backend/internal/pkg/openai/client.go

package openai

import (
	"bytes"
	"context"
	"encoding/base64"
	"encoding/json"
	"fmt"
	"image"
	_ "image/gif"
	"image/jpeg"
	_ "image/png"
	"io"
	"net/http"
	"net/url"
	"strings"
	"time"

	"with_you/internal/pkg/netutil"

	xdraw "golang.org/x/image/draw"
)

const moderationSystemPrompt = `你是中文社区的内容审核助手，负责对"帖子标题、正文、配图"做联合审核。目标是平衡社区安全与正常交流：必须拦截高风险违规内容，但不要误伤正常玩梗、二创、吐槽和轻度调侃。请只输出指定JSON。

审核流程：
1) 先判断是否命中硬性违规；
2) 再判断语境（玩笑/自嘲/朋友间互动/作品讨论）；
3) 做文图交叉判断（文本+图片合并理解）；
4) 给出 result 与简短 reason。

硬性违规（命中任一项必须 result=block）：
A. 宣传对立与煽动撕裂：
- 明确煽动群体对立、地域对立、性别对立、民族宗教对立，鼓动仇恨、排斥、报复。
B. 严重人身攻击与网暴引导：
- 持续性侮辱贬损、羞辱人格、号召围攻/骚扰/挂人/线下冲突。
C. 开盒/人肉/隐私暴露：
- 故意公开、拼接、索取他人可识别隐私信息（姓名+联系方式、身份证号、住址、学校单位、车牌、定位轨迹等）；
- 图片/截图中出现可识别隐私信息并伴随曝光意图，也按违规处理。
D. 其他高危违规：
- 违法犯罪、暴力威胁、极端仇恨、色情低俗、诈骗引流、恶意广告等。

可疑内容（需要人工复审，result=review）：
- 内容含义模糊，无法确定是否存在违规意图；
- 涉及敏感话题但未明确违规（如政治人物、社会热点事件等）；
- 存在轻微争议性表达，但不完全符合硬性违规标准；
- AI无法准确判断的文化/地域特定表达。

放行规则（以下通常 result=pass）：
- 正常玩梗、表情包、谐音梗、二次创作、无恶意的吐槽；
- 非定向、轻度口语化吐槽（无明确攻击对象、无网暴号召、无隐私暴露）；
- 对社会事件/作品的理性讨论、观点争论（即使语气尖锐，但未煽动对立或人身攻击）。

边界判定：
- 若只是"梗文化表达"且不指向现实伤害，优先通过；
- 若存在明确伤害意图（煽动、围攻、曝光隐私），必须拒绝；
- 对模糊内容不因个别粗口直接拒绝，需结合对象、意图、号召性和可执行性综合判断；
- 如果实在无法确定是否违规，标记为 review 状态交由人工复审。

reason 要求：
- result=block 时：中文10-30字，说明核心违规点；
- result=review 时：中文10-30字，说明可疑原因；
- result=pass 时：reason 为空字符串。

输出格式（严格）：
仅输出一行JSON对象，不要Markdown，不要额外解释：
{"result": "pass"/"review"/"block", "reason": "..."}`

const (
	defaultMaxImagesPerModerationRequest = 1
	maxModerationResultRetries           = 3
	maxChatCompletionRetries             = 3
	initialRetryBackoff                  = 500 * time.Millisecond
	maxDownloadImageBytes                = 10 * 1024 * 1024
	maxModerationImageSide               = 1280
	compressedJPEGQuality                = 72
	maxCompressedPayloadBytes            = 1536 * 1024
)

type ModerationResult string

const (
	ModerationResultPass   ModerationResult = "pass"
	ModerationResultReview ModerationResult = "review"
	ModerationResultBlock  ModerationResult = "block"
)

type ModerationResponse struct {
	Result ModerationResult `json:"result"`
	Reason string           `json:"reason"`
}

