63 lines
2.1 KiB
Go
63 lines
2.1 KiB
Go
package tests
|
|
|
|
import (
|
|
"testing"
|
|
|
|
"github.com/jfraeys/fetch_ml/internal/api/helpers"
|
|
"github.com/jfraeys/fetch_ml/internal/queue"
|
|
)
|
|
|
|
// TestDuplicateDetection verifies the duplicate detection logic
|
|
func TestDuplicateDetection(t *testing.T) {
|
|
// Test 1: Same args + same commit = duplicate (would be detected)
|
|
commitID := "abc123def4567890abc1"
|
|
args1 := "--epochs 10 --lr 0.001"
|
|
args2 := "--epochs 10 --lr 0.001" // Same args
|
|
|
|
hash1 := helpers.ComputeParamsHash(args1)
|
|
hash2 := helpers.ComputeParamsHash(args2)
|
|
|
|
if hash1 != hash2 {
|
|
t.Error("Same args should produce same hash")
|
|
}
|
|
t.Logf("Same args produce same hash: %s", hash1)
|
|
|
|
// Test 2: Different args = not duplicate
|
|
args3 := "--epochs 20 --lr 0.01" // Different
|
|
hash3 := helpers.ComputeParamsHash(args3)
|
|
|
|
if hash1 == hash3 {
|
|
t.Error("Different args should produce different hashes")
|
|
}
|
|
t.Logf("Different args produce different hashes: %s vs %s", hash1, hash3)
|
|
|
|
// Test 3: Same dataset specs = same dataset_id
|
|
ds1 := []queue.DatasetSpec{{Name: "mnist", Checksum: "sha256:abc123"}}
|
|
ds2 := []queue.DatasetSpec{{Name: "mnist", Checksum: "sha256:abc123"}}
|
|
|
|
id1 := helpers.ComputeDatasetID(ds1, nil)
|
|
id2 := helpers.ComputeDatasetID(ds2, nil)
|
|
|
|
if id1 != id2 {
|
|
t.Error("Same dataset specs should produce same ID")
|
|
}
|
|
t.Logf("Same dataset specs produce same ID: %s", id1)
|
|
|
|
// Test 4: Different dataset = different ID
|
|
ds3 := []queue.DatasetSpec{{Name: "cifar10", Checksum: "sha256:def456"}}
|
|
id3 := helpers.ComputeDatasetID(ds3, nil)
|
|
|
|
if id1 == id3 {
|
|
t.Error("Different datasets should produce different IDs")
|
|
}
|
|
t.Logf("Different datasets produce different IDs: %s vs %s", id1, id3)
|
|
|
|
// Test 5: Composite key logic
|
|
t.Log("\n=== Composite Key Detection ===")
|
|
t.Logf("Job 1: commit=%s, dataset_id=%s, params_hash=%s", commitID, id1, hash1)
|
|
t.Logf("Job 2: commit=%s, dataset_id=%s, params_hash=%s", commitID, id2, hash2)
|
|
t.Log("→ These would be detected as DUPLICATE (same commit + dataset + params)")
|
|
|
|
t.Logf("Job 3: commit=%s, dataset_id=%s, params_hash=%s", commitID, id1, hash3)
|
|
t.Log("→ This would NOT be duplicate (different params)")
|
|
}
|