package rag

import (
	"context"
	"encoding/json"
	"fmt"
	"strings"

	"agent-desk/internal/models"
	"agent-desk/internal/pkg/i18nx"
)

// faqContent keeps application-owned labels separate from source text. Never
// translate, trim, or replace substrings in the source question or answer.
type faqContent struct {
	question         string
	similarQuestions []string
	answer           string
}

func sourceFAQContent(faq models.KnowledgeFAQ) faqContent {
	content := faqContent{question: faq.Question, answer: faq.Answer}
	if faq.SimilarQuestions != "" {
		_ = json.Unmarshal([]byte(faq.SimilarQuestions), &content.similarQuestions)
	}
	return content
}

func (content faqContent) render(locale string) string {
	parts := []string{i18nx.Getf(locale, "knowledge.rag.faq.question", content.question)}
	if len(content.similarQuestions) > 0 {
		similar := joinNonEmptySimilarQuestions(content.similarQuestions, i18nx.Getf(locale, "knowledge.rag.faq.separator"))
		parts = append(parts, i18nx.Getf(locale, "knowledge.rag.faq.similarQuestions", similar))
	}
	parts = append(parts, i18nx.Getf(locale, "knowledge.rag.faq.answer", content.answer))
	return strings.Join(parts, "\n")
}

// Index generation deliberately uses the configured default, not the locale of
// whichever administrator requested a build. A locale read never re-embeds data.
func buildFAQChunkContent(faq models.KnowledgeFAQ) string {
	return sourceFAQContent(faq).render(i18nx.DefaultLocale)
}

// legacyChinese renders the exact pre-localization serialization. This is a
// recognition format, not UI copy. Keep it frozen when catalog wording changes.
func (content faqContent) legacyChinese() string {
	parts := []string{fmt.Sprintf("问题：%s", content.question)}
	if len(content.similarQuestions) > 0 {
		parts = append(parts, "相似问："+joinNonEmptySimilarQuestions(content.similarQuestions, "；"))
	}
	parts = append(parts, "回答："+content.answer)
	return strings.Join(parts, "\n")
}

// Only a full byte-for-byte match against a generated representation of the
// current source proves which text belongs to us. Stale/manual/malformed chunks
// stay as recorded. In particular, do not infer labels by splitting on Chinese
// words: the source content may itself contain those words or arbitrary lines.
// This function is for live retrieval only, never historical log replay.
func renderVerifiedFAQChunkContent(ctx context.Context, stored string, faq models.KnowledgeFAQ) string {
	content := sourceFAQContent(faq)
	if stored == content.legacyChinese() {
		return content.render(i18nx.LocaleFromContext(ctx))
	}
	for _, locale := range []string{i18nx.LocaleEnUS, i18nx.LocaleRuRU, i18nx.LocaleZhCN} {
		if stored == content.render(locale) {
			return content.render(i18nx.LocaleFromContext(ctx))
		}
	}
	return stored
}

func joinNonEmptySimilarQuestions(items []string, separator string) string {
	nonEmpty := make([]string, 0, len(items))
	for _, item := range items {
		if item != "" {
			nonEmpty = append(nonEmpty, item)
		}
	}
	return strings.Join(nonEmpty, separator)
}