renames_test.go 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374
  1. package plumbing
  2. import (
  3. "bytes"
  4. "compress/gzip"
  5. "io/ioutil"
  6. "os"
  7. "path"
  8. "sync"
  9. "testing"
  10. "time"
  11. "github.com/stretchr/testify/assert"
  12. "gopkg.in/src-d/go-git.v4/plumbing"
  13. "gopkg.in/src-d/go-git.v4/plumbing/object"
  14. "gopkg.in/src-d/hercules.v10/internal/core"
  15. "gopkg.in/src-d/hercules.v10/internal/test"
  16. )
  17. func fixtureRenameAnalysis() *RenameAnalysis {
  18. ra := RenameAnalysis{SimilarityThreshold: 80}
  19. ra.Initialize(test.Repository)
  20. return &ra
  21. }
  22. func TestRenameAnalysisMeta(t *testing.T) {
  23. ra := fixtureRenameAnalysis()
  24. assert.Equal(t, ra.Name(), "RenameAnalysis")
  25. assert.Equal(t, len(ra.Provides()), 1)
  26. assert.Equal(t, ra.Provides()[0], DependencyTreeChanges)
  27. assert.Equal(t, len(ra.Requires()), 2)
  28. assert.Equal(t, ra.Requires()[0], DependencyBlobCache)
  29. assert.Equal(t, ra.Requires()[1], DependencyTreeChanges)
  30. opts := ra.ListConfigurationOptions()
  31. assert.Len(t, opts, 2)
  32. assert.Equal(t, opts[0].Name, ConfigRenameAnalysisSimilarityThreshold)
  33. assert.Equal(t, opts[1].Name, ConfigRenameAnalysisTimeout)
  34. ra.SimilarityThreshold = 0
  35. assert.NoError(t, ra.Configure(map[string]interface{}{
  36. ConfigRenameAnalysisSimilarityThreshold: 70,
  37. ConfigRenameAnalysisTimeout: 1000,
  38. }))
  39. assert.Equal(t, ra.SimilarityThreshold, 70)
  40. assert.Equal(t, ra.Timeout, time.Second)
  41. logger := core.NewLogger()
  42. assert.NoError(t, ra.Configure(map[string]interface{}{
  43. core.ConfigLogger: logger,
  44. }))
  45. assert.Equal(t, logger, ra.l)
  46. assert.Equal(t, ra.SimilarityThreshold, 70)
  47. assert.Equal(t, ra.Timeout, time.Second)
  48. }
  49. func TestRenameAnalysisRegistration(t *testing.T) {
  50. summoned := core.Registry.Summon((&RenameAnalysis{}).Name())
  51. assert.Len(t, summoned, 1)
  52. assert.Equal(t, summoned[0].Name(), "RenameAnalysis")
  53. summoned = core.Registry.Summon((&RenameAnalysis{}).Provides()[0])
  54. assert.True(t, len(summoned) >= 1)
  55. matched := false
  56. for _, tp := range summoned {
  57. matched = matched || tp.Name() == "RenameAnalysis"
  58. }
  59. assert.True(t, matched)
  60. }
  61. func TestRenameAnalysisInitializeInvalidThreshold(t *testing.T) {
  62. ra := RenameAnalysis{SimilarityThreshold: -10}
  63. ra.Initialize(test.Repository)
  64. assert.Equal(t, ra.SimilarityThreshold, RenameAnalysisDefaultThreshold)
  65. ra = RenameAnalysis{SimilarityThreshold: 110}
  66. ra.Initialize(test.Repository)
  67. assert.Equal(t, ra.SimilarityThreshold, RenameAnalysisDefaultThreshold)
  68. ra = RenameAnalysis{SimilarityThreshold: 0}
  69. ra.Initialize(test.Repository)
  70. ra = RenameAnalysis{SimilarityThreshold: 100}
  71. ra.Initialize(test.Repository)
  72. }
  73. func TestRenameAnalysisConsume(t *testing.T) {
  74. ra := fixtureRenameAnalysis()
  75. changes := make(object.Changes, 3)
  76. // 2b1ed978194a94edeabbca6de7ff3b5771d4d665
  77. treeFrom, _ := test.Repository.TreeObject(plumbing.NewHash(
  78. "96c6ece9b2f3c7c51b83516400d278dea5605100"))
  79. treeTo, _ := test.Repository.TreeObject(plumbing.NewHash(
