renames_test.go 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357
  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. facts := map[string]interface{}{}
  34. facts[ConfigRenameAnalysisSimilarityThreshold] = 70
  35. ra.Configure(facts)
  36. assert.Equal(t, ra.SimilarityThreshold, 70)
  37. delete(facts, ConfigRenameAnalysisSimilarityThreshold)
  38. ra.Configure(facts)
  39. assert.Equal(t, ra.SimilarityThreshold, 70)
  40. }
  41. func TestRenameAnalysisRegistration(t *testing.T) {
  42. summoned := core.Registry.Summon((&RenameAnalysis{}).Name())
  43. assert.Len(t, summoned, 1)
  44. assert.Equal(t, summoned[0].Name(), "RenameAnalysis")
  45. summoned = core.Registry.Summon((&RenameAnalysis{}).Provides()[0])
  46. assert.True(t, len(summoned) >= 1)
  47. matched := false
  48. for _, tp := range summoned {
  49. matched = matched || tp.Name() == "RenameAnalysis"
  50. }
  51. assert.True(t, matched)
  52. }
  53. func TestRenameAnalysisInitializeInvalidThreshold(t *testing.T) {
  54. ra := RenameAnalysis{SimilarityThreshold: -10}
  55. ra.Initialize(test.Repository)
  56. assert.Equal(t, ra.SimilarityThreshold, RenameAnalysisDefaultThreshold)
  57. ra = RenameAnalysis{SimilarityThreshold: 110}
  58. ra.Initialize(test.Repository)
  59. assert.Equal(t, ra.SimilarityThreshold, RenameAnalysisDefaultThreshold)
  60. ra = RenameAnalysis{SimilarityThreshold: 0}
  61. ra.Initialize(test.Repository)
  62. ra = RenameAnalysis{SimilarityThreshold: 100}
  63. ra.Initialize(test.Repository)
  64. }
  65. func TestRenameAnalysisConsume(t *testing.T) {
  66. ra := fixtureRenameAnalysis()
  67. changes := make(object.Changes, 3)
  68. // 2b1ed978194a94edeabbca6de7ff3b5771d4d665
  69. treeFrom, _ := test.Repository.TreeObject(plumbing.NewHash(
  70. "96c6ece9b2f3c7c51b83516400d278dea5605100"))
  71. treeTo, _ := test.Repository.TreeObject(plumbing.NewHash(
  72. "251f2094d7b523d5bcc60e663b6cf38151bf8844"))
  73. changes[0] = &object.Change{From: object.ChangeEntry{
  74. Name: "analyser.go",
  75. Tree: treeFrom,
  76. TreeEntry: object.TreeEntry{
  77. Name: "analyser.go",
  78. Mode: 0100644,
  79. Hash: plumbing.NewHash("baa64828831d174f40140e4b3cfa77d1e917a2c1"),
  80. },
  81. }, To: object.ChangeEntry{},
  82. }
  83. changes[1] = &object.Change{From: object.ChangeEntry{}, To: object.ChangeEntry{
  84. Name: "burndown.go",
  85. Tree: treeTo,
  86. TreeEntry: object.TreeEntry{
  87. Name: "burndown.go",
  88. Mode: 0100644,
  89. Hash: plumbing.NewHash("29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2"),
  90. },
  91. },
  92. }
  93. changes[2] = &object.Change{From: object.ChangeEntry{
  94. Name: "cmd/hercules/main.go",
  95. Tree: treeFrom,
  96. TreeEntry: object.TreeEntry{
  97. Name: "cmd/hercules/main.go",
  98. Mode: 0100644,
  99. Hash: plumbing.NewHash("c29112dbd697ad9b401333b80c18a63951bc18d9"),
  100. },
  101. }, To: object.ChangeEntry{
  102. Name: "cmd/hercules/main.go",
  103. Tree: treeTo,
  104. TreeEntry: object.TreeEntry{
  105. Name: "cmd/hercules/main.go",
  106. Mode: 0100644,
  107. Hash: plumbing.NewHash("f7d918ec500e2f925ecde79b51cc007bac27de72"),
  108. },
  109. },
  110. }
  111. cache := map[plumbing.Hash]*CachedBlob{}
  112. AddHash(t, cache, "baa64828831d174f40140e4b3cfa77d1e917a2c1")
  113. AddHash(t, cache, "29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2")
  114. AddHash(t, cache, "c29112dbd697ad9b401333b80c18a63951bc18d9")
  115. AddHash(t, cache, "f7d918ec500e2f925ecde79b51cc007bac27de72")
  116. deps := map[string]interface{}{}
  117. deps[DependencyBlobCache] = cache
  118. deps[DependencyTreeChanges] = changes
  119. ra.SimilarityThreshold = 37
  120. res, err := ra.Consume(deps)
  121. assert.Nil(t, err)
  122. renamed := res[DependencyTreeChanges].(object.Changes)
