How can I insert a value into a numpy record array?

Question

This question has been edited as to make more sense.

The original question is how to insert values into a numpy record array, and I have had som success but still have an issue. Based off of the website below I have been inserting values into a record array.

Python code

instance_format={
    'names' : ('name','offset'),
    'formats' : ('U100','U30')}

instance=np.zeros(20,dtype=instance_format)

#I am placing values in the array similar to this
instance[0]['name']="Wire 1"
instance[1]['name']="Wire 2"
instance[2]['name']="Wire 3"


instance[0]['offset']="0x103"
instance[1]['offset']="0x104"
instance[2]['offset']="0x105"

#Here is the insertion statement that works
instance1 = np.insert(instance1,1,"Module one")

print(instance1)

Output

[('One Wire 1', '0x103')
 ('Module One',  'Module One')
 ('One Wire 2', '0x104')
 ('One Wire 3', '0x105')

So the insert statement works, however it inserts it both in the name and the offset field. I want to insert it just in the name field. How do I this?

Thanks

hpaulj · Accepted Answer · 2015-11-19 23:02:00Z

Your instance

In [470]: instance
Out[470]: 
array([('', '', ''), ('', '', ''), ('', '', ''), ('', '', ''),
       ('', '', ''), ('', '', ''), ('', '', ''), ('', '', ''),
       ('', '', ''), ('', '', ''), ('', '', ''), ('', '', ''),
       ('', '', ''), ('', '', ''), ('', '', ''), ('', '', ''),
       ('', '', ''), ('', '', ''), ('', '', ''), ('', '', '')], 
      dtype=[('name', '<U100'), ('module', '<U100'), ('offset', '<U30')])

does not look like

 ['One Wire Instance 1', 'One Wire Instance 2', 'One Wire Instance 3']

Are you talking about one record of instance, which would display as

 ('One Wire Instance 1', 'One Wire Instance 2', 'One Wire Instance 3')

with each string being the name, module, and offset.

Or are these 3 strings e.g. instance['name'][:3], the 'name' field from 3 records?

Inserting a new record into the instance array is one thing, adding a new field to the array is quite another.

To use np.insert with a structured array, you need provide a 1 element array with the correct dtype.

With your new instance:

In [580]: newone = np.array(("module one",'',''),dtype=instance.dtype)
In [581]: newone
Out[581]: 
array(('module one', '', ''), 
      dtype=[('name', '<U100'), ('module', '<U100'), ('offset', '<U30')])

In [582]: np.insert(instance,1,newone)
Out[582]: 
array([('Wire 1', '', '0x103'), ('module one', '', ''),
       ('Wire 2', '', '0x104'), ('Wire 3', '', '0x105')], 
      dtype=[('name', '<U100'), ('module', '<U100'), ('offset', '<U30')])

np.insert is just a function that performs these steps:

In [588]: instance2 = np.zeros((4,),dtype=instance.dtype)
In [589]: instance2[:1]=instance[:1]
In [590]: instance2[2:]=instance[1:3]
In [591]: instance2
Out[591]: 
array([('Wire 1', '', '0x103'), ('', '', ''), ('Wire 2', '', '0x104'),
       ('Wire 3', '', '0x105')], 
      dtype=[('name', '<U100'), ('module', '<U100'), ('offset', '<U30')])
In [592]: instance2[1]=newone
In [593]: instance2
Out[593]: 
array([('Wire 1', '', '0x103'), ('module one', '', ''),
       ('Wire 2', '', '0x104'), ('Wire 3', '', '0x105')], 
      dtype=[('name', '<U100'), ('module', '<U100'), ('offset', '<U30')])

It creates a new array of the correct target size, copies elements from the original array, and puts the new array into the empty slot.

Alex Yu · Accepted Answer · 2015-11-17 23:42:58Z

0

I can't understand what you mean by:

I want to insert the name "Reserved" in the second element which would make the array have the following contents ['One Wire Instance 1','Reserved' , 'One Wire Instance 2', 'One Wire Instance 3']

Do you want:

instance[1] = 'Reserved','', ''

?

answered Nov 17, 2015 at 23:42

Alex Yu

3,5871 gold badge27 silver badges42 bronze badges

1 Comment

GoldenEagle Over a year ago

Hi please check the edits, and since arrays are indexed at 0 it would be instance[1]= ' ','Reserved'

Collectives™ on Stack Overflow

How can I insert a value into a numpy record array?

2 Answers 2

1 Comment

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related