  80. "251f2094d7b523d5bcc60e663b6cf38151bf8844"))
  81. changes[0] = &object.Change{From: object.ChangeEntry{
  82. Name: "analyser.go",
  83. Tree: treeFrom,
  84. TreeEntry: object.TreeEntry{
  85. Name: "analyser.go",
  86. Mode: 0100644,
  87. Hash: plumbing.NewHash("baa64828831d174f40140e4b3cfa77d1e917a2c1"),
  88. },
  89. }, To: object.ChangeEntry{},
  90. }
  91. changes[1] = &object.Change{From: object.ChangeEntry{}, To: object.ChangeEntry{
  92. Name: "burndown.go",
  93. Tree: treeTo,
  94. TreeEntry: object.TreeEntry{
  95. Name: "burndown.go",
  96. Mode: 0100644,
  97. Hash: plumbing.NewHash("29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2"),
  98. },
  99. },
  100. }
  101. changes[2] = &object.Change{From: object.ChangeEntry{
  102. Name: "cmd/hercules/main.go",
  103. Tree: treeFrom,
  104. TreeEntry: object.TreeEntry{
  105. Name: "cmd/hercules/main.go",
  106. Mode: 0100644,
  107. Hash: plumbing.NewHash("c29112dbd697ad9b401333b80c18a63951bc18d9"),
  108. },
  109. }, To: object.ChangeEntry{
  110. Name: "cmd/hercules/main.go",
  111. Tree: treeTo,
  112. TreeEntry: object.TreeEntry{
  113. Name: "cmd/hercules/main.go",
  114. Mode: 0100644,
  115. Hash: plumbing.NewHash("f7d918ec500e2f925ecde79b51cc007bac27de72"),
  116. },
  117. },
  118. }
  119. cache := map[plumbing.Hash]*CachedBlob{}
  120. AddHash(t, cache, "baa64828831d174f40140e4b3cfa77d1e917a2c1")
  121. AddHash(t, cache, "29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2")
  122. AddHash(t, cache, "c29112dbd697ad9b401333b80c18a63951bc18d9")
  123. AddHash(t, cache, "f7d918ec500e2f925ecde79b51cc007bac27de72")
  124. deps := map[string]interface{}{}
  125. deps[DependencyBlobCache] = cache
  126. deps[DependencyTreeChanges] = changes
  127. ra.SimilarityThreshold = 37
  128. res, err := ra.Consume(deps)
  129. assert.Nil(t, err)
  130. renamed := res[DependencyTreeChanges].(object.Changes)
  131. assert.Equal(t, len(renamed), 2)
  132. ra.SimilarityThreshold = 39
  133. res, err = ra.Consume(deps)
  134. assert.Nil(t, err)
  135. renamed = res[DependencyTreeChanges].(object.Changes)
  136. assert.Equal(t, len(renamed), 3)
  137. ra.SimilarityThreshold = 37
  138. ra.Timeout = time.Microsecond
  139. res, err = ra.Consume(deps)
  140. assert.Nil(t, err)
  141. renamed = res[DependencyTreeChanges].(object.Changes)
  142. assert.Equal(t, len(renamed), 3)
  143. }
  144. func TestSortableChanges(t *testing.T) {
  145. changes := sortableChanges{
  146. sortableChange{
  147. nil, plumbing.NewHash("0000000000000000000000000000000000000000"),
  148. }, sortableChange{
  149. nil, plumbing.NewHash("ffffffffffffffffffffffffffffffffffffffff"),
  150. },
  151. }
  152. assert.True(t, changes.Less(0, 1))
  153. assert.False(t, changes.Less(1, 0))
  154. assert.False(t, changes.Less(0, 0))
  155. changes.Swap(0, 1)
  156. assert.Equal(t, changes[0].hash.String(), "ffffffffffffffffffffffffffffffffffffffff")
  157. assert.Equal(t, changes[1].hash.String(), "0000000000000000000000000000000000000000")
  158. }
  159. func TestSortableBlobs(t *testing.T) {
  160. blobs := sortableBlobs{
  161. sortableBlob{
  162. nil, int64(0),
  163. }, sortableBlob{
  164. nil, int64(1),
  165. },
  166. }