type Client interface {
	IsEnabled() bool
	Config() Config
	ModeratePost(ctx context.Context, title, content string, images []string) (*ModerationResponse, error)
	ModerateComment(ctx context.Context, content string, images []string) (*ModerationResponse, error)
	ModerateImage(ctx context.Context, imageURL string) (*ModerationResponse, error)
	ModerateBio(ctx context.Context, bio string) (*ModerationResponse, error)
}

type clientImpl struct {
	cfg        Config
	httpClient *http.Client
}

func NewClient(cfg Config) Client {
	timeout := cfg.RequestTimeoutSeconds
	if timeout <= 0 {
		timeout = 30
	}
	return &clientImpl{
		cfg:        cfg,
		httpClient: netutil.NewSafeHTTPClient(time.Duration(timeout) * time.Second),
	}
}

func (c *clientImpl) IsEnabled() bool {
	return c.cfg.Enabled && c.cfg.APIKey != "" && c.cfg.BaseURL != ""
}

func (c *clientImpl) Config() Config {
	return c.cfg
}

func (c *clientImpl) ModeratePost(ctx context.Context, title, content string, images []string) (*ModerationResponse, error) {
	if !c.IsEnabled() {
		return &ModerationResponse{Result: ModerationResultPass}, nil
	}

	var prompt strings.Builder
	if strings.TrimSpace(title) != "" {
		prompt.WriteString("帖子标题：" + title)
	}
	if strings.TrimSpace(content) != "" {
		if prompt.Len() > 0 {
			prompt.WriteString("\n")
		}
		prompt.WriteString("帖子内容：" + content)
	}
	if prompt.Len() == 0 && len(normalizeImageURLs(images)) > 0 {
		prompt.WriteString("帖子内容：[仅包含图片]")
	}
	if prompt.Len() == 0 {
		prompt.WriteString("帖子内容：[空]")
	}

	return c.moderateContentInBatches(ctx, prompt.String(), images)
}

func (c *clientImpl) ModerateComment(ctx context.Context, content string, images []string) (*ModerationResponse, error) {
	if !c.IsEnabled() {
		return &ModerationResponse{Result: ModerationResultPass}, nil
	}

	var prompt strings.Builder
	if strings.TrimSpace(content) != "" {
		prompt.WriteString("评论内容：" + content)
	} else if len(normalizeImageURLs(images)) > 0 {
		prompt.WriteString("评论内容：[仅包含图片]")
	} else {
		prompt.WriteString("评论内容：[空]")
	}

	return c.moderateContentInBatches(ctx, prompt.String(), images)
}

const imageModerationSystemPrompt = `你是图片内容审核助手，负责审核用户头像和背景图片。目标是确保图片内容符合社区规范。请只输出指定JSON。

硬性违规（命中任一项必须 result=block）：
- 色情低俗、裸露、性暗示内容
- 暴力血腥、恐怖画面
- 违法犯罪相关内容
- 明确的政治敏感宣传
- 恶意广告、诈骗信息
- 仇恨歧视符号或标志

可疑内容（result=review）：
- 图片含义模糊，无法确定是否存在违规
- 轻微争议性内容但不完全违规
- AI无法准确判断的内容

放行规则（result=pass）：
- 正常风景、人物、动物、动漫等图片
- 表情包、梗图
- 正常的艺术创作、摄影作品

输出格式（严格）：
仅输出一行JSON对象，不要Markdown，不要额外解释：
{"result": "pass"/"review"/"block", "reason": "..."}`

const bioModerationSystemPrompt = `你是用户个性签名内容审核助手，负责审核用户个人简介/签名。目标是平衡社区安全与正常自我表达。请只输出指定JSON。

硬性违规（命中任一项必须 result=block）：
- 明确的违法犯罪、暴力威胁内容
- 色情低俗、性暗示描述
- 诈骗引流、恶意广告
- 明确煽动群体对立、仇恨
- 恶意公开他人隐私信息

