Struct Ranking 

pub struct Ranking { /* private fields */ }

Sorted decided candidates and every unscored candidate’s evidence.

Implementations

impl Ranking

pub fn question(&self) -> &Score

Original scoring rubric.

pub fn ranked(&self) -> &[Ranked]

Candidates sorted by descending score, with exact ties in retrieval order.

pub fn unscored(&self) -> &IndexMap<String, ItemResult>

Undecided answers and provider failures, in retrieval order.

pub fn complete(&self) -> bool

Whether every candidate has a decided score.

pub fn models(&self) -> BTreeSet<&str>

Nonempty model identities actually reported, including undecided answers.

pub fn top(&self, k: usize, allow_partial: bool) -> Result<&[Ranked]>

Return at most k candidates, checking completeness and reported-model consistency.

allow_partial permits unscored candidates; it never permits mixed models. Unknown model metadata stays unknown rather than claiming one common model.

Errors

Returns InvalidArgument if candidates are unscored and allow_partial is false, or if answers report more than one nonempty model identity. The mixed-model check applies even when partial results are allowed.

pub fn reweighted(&self, weights: &[f64]) -> Result<Self>

Re-sort decided answers using finite strictly increasing rubric weights.

Uses normalized sum(weight * probability) / sum(probability) and retains the original answers and unscored entries. No provider call occurs.

Errors

Returns InvalidArgument if weights do not match the rubric length, are nonfinite or not strictly increasing, or weighted arithmetic is nonfinite.

Trait Implementations

impl Clone for Ranking

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Ranking

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

Formats the value using the given formatter.