yaju-is-always-watching-you/internal/detector/detector.go

// Package detector implements plain substring matching against a configured
// phrase list (the "goroku" list). Entries are literal fixed phrases rather
// than regexes because the source material is a closed set of copypasta
// lines, not a pattern language.
package detector

import "strings"

type Detector struct {
	phrases []string
}

func New(phrases []string) *Detector {
	uniq := make([]string, 0, len(phrases))
	seen := make(map[string]struct{}, len(phrases))
	for _, p := range phrases {
		p = strings.TrimSpace(p)
		if p == "" {
			continue
		}
		if _, ok := seen[p]; ok {
			continue
		}
		seen[p] = struct{}{}
		uniq = append(uniq, p)
	}
	return &Detector{phrases: uniq}
}

// Detect returns every registered phrase found in text, in registration order.
func (d *Detector) Detect(text string) []string {
	var found []string
	for _, p := range d.phrases {
		if strings.Contains(text, p) {
			found = append(found, p)
		}
	}
	return found
}

// Redact returns text with every matched phrase replaced by a same-width
// mask, along with the list of phrases that were matched.
func (d *Detector) Redact(text string) (string, []string) {
	matched := d.Detect(text)
	redacted := text
	for _, p := range matched {
		mask := strings.Repeat("○", len([]rune(p)))
		redacted = strings.ReplaceAll(redacted, p, mask)
	}
	return redacted, matched
}