renames_test.go 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362
  1. package plumbing
  2. import (
  3. "bytes"
  4. "compress/gzip"
  5. "io/ioutil"
  6. "os"
  7. "path"
  8. "sync"
  9. "testing"
  10. "github.com/stretchr/testify/assert"
  11. "gopkg.in/src-d/go-git.v4/plumbing"
  12. "gopkg.in/src-d/go-git.v4/plumbing/object"
  13. "gopkg.in/src-d/hercules.v10/internal/core"
  14. "gopkg.in/src-d/hercules.v10/internal/test"
  15. )
  16. func fixtureRenameAnalysis() *RenameAnalysis {
  17. ra := RenameAnalysis{SimilarityThreshold: 80}
  18. ra.Initialize(test.Repository)
  19. return &ra
  20. }
  21. func TestRenameAnalysisMeta(t *testing.T) {
  22. ra := fixtureRenameAnalysis()
  23. assert.Equal(t, ra.Name(), "RenameAnalysis")
  24. assert.Equal(t, len(ra.Provides()), 1)
  25. assert.Equal(t, ra.Provides()[0], DependencyTreeChanges)
  26. assert.Equal(t, len(ra.Requires()), 2)
  27. assert.Equal(t, ra.Requires()[0], DependencyBlobCache)
  28. assert.Equal(t, ra.Requires()[1], DependencyTreeChanges)
  29. opts := ra.ListConfigurationOptions()
  30. assert.Len(t, opts, 1)
  31. assert.Equal(t, opts[0].Name, ConfigRenameAnalysisSimilarityThreshold)
  32. ra.SimilarityThreshold = 0
  33. assert.NoError(t, ra.Configure(map[string]interface{}{
  34. ConfigRenameAnalysisSimilarityThreshold: 70,
  35. }))
  36. assert.Equal(t, ra.SimilarityThreshold, 70)
  37. logger := core.NewLogger()
  38. assert.NoError(t, ra.Configure(map[string]interface{}{
  39. core.ConfigLogger: logger,
  40. }))
  41. assert.Equal(t, logger, ra.l)
  42. assert.Equal(t, ra.SimilarityThreshold, 70)
  43. }
  44. func TestRenameAnalysisRegistration(t *testing.T) {
  45. summoned := core.Registry.Summon((&RenameAnalysis{}).Name())
  46. assert.Len(t, summoned, 1)
  47. assert.Equal(t, summoned[0].Name(), "RenameAnalysis")
  48. summoned = core.Registry.Summon((&RenameAnalysis{}).Provides()[0])
  49. assert.True(t, len(summoned) >= 1)
  50. matched := false
  51. for _, tp := range summoned {
  52. matched = matched || tp.Name() == "RenameAnalysis"
  53. }
  54. assert.True(t, matched)
  55. }
  56. func TestRenameAnalysisInitializeInvalidThreshold(t *testing.T) {
  57. ra := RenameAnalysis{SimilarityThreshold: -10}
  58. ra.Initialize(test.Repository)
  59. assert.Equal(t, ra.SimilarityThreshold, RenameAnalysisDefaultThreshold)
  60. ra = RenameAnalysis{SimilarityThreshold: 110}
  61. ra.Initialize(test.Repository)
  62. assert.Equal(t, ra.SimilarityThreshold, RenameAnalysisDefaultThreshold)
  63. ra = RenameAnalysis{SimilarityThreshold: 0}
  64. ra.Initialize(test.Repository)
  65. ra = RenameAnalysis{SimilarityThreshold: 100}
  66. ra.Initialize(test.Repository)
  67. }
  68. func TestRenameAnalysisConsume(t *testing.T) {
  69. ra := fixtureRenameAnalysis()
  70. changes := make(object.Changes, 3)
  71. // 2b1ed978194a94edeabbca6de7ff3b5771d4d665
  72. treeFrom, _ := test.Repository.TreeObject(plumbing.NewHash(
  73. "96c6ece9b2f3c7c51b83516400d278dea5605100"))
  74. treeTo, _ := test.Repository.TreeObject(plumbing.NewHash(
  75. "251f2094d7b523d5bcc60e663b6cf38151bf8844"))
  76. changes[0] = &object.Change{From: object.ChangeEntry{