可疑内容（result=review）：
- 内容含义模糊，无法确定违规意图
- 涉及敏感话题但未明确违规
- AI无法准确判断的表达

放行规则（result=pass）：
- 正常的自我介绍、心情表达
- 玩梗、幽默、吐槽
- 引用、歌词、诗句等
- 理性的观点表达

输出格式（严格）：
仅输出一行JSON对象，不要Markdown，不要额外解释：
{"result": "pass"/"review"/"block", "reason": "..."}`

func (c *clientImpl) ModerateImage(ctx context.Context, imageURL string) (*ModerationResponse, error) {
	if !c.IsEnabled() {
		return &ModerationResponse{Result: ModerationResultPass}, nil
	}

	if strings.TrimSpace(imageURL) == "" {
		return &ModerationResponse{Result: ModerationResultPass}, nil
	}

	userPrompt := "请审核以下用户图片"
	cleanImages := normalizeImageURLs([]string{imageURL})
	optimizedImages := c.optimizeImagesForModeration(ctx, cleanImages)

	var lastErr error
	for attempt := 0; attempt < maxModerationResultRetries; attempt++ {
		replyText, err := c.chatCompletion(ctx, c.cfg.ModerationModel, imageModerationSystemPrompt, userPrompt, optimizedImages, 0.1, 220)
		if err != nil {
			lastErr = err
		} else {
			parsed := struct {
				Result string `json:"result"`
				Reason string `json:"reason"`
			}{}
			if err := json.Unmarshal([]byte(extractJSONObject(replyText)), &parsed); err != nil {
				lastErr = fmt.Errorf("failed to parse moderation result: %w", err)
			} else {
				result := ModerationResultPass
				switch parsed.Result {
				case "review":
					result = ModerationResultReview
				case "block":
					result = ModerationResultBlock
				}
				return &ModerationResponse{
					Result: result,
					Reason: parsed.Reason,
				}, nil
			}
		}

		if attempt == maxModerationResultRetries-1 {
			break
		}
		if sleepErr := sleepWithBackoff(ctx, attempt); sleepErr != nil {
			return nil, sleepErr
		}
	}

	return nil, fmt.Errorf("image moderation failed after %d attempts: %w", maxModerationResultRetries, lastErr)
}

func (c *clientImpl) ModerateBio(ctx context.Context, bio string) (*ModerationResponse, error) {
	if !c.IsEnabled() {
		return &ModerationResponse{Result: ModerationResultPass}, nil
	}

	if strings.TrimSpace(bio) == "" {
		return &ModerationResponse{Result: ModerationResultPass}, nil
	}

	userPrompt := "个性签名内容：" + bio
	return c.moderateWithCustomPrompt(ctx, bioModerationSystemPrompt, userPrompt)
}

func (c *clientImpl) moderateWithCustomPrompt(ctx context.Context, systemPrompt string, userPrompt string) (*ModerationResponse, error) {
	var lastErr error
	for attempt := 0; attempt < maxModerationResultRetries; attempt++ {
		replyText, err := c.chatCompletion(ctx, c.cfg.ModerationModel, systemPrompt, userPrompt, nil, 0.1, 220)
		if err != nil {
			lastErr = err
		} else {
			parsed := struct {
				Result string `json:"result"`
				Reason string `json:"reason"`
			}{}
			if err := json.Unmarshal([]byte(extractJSONObject(replyText)), &parsed); err != nil {
				lastErr = fmt.Errorf("failed to parse moderation result: %w", err)
			} else {
				result := ModerationResultPass
				switch parsed.Result {
				case "review":
					result = ModerationResultReview
				case "block":
					result = ModerationResultBlock
				}
				return &ModerationResponse{
					Result: result,
					Reason: parsed.Reason,
				}, nil
			}
		}

		if attempt == maxModerationResultRetries-1 {
			break
		}
		if sleepErr := sleepWithBackoff(ctx, attempt); sleepErr != nil {
			return nil, sleepErr
		}
	}

	return nil, fmt.Errorf("moderation failed after %d attempts: %w", maxModerationResultRetries, lastErr)
}

