Browse Source

Added file path similarity to scoring metric in rename detection

The scoring method was not taking into account the similarity of
the file paths and file names. I changed the metric so that it is 99%
based on content (which used to be 100% of the old metric), and 1%
based on path similarity. Of that 1%, half (.5% of the total final
score) is based on the actual file names (e.g. "foo.java"), and half
on the directory (e.g. "src/com/foo/bar/").

Change-Id: I94f0c23bf6413c491b10d5625f6ad7d2ecfb4def
stable-0.9
Jeff Schumacher 15 years ago
parent
commit
9a48de86d8
  1. 12
      org.eclipse.jgit.test/tst/org/eclipse/jgit/diff/RenameDetectorTest.java
  2. 8
      org.eclipse.jgit.test/tst/org/eclipse/jgit/diff/SimilarityIndexTest.java
  3. 6
      org.eclipse.jgit/src/org/eclipse/jgit/diff/SimilarityIndex.java
  4. 56
      org.eclipse.jgit/src/org/eclipse/jgit/diff/SimilarityRenameDetector.java

12
org.eclipse.jgit.test/tst/org/eclipse/jgit/diff/RenameDetectorTest.java

@ -124,8 +124,8 @@ public class RenameDetectorTest extends RepositoryTestCase {
}
public void testInexactRename_OnePair() throws Exception {
ObjectId aId = blob("foo\nbar\nbaz\n");
ObjectId bId = blob("foo\nbar\nblah\n");
ObjectId aId = blob("foo\nbar\nbaz\nblarg\n");
ObjectId bId = blob("foo\nbar\nbaz\nblah\n");
DiffEntry a = DiffEntry.add(PATH_A, aId);
DiffEntry b = DiffEntry.delete(PATH_Q, bId);
@ -135,12 +135,12 @@ public class RenameDetectorTest extends RepositoryTestCase {
List<DiffEntry> entries = rd.compute();
assertEquals(1, entries.size());
assertRename(b, a, 61, entries.get(0));
assertRename(b, a, 66, entries.get(0));
}
public void testInexactRename_OneRenameTwoUnrelatedFiles() throws Exception {
ObjectId aId = blob("foo\nbar\nbaz\n");
ObjectId bId = blob("foo\nbar\nblah\n");
ObjectId aId = blob("foo\nbar\nbaz\nblarg\n");
ObjectId bId = blob("foo\nbar\nbaz\nblah\n");
DiffEntry a = DiffEntry.add(PATH_A, aId);
DiffEntry b = DiffEntry.delete(PATH_Q, bId);
@ -158,7 +158,7 @@ public class RenameDetectorTest extends RepositoryTestCase {
assertEquals(3, entries.size());
assertSame(c, entries.get(0));
assertSame(d, entries.get(1));
assertRename(b, a, 61, entries.get(2));
assertRename(b, a, 66, entries.get(2));
}
public void testInexactRename_LastByteDifferent() throws Exception {

8
org.eclipse.jgit.test/tst/org/eclipse/jgit/diff/SimilarityIndexTest.java

@ -78,8 +78,8 @@ public class SimilarityIndexTest extends TestCase {
assertEquals(8, src.common(dst));
assertEquals(8, dst.common(src));
assertEquals(100, src.score(dst));
assertEquals(100, dst.score(src));
assertEquals(100, src.score(dst, 100));
assertEquals(100, dst.score(src, 100));
}
public void testCommonScore_EmptyFiles() {
@ -102,8 +102,8 @@ public class SimilarityIndexTest extends TestCase {
assertEquals(6, src.common(dst));
assertEquals(6, dst.common(src));
assertEquals(75, src.score(dst));
assertEquals(75, dst.score(src));
assertEquals(75, src.score(dst, 100));
assertEquals(75, dst.score(src, 100));
}
private static SimilarityIndex hash(String text) {

6
org.eclipse.jgit/src/org/eclipse/jgit/diff/SimilarityIndex.java

@ -142,11 +142,11 @@ class SimilarityIndex {
Arrays.sort(idHash);
}
int score(SimilarityIndex dst) {
int score(SimilarityIndex dst, int maxScore) {
long max = Math.max(fileSize, dst.fileSize);
if (max == 0)
return 100;
return (int) ((common(dst) * 100L) / max);
return maxScore;
return (int) ((common(dst) * maxScore) / max);
}
int common(SimilarityIndex dst) {

56
org.eclipse.jgit/src/org/eclipse/jgit/diff/SimilarityRenameDetector.java

@ -260,7 +260,14 @@ class SimilarityRenameDetector {
}
SimilarityIndex d = hash(dstEnt.newId.toObjectId());
int score = s.score(d);
int contentScore = s.score(d, 10000);
// nameScore returns a value between 0 and 100, but we want it
// to be in the same range as the content score. This allows it
// to be dropped into the pretty formula for the final score.
int nameScore = nameScore(srcEnt.oldName, dstEnt.newName) * 100;
int score = (contentScore * 99 + nameScore * 1) / 10000;
if (score < renameScore) {
pm.update(1);
@ -280,6 +287,53 @@ class SimilarityRenameDetector {
return mNext;
}
private int nameScore(String a, String b) {
int aDirLen = a.lastIndexOf("/") + 1;
int bDirLen = b.lastIndexOf("/") + 1;
int dirMin = Math.min(aDirLen, bDirLen);
int dirMax = Math.max(aDirLen, bDirLen);
final int dirScoreLtr;
final int dirScoreRtl;
if (dirMax == 0) {
dirScoreLtr = 100;
dirScoreRtl = 100;
} else {
int dirSim = 0;
for (; dirSim < dirMin; dirSim++) {
if (a.charAt(dirSim) != b.charAt(dirSim))
break;
}
dirScoreLtr = (dirSim * 100) / dirMax;
if (dirScoreLtr == 100) {
dirScoreRtl = 100;
} else {
for (dirSim = 0; dirSim < dirMin; dirSim++) {
if (a.charAt(aDirLen - 1 - dirSim) != b.charAt(bDirLen - 1
- dirSim))
break;
}
dirScoreRtl = (dirSim * 100) / dirMax;
}
}
int fileMin = Math.min(a.length() - aDirLen, b.length() - bDirLen);
int fileMax = Math.max(a.length() - aDirLen, b.length() - bDirLen);
int fileSim = 0;
for (; fileSim < fileMin; fileSim++) {
if (a.charAt(a.length() - 1 - fileSim) != b.charAt(b.length() - 1
- fileSim))
break;
}
int fileScore = (fileSim * 100) / fileMax;
return (((dirScoreLtr + dirScoreRtl) * 25) + (fileScore * 50)) / 100;
}
private SimilarityIndex hash(ObjectId objectId) throws IOException {
SimilarityIndex r = new SimilarityIndex();
r.hash(repo.openObject(objectId));

Loading…
Cancel
Save