store.py 10.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259
  1. #!/usr/bin/env python
  2. import os
  3. import fcntl
  4. import hashlib
  5. import tempfile
  6. import shutil
  7. import unittest
  8. import uuid
  9. class Store(object):
  10. """
  11. """
  12. class DoesNotExist(KeyError):
  13. """"""
  14. class AlreadyExists(KeyError):
  15. """"""
  16. IDLE = 'Idle'
  17. OPEN = 'Open'
  18. ACTIVE = 'Active'
  19. VERSION = 'DEDUPSTORE VERSION 1'
  20. def __init__(self, path):
  21. self.tid = '-1'
  22. self.state = Store.IDLE
  23. if not os.path.exists(path):
  24. self.create(path)
  25. self.open(path)
  26. def create(self, path):
  27. os.mkdir(path)
  28. open(os.path.join(path, 'version'), 'wb').write(self.VERSION)
  29. open(os.path.join(path, 'uuid'), 'wb').write(str(uuid.uuid4()))
  30. open(os.path.join(path, 'tid'), 'wb').write('0')
  31. os.mkdir(os.path.join(path, 'data'))
  32. def open(self, path):
  33. self.path = path
  34. if not os.path.isdir(path):
  35. raise Exception('%s Does not look like a store')
  36. version_path = os.path.join(path, 'version')
  37. if not os.path.exists(version_path) or open(version_path, 'rb').read() != self.VERSION:
  38. raise Exception('%s Does not look like a store2')
  39. self.uuid = open(os.path.join(path, 'uuid'), 'rb').read()
  40. self.lock_fd = open(os.path.join(path, 'lock'), 'w')
  41. fcntl.flock(self.lock_fd, fcntl.LOCK_EX)
  42. self.tid = int(open(os.path.join(path, 'tid'), 'r').read())
  43. self.recover()
  44. def recover(self):
  45. if os.path.exists(os.path.join(self.path, 'txn-active')):
  46. shutil.rmtree(os.path.join(self.path, 'txn-active'))
  47. if os.path.exists(os.path.join(self.path, 'txn-commit')):
  48. self.apply_txn()
  49. if os.path.exists(os.path.join(self.path, 'txn-applied')):
  50. shutil.rmtree(os.path.join(self.path, 'txn-applied'))
  51. self.state = Store.OPEN
  52. self.txn_delete = []
  53. self.txn_write = []
  54. def close(self):
  55. self.recover()
  56. self.lock_fd.close()
  57. self.state = Store.IDLE
  58. def commit(self):
  59. """
  60. """
  61. if self.state == Store.OPEN:
  62. return
  63. assert self.state == Store.ACTIVE
  64. with open(os.path.join(self.path, 'txn-active', 'delete_index'), 'wb') as fd:
  65. fd.write('\n'.join(self.txn_delete))
  66. with open(os.path.join(self.path, 'txn-active', 'write_index'), 'wb') as fd:
  67. fd.write('\n'.join(self.txn_write))
  68. with open(os.path.join(self.path, 'txn-active', 'tid'), 'wb') as fd:
  69. fd.write(str(self.tid + 1))
  70. os.rename(os.path.join(self.path, 'txn-active'),
  71. os.path.join(self.path, 'txn-commit'))
  72. self.recover()
  73. def apply_txn(self):
  74. assert os.path.isdir(os.path.join(self.path, 'txn-commit'))
  75. tid = int(open(os.path.join(self.path, 'txn-commit', 'tid'), 'rb').read())
  76. assert tid == self.tid + 1
  77. delete_list = [line.strip() for line in
  78. open(os.path.join(self.path, 'txn-commit', 'delete_index'), 'rb').readlines()]
  79. for name in delete_list:
  80. path = os.path.join(self.path, 'data', name)
  81. os.unlink(path)
  82. write_list = [line.strip() for line in
