Array::2D - Methods for simple array-of-arrays data structures
This documentation refers to version 0.001_005
This is alpha software. Method names and behaviors are subject to change. The test suite has significant omissions.
use Array::2D; my $array2d = Array::2D->new( [ qw/a b c/ ] , [ qw/w x y/ ] ); # $array2d contains # a b c # w x y $array2d->push_col (qw/d z/); # a b c d # w x y z say $array2d->[0][1]; # prints "b"
Array::2D is a module that adds useful methods to Perl's standard array of arrays ("AoA") data structure, as described in Perl's perldsc documentation and Perl's perllol documentation. That is, an array that contains other arrays:
[ [ 1, 2, 3 ] , [ 4, 5, 6 ] , ]
This module provides methods for using that standard construction.
Most of the time, it's good practice to avoid having programs that use a module know about the internal construction of an object. However, this module is not like that. It assumes that the data structure is accessible outside the module's code, and may be altered by other code. The module will never change the data structure to include anything else. Therefore, it is perfectly reasonable to use the normal reference syntax to access items inside the array. A construction like $array2d->[0][1] for accessing a single element, or @{$array2d} to get the list of rows, is perfectly appropriate. This module exists because the reference-based implementation of multidimensional arrays in Perl makes it difficult to access, for example, a single column, or a two-dimensional slice, without writing lots of extra code.
$array2d->[0][1]
@{$array2d}
Array::2D uses "row" for the first dimension, and "column" or "col" for the second dimension. This does mean that the order of (row, column) is the opposite of the usual (x,y) algebraic order.
Because this object is just an array of arrays, most of the methods referring to rows are here mainly for completeness, and aren't much more useful than the native Perl construction (e.g., $array2d->last_row() is just a slower way of doing $#{$array2d}.) They will also typically be much slower.
$array2d->last_row()
$#{$array2d}
On the other hand, most of the methods referring to columns are useful, since there's no simple way of fetching a column or columns in Perl.
Because it is intended that the structure can be altered by standard Perl constructions, there is no guarantee that the object is either completely padded out so that every value within the structure's height and width has a value (undefined or not), alternatively completely pruned so that there are as few undefined values as possible. The only padding that must exist is padding to ensure that the row and column indexes are correct for all defined values.
Other Perl code could change the padding state at any time, or leave it in an intermediate state (where some padding exists, but the padding is not complete).
For example, the following would be valid:
$array2d = [ [ undef, 1, 2 ], 3 ], [ 4, 6, ], ];
The columns would be returned as (undef, 3, 4), (1, undef, 6), and (2).
There are methods to set padding -- the prune() method will eliminate padding, and the pad method will pad out the array to the highest row and column with a defined value.
prune()
pad
Methods that retrieve data will prune the data before returning it.
Methods that delete rows or columns (del_*, shift_*, pop_*, and in void context, slice) will prune not only the returned data but also the array itself.
Some general notes:
Except for constructor methods, all methods can be called as an object method on a blessed Array::2D object:
$array_obj->clone();
Or as a class method, if one supplies the array of arrays as the first argument:
Array::2D->clone($array);
In the latter case, the array of arrays need not be blessed (and will not be blessed by Array::2D).
In all cases where an array of arrays is specified as an argument (aoa_ref), this can be either an Array::2D object or a regular array of arrays data structure that is not an object.
Where rows are columns are removed from the array (as with any of the pop_*, shift_*, del_* methods), time-consuming assemblage of return values is ommitted in void context.
pop_*
shift_*
del_*
Some care is taken to ensure that rows are not autovivified. Normally, if the highest row in an arrayref-of-arrayrefs is 2, and a program attempts to read the value of $aoa->[3]->[$anything], Perl will create an empty third row. This module avoids autovification from just reading data. This is the only advantage of methods like element, row, etc. compared to regular Perl constructions.
element
row
It is assumed that row and column indexes passed to the methods are integers. If they are negative, they will count from the end instead of the beginning, as in regular Perl array subscripts. Specifying a negative index that is off the beginning of the array (e.g., specifying column -6 on an array whose width is 5) will cause an exception to be thrown. This is different than specifying an index is off the end of the array -- reading column #5 of a three-column array will return an empty column, and trying to write to tha column will pad out the intervening columns with undefined values.
The behavior of the module when anything other than an integer is passed in (strings, undef, floats, NaN, objects, etc.) is unspecified. Don't do that.
Returns a new Array::2D object. It accepts a list of array references as arguments, which become the rows of the object.
