import numpy as np
from scipy import ndimage as nd
[docs]def isiterable(obj):
try:
iter(obj)
return True
except:
return False
[docs]class array_dict:
def __init__(self, values= np.array([]), keys = None):
if isinstance(values,array_dict):
from copy import deepcopy
self._values = deepcopy(values._values)
self._keys = deepcopy(values._keys)
self._index = deepcopy(values._index)
elif isinstance(values,dict):
self._values = np.array(list(values.values()))
self._keys = np.array(list(values.keys()))
#self._index = np.array(nd.sum(np.arange(len(values)),self._keys,index=np.arange(max(self._keys)+1)),int)
self._index = np.zeros(self._keys.max()+1,int)
self._index[self._keys] = np.arange(len(self._keys))
elif isinstance(values,list) and isinstance(values[0],tuple):
self._values = np.array(values)[:,1]
self._keys = np.array(np.array(values)[:,0],int)
#self._index = np.array(nd.sum(np.arange(len(values)),self._keys,index=np.arange(max(self._keys)+1)),int)
self._index = np.zeros(self._keys.max()+1,int)
self._index[self._keys] = np.arange(len(self._keys))
else:
self._values = np.array(list(values))
if not keys is None:
assert len(keys) == len(values)
self._keys = np.array(list(keys))
#self._index = np.array(nd.sum(np.arange(len(values)),self._keys,index=np.arange(max(self._keys)+1)),int)
self._index = np.zeros(self._keys.max()+1,int)
self._index[self._keys] = np.arange(len(self._keys))
else:
if len(values)>0:
self._keys = np.arange(len(values))
self._index = np.arange(len(values))
else:
self._keys = np.array([],int)
self._index = np.array([],int)
def __getitem__(self,key):
if self._index[key] > 0:
return self._values[self._index[key]]
elif key in self._keys:
return self._values[0]
else:
raise KeyError(str(key))
def __setitem__(self,key, value):
import numpy as np
if key in self._keys:
self._values[self._index[key]] = value
else:
newindex = len(self._values)
self._keys = np.append(self._keys,np.array([key]),axis=0)
if key<len(self._index):
self._index[key] = newindex
else:
new_index = np.zeros(key+1,int)
new_index[0:len(self._index)] = self._index
new_index[key] = newindex
self._index = new_index
if len(self) == 0:
self._values = np.array([value])
else:
self._values = np.append(self._values,np.array([value]),axis=0)
def __delitem__(self, key):
index = self._index[key]
self._values = np.delete(self._values,index,axis=0)
self._index[self._keys[index:]] -= 1
self._keys = np.delete(self._keys,index,axis=0)
self._index[key] = 0
def __len__(self):
return len(self._values)
def __str__(self):
self.dict_string = "{"
for i,k in enumerate(self._keys):
self.dict_string = self.dict_string+str(k)+": "+str(self._values[self._index[k]])
if i < self._values.size-1 :
self.dict_string = self.dict_string+", "
self.dict_string = self.dict_string+"}\n"
return self.dict_string
def __repr__(self):
self.dict_string = "{"
if len(self)<10:
for i,k in enumerate(self._keys):
self.dict_string = self.dict_string+str(k)+": "+str(self._values[i])
if i < self._values.size-1 :
self.dict_string = self.dict_string+", "
else:
for i in range(3):
self.dict_string = self.dict_string+str(self._keys[i])+": "+str(self._values[i])+","
self.dict_string = self.dict_string+"..."
for i in range(3):
self.dict_string = self.dict_string+","+str(self._keys[len(self)-3+i])+": "+str(self._values[len(self)-3+i])
self.dict_string = self.dict_string+"}\n"
return self.dict_string
[docs] def to_dict(self):
return dict([(tuple(k),self._values[i]) if isiterable(k) else (k,self._values[i]) for i,k in enumerate(self._keys)])
[docs] def values(self,keys=None):
if keys is None:
return self._values[self._index[self._keys]]
else:
if not isinstance(keys,np.ndarray):
keys = np.array(keys)
if keys.dtype == np.dtype('O'):
return np.array([self.values(k) for k in keys])
else:
return self._values[self._index[keys]]
[docs] def keys(self) :
return self._keys
[docs] def items(self):
return [(k,self._values[i]) for i,k in enumerate(self._keys)]
[docs] def update(self,values,keys=None,ignore_missing_keys=True,erase_missing_keys=True):
if keys is None or (len(keys) == len(self._keys) and (keys==self.keys()).all()):
assert len(self) == len(values)
self._values = values
else:
assert len(keys) == len(values)
keys_to_assign = np.intersect1d(keys,self._keys)
if len(keys_to_assign)!=len(keys):
key_range = np.array(nd.sum(np.arange(keys.size),keys,index=keys_to_assign),int)
self._values[self._index[keys_to_assign]] = values[key_range]
if ignore_missing_keys:
print("Warning : some keys were missing from dictionary! (values ignored)")
else:
print("Warning : missing keys were added to the dictionary!")
missed_keys = np.array(tuple(set(keys).difference(set(self._keys))))
newindex = np.arange(len(self._values),len(self._values)+len(missed_keys))
self._keys = np.append(self._keys,missed_keys,axis=0)
if missed_keys.max()<len(self._index):
self._index[missed_keys] = newindex
else:
new_index = np.zeros(missed_keys.max()+1,int)
new_index[0:len(self._index)] = self._index
new_index[missed_keys] = newindex
self._index = new_index
key_range = np.array(nd.sum(np.arange(keys.size),keys,index=missed_keys),int)
if len(self) == 0:
self._values = np.array(values[key_range])
else:
self._values = np.append(self._values,values[key_range],axis=0)
else:
if len(keys_to_assign)!=len(self._keys) or (keys!=self.keys()).any():
if erase_missing_keys:
print("Warning : missing keys were erased from the dictionary!")
self._values = values
self._keys = np.array(keys)
self._index = np.zeros(self._keys.max()+1,int)
self._index[self._keys] = np.arange(len(self._keys))
else:
self._values[self._index[keys]] = values
else:
self._values[self._index[keys]] = values
[docs] def delete(self,keys_to_delete):
keys_to_delete = np.intersect1d(keys_to_delete,self._keys)
if len(keys_to_delete)>0:
indices_to_delete = self._index[keys_to_delete]
self._values = np.delete(self._values,indices_to_delete,0)
self._keys = np.delete(self._keys,indices_to_delete,0)
self._index[keys_to_delete] = 0
self._index[self._keys] = np.arange(len(self._keys))
[docs] def has_key(self, key):
return key in self._keys
# def values(self):
# return self.to_dict().values()
# def keys(self):
# return self.to_dict().keys()
# def items(self):
# return self.to_dict().items()
[docs] def keys_where(self,criterion):
if isinstance(criterion,str):
return self._keys[np.where(eval('self._values'+criterion))]
elif isinstance(criterion,tuple) & isinstance(criterion[0],str):
expression = '(self._values'+criterion[0]+')'
for c in criterion[1:]:
expression += '& (self._values'+c+')'
return self._keys()[np.where(eval(expression))]
else:
assert len(criterion) == len(self._keys)
return self._keys[np.where(criterion)]
