140 lines
4.4 KiB
Python
140 lines
4.4 KiB
Python
"""
|
|
Tests for scripts/hash_dedup.py — Bounded hash deduplication.
|
|
"""
|
|
|
|
import json
|
|
import os
|
|
import shutil
|
|
import tempfile
|
|
import unittest
|
|
from datetime import datetime, timedelta
|
|
from pathlib import Path
|
|
|
|
import sys
|
|
sys.path.insert(0, str(Path(__file__).parent.parent / "scripts"))
|
|
from hash_dedup import HashDedup
|
|
|
|
|
|
class TestHashDedup(unittest.TestCase):
|
|
def setUp(self):
|
|
self.tmpdir = tempfile.mkdtemp()
|
|
self.dedup = HashDedup(self.tmpdir)
|
|
|
|
def tearDown(self):
|
|
shutil.rmtree(self.tmpdir)
|
|
|
|
def test_compute_hash(self):
|
|
h = HashDedup.compute_hash("test content")
|
|
self.assertEqual(len(h), 64) # SHA-256 hex
|
|
self.assertTrue(all(c in '0123456789abcdef' for c in h))
|
|
|
|
def test_same_content_same_hash(self):
|
|
h1 = HashDedup.compute_hash("hello")
|
|
h2 = HashDedup.compute_hash("hello")
|
|
self.assertEqual(h1, h2)
|
|
|
|
def test_different_content_different_hash(self):
|
|
h1 = HashDedup.compute_hash("hello")
|
|
h2 = HashDedup.compute_hash("world")
|
|
self.assertNotEqual(h1, h2)
|
|
|
|
def test_add_new(self):
|
|
result = self.dedup.add("new content")
|
|
self.assertTrue(result)
|
|
|
|
def test_add_duplicate(self):
|
|
self.dedup.add("content")
|
|
result = self.dedup.add("content")
|
|
self.assertFalse(result)
|
|
|
|
def test_is_duplicate_false(self):
|
|
self.assertFalse(self.dedup.is_duplicate("unknown"))
|
|
|
|
def test_is_duplicate_true(self):
|
|
self.dedup.add("known content")
|
|
self.assertTrue(self.dedup.is_duplicate("known content"))
|
|
|
|
def test_add_batch(self):
|
|
items = ["a", "b", "c"]
|
|
added = self.dedup.add_batch(items)
|
|
self.assertEqual(added, 3)
|
|
|
|
def test_add_batch_deduplicates(self):
|
|
items = ["a", "b", "a", "c", "b"]
|
|
added = self.dedup.add_batch(items)
|
|
self.assertEqual(added, 3)
|
|
|
|
def test_creates_date_file(self):
|
|
self.dedup.add("test")
|
|
today = datetime.utcnow().strftime("%Y-%m-%d")
|
|
path = Path(self.tmpdir) / f"{today}.json"
|
|
self.assertTrue(path.exists())
|
|
|
|
def test_file_format(self):
|
|
self.dedup.add("test")
|
|
today = datetime.utcnow().strftime("%Y-%m-%d")
|
|
path = Path(self.tmpdir) / f"{today}.json"
|
|
with open(path) as f:
|
|
data = json.load(f)
|
|
self.assertEqual(data["date"], today)
|
|
self.assertEqual(data["count"], 1)
|
|
self.assertEqual(len(data["hashes"]), 1)
|
|
|
|
def test_cleanup_removes_old(self):
|
|
# Create fake old file
|
|
old_date = (datetime.utcnow() - timedelta(days=10)).strftime("%Y-%m-%d")
|
|
old_path = Path(self.tmpdir) / f"{old_date}.json"
|
|
with open(old_path, 'w') as f:
|
|
json.dump({"date": old_date, "count": 0, "hashes": []}, f)
|
|
|
|
removed = self.dedup.cleanup(keep_days=7)
|
|
self.assertEqual(removed, 1)
|
|
self.assertFalse(old_path.exists())
|
|
|
|
def test_cleanup_keeps_recent(self):
|
|
recent_date = (datetime.utcnow() - timedelta(days=3)).strftime("%Y-%m-%d")
|
|
recent_path = Path(self.tmpdir) / f"{recent_date}.json"
|
|
with open(recent_path, 'w') as f:
|
|
json.dump({"date": recent_date, "count": 0, "hashes": []}, f)
|
|
|
|
removed = self.dedup.cleanup(keep_days=7)
|
|
self.assertEqual(removed, 0)
|
|
self.assertTrue(recent_path.exists())
|
|
|
|
def test_cleanup_ignores_non_date_files(self):
|
|
junk = Path(self.tmpdir) / "not-a-date.json"
|
|
with open(junk, 'w') as f:
|
|
f.write("{}")
|
|
|
|
removed = self.dedup.cleanup(keep_days=1)
|
|
self.assertEqual(removed, 0)
|
|
self.assertTrue(junk.exists())
|
|
|
|
def test_stats_empty(self):
|
|
stats = self.dedup.stats()
|
|
self.assertEqual(stats["file_count"], 0)
|
|
self.assertEqual(stats["total_hashes"], 0)
|
|
|
|
def test_stats_with_data(self):
|
|
self.dedup.add("one")
|
|
self.dedup.add("two")
|
|
stats = self.dedup.stats()
|
|
self.assertEqual(stats["file_count"], 1)
|
|
self.assertEqual(stats["total_hashes"], 2)
|
|
self.assertEqual(stats["today_count"], 2)
|
|
|
|
def test_max_hashes_per_file(self):
|
|
dedup = HashDedup(self.tmpdir, max_hashes_per_file=3)
|
|
for i in range(10):
|
|
dedup.add(f"content-{i}")
|
|
|
|
today = datetime.utcnow().strftime("%Y-%m-%d")
|
|
path = Path(self.tmpdir) / f"{today}.json"
|
|
with open(path) as f:
|
|
data = json.load(f)
|
|
self.assertLessEqual(len(data["hashes"]), 3)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
unittest.main()
|