If it receives only one argument, and that argument is an array of arrays -- that is, a reference to an unblessed array, and in turn that array only contains references to unblessed arrays -- then the arrayrefs contained in that structure are made into the rows of a new Array::2D object.
If you want it to bless an existing arrayref-of-arrayrefs, use bless(). If you don't want to reuse the existing arrayrefs as the rows inside the object, use clone().
bless()
clone()
If you think it's possible that the detect-an-AoA-structure could give a false positive (you want a new object that might have only one row, where each entry in that row is an reference to an unblessed array), use Array::2D->bless ( [ @your_rows ] ).
Array::2D->bless ( [ @your_rows ] )
Just like new(), except that if passed a single arrayref which contains only other arrayrefs, it will bless the outer arrayref and return it. This saves the time and memory needed to copy the rows.
Note that this blesses the original array, so any other references to this data structure will become a reference to the object, too.
Returns a new, empty Array::2D object.
Takes a flat list and returns it as an Array::2D object, where each row has the number of elements specified. So, for example,
Array::2D->new_across (3, qw/a b c d e f g h i j/)
returns
[ [ a, b, c] , [ d, e, f] , [ g, h, i] , [ j ], ]
Takes a flat list and returns it as an Array::2D object, where each column has the number of elements specified. So, for example,
Array::2D->new_down (3, qw/a b c d e f g h i j/)
[ [ a, d, g, j ] , [ b, e, h ] , [ c, f, i ] , ]
A combination of new_down() and tabulate_equal_width(). Takes three named arguments:
new_down()
tabulate_equal_width()
A one-dimensional list of scalars.
A scalar to be passed to ->tabulate_equal_width(). The default is a single space.
The width of the terminal. If not specified, defaults to 80.
The method determines the number of text columns required, creates an Array::2D object of that number of text columns using new_down, and then returns first the object and then the results of ->tabulate_equal_width() on that object.
See Tabulating into Columnar Output below for information on how the widths of text in text columns are determined.
Returns a new object from a string containing tab-separated values. The string is first split into lines and then split into values by tabs.
Lines can be separated by by carriage returns, line feeds, a CR/LF pair, or other characters matching Perl's \R (see perlrebackslash).
If multiple strings are provided, they will be considered additional lines. So, if one has already read a TSV file, one can pass the entire contents, the series of lines in the TSV file, or a combination of two.
Note that this is not a routine that reads TSV files, just TSV strings, which may or may not have been read from a file. See new_from_file() for a method that reads TSV files (and other kinds).
new_from_file()
This method requires that Spreadsheet::ParseXLSX be installed on the local system.
Returns a new object from a worksheet in an Excel XLSX file, consisting of the rows and columns of that sheet. The sheet_requested parameter is passed directly to the ->worksheet method of Spreadsheet::ParseXLSX, which accepts a name or an index. If nothing is passed, it requests sheet 0 (the first sheet).
->worksheet
Spreadsheet::ParseXLSX
Returns a new object from a file on disk, specified as filespec.
If filetype is present, then it must be either 'xlsx' or 'tsv', and it will read the file assuming it is of that type.
If no filetype is present, it will attempt to use the file's extension to determine the proper filetype. Any file whose extension is '.xlsx' will be treated as type 'xlsx', and any file whose extension is either '.tab' or '.tsv' will be treated as type 'tsv'.
For the moment, it will also assume that a file whose extension is '.txt' is of type 'tsv'. It should be assumed that future versions may attempt to determine whether the file is more likely to be a comma-separated values file instead. To ensure that the file will be treated as tab-separated, pass in a filetype explicitly.
If the file type is 'xlsx', this method passes that file on to new_from_xlsx() and requests the first worksheet.
new_from_xlsx()
If the file type is 'tsv', it slurps the file in memory and passes the result to new_from_tsv. This uses File::Slurper, which mus be installed on the system.
new_from_tsv
Returns new object which has copies of the data in the 2D array object. The 2D array will be different, but if any of the elements of the 2D array are themselves references, they will refer to the same things as in the original 2D array.
Returns an unblessed array containing the same rows as the 2D array object. If called as a class method and given an argument that is already unblessed, will return the argument. Otherwise will create a new, unblessed array.
This is usually pointless, as Perl lets you ignore the object-ness of any object and access the data inside, but sometimes certain modules don't like to break object encapsulation, and this will allow getting around that.
Note that while modifying the elements inside the rows will modify the original 2D array, modifying the outer arrayref will not (unless that arrayref was not blessed in the first place). So:
my $unblessed = $array2d->unblessed; $unblessed->[0][0] = 'Up in the corner'; # modifies original object $unblessed->[0] = [ 'Up in the corner ' , 'Yup']; # does not modify original object
This can be confusing, so it's best to avoid modifying the result of unblessed. Use clone_unblessed instead.
unblessed
clone_unblessed
Returns a new, unblessed, array of arrays containing copies of the data in the 2D array object.