  83. open(os.path.join(self.path, 'txn-commit', 'write_index'), 'rb').readlines()]
  84. for name in write_list:
  85. destname = os.path.join(self.path, 'data', name)
  86. if not os.path.exists(os.path.dirname(destname)):
  87. os.makedirs(os.path.dirname(destname))
  88. os.rename(os.path.join(self.path, 'txn-commit', 'write', name), destname)
  89. with open(os.path.join(self.path, 'tid'), 'wb') as fd:
  90. fd.write(str(tid))
  91. os.rename(os.path.join(self.path, 'txn-commit'),
  92. os.path.join(self.path, 'txn-applied'))
  93. shutil.rmtree(os.path.join(self.path, 'txn-applied'))
  94. self.tid = tid
  95. def rollback(self):
  96. """
  97. """
  98. self.recover()
  99. def prepare_txn(self):
  100. if self.state == Store.ACTIVE:
  101. return os.path.join(self.path, 'txn-active')
  102. elif self.state == Store.OPEN:
  103. os.mkdir(os.path.join(self.path, 'txn-active'))
  104. os.mkdir(os.path.join(self.path, 'txn-active', 'write'))
  105. self.state = Store.ACTIVE
  106. def _filename(self, ns, id, base=''):
  107. ns = ns.encode('hex')
  108. id = id.encode('hex')
  109. return os.path.join(base, ns, id[:2], id[2:4], id[4:])
  110. def get(self, ns, id):
  111. """
  112. """
  113. path = self._filename(ns, id)
  114. if path in self.txn_write:
  115. filename = os.path.join(self.path, 'txn-active', 'write', path)
  116. return open(filename, 'rb').read()
  117. if path in self.txn_delete:
  118. raise Store.DoesNotExist('Object %s:%s does not exist' % (ns.encode('hex'), id.encode('hex')))
  119. filename = self._filename(ns, id, os.path.join(self.path, 'data'))
  120. if os.path.exists(filename):
  121. return open(filename, 'rb').read()
  122. else:
  123. raise Store.DoesNotExist('Object %s:%s does not exist' % (ns.encode('hex'), id.encode('hex')))
  124. def put(self, ns, id, data):
  125. """
  126. """
  127. self.prepare_txn()
  128. path = self._filename(ns, id)
  129. filename = self._filename(ns, id, os.path.join(self.path, 'data'))
  130. if (path in self.txn_write or
  131. (path not in self.txn_delete and os.path.exists(filename))):
  132. raise Store.AlreadyExists('Object already exists: %s:%s' % (ns.encode('hex'), id.encode('hex')))
  133. if path in self.txn_delete:
  134. self.txn_delete.remove(path)
  135. if path not in self.txn_write:
  136. self.txn_write.append(path)
  137. filename = self._filename(ns, id, os.path.join(self.path, 'txn-active', 'write'))
  138. if not os.path.exists(os.path.dirname(filename)):
  139. os.makedirs(os.path.dirname(filename))
  140. with open(filename, 'wb') as fd:
  141. fd.write(data)
  142. def delete(self, ns, id):
  143. """
  144. """
  145. self.prepare_txn()
  146. path = self._filename(ns, id)
  147. if path in self.txn_write:
  148. filename = self._filename(ns, id, os.path.join(self.path, 'txn-active', 'write'))
  149. self.txn_write.remove(path)
  150. os.unlink(filename)
  151. else:
  152. filename = os.path.join(self.path, 'data', path)
  153. if os.path.exists(filename):
  154. self.txn_delete.append(path)
  155. else:
  156. raise Store.DoesNotExist('Object does not exist: %s' % hash.encode('hex'))
  157. def list(self, ns, prefix='', marker=None, max_keys=1000000):
  158. for x in self.foo(os.path.join(self.path, 'data', ns.encode('hex')),
