NCBACK-34 add logic to find similar image in db
Test Workflow / test (push) Successful in 3s

This commit is contained in:
vladp
2026-01-13 22:40:52 +07:00
parent 344ec075d4
commit 8ac73ff045
18 changed files with 394 additions and 188 deletions
@@ -0,0 +1,62 @@
package ru.soune.nocopy.service;
import lombok.RequiredArgsConstructor;
import org.springframework.stereotype.Service;
import ru.soune.nocopy.dto.file.SimilarFileResponse;
import ru.soune.nocopy.repository.ImageHashRepository;
import ru.soune.nocopy.repository.ImageSimilarityRepository;
import ru.soune.nocopy.repository.SimilarImageProjection;
import java.util.List;
@Service
@RequiredArgsConstructor
public class FileSimilarityService {
private final ImageSimilarityRepository repository;
private final ImageHashRepository hashRepository;
public List<SimilarFileResponse> findSimilarFiles(String fileId) {
var imageHashEntity = hashRepository.findById(fileId)
.orElseThrow(() -> new RuntimeException("Hash not found"));
Integer hash64Hi = imageHashEntity.getHash64Hi();
Integer hash64Lo = imageHashEntity.getHash64Lo();
List<SimilarImageProjection> candidates =
repository.findCandidates(fileId);
return candidates.stream()
.map(c -> {
Integer cHi = c.getHash64Hi();
Integer cLo = c.getHash64Lo();
int hamming = hamming64(hash64Hi, hash64Lo, cHi, cLo);
String level;
if (hamming <= 5) {
level = "DUPLICATE";
} else if (hamming <= 12) {
level = "SIMILAR";
} else {
level = "DIFFERENT";
}
return SimilarFileResponse.builder()
.fileId(c.getSimilarFileId())
.originalFileName(c.getOriginalFileName())
.fileSize(c.getFileSize())
.hammingDistance(hamming)
.similarityLevel(level)
.build();
})
.sorted((a, b) ->
Integer.compare(a.getHammingDistance(), b.getHammingDistance()))
.toList();
}
private int hamming64(int aHi, int aLo, int bHi, int bLo) {
return Integer.bitCount(aHi ^ bHi)
+ Integer.bitCount(aLo ^ bLo);
}
}