  123. assert.Equal(t, len(renamed), 2)
  124. ra.SimilarityThreshold = 39
  125. res, err = ra.Consume(deps)
  126. assert.Nil(t, err)
  127. renamed = res[DependencyTreeChanges].(object.Changes)
  128. assert.Equal(t, len(renamed), 3)
  129. }
  130. func TestSortableChanges(t *testing.T) {
  131. changes := sortableChanges{
  132. sortableChange{
  133. nil, plumbing.NewHash("0000000000000000000000000000000000000000"),
  134. }, sortableChange{
  135. nil, plumbing.NewHash("ffffffffffffffffffffffffffffffffffffffff"),
  136. },
  137. }
  138. assert.True(t, changes.Less(0, 1))
  139. assert.False(t, changes.Less(1, 0))
  140. assert.False(t, changes.Less(0, 0))
  141. changes.Swap(0, 1)
  142. assert.Equal(t, changes[0].hash.String(), "ffffffffffffffffffffffffffffffffffffffff")
  143. assert.Equal(t, changes[1].hash.String(), "0000000000000000000000000000000000000000")
  144. }
  145. func TestSortableBlobs(t *testing.T) {
  146. blobs := sortableBlobs{
  147. sortableBlob{
  148. nil, int64(0),
  149. }, sortableBlob{
  150. nil, int64(1),
  151. },
  152. }
  153. assert.True(t, blobs.Less(0, 1))
  154. assert.False(t, blobs.Less(1, 0))
  155. assert.False(t, blobs.Less(0, 0))
  156. blobs.Swap(0, 1)
  157. assert.Equal(t, blobs[0].size, int64(1))
  158. assert.Equal(t, blobs[1].size, int64(0))
  159. }
  160. func TestRenameAnalysisFork(t *testing.T) {
  161. ra1 := fixtureRenameAnalysis()
  162. clones := ra1.Fork(1)
  163. assert.Len(t, clones, 1)
  164. ra2 := clones[0].(*RenameAnalysis)
  165. assert.True(t, ra1 == ra2)
  166. ra1.Merge([]core.PipelineItem{ra2})
  167. }
  168. func TestRenameAnalysisSizesAreClose(t *testing.T) {
  169. ra := fixtureRenameAnalysis()
  170. assert.True(t, ra.sizesAreClose(941, 963))
  171. assert.True(t, ra.sizesAreClose(941, 1150))
  172. assert.True(t, ra.sizesAreClose(941, 803))
  173. assert.False(t, ra.sizesAreClose(1320, 1668))
  174. }
  175. func TestRenameAnalysisSortRenameCandidates(t *testing.T) {
  176. candidates := []int{0, 1, 2, 3}
  177. sortRenameCandidates(candidates, "test_regression.py", func(i int) string {
  178. return []string{"gather_nd_op.h", "test.py", "test_file_system.cc", "regression.py"}[i]
  179. })
  180. assert.Equal(t, candidates[0], 3)
  181. assert.Equal(t, candidates[1], 1)
  182. }
  183. func TestBlobsAreCloseFlakyBug(t *testing.T) {
  184. gitBlob1, err := test.Repository.BlobObject(plumbing.NewHash(
  185. "29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2"))
  186. if err != nil {
  187. t.Fatalf("get 29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2 %v", err)
  188. }
  189. gitBlob2, err := test.Repository.BlobObject(plumbing.NewHash(
  190. "baa64828831d174f40140e4b3cfa77d1e917a2c1"))
  191. if err != nil {
  192. t.Fatalf("get baa64828831d174f40140e4b3cfa77d1e917a2c1 %v", err)
  193. }
  194. blob1 := &CachedBlob{*gitBlob1, nil}
  195. blob2 := &CachedBlob{*gitBlob2, nil}
  196. err = blob1.Cache()
  197. if err != nil {
  198. t.Fatalf("read 29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2 %v", err)
  199. }
  200. err = blob2.Cache()
  201. if err != nil {
  202. t.Fatalf("read baa64828831d174f40140e4b3cfa77d1e917a2c1 %v", err)
  203. }
  204. wg := sync.WaitGroup{}
  205. gr := 10 // number of concurrent goroutines
  206. wg.Add(gr)
  207. for i := 0; i < gr; i++ {
  208. go func() {
  209. ra := fixtureRenameAnalysis()
  210. ra.SimilarityThreshold = 37
  211. result, err := ra.blobsAreClose(blob1, blob2)
  212. assert.Nil(t, err)
  213. assert.True(t, result)
  214. result, err = ra.blobsAreClose(blob2, blob1)
  215. assert.Nil(t, err)
  216. assert.True(t, result)
  217. ra.SimilarityThreshold = 39
  218. result, err = ra.blobsAreClose(blob1, blob2)
  219. assert.Nil(t, err)
  220. assert.False(t, result)