The array of arrays will be different, but if any of the elements of the 2D array are themselves references, they will refer to the same things as in the original 2D array.
Transposes the array: the elements that used to be in rows are now in columns, and vice versa.
In void context, alters the original. Otherwise, creates a new Array::2D object and returns that.
The result of transpose() is pruned.
Returns the array as a single, one-dimensional flat list of all the defined values. Note that it does not flatten any arrayrefs that are deep inside the 2D structure -- just the rows and columns of the structure itself.
Returns a true value if the array is empty, false otherwise.
Returns the number of rows in the array. The same as scalar @$array.
scalar @$array
Returns the number of columns in the array. (The number of elements in the longest row.)
Returns the index of the last row of the array. If the array is empty, returns -1. The same as $#{$array}.
$#{$array}
Returns the index of the last column of the array. (The index of the last element in the longest row.) If the array is empty, returns -1.
Returns the element in the given row and column. A slower way of saying $array2d->[row_idx][col_idx], except that it avoids autovivification. Like that construct, it will return undef if the element does not already exist.
$array2d->[row_idx][col_idx]
Returns the elements in the given row. A slower way of saying @{$array2d->[row_idx]}, except that it avoids autovivification.
@{$array2d->[row_idx]}
Returns the elements in the given column.
Returns a new Array::2D object with all the columns of the specified rows.
Note that duplicates are not de-duplicated, so the result of $obj->rows(1,1,1) will be three copies of the same row.
Returns a new Array::2D object with the specified columns. This is transposed from the original array's order, so each column requested will be in its own row.
$array = [ [ qw/ a b c d / ], [ qw/ j k l m / ], [ qw/ w x y z / ], ]; my $cols = Array::2D->cols($array, 1, 2); # $cols = bless [ [ qw/ b k x / ] , [ qw/ c l y / ] ], 'Array::2D';
Note that duplicates are not de-duplicated, so the result of $obj->cols(1,1,1) will retrieve three copies of the same column.
Returns a new Array::2D object with the specified columns of each row. Unlike cols(), the result of this method is not transposed.
cols()
$array = [ [ qw/ a b c d / ], [ qw/ j k l m / ], [ qw/ w x y z / ], ]; my $sliced_cols = Array::2D->slice_cols($array, 1, 2); # $sliced_cols = bless [ # [ qw/ b c / ] , # [ qw/ k l / ] , # [ qw/ x y / ] , # ], 'Array::2D';
Note that duplicates are not de-duplicated, so the result of $obj->slice_cols(1,1,1) will retrieve three copies of the same column.
Takes a two-dimensional slice of the array; like cutting a rectangle out of the array.
In void context, alters the original array, which then will contain only the area specified; otherwise, creates a new Array::2D object and returns the object.
Negative indicies are treated as though they mean that many from the end: the last item is -1, the second-to-last is -2, and so on.
Slices are always returned in the order of the original array, so $obj->slice(0,1,0,1) is the same as $obj->slice(1,0,1,0).
None of these methods return anything. At some point it might be worthwhile to have them return the old values of whatever they changed (when not called in void context), but they don't do that yet.
Sets the element in the given row and column to the given value. Just a slower way of saying $array2d->[row_idx][col_idx] = value.
$array2d->[row_idx][col_idx] = value
Sets the given row to the given set of values. A slower way of saying {$array2d->[row_idx] = [ @values ].
{$array2d->[row_idx] = [ @values ]
Sets the given column to the given set of values. If more values are given than there are rows, will add rows; if fewer values than there are rows, will set the entries in the remaining rows to undef.
undef
Sets the rows starting at the given start row index to the rows given. So, for example, $obj->set_rows(1, $row_ref_a, $row_ref_b) will set row 1 of the object to be the elements of $row_ref_a and row 2 to be the elements of $row_ref_b.
The arguments after start_row_idx are passed to new(), so it accepts any of the arguments that new() accepts.
new()
Returns the height of the array.
Sets the columns starting at the given start column index to the columns given. So, for example, $obj->set_cols(1, $col_ref_a, $col_ref_b) will set column 1 of the object to be the elemnents of $col_ref_a and column 2 to be the elements of $col_ref_b.
Sets a rectangular segment of the object to have the values of the supplied rows or array of arrays, beginning at the supplied first row and first column. The arguments after the row and columns are passed to new(), so it accepts any of the arguments that new() accepts.