  77. Name: "analyser.go",
  78. Tree: treeFrom,
  79. TreeEntry: object.TreeEntry{
  80. Name: "analyser.go",
  81. Mode: 0100644,
  82. Hash: plumbing.NewHash("baa64828831d174f40140e4b3cfa77d1e917a2c1"),
  83. },
  84. }, To: object.ChangeEntry{},
  85. }
  86. changes[1] = &object.Change{From: object.ChangeEntry{}, To: object.ChangeEntry{
  87. Name: "burndown.go",
  88. Tree: treeTo,
  89. TreeEntry: object.TreeEntry{
  90. Name: "burndown.go",
  91. Mode: 0100644,
  92. Hash: plumbing.NewHash("29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2"),
  93. },
  94. },
  95. }
  96. changes[2] = &object.Change{From: object.ChangeEntry{
  97. Name: "cmd/hercules/main.go",
  98. Tree: treeFrom,
  99. TreeEntry: object.TreeEntry{
  100. Name: "cmd/hercules/main.go",
  101. Mode: 0100644,
  102. Hash: plumbing.NewHash("c29112dbd697ad9b401333b80c18a63951bc18d9"),
  103. },
  104. }, To: object.ChangeEntry{
  105. Name: "cmd/hercules/main.go",
  106. Tree: treeTo,
  107. TreeEntry: object.TreeEntry{
  108. Name: "cmd/hercules/main.go",
  109. Mode: 0100644,
  110. Hash: plumbing.NewHash("f7d918ec500e2f925ecde79b51cc007bac27de72"),
  111. },
  112. },
  113. }
  114. cache := map[plumbing.Hash]*CachedBlob{}
  115. AddHash(t, cache, "baa64828831d174f40140e4b3cfa77d1e917a2c1")
  116. AddHash(t, cache, "29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2")
  117. AddHash(t, cache, "c29112dbd697ad9b401333b80c18a63951bc18d9")
  118. AddHash(t, cache, "f7d918ec500e2f925ecde79b51cc007bac27de72")
  119. deps := map[string]interface{}{}
  120. deps[DependencyBlobCache] = cache
  121. deps[DependencyTreeChanges] = changes
  122. ra.SimilarityThreshold = 37
  123. res, err := ra.Consume(deps)
  124. assert.Nil(t, err)
  125. renamed := res[DependencyTreeChanges].(object.Changes)
  126. assert.Equal(t, len(renamed), 2)
  127. ra.SimilarityThreshold = 39
  128. res, err = ra.Consume(deps)
  129. assert.Nil(t, err)
  130. renamed = res[DependencyTreeChanges].(object.Changes)
  131. assert.Equal(t, len(renamed), 3)
  132. }
  133. func TestSortableChanges(t *testing.T) {
  134. changes := sortableChanges{
  135. sortableChange{
  136. nil, plumbing.NewHash("0000000000000000000000000000000000000000"),
  137. }, sortableChange{
  138. nil, plumbing.NewHash("ffffffffffffffffffffffffffffffffffffffff"),
  139. },
  140. }
  141. assert.True(t, changes.Less(0, 1))
  142. assert.False(t, changes.Less(1, 0))
  143. assert.False(t, changes.Less(0, 0))
  144. changes.Swap(0, 1)
  145. assert.Equal(t, changes[0].hash.String(), "ffffffffffffffffffffffffffffffffffffffff")
  146. assert.Equal(t, changes[1].hash.String(), "0000000000000000000000000000000000000000")
  147. }
  148. func TestSortableBlobs(t *testing.T) {
  149. blobs := sortableBlobs{
  150. sortableBlob{
  151. nil, int64(0),
  152. }, sortableBlob{
  153. nil, int64(1),
  154. },
  155. }
  156. assert.True(t, blobs.Less(0, 1))
  157. assert.False(t, blobs.Less(1, 0))
  158. assert.False(t, blobs.Less(0, 0))
  159. blobs.Swap(0, 1)
  160. assert.Equal(t, blobs[0].size, int64(1))
  161. assert.Equal(t, blobs[1].size, int64(0))
