def key(self, obj):
"""Return a cache key (relative path to file in cache) for an object"""
if isnumpy(obj):
# Key is byte view sha1 hash with .h5 extension
byteview = obj.view(numpy.uint8)
key = str(hashlib.sha1(byteview).hexdigest()) + '.h5'
elif isurl(obj):
# key is URL filename with an appended hash (for uniqueness)
p = urlparse.urlsplit(obj)
urlquery = urlparse.urlunsplit([p[0],p[1],p[2],p[3],None])
urlpath = urlparse.urlunsplit([p[0],p[1],p[2],None,None])
urlhash = self._hash(obj)
(filename, ext) = splitextension(path.basename(urlpath))
key = str(urlhash) + str(ext)
elif os.path.isfile(obj):
# within cache?
filebase = obj.split(self.root(),1)
if len(filebase) == 2:
# key is subpath within cache
key = filebase[1][1:]
else:
# key is filename with unique appended hash
(head, tail) = os.path.split(obj)
(filename, ext) = splitextension(tail)
namehash = hashlib.sha1(tail).hexdigest()
key = filename + '_' + str(namehash[0:7]) + ext
elif (path.isfile(self.abspath(obj)) or path.isdir(self.abspath(obj))):
key = obj # Already a cache key
elif isstring(obj):
key = obj # Use arbitrary string if not file or url
else:
raise CacheError('[bobo.cache][ERROR]: Unsupported object for constructing key')
return key
def put(self, obj, key=None, timeout=None, sha1=None):
"""Put a URI or numpy object into cache with the provided cache key"""
if key is None:
key = self.key(obj)
if self.iscached(key):
raise CacheError('[bobo.cache][Error]: Key collision! Existing object in cache with key "%s"' % key)
# Numpy object - export to file in cache with provided key
if isnumpy(obj):
quietprint('[bobo.cache][PUT]: Exporting numpy object to cache with key "' + key + '"')
f = h5py.File(self.abspath(key), 'a')
f[key] = obj
f.close()
# URL - download and save to cache with provided key
elif isurl(obj):
quietprint('[bobo.cache][PUT]: "%s" key "%s"' % (obj, key))
filename = self._download(obj, timeout=timeout)
shutil.move(filename, self.abspath(key))
# Unsupported type!
else:
raise CacheError('[bobo.cache][ERROR]: Unsupported object type for PUT')
# Return cache key
return key
