NAME

CSV::Table - Provides routines for querying and modifying a CSV file with or without a header row.

SYNOPSIS

For example, using an MxN row/column matrix for data plus a header row in a file with the first three lines being:

name, age, ...
John, 40,  ...
Sally, 38, ...
...

Handle the file with CSV::Table in a Raku program:

use CSV::Table;
# with indexing from zero
my $t = CSV::Table.new: :csv($my-csv-file);
say $t.fields;       # OUTPUT: «M␤»     # zero if no header row
say $t.rows;         # OUTPUT: «N-1␤»   # N if no header row
say $t.cols;         # OUTPUT: «M␤»
say $t.field[0];     # OUTPUT: «name␤»  # (Any) if no header row
say $t.cell[0][0];   # OUTPUT: «John␤»

There are multiple ways to query a data cell:

• by row and column

say $t.cell[1][0]    # OUTPUT: «Sally␤»
say $t.rowcol(1, 0); # OUTPUT: «Sally␤»
say $t.rc(1, 0);     # OUTPUT: «Sally␤»
say $t.ij(1, 0);     # OUTPUT: «Sally␤»

• by column and row

say $t.colrow(0, 1); # OUTPUT: «Sally␤»
say $t.cr(0, 1);     # OUTPUT: «Sally␤»
say $t.ji(0, 1);     # OUTPUT: «Sally␤»

You can change the value of any cell:

$t.cell[0][1] = 48;
$t.rowcol(0, 1, 50);

You can also change the names of fields, but, unless you also change the corresponding field names in the data hashes, you will most likely have problems. It would be much easier to modify the original CSV file.

You can choose to have the changed data ($t.save) any time, but you will be asked to confirm the save.

You can also save the data in a new file: $t.save: $stem. Where $stem is the desired basename without a suffix. The new files will have the '.csv' and '-raw.csv' names (or your desired 'raw' file string).

You can define a CSV table with row names using the '$has-row-names' option to the constructor and query the table by row name and column name.

say $t.rowcol("water", "Feb"); # OUTPUT: «10.80␤»

DESCRIPTION

CSV::Table is a class enabling access to a CSV table's contents. Tables with a header row must have unique field names. Input files are read immediately, so very large files may overwhelm system resources.

By default, text in a cell is 'normalized', that is, it is trimmed of leading and trailing whitespace and multiple contiguous interior whitespaces are collapsed into single space characters (' '). In this context, 'whitespace' is one or more characters in the set (" ", "\t", "\n"). Exceptions to that rule occur when the user wishes to use a newline in a cell or a tab is used as a cell separator. In those cases, some other character must be chosen as a line-ending or cell separator, and the newline is not considered to be a whitespace character for the normalization algorithm while any tab not used as a cell separator is treated as whitespace. (See more details and examples below.)

It can handle the following which other CSV handlers may not:

• With a header line

  • normalizing field names

  • data lines with fewer fields than a header (missing values assumed to be "" or the user can choose the default value in the constructor)

  • data lines with more fields than its header (fatal, but reported)

Note header lines with any empty fields causes an exception. This is a valid header line:

field0, field1, field2

This header line is not valid (notice the ending comma has no text following it):

field0, field1, field2,

• Without a header line

  • data lines are padded with empty fields to the maximum number of fields found in the file (or the user's chosen value)

• Either case

  • May use newlines in a cell if a different line ending is specified

As simple as it is, it also has some uncommon features that are very useful:

• Its slice method enables extraction of an M'xN' sub-array of the data cells

  The slice arguments are two range values defining the rows and columns zero-indexed cell indices of the desired sub-array.

• Comment lines are allowed

  This feature, which is not usual in CSV parsers, is to ignore comment lines interspersed between data lines (such lines may have leading whitespace). Data lines may also have inline comments following a comment character. The comment character is user-definable and its presence invalidates its use as a field separator. The default comment character is '#'. Its use demonstrated:

    # a comment preceding a data line
  1, 2, 3 # an inline comment following the data
   # ending
  # comments

  Note comments are preserved and restored when the CSV file is saved.

• Save

  There is a save method which saves the current state of the CSV file (including comments) as well as saving a "raw" CSV file without the comments so the file can be used with conventional CSV handlers such as LibreOffice or Excel. The method can also be used to change the names of the output files.