func (c *clientImpl) moderateContentInBatches(ctx context.Context, contentPrompt string, images []string) (*ModerationResponse, error) {
	cleanImages := normalizeImageURLs(images)
	optimizedImages := c.optimizeImagesForModeration(ctx, cleanImages)
	maxImagesPerRequest := c.maxImagesPerModerationRequest()
	totalBatches := 1
	if len(optimizedImages) > 0 {
		totalBatches = (len(optimizedImages) + maxImagesPerRequest - 1) / maxImagesPerRequest
	}

	for i := 0; i < totalBatches; i++ {
		start := i * maxImagesPerRequest
		end := start + maxImagesPerRequest
		if end > len(optimizedImages) {
			end = len(optimizedImages)
		}

		batchImages := []string{}
		if len(optimizedImages) > 0 {
			batchImages = optimizedImages[start:end]
		}

		resp, err := c.moderateSingleBatch(ctx, contentPrompt, batchImages, i+1, totalBatches)
		if err != nil {
			return nil, err
		}
		if resp.Result != ModerationResultPass {
			if totalBatches > 1 && strings.TrimSpace(resp.Reason) != "" {
				resp.Reason = fmt.Sprintf("第%d/%d批图片未通过：%s", i+1, totalBatches, resp.Reason)
			}
			return resp, nil
		}
	}

	return &ModerationResponse{Result: ModerationResultPass}, nil
}

func (c *clientImpl) moderateSingleBatch(
	ctx context.Context,
	contentPrompt string,
	images []string,
	batchNo, totalBatches int,
) (*ModerationResponse, error) {
	userPrompt := fmt.Sprintf(
		"%s\n图片批次：%d/%d（本次仅提供当前批次图片）",
		contentPrompt,
		batchNo,
		totalBatches,
	)

	var lastErr error
	for attempt := 0; attempt < maxModerationResultRetries; attempt++ {
		replyText, err := c.chatCompletion(ctx, c.cfg.ModerationModel, moderationSystemPrompt, userPrompt, images, 0.1, 220)
		if err != nil {
			lastErr = err
		} else {
			parsed := struct {
				Result string `json:"result"`
				Reason string `json:"reason"`
			}{}
			if err := json.Unmarshal([]byte(extractJSONObject(replyText)), &parsed); err != nil {
				lastErr = fmt.Errorf("failed to parse moderation result: %w", err)
			} else {
				result := ModerationResultPass
				switch parsed.Result {
				case "review":
					result = ModerationResultReview
				case "block":
					result = ModerationResultBlock
				}
				return &ModerationResponse{
					Result: result,
					Reason: parsed.Reason,
				}, nil
			}
		}

		if attempt == maxModerationResultRetries-1 {
			break
		}
		if sleepErr := sleepWithBackoff(ctx, attempt); sleepErr != nil {
			return nil, sleepErr
		}
	}

	return nil, fmt.Errorf(
		"moderation batch %d/%d failed after %d attempts: %w",
		batchNo,
		totalBatches,
		maxModerationResultRetries,
		lastErr,
	)
}

type chatCompletionsRequest struct {
	Model          string                `json:"model"`
	Messages       []chatMessage         `json:"messages"`
	Temperature    float64               `json:"temperature,omitempty"`
	MaxTokens      int                   `json:"max_tokens,omitempty"`
	EnableThinking *bool                 `json:"enable_thinking,omitempty"`
	ThinkingBudget *int                  `json:"thinking_budget,omitempty"`
	ResponseFormat *responseFormatConfig `json:"response_format,omitempty"`
}

type responseFormatConfig struct {
	Type string `json:"type"`
}

type chatMessage struct {
	Role    string `json:"role"`
	Content any    `json:"content"`
}

type contentPart struct {
	Type     string        `json:"type"`
	Text     string        `json:"text,omitempty"`
	ImageURL *imageURLPart `json:"image_url,omitempty"`
}