  159. prefix, marker, '', max_keys):
  160. yield x
  161. def foo(self, path, prefix, marker, base, max_keys):
  162. n = 0
  163. for name in sorted(os.listdir(path)):
  164. if n >= max_keys:
  165. return
  166. dirs = []
  167. names = []
  168. id = name.decode('hex')
  169. if os.path.isdir(os.path.join(path, name)):
  170. if prefix and not id.startswith(prefix[:len(id)]):
  171. continue
  172. for x in self.foo(os.path.join(path, name),
  173. prefix[len(id):], marker,
  174. base + id, max_keys - n):
  175. yield x
  176. n += 1
  177. else:
  178. if prefix and not id.startswith(prefix):
  179. continue
  180. if not marker or base + id >= marker:
  181. yield base + id
  182. n += 1
  183. class StoreTestCase(unittest.TestCase):
  184. def setUp(self):
  185. self.tmppath = tempfile.mkdtemp()
  186. self.store = Store(os.path.join(self.tmppath, 'store'))
  187. def tearDown(self):
  188. shutil.rmtree(self.tmppath)
  189. def test1(self):
  190. self.assertEqual(self.store.tid, 0)
  191. self.assertEqual(self.store.state, Store.OPEN)
  192. self.store.put('SOMENS', 'SOMEID', 'SOMEDATA')
  193. self.assertRaises(Store.AlreadyExists, lambda: self.store.put('SOMENS', 'SOMEID', 'SOMEDATA'))
  194. self.assertEqual(self.store.get('SOMENS', 'SOMEID'), 'SOMEDATA')
  195. self.store.rollback()
  196. self.assertRaises(Store.DoesNotExist, lambda: self.store.get('SOMENS', 'SOMEID'))
  197. self.assertEqual(self.store.tid, 0)
  198. def test2(self):
  199. self.assertEqual(self.store.tid, 0)
  200. self.assertEqual(self.store.state, Store.OPEN)
  201. self.store.put('SOMENS', 'SOMEID', 'SOMEDATA')
  202. self.assertEqual(self.store.get('SOMENS', 'SOMEID'), 'SOMEDATA')
  203. self.store.commit()
  204. self.assertEqual(self.store.tid, 1)
  205. self.assertEqual(self.store.get('SOMENS', 'SOMEID'), 'SOMEDATA')
  206. self.store.delete('SOMENS', 'SOMEID')
  207. self.assertRaises(Store.DoesNotExist, lambda: self.store.get('SOMENS', 'SOMEID'))
  208. self.store.rollback()
  209. self.assertEqual(self.store.get('SOMENS', 'SOMEID'), 'SOMEDATA')
  210. self.store.delete('SOMENS', 'SOMEID')
  211. self.assertRaises(Store.DoesNotExist, lambda: self.store.get('SOMENS', 'SOMEID'))
  212. self.store.commit()
  213. self.assertEqual(self.store.tid, 2)
  214. self.assertRaises(Store.DoesNotExist, lambda: self.store.get('SOMENS', 'SOMEID'))
  215. def test_list(self):
  216. self.store.put('SOMENS', 'SOMEID12', 'SOMEDATA')
  217. self.store.put('SOMENS', 'SOMEID', 'SOMEDATA')
  218. self.store.put('SOMENS', 'SOMEID1', 'SOMEDATA')
  219. self.store.put('SOMENS', 'SOMEID123', 'SOMEDATA')
  220. self.store.commit()
  221. self.assertEqual(list(self.store.list('SOMENS', max_keys=3)),
  222. ['SOMEID', 'SOMEID1', 'SOMEID12'])
  223. self.assertEqual(list(self.store.list('SOMENS', marker='SOMEID12')),
  224. ['SOMEID12', 'SOMEID123'])
  225. self.assertEqual(list(self.store.list('SOMENS', prefix='SOMEID1', max_keys=2)),
  226. ['SOMEID1', 'SOMEID12'])
  227. self.assertEqual(list(self.store.list('SOMENS', prefix='SOMEID1', marker='SOMEID12')),
  228. ['SOMEID12', 'SOMEID123'])
  229. if __name__ == '__main__':
  230. unittest.main()