  167. assert.True(t, blobs.Less(0, 1))
  168. assert.False(t, blobs.Less(1, 0))
  169. assert.False(t, blobs.Less(0, 0))
  170. blobs.Swap(0, 1)
  171. assert.Equal(t, blobs[0].size, int64(1))
  172. assert.Equal(t, blobs[1].size, int64(0))
  173. }
  174. func TestRenameAnalysisFork(t *testing.T) {
  175. ra1 := fixtureRenameAnalysis()
  176. clones := ra1.Fork(1)
  177. assert.Len(t, clones, 1)
  178. ra2 := clones[0].(*RenameAnalysis)
  179. assert.True(t, ra1 == ra2)
  180. ra1.Merge([]core.PipelineItem{ra2})
  181. }
  182. func TestRenameAnalysisSizesAreClose(t *testing.T) {
  183. ra := fixtureRenameAnalysis()
  184. assert.True(t, ra.sizesAreClose(941, 963))
  185. assert.True(t, ra.sizesAreClose(941, 1150))
  186. assert.True(t, ra.sizesAreClose(941, 803))
  187. assert.False(t, ra.sizesAreClose(1320, 1668))
  188. }
  189. func TestRenameAnalysisSortRenameCandidates(t *testing.T) {
  190. candidates := []int{0, 1, 2, 3}
  191. sortRenameCandidates(candidates, "test_regression.py", func(i int) string {
  192. return []string{"gather_nd_op.h", "test.py", "test_file_system.cc", "regression.py"}[i]
  193. })
  194. assert.Equal(t, candidates[0], 3)
  195. assert.Equal(t, candidates[1], 1)
  196. }
  197. func TestBlobsAreCloseFlakyBug(t *testing.T) {
  198. gitBlob1, err := test.Repository.BlobObject(plumbing.NewHash(
  199. "29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2"))
  200. if err != nil {
  201. t.Fatalf("get 29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2 %v", err)
  202. }
  203. gitBlob2, err := test.Repository.BlobObject(plumbing.NewHash(
  204. "baa64828831d174f40140e4b3cfa77d1e917a2c1"))
  205. if err != nil {
  206. t.Fatalf("get baa64828831d174f40140e4b3cfa77d1e917a2c1 %v", err)
  207. }
  208. blob1 := &CachedBlob{*gitBlob1, nil}
  209. blob2 := &CachedBlob{*gitBlob2, nil}
  210. err = blob1.Cache()
  211. if err != nil {
  212. t.Fatalf("read 29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2 %v", err)
  213. }
  214. err = blob2.Cache()
  215. if err != nil {
  216. t.Fatalf("read baa64828831d174f40140e4b3cfa77d1e917a2c1 %v", err)
  217. }
  218. wg := sync.WaitGroup{}
  219. gr := 10 // number of concurrent goroutines
  220. wg.Add(gr)
  221. for i := 0; i < gr; i++ {
  222. go func() {
  223. ra := fixtureRenameAnalysis()
  224. ra.SimilarityThreshold = 37
  225. result, err := ra.blobsAreClose(blob1, blob2)
  226. assert.Nil(t, err)
  227. assert.True(t, result)
  228. result, err = ra.blobsAreClose(blob2, blob1)
  229. assert.Nil(t, err)
  230. assert.True(t, result)
  231. ra.SimilarityThreshold = 39
  232. result, err = ra.blobsAreClose(blob1, blob2)
  233. assert.Nil(t, err)
  234. assert.False(t, result)
  235. result, err = ra.blobsAreClose(blob2, blob1)
  236. assert.Nil(t, err)
  237. assert.False(t, result)
  238. wg.Done()
  239. }()
  240. }
  241. wg.Wait()
  242. }
  243. func TestBlobsAreCloseText(t *testing.T) {
  244. blob1 := &CachedBlob{Data: []byte("hello, world!")}
  245. blob2 := &CachedBlob{Data: []byte("hello, world?")}
  246. blob1.Size = int64(len(blob1.Data))
  247. blob2.Size = int64(len(blob2.Data))
  248. ra := fixtureRenameAnalysis()
  249. result, err := ra.blobsAreClose(blob1, blob2)
  250. assert.Nil(t, err)
  251. assert.True(t, result)
