haixunMaster/haixun-backend/internal/library/knowledge/patrol_phrase.go

package knowledge

import (
	"strings"
	"unicode/utf8"
)

const (
	patrolIntentRelevance = "relevance"
	patrolIntentRecency   = "recency"
)

var patrolTopicAnchors = []string{
	"化療", "乳癌", "荷爾蒙", "標靶", "康復", "癌症", "病友", "敏感", "無香", "抗敏",
	"香料", "香精", "皮膚", "乾癢", "沐浴", "洗髮", "洗面", "卸妝", "防曬", "懷孕",
	"換季", "屏障", "過敏", "搔癢", "紅腫",
}

var patrolFillers = []string{
	"要", "什麼", "嗎", "怎麼", "請問", "有人", "適合", "用的", "分享", "經驗", "挑選",
	"可以", "不能", "需要", "應該", "到底", "真的", "覺得", "知道", "告訴", "請益",
	"推薦嗎", "好用嗎", "用過", "想", "還是", "會不會", "是不是", "有沒有", "如何", "為什麼",
}

// PatrolTagFromQuestion keeps research-map questions when already search-shaped.
func PatrolTagFromQuestion(raw string) string {
	raw = strings.TrimSpace(raw)
	raw = strings.Join(strings.Fields(raw), " ")
	if raw == "" {
		return ""
	}
	runes := utf8.RuneCountInString(raw)
	if runes >= minPatrolTagRunes && runes <= maxPatrolTagRunes {
		if looksLikeThreadsSearch(raw) {
			return raw
		}
		if runes >= 8 && (strings.Contains(raw, " ") || productCategoryHint(raw, "") != "") {
			phrase := ensurePatrolIntent(raw, patrolIntentRelevance)
			if utf8.RuneCountInString(phrase) <= maxPatrolTagRunes && !isMechanicalTag(phrase) {
				return phrase
			}
		}
	}
	return humanizePatrolPhrase(raw, patrolIntentRelevance)
}

// PatrolTagFromPillar compresses pillar phrases but keeps more context than generic labels.
func PatrolTagFromPillar(raw string) string {
	raw = strings.TrimSpace(raw)
	if raw == "" {
		return ""
	}
	runes := utf8.RuneCountInString(raw)
	if runes >= minPatrolTagRunes && runes <= maxPatrolTagRunes && strings.Contains(raw, " ") {
		return ensurePatrolIntent(raw, patrolIntentRelevance)
	}
	return humanizePatrolPhrase(raw, patrolIntentRelevance)
}

func humanizePatrolPhrase(raw, intent string) string {
	raw = strings.TrimSpace(raw)
	raw = strings.Join(strings.Fields(raw), " ")
	if raw == "" {
		return ""
	}

	runes := utf8.RuneCountInString(raw)
	if runes <= maxPatrolTagRunes && runes >= minPatrolTagRunes && looksLikeThreadsSearch(raw) {
		return raw
	}

	phrase := compressPatrolKeywords(raw)
	if phrase == "" {
		phrase = truncateRunes(raw, maxPatrolTagRunes)
	}
	phrase = ensurePatrolIntent(phrase, intent)
	if utf8.RuneCountInString(phrase) > maxPatrolTagRunes {
		phrase = truncateRunes(phrase, maxPatrolTagRunes)
	}
	if utf8.RuneCountInString(phrase) < minPatrolTagRunes {
		return ""
	}
	if isMechanicalTag(phrase) {
		return ""
	}
	return phrase
}

func looksLikeThreadsSearch(text string) bool {
	if strings.Contains(text, " ") {
		return true
	}
	for _, suffix := range []string{"推薦", "請問", "怎麼辦", "好用嗎", "有人", "求助"} {
		if strings.Contains(text, suffix) {
			return true
		}
	}
	return false
}

func compressPatrolKeywords(text string) string {
	category := productCategoryHint(text, "")
	anchors := []string{}
	seen := map[string]struct{}{}
	for _, anchor := range patrolTopicAnchors {
		if !strings.Contains(text, anchor) {
			continue
		}
		if _, ok := seen[anchor]; ok {
			continue
		}
		seen[anchor] = struct{}{}
		anchors = append(anchors, anchor)
		if len(anchors) >= 2 {
			break
		}
	}

	parts := append([]string{}, anchors...)
	if category != "" {
		parts = append(parts, category)
	}
	if len(parts) == 0 {
		for _, chunk := range splitPatrolChunks(text) {
			if isPatrolFiller(chunk) {
				continue
			}
			parts = append(parts, chunk)
			if len(parts) >= 2 {
				break
			}
		}
	}
	if len(parts) == 0 {
		return ""
	}
	phrase := strings.Join(parts, " ")
	if utf8.RuneCountInString(phrase) > maxPatrolTagRunes {
		return truncateRunes(phrase, maxPatrolTagRunes)
	}
	return phrase
}

func splitPatrolChunks(text string) []string {
	text = strings.TrimSpace(text)
	if text == "" {
		return nil
	}
	if strings.Contains(text, " ") {
		return strings.Fields(text)
	}
	runes := []rune(text)
	if len(runes) <= 6 {
		return []string{text}
	}
	// Long continuous Chinese: take leading topic chunk + trailing product-ish chunk.
	head := string(runes[:minInt(4, len(runes))])
	tail := string(runes[maxInt(0, len(runes)-4):])
	if head == tail {
		return []string{head}
	}
	return []string{head, tail}
}

func ensurePatrolIntent(phrase, intent string) string {
	phrase = strings.TrimSpace(phrase)
	if phrase == "" {
		return ""
	}
	if strings.ContainsAny(phrase, "推薦請問怎麼辦好用嗎有人求助") {
		return phrase
	}
	suffix := " 推薦"
	if intent == patrolIntentRecency {
		suffix = " 請問"
	}
	if utf8.RuneCountInString(phrase+suffix) <= maxPatrolTagRunes {
		return phrase + suffix
	}
	return phrase
}

func isPatrolFiller(chunk string) bool {
	chunk = strings.TrimSpace(chunk)
	if chunk == "" || utf8.RuneCountInString(chunk) < 2 {
		return true
	}
	for _, filler := range patrolFillers {
		if chunk == filler {
			return true
		}
	}
	return false
}

func truncateRunes(text string, max int) string {
	runes := []rune(strings.TrimSpace(text))
	if len(runes) <= max {
		return string(runes)
	}
	return string(runes[:max])
}

func maxInt(values ...int) int {
	max := values[0]
	for _, v := range values[1:] {
		if v > max {
			max = v
		}
	}
	return max
}