bandstore.py 9.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255
  1. import os
  2. import tempfile
  3. import shutil
  4. import unittest
  5. import sqlite3
  6. import uuid
  7. import fcntl
  8. Binary = sqlite3.Binary
  9. class BandStore(object):
  10. """
  11. """
  12. class DoesNotExist(KeyError):
  13. """"""
  14. class AlreadyExists(KeyError):
  15. """"""
  16. IDLE = 'Idle'
  17. OPEN = 'Open'
  18. ACTIVE = 'Active'
  19. BAND_LIMIT = 1 * 1024 * 1024
  20. def __init__(self, path):
  21. self.read_fd = None
  22. self.write_fd = None
  23. if not os.path.exists(path):
  24. self.create(path)
  25. self.open(path)
  26. def open(self, path):
  27. if not os.path.isdir(path):
  28. raise Exception('%s Does not look like a store')
  29. db_path = os.path.join(path, 'dedupestore.db')
  30. if not os.path.exists(db_path):
  31. raise Exception('%s Does not look like a store2')
  32. self.lock_fd = open(os.path.join(path, 'lock'), 'w')
  33. fcntl.flock(self.lock_fd, fcntl.LOCK_EX)
  34. self.path = path
  35. self.cnx = sqlite3.connect(db_path)
  36. self.cnx.text_factory = str
  37. self.cursor = self.cnx.cursor()
  38. self._begin()
  39. def _begin(self):
  40. if self.read_fd:
  41. self.read_fd.close()
  42. self.read_fd = None
  43. if self.write_fd:
  44. self.write_fd.close()
  45. self.write_fd = None
  46. row = self.cursor.execute('SELECT uuid, tid, nextband, version, '
  47. 'bandlimit FROM system').fetchone()
  48. self.uuid, self.tid, self.nextband, version, self.bandlimit = row
  49. assert version == 1
  50. self.state = self.OPEN
  51. self.read_band = None
  52. self.write_band = None
  53. self.to_delete = set()
  54. band = self.nextband
  55. while os.path.exists(self.band_filename(band)):
  56. os.unlink(self.band_filename(band))
  57. band += 1
  58. def create(self, path):
  59. os.mkdir(path)
  60. os.mkdir(os.path.join(path, 'bands'))
  61. cnx = sqlite3.connect(os.path.join(path, 'dedupestore.db'))
  62. cnx.execute('CREATE TABLE objects(ns BINARY NOT NULL, id BINARY NOT NULL, '
  63. 'band NOT NULL, offset NOT NULL, size NOT NULL)')
  64. cnx.execute('CREATE UNIQUE INDEX objects_pk ON objects(ns, id)')
  65. cnx.execute('CREATE TABLE system(uuid NOT NULL, tid NOT NULL, '
  66. 'nextband NOT NULL, version NOT NULL, bandlimit NOT NULL)')
  67. cnx.execute('INSERT INTO system VALUES(?,?,?,?,?)',
  68. (uuid.uuid1().hex, 0, 0, 1, self.BAND_LIMIT))
  69. cnx.commit()
  70. def close(self):
  71. self.rollback()
  72. self.cnx.close()
  73. self.lock_fd.close()
  74. os.unlink(os.path.join(self.path, 'lock'))
  75. def commit(self):
  76. """
  77. """
  78. self.band = None
  79. self.delete_bands()
  80. self.tid += 1
  81. self._begin()
  82. def delete_bands(self):
  83. for b in self.to_delete:
  84. objects = self.cursor.execute('SELECT ns, id, offset, size '
  85. 'FROM objects WHERE band=? ORDER BY offset',
  86. (b,)).fetchall()
  87. for o in objects:
  88. band, offset, size = self.store_data(self.retrieve_data(b, *o[2:]))
  89. self.cursor.execute('UPDATE objects SET band=?, offset=?, size=? '
  90. 'WHERE ns=? AND id=?', (band, offset, size,
  91. Binary(o[0]), Binary(o[1])))
  92. self.cursor.execute('UPDATE system SET tid=tid+1, nextband=?',
  93. (self.nextband,))
  94. self.cnx.commit()
  95. for b in self.to_delete:
  96. os.unlink(self.band_filename(b))
  97. def rollback(self):
  98. """
  99. """
  100. self.cnx.rollback()
  101. self._begin()
  102. def get(self, ns, id):
  103. """
  104. """
  105. self.cursor.execute('SELECT band, offset, size FROM objects WHERE ns=? and id=?',
  106. (Binary(ns), Binary(id)))
  107. row = self.cursor.fetchone()
  108. if row:
  109. return self.retrieve_data(*row)
  110. else:
  111. raise self.DoesNotExist
  112. def band_filename(self, band):
  113. return os.path.join(self.path, 'bands', str(band / 1000), str(band))
  114. def retrieve_data(self, band, offset, size):
  115. if self.write_band == band:
  116. self.write_fd.flush()
  117. if self.read_band != band:
  118. self.read_band = band
  119. if self.read_fd:
  120. self.read_fd.close()
  121. self.read_fd = open(self.band_filename(band), 'rb')
  122. self.read_fd.seek(offset)
  123. return self.read_fd.read(size)
  124. def store_data(self, data):
  125. if self.write_band is None:
  126. self.write_band = self.nextband
  127. self.nextband += 1
