📏 Distance & Comparison

cvss.DistanceCalculator measures how far apart two CVSS vectors are. Five metrics are available, each in three flavors: plain (silent 0 on incomplete base metrics), WithEnv (adds environmental metrics), and Checked (returns an error instead of silently returning 0).

Synopsis

dc := cvss.NewDistanceCalculator(a, b)
fmt.Println(dc.EuclideanDistance())          // base + temporal
fmt.Println(dc.EuclideanDistanceWithEnv())   // + environmental
e, err := dc.EuclideanDistanceChecked()      // error if base incomplete

Metric relationships

Metric	What it counts	Range	Notes
Euclidean	√(Σ score-diff²) over metric dims	≥ 0	PR/UI diffs use context-adjusted scores
Manhattan	Σ |score-diff| over metric dims	≥ 0	same context adjustments
Hamming	count of differing metrics	int ≥ 0	compares short values, not scores
Jaccard similarity	same / total metrics	0–1	1 = identical; uses Hamming internally
Score difference	|score(a) − score(b)|	0–10	uses Calculate, so the most refined score

API Reference

func NewDistanceCalculator(vector1, vector2 *Cvss3x) *DistanceCalculator

Plain (silent 0 on incomplete base metrics)

func (dc *DistanceCalculator) EuclideanDistance() float64
func (dc *DistanceCalculator) ManhattanDistance() float64
func (dc *DistanceCalculator) HammingDistance() int
func (dc *DistanceCalculator) JaccardSimilarity() float64
func (dc *DistanceCalculator) ScoreDifference() float64

With environmental metrics

func (dc *DistanceCalculator) EuclideanDistanceWithEnv() float64
func (dc *DistanceCalculator) ManhattanDistanceWithEnv() float64
func (dc *DistanceCalculator) HammingDistanceWithEnv() int
func (dc *DistanceCalculator) JaccardSimilarityWithEnv() float64

These add the 11 environmental dimensions (CR/IR/AR + MAV..MA) when both vectors have environmental metrics; otherwise they behave like the plain variants.

Checked (return error instead of silent 0)

func (dc *DistanceCalculator) EuclideanDistanceChecked() (float64, error)
func (dc *DistanceCalculator) ManhattanDistanceChecked() (float64, error)
func (dc *DistanceCalculator) ScoreDifferenceChecked() (float64, error)
func (dc *DistanceCalculator) EuclideanDistanceWithEnvChecked() (float64, error)
func (dc *DistanceCalculator) ManhattanDistanceWithEnvChecked() (float64, error)

The *Checked variants return errIncompleteMetrics ("base metrics incomplete, cannot compute distance") when either vector is missing required base metrics, instead of silently returning 0.

Jaccard is a similarity, not a distance

JaccardSimilarity returns 1.0 for identical vectors and 0.0 for fully disjoint ones. Convert to a distance with 1 - jaccard if you need a dissimilarity.

ScoreDifference returns 0 on either nil vector

ScoreDifference silently returns 0.0 if either vector is nil or if either Calculate errors. Use ScoreDifferenceChecked to surface those cases.

Example

package main

import (
    "fmt"

    "github.com/scagogogo/cvss-skills/pkg/cvss"
    "github.com/scagogogo/cvss-skills/pkg/parser"
)

func main() {
    a, _ := parser.ParseString("CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H")
    b, _ := parser.ParseString("CVSS:3.1/AV:L/AC:H/PR:H/UI:R/S:C/C:L/I:L/A:N")

    dc := cvss.NewDistanceCalculator(a, b)
    fmt.Printf("Euclidean : %.4f\n", dc.EuclideanDistance())
    fmt.Printf("Manhattan : %.4f\n", dc.ManhattanDistance())
    fmt.Printf("Hamming   : %d\n", dc.HammingDistance())        // 8
    fmt.Printf("Jaccard   : %.4f\n", dc.JaccardSimilarity())     // 0/8 -> 0.0
    fmt.Printf("ScoreDiff : %.1f\n", dc.ScoreDifference())       // |9.8 - 5.0| ...

    // Checked variant surfaces incomplete base metrics.
    partial, _ := parser.ParseRelaxed("AV:N/AC:L", "3.1") // only 2 of 8
    dc2 := cvss.NewDistanceCalculator(a, partial)
    if _, err := dc2.EuclideanDistanceChecked(); err != nil {
        fmt.Println("checked error:", err)
    }
}

Related

• pkg/cvss — Diff for a per-metric breakdown, Equal for equality
• Scoring (calculator) — used by ScoreDifference
• Impact & Sensitivity — single-vector "which metric matters most" analysis