Struct Verdict 

pub struct Verdict<'a> {
    pub text: Cow<'a, str>,
    pub valid: bool,
    pub risk: f32,
}

One scanner’s outcome. Mirrors LLM Guard’s (sanitized, valid, risk).

text is a Cow so a pure detector (the common case) returns the input borrowed with zero allocation, while a transform returns the rewrite owned.

Fields

text: Cow<'a, str>

Possibly-rewritten text. For a pure detector, borrows the input.

valid: bool

false ⇒ the stack should reject (honoured for Block scanners).

risk: f32

0.0 safe … 1.0 high. Verdict::NOT_A_RISK (-1.0) means “not a risk judgment” — a transformer that does not score.

Implementations

impl<'a> Verdict<'a>

pub const NOT_A_RISK: f32 = -1.0

The sentinel risk a pure transformer returns: “not a risk judgment”, so stack-level risk aggregation can skip it honestly.

pub fn transformed(text: Cow<'a, str>) -> Self

A transformer verdict: text rewritten (or borrowed unchanged), always valid, Verdict::NOT_A_RISK.

pub fn is_risk_judgment(&self) -> bool

Whether this verdict carries a risk judgment (a detector) rather than being a pure transform (Verdict::NOT_A_RISK).

pub fn detected(text: &'a str, risk: f32, threshold: f32) -> Self

A detector verdict over borrowed input: valid = risk < threshold.

Trait Implementations

impl<'a> Clone for Verdict<'a>

fn clone(&self) -> Verdict<'a>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<'a> Debug for Verdict<'a>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'a> PartialEq for Verdict<'a>

fn eq(&self, other: &Verdict<'a>) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl<'a> StructuralPartialEq for Verdict<'a>