Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.8/site-packages/pickleshare.py: 22%

194 statements  

« prev     ^ index     » next       coverage.py v7.4.4, created at 2024-04-20 06:09 +0000

1#!/usr/bin/env python 

2 

3""" PickleShare - a small 'shelve' like datastore with concurrency support 

4 

5Like shelve, a PickleShareDB object acts like a normal dictionary. Unlike 

6shelve, many processes can access the database simultaneously. Changing a 

7value in database is immediately visible to other processes accessing the 

8same database. 

9 

10Concurrency is possible because the values are stored in separate files. Hence 

11the "database" is a directory where *all* files are governed by PickleShare. 

12 

13Example usage:: 

14 

15 from pickleshare import * 

16 db = PickleShareDB('~/testpickleshare') 

17 db.clear() 

18 print "Should be empty:",db.items() 

19 db['hello'] = 15 

20 db['aku ankka'] = [1,2,313] 

21 db['paths/are/ok/key'] = [1,(5,46)] 

22 print db.keys() 

23 del db['aku ankka'] 

24 

25This module is certainly not ZODB, but can be used for low-load 

26(non-mission-critical) situations where tiny code size trumps the 

27advanced features of a "real" object database. 

28 

29Installation guide: pip install pickleshare 

30 

31Author: Ville Vainio <vivainio@gmail.com> 

32License: MIT open source license. 

33 

34""" 

35 

36from __future__ import print_function 

37 

38 

39__version__ = "0.7.5" 

40 

41try: 

42 from pathlib import Path 

43except ImportError: 

44 # Python 2 backport 

45 from pathlib2 import Path 

46 

47import os,stat,time 

48try: 

49 import collections.abc as collections_abc 

50except ImportError: 

51 import collections as collections_abc 

52try: 

53 import cPickle as pickle 

54except ImportError: 

55 import pickle 

56import errno 

57import sys 

58 

59if sys.version_info[0] >= 3: 

60 string_types = (str,) 

61else: 

62 string_types = (str, unicode) 

63 

64def gethashfile(key): 

65 return ("%02x" % abs(hash(key) % 256))[-2:] 

66 

67_sentinel = object() 

68 

69class PickleShareDB(collections_abc.MutableMapping): 

70 """ The main 'connection' object for PickleShare database """ 

71 def __init__(self,root): 

72 """ Return a db object that will manage the specied directory""" 

73 if not isinstance(root, string_types): 

74 root = str(root) 

75 root = os.path.abspath(os.path.expanduser(root)) 

76 self.root = Path(root) 

77 if not self.root.is_dir(): 

78 # catching the exception is necessary if multiple processes are concurrently trying to create a folder 

79 # exists_ok keyword argument of mkdir does the same but only from Python 3.5 

80 try: 

81 self.root.mkdir(parents=True) 

82 except OSError as e: 

83 if e.errno != errno.EEXIST: 

84 raise 

85 # cache has { 'key' : (obj, orig_mod_time) } 

86 self.cache = {} 

87 

88 

89 def __getitem__(self,key): 

90 """ db['key'] reading """ 

91 fil = self.root / key 

92 try: 

93 mtime = (fil.stat()[stat.ST_MTIME]) 

94 except OSError: 

95 raise KeyError(key) 

96 

97 if fil in self.cache and mtime == self.cache[fil][1]: 

98 return self.cache[fil][0] 

99 try: 

100 # The cached item has expired, need to read 

101 with fil.open("rb") as f: 

102 obj = pickle.loads(f.read()) 

103 except: 

104 raise KeyError(key) 

105 

106 self.cache[fil] = (obj,mtime) 

107 return obj 

108 

109 def __setitem__(self,key,value): 

110 """ db['key'] = 5 """ 

111 fil = self.root / key 

112 parent = fil.parent 

113 if parent and not parent.is_dir(): 

114 parent.mkdir(parents=True) 

115 # We specify protocol 2, so that we can mostly go between Python 2 

116 # and Python 3. We can upgrade to protocol 3 when Python 2 is obsolete. 

117 with fil.open('wb') as f: 

118 pickle.dump(value, f, protocol=2) 

119 try: 

120 self.cache[fil] = (value, fil.stat().st_mtime) 

121 except OSError as e: 

122 if e.errno != errno.ENOENT: 

123 raise 

124 

125 def hset(self, hashroot, key, value): 

126 """ hashed set """ 

127 hroot = self.root / hashroot 

128 if not hroot.is_dir(): 

129 hroot.mkdir() 

130 hfile = hroot / gethashfile(key) 

131 d = self.get(hfile, {}) 

132 d.update( {key : value}) 

133 self[hfile] = d 

134 

135 

136 

137 def hget(self, hashroot, key, default = _sentinel, fast_only = True): 

138 """ hashed get """ 

139 hroot = self.root / hashroot 

140 hfile = hroot / gethashfile(key) 

141 

142 d = self.get(hfile, _sentinel ) 

143 #print "got dict",d,"from",hfile 

144 if d is _sentinel: 

145 if fast_only: 

146 if default is _sentinel: 

147 raise KeyError(key) 

148 

149 return default 

150 

151 # slow mode ok, works even after hcompress() 

152 d = self.hdict(hashroot) 

153 

154 return d.get(key, default) 

155 

156 def hdict(self, hashroot): 

157 """ Get all data contained in hashed category 'hashroot' as dict """ 

158 hfiles = self.keys(hashroot + "/*") 

159 hfiles.sort() 

160 last = len(hfiles) and hfiles[-1] or '' 

161 if last.endswith('xx'): 

162 # print "using xx" 

163 hfiles = [last] + hfiles[:-1] 

164 

165 all = {} 

166 

167 for f in hfiles: 

168 # print "using",f 

169 try: 

170 all.update(self[f]) 

171 except KeyError: 

172 print("Corrupt",f,"deleted - hset is not threadsafe!") 