  128. if self.write_band % 1000 == 0:
  129. path = os.path.join(self.path, 'bands', str(self.write_band / 1000))
  130. if not os.path.exists(path):
  131. os.mkdir(path)
  132. assert not os.path.exists(self.band_filename(self.write_band))
  133. self.write_fd = open(self.band_filename(self.write_band), 'ab')
  134. band = self.write_band
  135. offset = self.write_fd.tell()
  136. self.write_fd.write(data)
  137. if offset + len(data) > self.bandlimit:
  138. self.write_band = None
  139. return band, offset, len(data)
  140. def put(self, ns, id, data):
  141. """
  142. """
  143. try:
  144. band, offset, size = self.store_data(data)
  145. self.cursor.execute('INSERT INTO objects (ns, id, band, offset, size) '
  146. 'VALUES(?, ?, ?, ?, ?)',
  147. (Binary(ns), Binary(id), band, offset, size))
  148. except sqlite3.IntegrityError:
  149. raise self.AlreadyExists
  150. def delete(self, ns, id):
  151. """
  152. """
  153. self.cursor.execute('SELECT band FROM objects WHERE ns=? and id=?',
  154. (Binary(ns), Binary(id)))
  155. row = self.cursor.fetchone()
  156. if not row:
  157. raise self.DoesNotExist
  158. self.cursor.execute('DELETE FROM objects WHERE ns=? AND id=?',
  159. (Binary(ns), Binary(id)))
  160. self.to_delete.add(row[0])
  161. def list(self, ns, prefix='', marker=None, max_keys=1000000):
  162. """
  163. """
  164. sql = 'SELECT id FROM objects WHERE ns=:ns'
  165. args = dict(ns=Binary(ns))
  166. if prefix:
  167. args['prefix'] = Binary(prefix)
  168. args['end'] = Binary(prefix + chr(255))
  169. sql += ' AND id >= :prefix AND id < :end'
  170. if marker:
  171. sql += ' AND id >= :marker'
  172. args['marker'] = Binary(marker)
  173. for row in self.cursor.execute(sql + ' LIMIT ' + str(max_keys), args):
  174. yield str(row[0])
  175. class BandStoreTestCase(unittest.TestCase):
  176. def setUp(self):
  177. self.tmppath = tempfile.mkdtemp()
  178. self.store = BandStore(os.path.join(self.tmppath, 'store'))
  179. def tearDown(self):
  180. shutil.rmtree(self.tmppath)
  181. def test1(self):
  182. self.assertEqual(self.store.tid, 0)
  183. self.assertEqual(self.store.state, self.store.OPEN)
  184. self.store.put('SOMENS', 'SOMEID', 'SOMEDATA')
  185. self.assertRaises(self.store.AlreadyExists, lambda: self.store.put('SOMENS', 'SOMEID', 'SOMEDATA'))
  186. self.assertEqual(self.store.get('SOMENS', 'SOMEID'), 'SOMEDATA')
  187. self.store.rollback()
  188. self.assertRaises(self.store.DoesNotExist, lambda: self.store.get('SOMENS', 'SOMEID'))
  189. self.assertEqual(self.store.tid, 0)
  190. def test2(self):
  191. self.assertEqual(self.store.tid, 0)
  192. self.assertEqual(self.store.state, self.store.OPEN)
  193. self.store.put('SOMENS', 'SOMEID', 'SOMEDATA')
  194. self.assertEqual(self.store.get('SOMENS', 'SOMEID'), 'SOMEDATA')
  195. self.store.commit()
  196. self.assertEqual(self.store.tid, 1)
  197. self.assertEqual(self.store.get('SOMENS', 'SOMEID'), 'SOMEDATA')
  198. self.store.delete('SOMENS', 'SOMEID')
  199. self.assertRaises(self.store.DoesNotExist, lambda: self.store.get('SOMENS', 'SOMEID'))
  200. self.store.rollback()
  201. self.assertEqual(self.store.get('SOMENS', 'SOMEID'), 'SOMEDATA')
  202. self.store.delete('SOMENS', 'SOMEID')
  203. self.assertRaises(self.store.DoesNotExist, lambda: self.store.get('SOMENS', 'SOMEID'))
  204. self.store.commit()
  205. self.assertEqual(self.store.tid, 2)
  206. self.assertRaises(self.store.DoesNotExist, lambda: self.store.get('SOMENS', 'SOMEID'))
  207. def test_list(self):
  208. self.store.put('SOMENS', 'SOMEID12', 'SOMEDATA')
  209. self.store.put('SOMENS', 'SOMEID', 'SOMEDATA')
  210. self.store.put('SOMENS', 'SOMEID1', 'SOMEDATA')
  211. self.store.put('SOMENS', 'SOMEID123', 'SOMEDATA')
  212. self.store.commit()
  213. self.assertEqual(list(self.store.list('SOMENS', max_keys=3)),
  214. ['SOMEID', 'SOMEID1', 'SOMEID12'])
  215. self.assertEqual(list(self.store.list('SOMENS', marker='SOMEID12')),
  216. ['SOMEID12', 'SOMEID123'])
  217. self.assertEqual(list(self.store.list('SOMENS', prefix='SOMEID1', max_keys=2)),
  218. ['SOMEID1', 'SOMEID12'])
  219. self.assertEqual(list(self.store.list('SOMENS', prefix='SOMEID1', marker='SOMEID12')),
  220. ['SOMEID12', 'SOMEID123'])
  221. def suite():
  222. return unittest.TestLoader().loadTestsFromTestCase(BandStoreTestCase)
  223. if __name__ == '__main__':
  224. unittest.main()