NAME

CSV-Autoclass - Define a class with a CSV file and provide data for a list of class objects in the same file

SYNOPSIS

use CSV-Autoclass;

DESCRIPTION

CSV-Autoclass is a module with two accompanying programs. For each program, execute it without any arguments to see instructions.

• csv2class

  Converts a suitably-formatted CSV file into a class-generator module.

  For instance, given a CSV file named persons.csv, the program, by default, will generate module Person.rakumod which can be used by another included program, use-class, to demonstrate using the module.

  Note the convention is to expect the base CSV file name to be constructed of a lower-case, plural name, using only ASCII letters 'a..z' plus the suffx '.csv'. The resulting class name will be a capitalized and plural version of the base file's stem.

  Alternatively, the user can specify another class name by entering class=MyClassName as an argument to csv2class.

• use-class

  Uses a CSV file with class data to list all the data. It uses CSV-Autoclass routines provided for interrogating any suitable list of CSV-Autoclass-defined objects. It can be used as a template for creating programs that manipulate any CSV file with the same header attributes as one of the generated classes.

  Its help option has more details about its usage.

Notes

The header line in the CSV data file is currently designed to use alphanumeric characters which works fine with files designed by the user. However, files produced by outside entities, such as banks, stock markets, and government agencies, may use other symbols (such as '#') that cannot be used in Raku for class attribute names. In such cases, the names will be transformed into approximations which may include the zero-index number of the field's position in the header line.

One known example is the header line in the transactions files of the Hancock-Whitney Bank. That transaction file header is shown below.

Date,Check#,Transaction Type,Description,Debits (-),Credits(+)

It is transformed into this header for attribute naming:

Date,Check,TransactionType,Description,Debit,Credit

On the other hand, Synovus Bank has the following header for its transactions CSV file. The field names work fine after down-casing them:

Date,Account,Description,Category,Check,Credit,Debit

In both cases the class attribute names are clear and objects created from the CSV files should be good translations of the attribute values as strings. However, the meaning of the columns may not be obvious, nor may the transactions be unique if the files are concatenated erroneously by the user in processing the downloads.

Also note both banks have their transactions temporally ordered with the most recent one on top. Given that, the user is cautioned about rewriting those files for whatever reason. One suggestion is to add an index number column and start the first transaction in each month with one and increment by one for succeeding transactions

Possible improvements

• In like manner to module App::Mi6, add an INI file to the user's $*HOME directory to be used for defining translations for CSV file header field names. Such translations could be modified by the user if the user wishes to improve the transformation.

• Use a database for storage instead of a CSV file.

TODO

• Make field name translations work.

• Add extensive tests.

AUTHOR

Tom Browder tbrowder@acm.org

COPYRIGHT AND LICENSE

© 2022 Tom Browder

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