rynaro / estratto Goto Github PK

In version 1.0.0 we only accept file paths as parameters in Document entry point.
One great improvement to gem is accept the File, and YAML objects from caller system. This feature makes gem more flexible for integration.

Third-party files sometimes has lack of data integrity. It's not uncommon, you need to parse a date (like birthdate), and third-party file does not contain this information in the specified range.
Today, estratto raises error trying to parse the date format, 'cause ' ' isnt valid date to parse.

Proposal change:

In formats node, we'll insert a new property called allow_empty receiving a boolean value. When this value was true, estratto will return nil value to avoid any problems, and keeping data integrity. (Using premise, if we don't get data from range, a nil data can describe the actual state)
If value is false or allow_empty omitted, estratto will raise errors as default behaviour.

Technical usage proposal:

        - name: birthdate
          range: 137..144
          type: DateTime
          formats:
            allow_empty: true
            format: '%Y%m%d'

Estratto uses the CharlockHolmes gem which depends on ICU to check the input file encoding.
It is necessary to add a section on README describing this dependency and provide install instructions for libicu-dev otherwise the user may face the following error message:

Gem::Ext::BuildError: ERROR: Failed to build gem native extension.

Estratto was builted with intent to deal with multi-layout as primary context, but one of most common layout type of fixed width.

In mono layout type generally has header, and footer, that these lines do not shares the same pattern of principal content lines.

Suggested initial point of this feature:

layout:
  name: 'monolayout'
  multi-register: false # or nonexistent key
  header: true
  footer: true
  registers:
    - header:
     # content
    - register:
     # content
    - footer:
     # content

Add a safe_data: true or sanitize: true option to get rid off invalid characters.