io.EOF)
286         }
287         c.Check(rdr.Close(), check.IsNil)
288 }