All these methods return the new number of either rows or columns.
Adds the specified elements as a new row at the given index.
Adds the specified elements as a new column at the given index.
Takes the specified array of arrays and inserts them as new rows at the given index.
The arguments after the row index are passed to new(), so it accepts any of the arguments that new() accepts.
Takes the specified array of arrays and inserts them as new columns at the given index.
Adds the specified elements as the new first row.
Adds the specified elements as the new first column.
Takes the specified array of arrays and adds them as new rows before the beginning of the existing rows. Returns the new number of rows.
The arguments are passed to new(), so it accepts any of the arguments that new() accepts.
Takes the specified array of arrays and adds them as new columns, before the beginning of the existing columns. Returns the new number of columns.
Adds the specified elements as the new final row. Returns the new number of rows.
Adds the specified elements as the new final column. Returns the new number of columns.
Takes the specified array of arrays and adds them as new rows after the end of the existing rows. Returns the new number of rows.
Takes the specified array of arrays and adds them as new columns, after the end of the existing columns. Returns the new number of columns.
Removes the row of the object specified by the index and returns a list of the elements of that row.
Removes the column of the object specified by the index and returns a list of the elements of that column.
Removes the rows of the object specified by the indices. Returns an Array::2D object of those rows.
Removes the columns of the object specified by the indices. Returns an Array::2D object of those columns.
Removes the first row of the object and returns a list of the elements of that row.
Removes the first column of the object and returns a list of the elements of that column.
Removes the last row of the object and returns a list of the elements of that row.
Removes the last column of the object and returns a list of the elements of that column.
Padding, here, means empty values beyond the last defined value of each column or row. What counts as "empty" depends on the method being used.
Occasionally an array of arrays can end up with final rows or columns that are entirely undefined. For example:
my $obj = Array::2D->new ( [ qw/a b c/] , [ qw/f g h/ ]); $obj->[0][4] = 'e'; $obj->[3][0] = 'k'; # a b c undef e # f g h # (empty) # k $obj->pop_row(); $obj->pop_col(); # a b c undef # f g h # (empty)
That would yield an object with four columns, but in which the last column and last row (each with index 3) consists of only undefined values.
The prune method eliminates these entirely undefined or empty columns and rows at the end of the object.
prune
In void context, alters the original object. Otherwise, creates a new Array::2D object and returns the object.
Like prune, but treats not only undefined values as blank, but also empty strings.
Like prune, but treats not only undefined values as blank, but also strings that are empty or that consist solely of white space.
Like prune, but calls the <code_ref> for each element, setting $_ to each element. If the callback code returns true, the value is considered padding, and is removed if it's beyond the last non-padding value at the end of a column or row.
For example, this would prune values that were undefined, the empty string, or zero:
my $callback = sub { ! defined $_ or $_ eq q[] or $_ == 0; } $obj->prune_callback($callback);
Completely empty rows cannot be sent to the callback function, so those are always removed.
The opposite of prune(), this pads out the array so every column has the same number of elements. If provided, the added elements are given the value provided; otherwise, they are set to undef.
Each of these methods alters the original array in void context. If not in void context, creates a new Array::2D object and returns the object.
Calls the $code_ref for each element, aliasing $_ to each element in turn. This allows an operation to be performed on every element.
$code_ref
For example, this would lowercase every element in the array (assuming all values are defined):
$obj->apply(sub {lc});
If an entry in the array is undefined, it will still be passed to the callback.
For each invocation of the callback, @_ is set to the row and column indexes (0-based).
Removes white space, if present, from the beginning and end of each element in the array.
Removes white space from the end of each element in the array.
Replaces undefined values with the empty string.
Returns a hash reference. The values of the specified column of the array become the keys of the hash. The values of the hash are arrayrefs containing the elements of the rows of the array, with the value in the key column removed.
If the key column is not specified, the first column is used for the keys.
So:
$obj = Array::2D->new([qw/a 1 2/],[qw/b 3 4/]); $hashref = $obj->hash_of_rows(0); # $hashref = { a => [ '1' , '2' ] , b => [ '3' , '4' ] }
Like hash_of_rows, but accepts a key column and a value column, and the values are not whole rows but only single elements.
hash_of_rows
$obj = Array::2D->new([qw/a 1 2/],[qw/b 3 4/]); $hashref = $obj->hash_of_row_elements(0, 1); # $hashref = { a => '1' , b => '3' }
If neither key column nor value column are specified, column 0 will be used for the key and the column 1 will be used for the value.
If the key column is specified but the value column is not, then the first column that is not the key column will be used as the value column. (In other words, if the key column is column 0, then column 1 will be used as the value; otherwise column 0 will be used as the value.)