  162. }
  163. func TestRenameAnalysisFork(t *testing.T) {
  164. ra1 := fixtureRenameAnalysis()
  165. clones := ra1.Fork(1)
  166. assert.Len(t, clones, 1)
  167. ra2 := clones[0].(*RenameAnalysis)
  168. assert.True(t, ra1 == ra2)
  169. ra1.Merge([]core.PipelineItem{ra2})
  170. }
  171. func TestRenameAnalysisSizesAreClose(t *testing.T) {
  172. ra := fixtureRenameAnalysis()
  173. assert.True(t, ra.sizesAreClose(941, 963))
  174. assert.True(t, ra.sizesAreClose(941, 1150))
  175. assert.True(t, ra.sizesAreClose(941, 803))
  176. assert.False(t, ra.sizesAreClose(1320, 1668))
  177. }
  178. func TestRenameAnalysisSortRenameCandidates(t *testing.T) {
  179. candidates := []int{0, 1, 2, 3}
  180. sortRenameCandidates(candidates, "test_regression.py", func(i int) string {
  181. return []string{"gather_nd_op.h", "test.py", "test_file_system.cc", "regression.py"}[i]
  182. })
  183. assert.Equal(t, candidates[0], 3)
  184. assert.Equal(t, candidates[1], 1)
  185. }
  186. func TestBlobsAreCloseFlakyBug(t *testing.T) {
  187. gitBlob1, err := test.Repository.BlobObject(plumbing.NewHash(
  188. "29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2"))
  189. if err != nil {
  190. t.Fatalf("get 29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2 %v", err)
  191. }
  192. gitBlob2, err := test.Repository.BlobObject(plumbing.NewHash(
  193. "baa64828831d174f40140e4b3cfa77d1e917a2c1"))
  194. if err != nil {
  195. t.Fatalf("get baa64828831d174f40140e4b3cfa77d1e917a2c1 %v", err)
  196. }
  197. blob1 := &CachedBlob{*gitBlob1, nil}
  198. blob2 := &CachedBlob{*gitBlob2, nil}
  199. err = blob1.Cache()
  200. if err != nil {
  201. t.Fatalf("read 29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2 %v", err)
  202. }
  203. err = blob2.Cache()
  204. if err != nil {
  205. t.Fatalf("read baa64828831d174f40140e4b3cfa77d1e917a2c1 %v", err)
  206. }
  207. wg := sync.WaitGroup{}
  208. gr := 10 // number of concurrent goroutines
  209. wg.Add(gr)
  210. for i := 0; i < gr; i++ {
  211. go func() {
  212. ra := fixtureRenameAnalysis()
  213. ra.SimilarityThreshold = 37
  214. result, err := ra.blobsAreClose(blob1, blob2)
  215. assert.Nil(t, err)
  216. assert.True(t, result)
  217. result, err = ra.blobsAreClose(blob2, blob1)
  218. assert.Nil(t, err)
  219. assert.True(t, result)
  220. ra.SimilarityThreshold = 39
  221. result, err = ra.blobsAreClose(blob1, blob2)
  222. assert.Nil(t, err)
  223. assert.False(t, result)
  224. result, err = ra.blobsAreClose(blob2, blob1)
  225. assert.Nil(t, err)
  226. assert.False(t, result)
  227. wg.Done()
  228. }()
  229. }
  230. wg.Wait()
  231. }
  232. func TestBlobsAreCloseText(t *testing.T) {
  233. blob1 := &CachedBlob{Data: []byte("hello, world!")}
  234. blob2 := &CachedBlob{Data: []byte("hello, world?")}
  235. blob1.Size = int64(len(blob1.Data))
  236. blob2.Size = int64(len(blob2.Data))
  237. ra := fixtureRenameAnalysis()
  238. result, err := ra.blobsAreClose(blob1, blob2)
  239. assert.Nil(t, err)
  240. assert.True(t, result)
  241. blob1.Data = []byte("hello, mloncode")
  242. blob1.Size = int64(len(blob1.Data))
