The London Perl and Raku Workshop takes place on 26th Oct 2024. If your company depends on Perl, please consider sponsoring and/or attending.


DataExtract::FixedWidth - The one stop shop for parsing static column width text tables!


        ## We assume the columns have no spaces in the header.
        my $de = DataExtract::FixedWidth->new({ header_row => $header_row });
        ## We explicitly tell what column names to pick out of the header.
        my $de = DataExtract::FixedWidth->new({
                header_row => $header_row
                cols       => [qw/COL1NAME COL2NAME COL3NAME/, 'COL WITH SPACE IN NAME']
        ## We supply data to heuristic and assume
        ## * first row is the header (to avoid this assumption
        ##   set the header_row to undef. )
        ## * heurisitic's unpack_string is correct
        ## * unpack_string applied to header_row will tell us the columns
        my $de = DataExtract::FixedWidth->new({ heuristic => \@datarows });
        ## We supply data to heuristic, say we have no header, and the set columns
        ## just like the above except ->parse_hash will be be indexed by the
        ## provided columns and no row is designated as the header.
        my $de = DataExtract::FixedWidth->new({
                heuristic    => \@datarows
                , header_row => undef
                , columns    => [qw/ foo bar baz/]
        ## We supply data to heuristic, and we explicitly add the header_row
        ## with this method it doesn't have to occur in the data.
        ## The unpack string rendered will be applied to the first row to get
        ## the columns
        my $de = DataExtract::FixedWidth->new({
                heuristic    => \@datarows
                , header_row => $header_row
        ## We explicitly add the header_row, with this method it doesn't have
        ## to occur in the data. The unpack string rendered will be applied
        ## to the provided header_row to get the columns
        my $de = DataExtract::FixedWidth->new({
                unpack_string => $template
                , header_row  => $header_row
        $de->parse( $data_row );
        $de->parse_hash( $data_row );


This module parses any type of fixed width table -- these types of tables are often outputed by ghostscript, printf() displays with string padding (i.e. %-20s %20s etc), and most screen capture mechanisms. This module is using Moose all methods can be specified in the constructor.

In the below example, this module can discern the column names from the header. Or, you can supply them explicitly in the constructor; or, you can supply the rows in an ArrayRef to heuristic and pray for the best luck. This module is pretty abstracted and will deduce what it doesn't know in a decent fashion if all of the information is not provided.


After you have constructed, you can ->parse which will return an ArrayRef $de->parse('FOOBARBAZ THIS IS TEXT ANOTHER COL');

Or, you can use ->parse_hash() which returns a HashRef of the data indexed by the column headers. They can be determined in many ways with the data you provide.


The class constructor, ->new, has numerious forms. Some options it has are:

heuristics => \@lines

This will deduce the unpack format string from data. If you opt to use this method, and need parse_hash, the first row of the heurisitic is assumed to be the header_row. The unpack_string that results for the heuristic is applied to the header_row to determine the columns.

cols => \@cols

This will permit you to explicitly list the columns in the header row. This is especially handy if you have spaces in the column header. This option will make the header_row mandatory.

header_row => $string

If a cols option is not provided the assumption is that there are no spaces in the column header. The module can take care of the rest. The only way this column can be avoided is if we deduce the header from heuristics, or if you explicitly supply the unpack string and only use ->parse($line). If you are not going to supply a header, and you do not want to waste the first line on a header assumption, set the header_row => undef in the constructor.

verbose => 1|0

Right now, it simply display's warnings when it does something that might at first seem awkward. Like returning undef when it encouters a duplicate copy of a header row.


An astrisk, (*) in the option means that is the default.

->parse( $data_line )

Parses the data and returns an ArrayRef

->parse_hash( $data_line )

Parses the data and returns a HashRef, indexed by the cols (headers)


This option forces the unpack string to make the first column assume the characters to the left of the header column. So, in the below example the first column also includes the first char of the row, even though the word stock begins at the second character.

        CHAR NUMBERS: |1|2|3|4|5|6|7|8|9|10
        HEADER ROW  : | |S|T|O|C|K| |V|I|N

Trim the whitespace for the elements that ->parse($line) outputs.


Fixes columns that bleed into other columns, move over all non-whitespace characters preceding the first whitespace of the next column. This does not work with heurisitic because the unpack string makes the assumption the data is not mangeled.

So if ColumnA as is 'foob' and ColumnB is 'ar Hello world'

* ColumnA becomes 'foobar', and ColumnB becomes 'Hello world'


Simply undef all elements that return length(element) = 0, requires ->trim_whitespace.


Skips duplicate copies of the header_row if found in the data.


Returns a HashRef that displays the results of each column header and relative character position the column starts at. In the case of heuristic this is a simple ordinal number. In the case of non-heuristic provided data it is currently a cardinal character position.


Returns the CORE::unpack() template string that will be used internally by ->parse($line)


Git repo at


Copyright 2008 Evan, all rights reserved.

This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.


        Evan Carroll <me at>
        System Lord of the Internets


Please report any bugs or feature requests to bug-dataexract-fixedwidth at, or through the web interface at I will be notified, and then you'll automatically be notified of progress on your bug as I make changes.