FortranFormat usage

Basic usage

>>> from fortranformat import FortranRecordWriter
>>> format = FortranRecordWriter('(3I10)')
>>> format.write([0, 1, None])
'         0         1         0'

>>> from fortranformat import FortranRecordReader
>>> format = FortranRecordReader('(3I10)')
>>> format.read('         0         1')
[0, 1, None]

It supports all the edit descriptors found in F77 as well as repeat formats.

Configuration

RET_WRITTEN_VARS_ONLY

Default: False

In FORTRAN you must first declare a bunch of variables to be filled when you read in a file. So if you read a text file that has 5 values but you allocate an array of length 10 then the array will be half filled. In Python we don't need to pre-allocate so we can just return the values that we read, by default though we read to the end of the record and pad the results with the None value. You can set this value to True to return just the written values.

To return FORTRAN default values rather than None see RET_UNWRITTEN_VARS_NONE.

Example

>>> from fortranformat import config
>>> format = FortranRecordReader('(3I10)')
>>> config.RET_WRITTEN_VARS_ONLY = False  # default
>>> format.read('         0         1')
[0, 1, None]
>>> config.RET_WRITTEN_VARS_ONLY = True
>>> format.read('         0         1')
[0, 1]

RET_UNWRITTEN_VARS_NONE

Default: True

In FORTRAN you must first declare a bunch of variables to be filled when you read in a file. If there are not enough values read in to fill all the variables then FORTRAN will preserve their default values (e.g. 0 for a float). Rather than returning an array of zero's (which is misleading) the library by default returns None instead. If you need the more precise FORTRAN behaviour then set this to False to return the default FORTRAN values (e.g. 0).

To only return written values see RET_WRITTEN_VARS_ONLY.

Example

>>> format = FortranRecordReader('(3I10)')
>>> config.RET_UNWRITTEN_VARS_NONE = True  # default
>>> format.read('         0         1')
[0, 1, None]
>>> config.RET_UNWRITTEN_VARS_NONE = False
>>> format.read('         0         1')
[0, 1]

G_INPUT_TRIAL_EDS

Default: ['F', 'L', 'A']

There are cases where using the G descriptor on input can be ambiguous (e.g. is it a string or a boolean?). FORTRAN is okay with this because the variable used to capture the value has a type which Python doesn't really do. We specify here a preferential ordering of edit descriptors to try until we find something that fits. By default it tries reading it as a number, then a boolean, then a string. If you don't want to interpret the input, for example, as a string then you can set this to ['F', 'L'] and it will raise an error otherwise.

RECORD_SEPARATOR

Default: '\n'

When wrapping the records, this string is used to delimit the lines.

Example

>>> config.RECORD_SEPARATOR = '|'
>>> FortranRecordWriter('(2I10)').write([0, 0, 0, 0])
'         0         0|         0         0'

PROC_MAXINT

Default: 2**31 (32bit signed)

This mimics the overflow behaviour of FORTRAN as well as the encoding of negative number which is twos complement. Set this to the size of the maximum positive integer value for your platform. If you don't want overflow behaviour then set this to None

Example

>>> format = FortranRecordWriter('(Z10)')
>>> format.write([-10])
'  FFFFFFF6'
>>> config.PROC_MAXINT = None
>>> format.write([-10])
'        -A'

reset()

Call this to reset the configuration to its defaults

>>> config.RECORD_SEPARATOR = '|'
>>> format = FortranRecordWriter('(2I10)')
>>> format.write([0, 0, 0, 0])
'         0         0|         0         0'
>>> config.reset()
>>> format.write([0, 0, 0, 0])
'         0         0\n         0         0'