def __init__(self,root):
""" Return a db object that will manage the specied directory"""
self.root = Path(root).expanduser().abspath()
if not self.root.isdir():
self.root.makedirs()
# cache has { 'key' : (obj, orig_mod_time) }
self.cache = {}
def __init__(self, root):
""" Return a db object that will manage the specied directory"""
self.root = Path(root).expanduser().abspath()
if not self.root.isdir():
self.root.makedirs()
# cache has { 'key' : (obj, orig_mod_time) }
self.cache = {}
class PickleShareDB(UserDict.DictMixin):
""" The main 'connection' object for PickleShare database """
def __init__(self,root):
""" Return a db object that will manage the specied directory"""
self.root = Path(root).expanduser().abspath()
if not self.root.isdir():
self.root.makedirs()
# cache has { 'key' : (obj, orig_mod_time) }
self.cache = {}
def __getitem__(self,key):
""" db['key'] reading """
fil = self.root / key
try:
mtime = (fil.stat()[stat.ST_MTIME])
except OSError:
raise KeyError(key)
if fil in self.cache and mtime == self.cache[fil][1]:
return self.cache[fil][0]
try:
# The cached item has expired, need to read
obj = pickle.load(fil.open())
except:
raise KeyError(key)
self.cache[fil] = (obj,mtime)
return obj
def __setitem__(self,key,value):
""" db['key'] = 5 """
fil = self.root / key
parent = fil.parent
if parent and not parent.isdir():
parent.makedirs()
pickled = pickle.dump(value,fil.open('w'))
try:
self.cache[fil] = (value,fil.mtime)
except OSError,e:
if e.errno != 2:
raise
class PickleShareDB(UserDict.DictMixin):
""" The main 'connection' object for PickleShare database """
def __init__(self, root):
""" Return a db object that will manage the specied directory"""
self.root = Path(root).expanduser().abspath()
if not self.root.isdir():
self.root.makedirs()
# cache has { 'key' : (obj, orig_mod_time) }
self.cache = {}
def __getitem__(self, key):
""" db['key'] reading """
fil = self.root / key
try:
mtime = (fil.stat()[stat.ST_MTIME])
except OSError:
raise KeyError(key)
if fil in self.cache and mtime == self.cache[fil][1]:
return self.cache[fil][0]
try:
# The cached item has expired, need to read
obj = pickle.load(fil.open())
except:
raise KeyError(key)
self.cache[fil] = (obj, mtime)
return obj
def __setitem__(self, key, value):
""" db['key'] = 5 """
fil = self.root / key
parent = fil.parent
if parent and not parent.isdir():
parent.makedirs()
pickled = pickle.dump(value, fil.open('w'))
try:
self.cache[fil] = (value, fil.mtime)
except OSError, e:
if e.errno != 2:
raise
class PickleShareDB(UserDict.DictMixin):
""" The main 'connection' object for PickleShare database """
def __init__(self,root):
""" Return a db object that will manage the specied directory"""
self.root = Path(root).expanduser().abspath()
if not self.root.isdir():
self.root.makedirs()
# cache has { 'key' : (obj, orig_mod_time) }
self.cache = {}
def __getitem__(self,key):
""" db['key'] reading """
fil = self.root / key
try:
mtime = (fil.stat()[stat.ST_MTIME])
except OSError:
raise KeyError(key)
if fil in self.cache and mtime == self.cache[fil][1]:
return self.cache[fil][0]
try:
# The cached item has expired, need to read
obj = pickle.load(fil.open())
except:
raise KeyError(key)
self.cache[fil] = (obj,mtime)
return obj
def __setitem__(self,key,value):
""" db['key'] = 5 """
fil = self.root / key
parent = fil.parent
if parent and not parent.isdir():
parent.makedirs()
pickled = pickle.dump(value,fil.open('w'))
try:
self.cache[fil] = (value,fil.mtime)
except OSError as e:
if e.errno != 2:
raise
def hset(self, hashroot, key, value):
""" hashed set """
hroot = self.root / hashroot
if not hroot.isdir():
hroot.makedirs()
hfile = hroot / gethashfile(key)
d = self.get(hfile, {})
d.update( {key : value})
self[hfile] = d
def hget(self, hashroot, key, default = _sentinel, fast_only = True):
""" hashed get """
hroot = self.root / hashroot
hfile = hroot / gethashfile(key)
d = self.get(hfile, _sentinel )
#print "got dict",d,"from",hfile
if d is _sentinel:
if fast_only:
if default is _sentinel:
raise KeyError(key)
return default
# slow mode ok, works even after hcompress()
d = self.hdict(hashroot)
return d.get(key, default)
def hdict(self, hashroot):
""" Get all data contained in hashed category 'hashroot' as dict """
hfiles = self.keys(hashroot + "/*")
hfiles.sort()
last = len(hfiles) and hfiles[-1] or ''
if last.endswith('xx'):
# print "using xx"
hfiles = [last] + hfiles[:-1]
all = {}
for f in hfiles:
# print "using",f
try:
all.update(self[f])
except KeyError:
print("Corrupt",f,"deleted - hset is not threadsafe!")
del self[f]
self.uncache(f)
return all
def hcompress(self, hashroot):
""" Compress category 'hashroot', so hset is fast again
hget will fail if fast_only is True for compressed items (that were
hset before hcompress).
"""
hfiles = self.keys(hashroot + "/*")
all = {}
for f in hfiles:
# print "using",f
all.update(self[f])
self.uncache(f)
self[hashroot + '/xx'] = all
for f in hfiles:
p = self.root / f
if p.basename() == 'xx':
continue
p.remove()
def __delitem__(self,key):
""" del db["key"] """
fil = self.root / key
self.cache.pop(fil,None)
try:
fil.remove()
except OSError:
# notfound and permission denied are ok - we
# lost, the other process wins the conflict
pass
def _normalized(self, p):
""" Make a key suitable for user's eyes """
return str(self.root.relpathto(p)).replace('\\','/')
def keys(self, globpat = None):
""" All keys in DB, or all keys matching a glob"""
if globpat is None:
files = self.root.walkfiles()
else:
files = [Path(p) for p in glob.glob(self.root/globpat)]
return [self._normalized(p) for p in files if p.isfile()]
def uncache(self,*items):
""" Removes all, or specified items from cache
Use this after reading a large amount of large objects
to free up memory, when you won't be needing the objects
for a while.
"""
if not items:
self.cache = {}
for it in items:
self.cache.pop(it,None)
def waitget(self,key, maxwaittime = 60 ):
""" Wait (poll) for a key to get a value
Will wait for `maxwaittime` seconds before raising a KeyError.
The call exits normally if the `key` field in db gets a value
within the timeout period.
Use this for synchronizing different processes or for ensuring
that an unfortunately timed "db['key'] = newvalue" operation
in another process (which causes all 'get' operation to cause a
KeyError for the duration of pickling) won't screw up your program
logic.
"""
wtimes = [0.2] * 3 + [0.5] * 2 + [1]
tries = 0
waited = 0
while 1:
try:
val = self[key]
return val
except KeyError:
pass
if waited > maxwaittime:
raise KeyError(key)
time.sleep(wtimes[tries])
waited+=wtimes[tries]
if tries < len(wtimes) -1:
tries+=1
def getlink(self,folder):
""" Get a convenient link for accessing items """
return PickleShareLink(self, folder)
def __repr__(self):
return "PickleShareDB('%s')" % self.root