name

Required. Used for the hashref key if you pull records as hashes, and used in diagnostic messages.

header

A string or regex describing the column header you want to find in the spreadsheet. If you specify a regex, it is used directly. If you specify a string, it becomes the regex matching any string with the same words (\w+) and non-whitespace (\S+) characters in the same order, case insensitive, surrounded by any amount of non-alphanumeric garbage ("[\W_]*"). When no header is specified, the "name" is used as a string after first breaking it into words on underscore or camel-case or numeric boundaries.

This deserves some examples:

  Name           Implied Default Header
  "zipcode"      "zipcode"
  "ZipCode"      "Zip Code"
  "Zip_Code"     "zip Code"
  "zip5"         "zip 5"
  
  Header         Regex                                  Could Match...
  "ZipCode"      /^[\W_]*ZipCode[\W_]*$/i               "zipcode:"
  "zip_code"     /^[\W_]*zip_code[\W_]*$/i              "--ZIP_CODE--"
  "zip code"     /^[\W_]*zip[\W_]*code[\W_]*$/i         "ZIP\nCODE    "
  "zip-code"     /^[\W_]*zip[\W_]*-[\W_]*code[\W_]*$/i  "ZIP-CODE:"
  qr/Zip.*Code/  /Zip.*Code/                            "Post(Zip)Code"

If this default matching doesn't meet your needs or paranoia level, then you should always specify your own header regexes.

(If your data actually doesn't have any header at all and you want to brazenly assume the columns match the fields, see reader attribute "header_row_at" in Data::TableReader)

required

Whether or not this field must be found in order to detect a table. Defaults is true. Note this does not require the field of a row to contain data in order to read a record from the table; it just requires a column to exist.

trim

  # remove leading/trailing whitespace
  trim => 1
  
  # remove leading/trailing whitespace but also remove "N/A" and "NULL"
  trim => qr( ^ \s* N/A \s* $ | ^ \s* NULL \s* $ | ^ \s+ | \s+ $ )xi
  
  # custom search/replace in a coderef
  trim => sub { s/[\0-\1F\7F]+/ /g; s/^\s+//; s/\s+$//; };

If set to a non-reference, this is treated as a boolean of whether to remove leading and trailing whitespace. If set to a coderef, the coderef will be called for each value with $_ set to the current value; it should modify $_ as appropriate (return value is ignored). It can also be set to a regular expression of all the patterns to remove, as per "s/$regexp//g".

Default is 1, which is equivalent to a regular expression of "qr/(^\s+)|(\s+$)/".

blank

The value to extract when the spreadsheet cell is an empty string or undef. (after any processing done by "trim") Default is "undef". Another common value would be "".

type

A Type::Tiny type (or any object or class with a "validate" method) or a coderef which returns a validation error message (undef if it is valid).

  use Types::Standard;
  ...
     type => Maybe[Int]
  
  # or without Type::Tiny
     type => sub { $_[0] =~ /^\w+/? undef : "word-characters only" },

This is an optional feature and there is no default. The behavior of a validation failure depends on the options to TableReader.

array

Boolean of whether this field can be found multiple times in one table. Default is false. If true, the value of the field will always be an arrayref (even if only one column matched).

follows

Name (or arrayref of names) of a field which this field must follow, in a first-to-last ordering of the columns. This field must occur immediately after the named field(s), or after another field which also has a "follows" restriction and follows the named field(s).

The purpose of this attribute is to resolve ambiguous columns. Suppose you expect columns with the following headers:

  Father    |          |      |       | Mother    |          |      |      
  FirstName | LastName | Tel. | Email | FirstName | LastName | Tel. | Email

You can use "qr/Father\nFirstName/" to identify the first column, but after FirstName the rest are ambiguous. But, TableReader can figure it out if you say:

  { name => 'father_first', header => qr/Father\nFirstName/ },
  { name => 'father_last',  header => 'LastName', follows => 'father_first' },
  { name => 'father_tel',   header => 'Tel.',     follows => 'father_first' },
  { name => 'father_email', header => 'Email',    follows => 'father_first' },
  ..

and so on. Note how 'father_first' is used for each as the "follows" name; this way if any non-required fields (like maybe "Tel") are completely removed from the file, TableReader will still be able to find "LastName" and "Email".

You can also use this to accumulate an array of columns that lack headers:

  Scores |      |       |      |       |       |       | OtherData
  12%    | 35%  | 42%   | 18%  | 65%   | 99%   | 55%   | xyz
  { name => 'scores', array => 1, trim => 1 },
  { name => 'scores', array => 1, trim => 1, header => '', follows => 'scores' },

The second field definition has an empty header, which would normally make it rather ambiguous and potentially capture blank-header columns that might not be part of the array. But, because it must follow a column named 'scores' there's no ambiguity; you get exactly any column starting from the header 'Scores' until a column of any other header.

follows_list

Convenience accessor for "@{ ->follows }", useful because "follows" might only be a scalar.

header_regex

"header", coerced to a regex if it wasn't already