Table Of Contents

Previous topic

Constructing a table

Next topic

Modifying tables

This Page

Accessing Table Data

Accessing the data

The table data is stored in a NumPy structured array, which can be accessed by passing the column name a key. This returns the column in question as a NumPy array:


For convenience, columns with names that satisfy the python variable name requirements (essentially starting with a letter and containing no symbols apart from underscores) can be accessed directly as attributes of the table:


Since the returned data is a NumPy array, individual elements can be accessed using:




Both notations can be used to set data in the table, for example:

t.column_name[row_number] = 1


t['column_name'][row_number] = 1

are equivalent, and will set the element at row_number to 1

Accessing the metadata

The column metadata is stored in the columns attribute. To see an overview of the metadata, simply use:

>>> t.columns

The metadata for a specific column can then be accessed by specifying the column name as a key:

>>> t.columns['some_column']

or using the column number:

>>> t.columns[column_number]

The attributes of a column object are dtype, unit, description, null, and format.


While the unit, description and format for a column can be modified using the columns attribute, the dtype and null values should not be modified in this way as the changes will not propagate to the data array.

It is also possible to view a description of the table by using the describe method of the Table instance:

>>> t.describe()

In addition to the column metadata, the comments and keywords are available via the keywords and comments attributes of the Table instance, for example:

>>> instrument = t.keywords['instrument']

The keywords attribute is a dictionary, and the comments attribute is a list.

Accessing table rows

The row(...) method can be used to access a specific row in a table:

>>> row = t.row(row_number)

This returns the row as a NumPy record. The row can instead be returned as a tuple of elements with Python types, by using the python_types argument:

>>> row = t.row(row_number, python_types=True)

Two more powerful methods are available: rows and where. The rows method can be used to retrieve specific rows from a table as a new Table instance:

>>> t_new = t.rows([1,3,5,2,7,8])

Alternatively, the where method can be given a boolean array to determine which rows should be selected. This is in fact very powerful as the boolean array can actually be written as selection conditions:

>>> t_new = t.where(( > 10) & (t.ra < 45.4) & (t.flag == 'ok'))

Global Table properties

One can access the number of rows in a table by using the python len function:

>>> len(t)

In addition, the number of rows and columns can also be accessed with the shape attribute:

>>> t.shape

where the first number is the number of rows, and the second is the number of columns (note that a vector column counts as a single column).