Skip to main content
Version: 3.1.X

Data Validation with PFT

In ABCD and CISIS databases, data integrity is enforced using Validation Formats. Unlike standard display formats, these are executed during data entry (when creating or updating a record).

The Concept​

The validation logic works on a simple principle: "Silence is Success".

  • If the PFT executes and produces no output (empty string), the record is considered Valid and gets saved.
  • If the PFT produces any output (text), that output is treated as an Error Message, displayed to the user, and the save operation is blocked.

Data Entry Mode​

When a PFT is executed for validation, the system runs in Data Entry Mode. This has specific implications for how subfields and delimiters are handled compared to the standard Proof Mode.

  • Subfields: In validation, you often need to check specific subfields (e.g., v10^a) rather than the whole field.
  • Repetitions: Validation often iterates through repeatable fields to ensure every occurrence meets the criteria.

Common Validation Scenarios​

1. Mandatory Fields​

To ensure a field (e.g., Tag 10 - Title) is not empty.

/* If Tag 10 is absent, display error */
if a(v10) then 'Error: Title is mandatory!' fi

2. Conditional Mandatoriness​

A field is required only if another field has a specific value. Example: If "Publication Type" (v20) is "Article", then "Source Journal" (v30) is required.

if v20 = 'Article' and a(v30) then
   'Error: Source Journal is required for Articles.'
fi

3. Validating Repeatable Groups​

If a field is repeatable (e.g., Authors in Tag 70), you must ensure all authors have a valid format. You iterate through them using a repeatable group.

(
   if p(v70) and size(v70) < 3 then
      'Error: Author name "' v70 '" is too short.' /
   fi
)

Note: The / ensures that if multiple authors are invalid, each error appears on a new line.

4. Duplicate Checking​

To prevent duplicate records (e.g., ensuring ISBN in Tag 20 is unique).

  • Technique: Use the l() lookup function to see if the term already exists in the inverted file.
  • Constraint: You must exclude the current record from the check (otherwise, editing an existing record would trigger a "duplicate" error against itself).
/* v20 is ISBN. "ISBN_" is the prefix in FST */
if p(v20) then
   /* Find MFN of record containing this ISBN */
   proc('a9999~' l('ISBN_' v20) '~'),
   
   /* If found (val > 0) AND it is not the current record (mfn) */
   if val(v9999) > 0 and val(v9999) <> mfn then
      'Error: ISBN ' v20 ' already exists in record ' v9999
   fi
fi

5. Format/Pattern Validation​

Checking if a field follows a specific pattern (e.g., Year must be 4 digits).

if p(v64) then
   if val(v64) < 1000 or val(v64) > 3000 then
      'Error: Invalid Year ' v64
   fi
fi

6. Value Lists (Controlled Vocabulary)​

Ensuring a field contains one of the allowed values.

if p(v40) then
   if v40 <> 'ENG' and v40 <> 'SPA' and v40 <> 'POR' then
      'Error: Language ' v40 ' is not valid (Use ENG, SPA, POR).'
   fi
fi

Tip: For long lists, it is more efficient to use a ref() to a thesaurus database or select case structure.

Integration in ABCD​

In ABCD, these validation scripts are typically saved as .val files in the database directory (e.g., bases/marc/def/en/marc.val) or configured in the Advanced Editing parameters.