Projects/laydi
Projects
/
laydi
Archived
7
0
Fork 0
This repository has been archived on 2024-07-04. You can view files and clone it, but cannot push or open issues or pull requests.
laydi/system/dataset.py

254 lines
8.3 KiB
Python

from system import logger
from scipy import atleast_2d,asarray,ArrayType
class Dataset:
"""The Dataset base class.
A Dataset is an n-way array with defined string identifiers across
all dimensions.
example of use:
---
dim_name_rows = 'rows'
names_rows = ('row_a','row_b')
ids_1 = [dim_name_rows, names_rows]
dim_name_cols = 'cols'
names_cols = ('col_a','col_b','col_c','col_d')
ids_2 = [dim_name_cols, names_cols]
Array_X = rand(2,4)
data = Dataset(Array_X,(ids_1,ids_2),name="Testing")
dim_names = [dim for dim in data]
column_identifiers = [id for id in data['cols'].keys()]
column_index = [index for index in data['cols'].values()]
'cols' in data -> True
---
data = Dataset(rand(10,20)) (generates dims and ids (no links))
"""
def __init__(self,array=None,identifiers=None,shape=None,all_dims=[],**kwds):
self._name = kwds.get("name","Unnamed data")
self._dims = [] #existing dimensions in this dataset
self._map = {} # internal mapping for dataset: identifier <--> index
self.has_array = False
self.shape = None
if array==None:
if shape == None:
raise ValueError, "Must define shape if array is None"
else:
self.shape = shape
if identifiers!=None:
self._set_identifiers(identifiers,all_dims)
else:
ids = self._create_identifiers(shape,all_dims)
self._set_identifiers(ids,all_dims)
elif isinstance(array,ArrayType):
array = atleast_2d(asarray(array))
self.shape = array.shape
if shape != None:
if self.shape!=shape:
#logger.log("debug","Dataset and input shape mismatch")
raise ValueError
if identifiers!=None:
self._set_identifiers(identifiers,all_dims)
else:
ids = self._create_identifiers(self.shape,all_dims)
self._set_identifiers(ids,all_dims)
self._array = array
self.has_array = True
else:
raise ValueError, "array input must be of ArrayType or None"
self._all_dims = all_dims
def __str__self(self):
return self._name
def __iter__(self):
"""Returns an iterator over dimensions of dataset."""
return self._dims.__iter__()
def __contains__(self,dim):
"""Returns True if dim is a dimension name in dataset."""
# return self._dims.__contains__(dim)
return self._map.__contains__(dim)
def __len__(self):
"""Returns the number of dimensions in the dataset"""
return len(self._map)
def __getitem__(self,dim):
"""Return the identifers along the dimension dim."""
return self._map[dim]
def _create_identifiers(self,shape,all_dims):
"""Creates dimension names and identifier names, and returns
identifiers."""
dim_names = ['rows','cols']
ids = []
for axis,n in enumerate(shape):
if axis<2:
dim_suggestion = dim_names[axis]
else:
dim_suggestion = 'dim'
while dim_suggestion in all_dims:
dim_suggestion = self._suggest_dim_name(dim_suggestion,all_dims)
identifier_creation = [str(axis) + "_" + i for i in map(str,range(n))]
ids.append((dim_suggestion,identifier_creation))
all_dims.append(dim_suggestion)
return ids
def _set_identifiers(self,identifiers,all_dims):
"""Creates internal mapping of identifiers structure."""
for dim,ids in identifiers:
pos_map={}
if dim not in self._dims:
self._dims.append(dim)
all_dims.append(dim)
else:
raise ValueError, "Dimension names must be unique"
for pos,id in enumerate(ids):
pos_map[id] = pos
self._map[dim] = pos_map
shape_chk = [len(i) for j,i in identifiers]
if shape_chk != list(self.shape):
raise ValueError, "Shape input: %s and array: %s mismatch" %(shape_chk,self.shape)
def _suggest_dim_name(self,dim_name,all_dims):
"""Suggests a unique name for dim and returns it"""
c = 0
while dim_name in all_dims:
dim_name = dim_name + "_" + str(c)
c+=1
return dim_name
def asarray(self):
"""Returns the numeric array (data) of dataset"""
if not self.has_array:
raise ValueError, "Dataset is empty"
else:
return self._array
def add_array(self,array):
"""Adds array as an ArrayType object.
A one-dim array is transformed to a two-dim array (row-vector)
"""
if self.has_array:
raise ValueError, "Dataset has array"
else:
if (len(self._map)!=len(array.shape)):
raise ValueError, "range(array_dims) and range(dim_names) mismatch"
if self.shape!=array.shape:
raise ValueError, "Input array must be of similar dimensions as dataset"
self._array = atleast_2d(asarray(array))
self.has_array = True
def get_name(self):
"""Returns dataset name"""
return self._name
def get_matrix(self):
"""Returns internal numeric matrix for dataset."""
return self._array
def get_all_dims(self):
"""Returns all dimensions in project"""
return self._all_dims
def get_dim_names(self):
"""Returns dim names"""
return [dim for dim in self._dims]
def get_identifiers(self, dim, indices=None):
"""Returns identifiers along dim, sorted by position (index).
You can optionally provide a list of indices to get only the
identifiers of a given position.
"""
items = self._map[dim].items()
backitems=[ [v[1],v[0]] for v in items]
backitems.sort()
sorted_ids=[ backitems[i][1] for i in range(0,len(backitems))]
# we use id as scipy-arrays return a new array on boolean
# operations
if id(indices) != id(None):
return [sorted_ids[index] for index in indices]
else:
return sorted_ids
def get_indices(self, dim, idents):
"""Get indices for identifiers along dimension."""
reverse = {}
for key, value in self._map[dim].items():
reverse[value] = key
return [self._map[dim][key] for key in idents]
class CategoryDataset(Dataset):
"""The category dataset class.
A dataset for representing class information as binary
matrices (0/1-matrices).
There is support for using a less memory demanding, and
fast intersection look-ups by representing the binary matrix as a
dictionary in each dimension.
"""
def __init__(self):
Dataset.__init__(self)
self.has_collection = False
def as_array(self):
"""Returns data as binary matrix"""
if not self.has_array and self.has_collection:
#build numeric array
pass
def as_collection(self,dim):
"""Returns data as collection along dim"""
pass
def add_collection(self,input_dict):
"""Adds a category data as collection.
A collection is a datastructure that contains a dictionary for
each pair of dimension in dataset, keyed by identifiers and
values is a set of identifiers in the other dimension
"""
#build category data as double dicts
pass
class GraphDataset(Dataset):
"""The graph dataset class.
A dataset class for representing graphs using an adjacency matrix
(aka. restricted to square symmetric signed integers matrices)
If the library NetworkX is installed, there is support for
representing the graph as a NetworkX.Graph, or NetworkX.XGraph structure.
"""
def __init__(self):
Dataset.__init(self)
self.has_graph = False
class Selection:
"""Handles selected identifiers along each dimension of a dataset"""
def __init__(self):
self.current_selection={}