type imageURLPart struct {
	URL string `json:"url"`
}

type chatCompletionsResponse struct {
	Choices []struct {
		Message struct {
			Content string `json:"content"`
		} `json:"message"`
	} `json:"choices"`
}

func (c *clientImpl) chatCompletion(
	ctx context.Context,
	model string,
	systemPrompt string,
	userPrompt string,
	images []string,
	temperature float64,
	maxTokens int,
) (string, error) {
	if model == "" {
		return "", fmt.Errorf("model is empty")
	}

	cleanImages := normalizeImageURLs(images)

	userParts := []contentPart{
		{Type: "text", Text: userPrompt},
	}
	for _, image := range cleanImages {
		userParts = append(userParts, contentPart{
			Type:     "image_url",
			ImageURL: &imageURLPart{URL: image},
		})
	}

	reqBody := chatCompletionsRequest{
		Model: model,
		Messages: []chatMessage{
			{Role: "system", Content: systemPrompt},
			{Role: "user", Content: userParts},
		},
		Temperature: temperature,
		MaxTokens:   maxTokens,
	}
	falseVal := false
	reqBody.EnableThinking = &falseVal
	zero := 0
	reqBody.ThinkingBudget = &zero
	reqBody.ResponseFormat = &responseFormatConfig{Type: "json_object"}

	data, err := json.Marshal(reqBody)
	if err != nil {
		return "", fmt.Errorf("marshal request: %w", err)
	}

	baseURL := strings.TrimRight(c.cfg.BaseURL, "/")
	endpoint := baseURL + "/v1/chat/completions"
	if strings.HasSuffix(baseURL, "/v1") {
		endpoint = baseURL + "/chat/completions"
	}

	var lastErr error
	for attempt := 0; attempt < maxChatCompletionRetries; attempt++ {
		body, statusCode, err := c.doChatCompletionRequest(ctx, endpoint, data)
		if err != nil {
			lastErr = err
		} else if statusCode >= 400 {
			lastErr = fmt.Errorf("openai error status=%d body=%s", statusCode, string(body))
			if !isRetryableStatusCode(statusCode) {
				return "", lastErr
			}
		} else {
			var parsed chatCompletionsResponse
			if err := json.Unmarshal(body, &parsed); err != nil {
				return "", fmt.Errorf("decode response: %w", err)
			}
			if len(parsed.Choices) == 0 {
				return "", fmt.Errorf("empty response choices")
			}
			return parsed.Choices[0].Message.Content, nil
		}

		if attempt == maxChatCompletionRetries-1 {
			break
		}
		if sleepErr := sleepWithBackoff(ctx, attempt); sleepErr != nil {
			return "", sleepErr
		}
	}

	return "", lastErr
}

func (c *clientImpl) doChatCompletionRequest(ctx context.Context, endpoint string, data []byte) ([]byte, int, error) {
	req, err := http.NewRequestWithContext(ctx, http.MethodPost, endpoint, bytes.NewReader(data))
	if err != nil {
		return nil, 0, fmt.Errorf("create request: %w", err)
	}
	req.Header.Set("Content-Type", "application/json")
	req.Header.Set("Authorization", "Bearer "+c.cfg.APIKey)

	resp, err := c.httpClient.Do(req)
	if err != nil {
		return nil, 0, fmt.Errorf("request openai: %w", err)
	}
	defer resp.Body.Close()

	body, err := io.ReadAll(resp.Body)
	if err != nil {
		return nil, 0, fmt.Errorf("read response: %w", err)
	}
	return body, resp.StatusCode, nil
}

func isRetryableStatusCode(statusCode int) bool {
	if statusCode == http.StatusTooManyRequests {
		return true
	}
	return statusCode >= 500 && statusCode <= 599
}

func sleepWithBackoff(ctx context.Context, attempt int) error {
	delay := initialRetryBackoff * time.Duration(1<<attempt)
	timer := time.NewTimer(delay)
	defer timer.Stop()