  252. blob1.Data = []byte("hello, mloncode")
  253. blob1.Size = int64(len(blob1.Data))
  254. result, err = ra.blobsAreClose(blob1, blob2)
  255. assert.Nil(t, err)
  256. assert.False(t, result)
  257. }
  258. func TestBlobsAreCloseBinary(t *testing.T) {
  259. blob1 := &CachedBlob{}
  260. blob2 := &CachedBlob{}
  261. ra := fixtureRenameAnalysis()
  262. result, err := ra.blobsAreClose(blob1, blob2)
  263. assert.Nil(t, err)
  264. assert.True(t, result)
  265. blob1.Data = make([]byte, 100)
  266. blob2.Data = blob1.Data
  267. blob1.Size = 100
  268. blob2.Size = 100
  269. result, err = ra.blobsAreClose(blob1, blob2)
  270. assert.Nil(t, err)
  271. assert.True(t, result)
  272. blob1.Data = bytes.Repeat([]byte{1}, 100)
  273. blob2.Data = blob1.Data
  274. result, err = ra.blobsAreClose(blob1, blob2)
  275. assert.Nil(t, err)
  276. assert.True(t, result)
  277. for i := 0; i < 100; i++ {
  278. blob1.Data[i] = byte(i)
  279. }
  280. result, err = ra.blobsAreClose(blob1, blob2)
  281. assert.Nil(t, err)
  282. assert.True(t, result)
  283. blob2.Data = make([]byte, 100)
  284. for i := 0; i < 80; i++ {
  285. blob2.Data[i] = byte(i)
  286. }
  287. result, err = ra.blobsAreClose(blob1, blob2)
  288. assert.Nil(t, err)
  289. assert.True(t, result)
  290. blob2.Data[79] = 0
  291. result, err = ra.blobsAreClose(blob1, blob2)
  292. assert.Nil(t, err)
  293. assert.False(t, result)
  294. blob1.Data = []byte("hello, world!")
  295. blob1.Size = int64(len(blob2.Data))
  296. result, err = ra.blobsAreClose(blob1, blob2)
  297. assert.Nil(t, err)
  298. assert.False(t, result)
  299. blob1.Data, blob2.Data = blob2.Data, blob1.Data
  300. blob1.Size, blob2.Size = blob2.Size, blob1.Size
  301. result, err = ra.blobsAreClose(blob1, blob2)
  302. assert.Nil(t, err)
  303. assert.False(t, result)
  304. }
  305. func loadData(t *testing.T, name string) []byte {
  306. gzsource, err := os.Open(path.Join("..", "test_data", name))
  307. defer gzsource.Close()
  308. if err != nil {
  309. t.Errorf("open ../test_data/%s: %v", name, err)
  310. }
  311. gzreader, err := gzip.NewReader(gzsource)
  312. if err != nil {
  313. t.Errorf("gzip ../test_data/%s: %v", name, err)
  314. }
  315. data, err := ioutil.ReadAll(gzreader)
  316. if err != nil {
  317. t.Errorf("gzip ../test_data/%s: %v", name, err)
  318. }
  319. return data
  320. }
  321. func TestBlobsAreCloseBug1(t *testing.T) {
  322. data := loadData(t, "rename_bug1.xml.gz")
  323. blob1 := &CachedBlob{Data: data}
  324. blob2 := &CachedBlob{Data: data}
  325. blob1.Size = int64(len(data))
  326. blob2.Size = int64(len(data))
  327. ra := fixtureRenameAnalysis()
  328. result, err := ra.blobsAreClose(blob1, blob2)
  329. assert.Nil(t, err)
  330. assert.True(t, result)
  331. }
  332. func TestBlobsAreCloseBug2(t *testing.T) {
  333. data1 := loadData(t, "rename_bug2.base.gz")
  334. data2 := loadData(t, "rename_bug2.head.gz")
  335. blob1 := &CachedBlob{Data: data1}
  336. blob2 := &CachedBlob{Data: data2}
  337. blob1.Size = int64(len(data1))
  338. blob2.Size = int64(len(data2))
  339. ra := fixtureRenameAnalysis()
  340. result, err := ra.blobsAreClose(blob1, blob2)
  341. assert.Nil(t, err)
  342. assert.False(t, result)
  343. }