Predefined data

Cantabular makes certain data about the user’s query and result table available to the disclosure rules language code. This data is made available via predefined objects with named properties. The properties are accessed using a dot notation: object.propertyname.

The query and output table are presented to the DRL code without any filters that the user may have requested. This is because the user must not be able to circumvent disclosure rules by filtering (slicing) output tables.

Query predefined data

For the querytest clauses a single object named “query” is pre-defined which represents the unfiltered input query. This has properties:

• vars

  This is a list of variables specifying the query in the order supplied by the user, each of which has properties:

  • name

  • cats

    This is a list of all the categories in the variable (not just those selected by the user via filters), each of which has:

    • code

Output table predefined data

For the tabletest clauses a single object named “table” is pre-defined which represents the unfiltered perturbed output table. If the first variable in the query is one of the rule hierarchy variables then a slice of the output table is created for each category in that variable and the rules are evaluated for each slice. If the first variable in the query is not one of the partial publication hierarchy variables then the “slice” being checked is the entire output table.

The perturbed output table object “table” has properties:

• slice

  This is a list which identifies the table slice being processed. If a whole table is being evaluated then this will be an empty list. Each element in the list has properties:

  • code : the category code of the table slice being processed

  • varname : the same as the corresponding variable in query

  In this version of Cantabular the list cannot have more than one element.

• vars

  This is a list of variables specifying the (crosstab) query in the order supplied by the user, each of which has properties:

  • name

  • cats

    This is a list of categories, each of which has:

    • code

    • total (a.k.a. Marginal Total)

• cats

  This is a list of categories from all the variables in vars in order. It is provided because it is often necessary to iterate over all categories from all variables and providing this list avoids nested loops in these cases. It is also essential to the tabledef clause which involves filtering categories.

• cells

  This is a list of cells in the output table slice, each of which has properties:

  • count (i.e. the value in the cell: number of rows matching this combination of categories)

  • key (i.e. the cell key)

• margins

  This is a list of margin tables (summations of the output table for each variable in the query), each of which has properties:

  • vars (variables as above excluding the summation variable, properties as above)

  • cells (list of cells in the margin table, properties as above), plus:

    • disclosiveCellIndex (integer)

      The value is not negative if only one cell in the main table contributes to this margin cell. The value is the index in the main table of the disclosive cell. If more than one cell or no cells contribute to this margin cell then the value is minus one.