  221. result, err = ra.blobsAreClose(blob2, blob1)
  222. assert.Nil(t, err)
  223. assert.False(t, result)
  224. wg.Done()
  225. }()
  226. }
  227. wg.Wait()
  228. }
  229. func TestBlobsAreCloseText(t *testing.T) {
  230. blob1 := &CachedBlob{Data: []byte("hello, world!")}
  231. blob2 := &CachedBlob{Data: []byte("hello, world?")}
  232. blob1.Size = int64(len(blob1.Data))
  233. blob2.Size = int64(len(blob2.Data))
  234. ra := fixtureRenameAnalysis()
  235. result, err := ra.blobsAreClose(blob1, blob2)
  236. assert.Nil(t, err)
  237. assert.True(t, result)
  238. blob1.Data = []byte("hello, mloncode")
  239. blob1.Size = int64(len(blob1.Data))
  240. result, err = ra.blobsAreClose(blob1, blob2)
  241. assert.Nil(t, err)
  242. assert.False(t, result)
  243. }
  244. func TestBlobsAreCloseBinary(t *testing.T) {
  245. blob1 := &CachedBlob{}
  246. blob2 := &CachedBlob{}
  247. ra := fixtureRenameAnalysis()
  248. result, err := ra.blobsAreClose(blob1, blob2)
  249. assert.Nil(t, err)
  250. assert.True(t, result)
  251. blob1.Data = make([]byte, 100)
  252. blob2.Data = blob1.Data
  253. blob1.Size = 100
  254. blob2.Size = 100
  255. result, err = ra.blobsAreClose(blob1, blob2)
  256. assert.Nil(t, err)
  257. assert.True(t, result)
  258. blob1.Data = bytes.Repeat([]byte{1}, 100)
  259. blob2.Data = blob1.Data
  260. result, err = ra.blobsAreClose(blob1, blob2)
  261. assert.Nil(t, err)
  262. assert.True(t, result)
  263. for i := 0; i < 100; i++ {
  264. blob1.Data[i] = byte(i)
  265. }
  266. result, err = ra.blobsAreClose(blob1, blob2)
  267. assert.Nil(t, err)
  268. assert.True(t, result)
  269. blob2.Data = make([]byte, 100)
  270. for i := 0; i < 80; i++ {
  271. blob2.Data[i] = byte(i)
  272. }
  273. result, err = ra.blobsAreClose(blob1, blob2)
  274. assert.Nil(t, err)
  275. assert.True(t, result)
  276. blob2.Data[79] = 0
  277. result, err = ra.blobsAreClose(blob1, blob2)
  278. assert.Nil(t, err)
  279. assert.False(t, result)
  280. blob1.Data = []byte("hello, world!")
  281. blob1.Size = int64(len(blob2.Data))
  282. result, err = ra.blobsAreClose(blob1, blob2)
  283. assert.Nil(t, err)
  284. assert.False(t, result)
  285. blob1.Data, blob2.Data = blob2.Data, blob1.Data
  286. blob1.Size, blob2.Size = blob2.Size, blob1.Size
  287. result, err = ra.blobsAreClose(blob1, blob2)
  288. assert.Nil(t, err)
  289. assert.False(t, result)
  290. }
  291. func loadData(t *testing.T, name string) []byte {
  292. gzsource, err := os.Open(path.Join("..", "test_data", name))
  293. defer gzsource.Close()
  294. if err != nil {
  295. t.Errorf("open ../test_data/%s: %v", name, err)
  296. }
  297. gzreader, err := gzip.NewReader(gzsource)
  298. if err != nil {
  299. t.Errorf("gzip ../test_data/%s: %v", name, err)
  300. }
  301. data, err := ioutil.ReadAll(gzreader)
  302. if err != nil {
  303. t.Errorf("gzip ../test_data/%s: %v", name, err)
  304. }
  305. return data
  306. }
  307. func TestBlobsAreCloseBug1(t *testing.T) {
  308. data := loadData(t, "rename_bug1.xml.gz")
  309. blob1 := &CachedBlob{Data: data}
  310. blob2 := &CachedBlob{Data: data}
  311. blob1.Size = int64(len(data))
  312. blob2.Size = int64(len(data))
  313. ra := fixtureRenameAnalysis()
  314. result, err := ra.blobsAreClose(blob1, blob2)
  315. assert.Nil(t, err)
  316. assert.True(t, result)
  317. }
  318. func TestBlobsAreCloseBug2(t *testing.T) {
  319. data1 := loadData(t, "rename_bug2.base.gz")
  320. data2 := loadData(t, "rename_bug2.head.gz")
  321. blob1 := &CachedBlob{Data: data1}
  322. blob2 := &CachedBlob{Data: data2}
  323. blob1.Size = int64(len(data1))
  324. blob2.Size = int64(len(data2))
  325. ra := fixtureRenameAnalysis()
  326. result, err := ra.blobsAreClose(blob1, blob2)
  327. assert.Nil(t, err)
  328. assert.False(t, result)
  329. }