Module quality_intent 

Declared column intent for Data Quality: what a column must hold, said by the user, measured on the rows a run reads anyway.

The profile can say a column is nearly unique; only a declaration can say it is a key, so that a repeat is a defect and not a category. Every rule is opt-in, and a violation is counted as a fact out of the rows or values checked, never folded into a score.

Structs

ColumnCheck

What a column’s declared rules found.

ColumnIntent

What one column must hold. An empty intent declares nothing.

DeclaredIntent

Everything a study declares about its columns: the key, and each column’s rules.

IntentResults

What a run found of the declared intent.

KeyCheck

Whether the declared key was checked against every row in scope, or a sample’s.

Enums

NumberReading

How text read as a number is read.

ValueKind

What a column’s values are, once read as declared: what a range can compare and a set can list.

Constants

MAX_ALLOWED_VALUES

The most values an allowed set holds.

MAX_INTENT_EXAMPLES

Values outside the allowed set, or text that does not read as a number, kept as examples where the rows are in memory.

Functions

allows_set

Whether an allowed set applies: values compared as stored, text, whole numbers or true and false. A float or a time is a measurement, not a code.

check_intent

Why intent cannot be measured on a column of dtype, said on the form: a bound that is not a value of the column’s kind, or a minimum above the maximum.

format_allowed

An allowed set as it is typed: values joined by commas, quoted where a comma, a quote or outer spaces would change how it reads back.

parse_allowed

The values an allowed set holds, from what was typed: separated by commas, outer spaces dropped, each once, in the order typed. A value in double quotes is taken as it stands, commas and spaces included, with "" for a quote inside it.

reads_as_number

Whether the column can be read as a number: text.