  243. result, err = ra.blobsAreClose(blob1, blob2)
  244. assert.Nil(t, err)
  245. assert.False(t, result)
  246. }
  247. func TestBlobsAreCloseBinary(t *testing.T) {
  248. blob1 := &CachedBlob{}
  249. blob2 := &CachedBlob{}
  250. ra := fixtureRenameAnalysis()
  251. result, err := ra.blobsAreClose(blob1, blob2)
  252. assert.Nil(t, err)
  253. assert.True(t, result)
  254. blob1.Data = make([]byte, 100)
  255. blob2.Data = blob1.Data
  256. blob1.Size = 100
  257. blob2.Size = 100
  258. result, err = ra.blobsAreClose(blob1, blob2)
  259. assert.Nil(t, err)
  260. assert.True(t, result)
  261. blob1.Data = bytes.Repeat([]byte{1}, 100)
  262. blob2.Data = blob1.Data
  263. result, err = ra.blobsAreClose(blob1, blob2)
  264. assert.Nil(t, err)
  265. assert.True(t, result)
  266. for i := 0; i < 100; i++ {
  267. blob1.Data[i] = byte(i)
  268. }
  269. result, err = ra.blobsAreClose(blob1, blob2)
  270. assert.Nil(t, err)
  271. assert.True(t, result)
  272. blob2.Data = make([]byte, 100)
  273. for i := 0; i < 80; i++ {
  274. blob2.Data[i] = byte(i)
  275. }
  276. result, err = ra.blobsAreClose(blob1, blob2)
  277. assert.Nil(t, err)
  278. assert.True(t, result)
  279. blob2.Data[79] = 0
  280. result, err = ra.blobsAreClose(blob1, blob2)
  281. assert.Nil(t, err)
  282. assert.False(t, result)
  283. blob1.Data = []byte("hello, world!")
  284. blob1.Size = int64(len(blob2.Data))
  285. result, err = ra.blobsAreClose(blob1, blob2)
  286. assert.Nil(t, err)
  287. assert.False(t, result)
  288. blob1.Data, blob2.Data = blob2.Data, blob1.Data
  289. blob1.Size, blob2.Size = blob2.Size, blob1.Size
  290. result, err = ra.blobsAreClose(blob1, blob2)
  291. assert.Nil(t, err)
  292. assert.False(t, result)
  293. }
  294. func loadData(t *testing.T, name string) []byte {
  295. gzsource, err := os.Open(path.Join("..", "test_data", name))
  296. defer gzsource.Close()
  297. if err != nil {
  298. t.Errorf("open ../test_data/%s: %v", name, err)
  299. }
  300. gzreader, err := gzip.NewReader(gzsource)
  301. if err != nil {
  302. t.Errorf("gzip ../test_data/%s: %v", name, err)
  303. }
  304. data, err := ioutil.ReadAll(gzreader)
  305. if err != nil {
  306. t.Errorf("gzip ../test_data/%s: %v", name, err)
  307. }
  308. return data
  309. }
  310. func TestBlobsAreCloseBug1(t *testing.T) {
  311. data := loadData(t, "rename_bug1.xml.gz")
  312. blob1 := &CachedBlob{Data: data}
  313. blob2 := &CachedBlob{Data: data}
  314. blob1.Size = int64(len(data))
  315. blob2.Size = int64(len(data))
  316. ra := fixtureRenameAnalysis()
  317. result, err := ra.blobsAreClose(blob1, blob2)
  318. assert.Nil(t, err)
  319. assert.True(t, result)
  320. }
  321. func TestBlobsAreCloseBug2(t *testing.T) {
  322. data1 := loadData(t, "rename_bug2.base.gz")
  323. data2 := loadData(t, "rename_bug2.head.gz")
  324. blob1 := &CachedBlob{Data: data1}
  325. blob2 := &CachedBlob{Data: data2}
  326. blob1.Size = int64(len(data1))
  327. blob2.Size = int64(len(data2))
  328. ra := fixtureRenameAnalysis()
  329. result, err := ra.blobsAreClose(blob1, blob2)
  330. assert.Nil(t, err)
  331. assert.False(t, result)
  332. }