	select {
	case <-ctx.Done():
		return fmt.Errorf("request cancelled: %w", ctx.Err())
	case <-timer.C:
		return nil
	}
}

func normalizeImageURLs(images []string) []string {
	clean := make([]string, 0, len(images))
	for _, image := range images {
		trimmed := strings.TrimSpace(image)
		if trimmed == "" {
			continue
		}
		clean = append(clean, trimmed)
	}
	return clean
}

func extractJSONObject(raw string) string {
	text := strings.TrimSpace(raw)
	start := strings.Index(text, "{")
	end := strings.LastIndex(text, "}")
	if start >= 0 && end > start {
		return text[start : end+1]
	}
	return text
}

func (c *clientImpl) maxImagesPerModerationRequest() int {
	if c.cfg.ModerationMaxImagesPerRequest <= 0 {
		return defaultMaxImagesPerModerationRequest
	}
	if c.cfg.ModerationMaxImagesPerRequest > 1 {
		return 1
	}
	return c.cfg.ModerationMaxImagesPerRequest
}

func (c *clientImpl) optimizeImagesForModeration(ctx context.Context, images []string) []string {
	if len(images) == 0 {
		return images
	}

	optimized := make([]string, 0, len(images))
	for _, imageURL := range images {
		optimized = append(optimized, c.tryCompressImageForModeration(ctx, imageURL))
	}
	return optimized
}

func (c *clientImpl) tryCompressImageForModeration(ctx context.Context, imageURL string) string {
	parsed, err := url.Parse(imageURL)
	if err != nil || (parsed.Scheme != "http" && parsed.Scheme != "https") {
		return imageURL
	}
	if err := netutil.ValidateURL(imageURL); err != nil {
		return imageURL
	}

	req, err := http.NewRequestWithContext(ctx, http.MethodGet, imageURL, nil)
	if err != nil {
		return imageURL
	}

	resp, err := c.httpClient.Do(req)
	if err != nil {
		return imageURL
	}
	defer resp.Body.Close()

	if resp.StatusCode >= 400 {
		return imageURL
	}
	if !strings.HasPrefix(strings.ToLower(resp.Header.Get("Content-Type")), "image/") {
		return imageURL
	}

	originBytes, err := io.ReadAll(io.LimitReader(resp.Body, maxDownloadImageBytes))
	if err != nil || len(originBytes) == 0 {
		return imageURL
	}

	srcImg, _, err := image.Decode(bytes.NewReader(originBytes))
	if err != nil {
		return imageURL
	}

	dstImg := resizeIfNeeded(srcImg, maxModerationImageSide)
	var buf bytes.Buffer
	if err := jpeg.Encode(&buf, dstImg, &jpeg.Options{Quality: compressedJPEGQuality}); err != nil {
		return imageURL
	}

	compressed := buf.Bytes()
	if len(compressed) == 0 || len(compressed) > maxCompressedPayloadBytes {
		return imageURL
	}
	if len(compressed) >= int(float64(len(originBytes))*0.95) {
		return imageURL
	}

	return "data:image/jpeg;base64," + base64.StdEncoding.EncodeToString(compressed)
}

func resizeIfNeeded(src image.Image, maxSide int) image.Image {
	bounds := src.Bounds()
	w := bounds.Dx()
	h := bounds.Dy()
	if w <= 0 || h <= 0 || max(w, h) <= maxSide {
		return src
	}

	newW, newH := w, h
	if w >= h {
		newW = maxSide
		newH = int(float64(h) * (float64(maxSide) / float64(w)))
	} else {
		newH = maxSide
		newW = int(float64(w) * (float64(maxSide) / float64(h)))
	}
	if newW < 1 {
		newW = 1
	}
	if newH < 1 {
		newH = 1
	}

	dst := image.NewRGBA(image.Rect(0, 0, newW, newH))
	xdraw.CatmullRom.Scale(dst, dst.Bounds(), src, bounds, xdraw.Over, nil)
	return dst
}