Documentation
¶
Overview ¶
Package lens applies transparent, reusable filters and weighted signals to local corpus candidates. It does not fetch or infer signal values.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func Validate ¶
func Validate(def Definition) error
Validate checks whether a lens definition can be stored and applied.
Types ¶
type Candidate ¶
type Candidate struct {
ID string
Repository string
Kind string
State string
Language string
Archived bool
Assigned bool
Stars int
UpdatedAt time.Time
Signals map[string]float64
}
Candidate is a locally derived item and its named, unnormalized signals.
type Definition ¶
type Definition struct {
Name string `json:"name"`
Filter Filter `json:"filter"`
Weights map[string]float64 `json:"weights"`
MaxResultsPerRepo int `json:"max_results_per_repo"`
}
Definition is a reusable ranking policy. Negative weights express costs or risks; positive weights express desirable signals.
func (*Definition) UnmarshalJSON ¶
func (d *Definition) UnmarshalJSON(data []byte) error
UnmarshalJSON supports JSON lens definitions where updated_within may be expressed as a Go duration string (e.g. "720h") or as nanoseconds.
type Filter ¶
type Filter struct {
Kinds []string `json:"kinds,omitempty"`
States []string `json:"states,omitempty"`
Languages []string `json:"languages,omitempty"`
ExcludeArchived bool `json:"exclude_archived,omitempty"`
Unassigned bool `json:"unassigned,omitempty"`
UpdatedWithin time.Duration `json:"updated_within,omitempty"`
MinStars int `json:"min_stars,omitempty"`
}
Filter contains hard eligibility rules evaluated before normalization.
func (Filter) MarshalJSON ¶
MarshalJSON emits updated_within as a Go duration string so JSON lens definitions round-trip with the same format used for input.
func (*Filter) UnmarshalJSON ¶
UnmarshalJSON supports duration strings for updated_within.
type Result ¶
type Result struct {
Candidate Candidate
Score float64
Normalized map[string]float64
Contributions map[string]float64
}
Result explains both the normalized values and each weighted contribution.
func Rank ¶
Rank filters candidates, min-max normalizes each configured signal within the eligible population, and returns a stable descending ranking. A signal with no population variance contributes zero. Missing values also contribute zero and remain absent from Normalized, making incomplete evidence visible.
Source Files