If the Unicode::GCString module can be loaded, its columns method will be used to determine the width of each character. This will treat composed accented characters and double-width Asian characters correctly.
columns
Otherwise, Array::2D will use Perl's length function.
length
Returns an arrayref of strings, where each string consists of the elements of each row, padded with enough spaces to ensure that each column has a consistent width.
The columns will be separated by whatever string is passed to tabulate(). If nothing is passed, a single space will be used.
tabulate()
So, for example,
$obj = Array::2D->new([qw/a bbb cc/],[qw/dddd e f/]); $arrayref = $obj->tabulate(); # $arrayref = [ 'a bbb cc' , # 'dddd e f' # ];
Completely empty columns and rows will be removed.
Like tabulate(), but instead of each column having its own width, all columns have the same width.
Like tabulate(), but returns the data as a single string, using line feeds as separators of rows, suitable for sending to a terminal.
Returns a list of strings in list context, or an arrayref of strings in scalar context. The elements of each row are present in the string, separated by tab characters.
If there are any arguments, they will be used first as the first row of text. The idea is that these will be the headers of the columns. It's not really any different than putting the column headers as the first element of the data, but frequently these are stored separately. If there is only one element and it is a reference to an array, that array will be used as the first row of text.
If tabs are present in any element, they will be replaced by the Unicode Replacement Character, U+FFFD.
Returns a single string with the elements of each row delimited by tabs, and rows delimited by line feeds.
If tabs or line feeds are present in any element, they will be replaced by the Unicode Replacement Character, U+FFFD.
Accepts a file specification and creates a new file at that location containing the data in the 2D array.
This method uses named parameters.
This parameter is the file's type. Currently, the types recognized are 'tsv' for tab-separated values, and 'xlsx' for Excel XLSX. If the type is not given, it attempts to determine the type from the file extension, which can be (case-insensitively) 'xlsx' for Excel XLSX files or 'tab', 'tsv' or 'txt' for tab-separated value files.
(If other text file formats are someday added, either they will have to have different extensions, or an explicit type must be passed to force that type to have a ".txt" extension.
This mandatory parameter contains the file specification.
This parameter is optional. If present, it contains an array reference to be used as the first row in the ouptut file.
The idea is that these will be the headers of the columns. It's not really any different than putting the column headers as the first element of the data, but frequently these are stored separately.
Accepts a file specification and creates a new Excel XLSX file at that location, with one sheet, containing the data in the 2D array.
This parameter is optional. If present, it contains an array reference to be used as the first row in the Excel file.
The idea is that these will be the headers of the columns. It's not really any different than putting the column headers as the first element of the data, but frequently these are stored separately. At this point no attempt is made to make them bold or anything like that.
This parameter is optional. If present, it contains a hash reference, with format parameters as specified by Excel::Writer::XLSX.
A non-arrayref, or blessed object (other than an Array::2D object), was passed to the new constructor.
Spreadsheet::ParseExcel returned an error indicating that the sheet requested was not found.
A file other than an Excel (XLSX) or tab-delimited text files (with tab, tsv, or txt extensions) are recognized in ->new_from_file.
No filename, or a blank filename, was passed to these methods.
A method was called that requires an array, but there was no array passed in the argument list. Typically this would be when it was called as a class object, e.g., $class->set_row(qw/a b c/);
An invalid character for TSV data was found in the array when creating TSV data. It was replaced with the Unicode REPLACEMENT CHARACTER (U+FFFD).
This is just a list of things that would be nice -- there's no current plan to implement these.
splice_row() and splice_col()
Alternatives to the methods that result in padded rather than pruned data.
CSV, JSON, maybe other file types in new_from_file() and file().
file()
Some other modules that have some similarities include:
These all appear much more powerful than Array::2D and may better suit your use.
Another relevant module is Data::ShowTable, which creates output formats such as ASCII art boxes and HTML.
The last three are required only by those methods that use them (new_from_tsv(), new_from_xlsx(), and xlsx() respectively).
new_from_tsv()
xlsx()
Aaron Priven <apriven@actransit.org>
Copyright 2015, 2017
This program is free software; you can redistribute it and/or modify it under the terms of either:
the GNU General Public License as published by the Free Software Foundation; either version 1, or (at your option) any later version, or
the Artistic License version 2.0.
This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
To install Array::2D, copy and paste the appropriate command in to your terminal.
cpanm
cpanm Array::2D
CPAN shell
perl -MCPAN -e shell install Array::2D
For more information on module installation, please visit the detailed CPAN module installation guide.