173 del self[f] 

174 

175 self.uncache(f) 

176 

177 return all 

178 

179 def hcompress(self, hashroot): 

180 """ Compress category 'hashroot', so hset is fast again 

181 

182 hget will fail if fast_only is True for compressed items (that were 

183 hset before hcompress). 

184 

185 """ 

186 hfiles = self.keys(hashroot + "/*") 

187 all = {} 

188 for f in hfiles: 

189 # print "using",f 

190 all.update(self[f]) 

191 self.uncache(f) 

192 

193 self[hashroot + '/xx'] = all 

194 for f in hfiles: 

195 p = self.root / f 

196 if p.name == 'xx': 

197 continue 

198 p.unlink() 

199 

200 

201 

202 def __delitem__(self,key): 

203 """ del db["key"] """ 

204 fil = self.root / key 

205 self.cache.pop(fil,None) 

206 try: 

207 fil.unlink() 

208 except OSError: 

209 # notfound and permission denied are ok - we 

210 # lost, the other process wins the conflict 

211 pass 

212 

213 def _normalized(self, p): 

214 """ Make a key suitable for user's eyes """ 

215 return str(p.relative_to(self.root)).replace('\\','/') 

216 

217 def keys(self, globpat = None): 

218 """ All keys in DB, or all keys matching a glob""" 

219 

220 if globpat is None: 

221 files = self.root.rglob('*') 

222 else: 

223 files = self.root.glob(globpat) 

224 return [self._normalized(p) for p in files if p.is_file()] 

225 

226 def __iter__(self): 

227 return iter(self.keys()) 

228 

229 def __len__(self): 

230 return len(self.keys()) 

231 

232 def uncache(self,*items): 

233 """ Removes all, or specified items from cache 

234 

235 Use this after reading a large amount of large objects 

236 to free up memory, when you won't be needing the objects 

237 for a while. 

238 

239 """ 

240 if not items: 

241 self.cache = {} 

242 for it in items: 

243 self.cache.pop(it,None) 

244 

245 def waitget(self,key, maxwaittime = 60 ): 

246 """ Wait (poll) for a key to get a value 

247 

248 Will wait for `maxwaittime` seconds before raising a KeyError. 

249 The call exits normally if the `key` field in db gets a value 

250 within the timeout period. 

251 

252 Use this for synchronizing different processes or for ensuring 

253 that an unfortunately timed "db['key'] = newvalue" operation 

254 in another process (which causes all 'get' operation to cause a 

255 KeyError for the duration of pickling) won't screw up your program 

256 logic. 

257 """ 

258 

259 wtimes = [0.2] * 3 + [0.5] * 2 + [1] 

260 tries = 0 

261 waited = 0 

262 while 1: 

263 try: 

264 val = self[key] 

265 return val 

266 except KeyError: 

267 pass 

268 

269 if waited > maxwaittime: 

270 raise KeyError(key) 

271 

272 time.sleep(wtimes[tries]) 

273 waited+=wtimes[tries] 

274 if tries < len(wtimes) -1: 

275 tries+=1 

276 

277 def getlink(self,folder): 

278 """ Get a convenient link for accessing items """ 

279 return PickleShareLink(self, folder) 

280 

281 def __repr__(self): 

282 return "PickleShareDB('%s')" % self.root 

283 

284 

285 

286class PickleShareLink: 

287 """ A shortdand for accessing nested PickleShare data conveniently. 

288 

289 Created through PickleShareDB.getlink(), example:: 

290 

291 lnk = db.getlink('myobjects/test') 

292 lnk.foo = 2 

293 lnk.bar = lnk.foo + 5 

294 

295 """ 

296 def __init__(self, db, keydir ): 

297 self.__dict__.update(locals()) 

298 

299 def __getattr__(self,key): 

300 return self.__dict__['db'][self.__dict__['keydir']+'/' + key] 

301 def __setattr__(self,key,val): 

302 self.db[self.keydir+'/' + key] = val 

303 def __repr__(self): 

304 db = self.__dict__['db'] 

305 keys = db.keys( self.__dict__['keydir'] +"/*") 

306 return "<PickleShareLink '%s': %s>" % ( 

307 self.__dict__['keydir'], 

308 ";".join([Path(k).basename() for k in keys])) 

309 

310def main(): 

311 import textwrap 

312 usage = textwrap.dedent("""\ 

313 pickleshare - manage PickleShare databases 

314 

315 Usage: 

316 

317 pickleshare dump /path/to/db > dump.txt 

318 pickleshare load /path/to/db < dump.txt 

319 pickleshare test /path/to/db 

320 """) 

321 DB = PickleShareDB 

322 import sys 

323 if len(sys.argv) < 2: 

324 print(usage) 

325 return 

326 

327 cmd = sys.argv[1] 

328 args = sys.argv[2:] 

329 if cmd == 'dump': 

330 if not args: args= ['.'] 

331 db = DB(args[0]) 

332 import pprint 

333 pprint.pprint(db.items()) 

334 elif cmd == 'load': 

335 cont = sys.stdin.read() 

336 db = DB(args[0]) 

337 data = eval(cont) 

338 db.clear() 

339 for k,v in db.items(): 

340 db[k] = v 

341 elif cmd == 'testwait': 

342 db = DB(args[0]) 

343 db.clear() 

344 print(db.waitget('250')) 

345 elif cmd == 'test': 

346 test() 

347 stress() 

348 

349if __name__== "__main__": 

350 main() 

351 

352