• Text normalization

  Its results are to normalize text in a field or cell, that is: leading and trailing whitespace is trimmed and interior whitespace is collapsed to one space character (' ') between words. This is the default behavior, but it can be turned off if desired (normalize=False). In that event, data in all fields are still trimmed of leading and trailing whitespace unless you also set trim=False). Some examples:

  Cell contents (line ending not a newline): " Sally \n Jean "; normalized: "Sally\nJean".

  Cell contents (tab not a cell separator): " Sally \t Jean "; normalized: "Sally Jean".

• Automatic determination of separator character

  Unless the field separator is selected otherwise, the default is 'auto' and the first data line is searched for the most-used separator character from this list: |, ;, , and \t. Other non-space characters may be used but are probably not tested. File an issue if you want to add a separator not currently specified.

• Row names

  Row names can be used if the "$has-row-names" option is entered with the constructor. Then you can get cell values by row and column name as shown in this example repeated from above:

  say $t.rowcol("water", "Feb"); # OUTPUT: «10.80␤»

  Note the names are case sensitive. A non-case-sensitive option may be provided in a future release.

Limitations

It cannot currently handle:

• special characters

• backslashes

• binary data

• duplicate field names in a header line

• duplicate row names

Also, quoted words are not specially treated nor are unbalanced quote characters detected.

Constructor with default options

my $t = CSV::Table.new: :$csv,
                        :has-header=True,
                        :separator='auto',
                        :normalize=True,
                        :trim=True,
                        :comment-char='#',
                        :line-ending="\n",
                        :empty-cell-value="",
                        :has-row-names=False,
                        :raw-ending="-raw",
                        :config
                        ;

Following are the allowable values for the named arguments. The user is cautioned that unspecified values are probably not tested. File an issue if your value of choice is not specified and it can be added and tested for.

There are a lot of options, one or all of which can be defined in a YAML (or JSON) configuration file whose path is provided by the config option. The user may get a prefilled YAML config file by executing:

$ raku -e'use CSV::Table; CSV::Table.write-config'
See CSV::Table JSON configuration file 'config-csv-table.yml'

or

$ raku -e'use CSV::Table; CSV::Table.write-config(:type<json>)'
See CSV::Table JSON configuration file 'config-csv-table.json'

Alternatively, you can call the method on a CSV::Table object in the REPL:

$ raku
> use CSV::Table;
> CSV::Table.write-config
See CSV::Table YAML configuration file 'config-csv-table.yml'

• :$separator

  • auto [default]

  • comma (,)

  • pipe (|)

  • semicolon (;)

  • tab (\t)

• :$normalize

  • True [default]

  • False

• :$trim

  • True [default]

  • False

• :$comment-char

  • # [default]

  • others, including multiple characters, are possible

• :$has-header

  • True [default]

  • False

• :$line-ending

  • "\n" [default]

  • String

• :$empty-cell-value

  • '' [default]

  • String

• :$has-row-names

  • False [default]

  • True

Accessing the table

The following table shows how to access each cell in a table $t with a header row plus R rows and C columns of data. (In matrix terminology it is an M x N rectangular matrix with M rows and N columns.)

The table's data cells can also be accessed by field name and row number:

$t.col{$field-name}{$row-number}

If row names are provided, data cells can be accessed by row and column names:

$t.rowcol: $row-name, $field-name

Possible new features

The following features can be implemented fairly easily if users want it and file an issue.

• non-case-sensitive row and column names

• add new rows or columns

• delete rows or columns

• row sum and average

• column sum and average

• normalize comments

Other matrix-related features could be implemented, but most are available in the published modules Math::Libgsl::Matrix and Math::Matrix.

CREDITS

Thanks to @lizmat and @[Coke] for pushing for a more robust CSV handling module including quotes and newlines.

Thanks to @librasteve for the idea of the slice method and his suggestion of aliases slice2d and view for slice.

AUTHOR

Tom Browder [email protected]

COPYRIGHT AND LICENSE

© 2024 Tom Browder

This library is free software; you may redistribute it or modify it under the Artistic License 2.0.