ai-chatbot/app/api/canvas/text-detect/route.ts

import { NextResponse } from "next/server";
import { z } from "zod";
import { auth } from "@/app/(auth)/auth";
import { callQwen3Vl } from "@/lib/ai/qwen3-vl";
import { createLogger } from "@/lib/observability/logger";
import sharp from "sharp";

export const runtime = "nodejs";
export const maxDuration = 120;

const requestSchema = z.object({
  imageUrl: z.string().min(1, "图片地址不能为空"),
});

const OCR_IMAGE_FETCH_TIMEOUT_MS = 15000;
const OCR_IMAGE_MAX_SIDE = 1400;
const logger = createLogger("canvas-text-detect");

interface OcrItem {
  id: string;
  text: string;
  confidence?: number;
}

function parseJsonText(raw: string): any | null {
  if (!raw) return null;

  const trimmed = raw.trim();
  try {
    return JSON.parse(trimmed);
  } catch {
    // continue
  }

  const markdownJsonMatch = trimmed.match(/```json\s*([\s\S]*?)\s*```/i);
  if (markdownJsonMatch?.[1]) {
    try {
      return JSON.parse(markdownJsonMatch[1]);
    } catch {
      // continue
    }
  }

  const objectMatch = trimmed.match(/\{[\s\S]*\}/);
  if (objectMatch?.[0]) {
    try {
      return JSON.parse(objectMatch[0]);
    } catch {
      // continue
    }
  }

  return null;
}

function sanitizeItems(items: unknown[]): OcrItem[] {
  const result: OcrItem[] = [];

  for (let index = 0; index < items.length; index += 1) {
    const item = items[index] as Record<string, unknown>;
    const rawText = typeof item?.text === "string" ? item.text : "";
    const text = rawText.replace(/\s+/g, " ").trim();
    if (!text) continue;

    const confidenceValue =
      typeof item?.confidence === "number" && Number.isFinite(item.confidence)
        ? Math.min(Math.max(item.confidence, 0), 1)
        : undefined;

    result.push({
      id: String(item?.id || index + 1),
      text,
      confidence: confidenceValue,
    });
  }

  return result.slice(0, 50);
}

function fallbackItemsFromRaw(raw: string): OcrItem[] {
  const lines = raw
    .split("\n")
    .map((line) => line.replace(/^[-*\d.\s]+/, "").trim())
    .filter(Boolean)
    .slice(0, 20);

  return lines.map((text, index) => ({
    id: String(index + 1),
    text,
  }));
}

function shouldUseDataUrlFallback(error: unknown) {
  const message =
    error instanceof Error ? error.message.toLowerCase() : String(error || "").toLowerCase();
  return /abort|timeout|timed out|fetch failed|econnreset|etimedout|503|504/.test(message);
}

async function buildVisionDataUrl(imageUrl: string) {
  const controller = new AbortController();
  const timeout = setTimeout(() => controller.abort(), OCR_IMAGE_FETCH_TIMEOUT_MS);
  try {
    const response = await fetch(imageUrl, { method: "GET", signal: controller.signal });
    if (!response.ok) {
      throw new Error(`拉取图片失败: HTTP ${response.status}`);
    }

    const sourceBuffer = Buffer.from(await response.arrayBuffer());
    const source = sharp(sourceBuffer, { failOn: "none" }).rotate();
    const meta = await source.metadata();
    const hasAlpha = Boolean(meta.hasAlpha);

    const resized = source.resize({
      width: OCR_IMAGE_MAX_SIDE,
      height: OCR_IMAGE_MAX_SIDE,
      fit: "inside",
      withoutEnlargement: true,
    });

    if (hasAlpha) {
      const pngBuffer = await resized.png({ compressionLevel: 9 }).toBuffer();
      return `data:image/png;base64,${pngBuffer.toString("base64")}`;
    }

    const jpegBuffer = await resized.jpeg({ quality: 86, mozjpeg: true }).toBuffer();
    return `data:image/jpeg;base64,${jpegBuffer.toString("base64")}`;
  } finally {
    clearTimeout(timeout);
  }
}

export async function POST(request: Request) {
  try {
    const session = await auth();
    if (!session?.user?.id) {
      return NextResponse.json({ success: false, error: "未登录" }, { status: 401 });
    }

    const body = await request.json();
    const parsed = requestSchema.safeParse(body);
    if (!parsed.success) {
      return NextResponse.json({ success: false, error: parsed.error.errors[0]?.message || "参数错误" }, { status: 400 });
    }

    const imageUrl = parsed.data.imageUrl.trim();
    if (!/^https?:\/\//i.test(imageUrl)) {
      return NextResponse.json({ success: false, error: "仅支持 URL 图片，不允许 base64 传输" }, { status: 400 });
    }

    logger.debug("model_selected", { model: "qwen3-vl" });
    const prompt = [
      "你是图片文字识别助手。",
      "请识别图片中可见文字并按阅读顺序输出 JSON。",
      "只返回 JSON，不要额外说明。",
      "格式：{\"items\":[{\"id\":\"1\",\"text\":\"示例\",\"confidence\":0.95}]}",
      "如果没有识别到文字，返回 {\"items\":[]}。",
    ].join("\n");
    let text = "";
    let model = "";

    try {
      const directResult = await callQwen3Vl({
        imageUrl,
        prompt,
        temperature: 0.1,
        maxTokens: 600,
        timeoutMs: 45000,
        retries: 1,
      });
      text = directResult.text;
      model = directResult.model;
    } catch (error) {
      if (!shouldUseDataUrlFallback(error)) {
        throw error;
      }

      logger.warn("vision_url_fallback", {
        reason: error instanceof Error ? error.message : "unknown",
      });
      const dataUrl = await buildVisionDataUrl(imageUrl);
      const fallbackResult = await callQwen3Vl({
        imageUrl: dataUrl,
        prompt,
        temperature: 0.1,
        maxTokens: 600,
        timeoutMs: 60000,
        retries: 1,
      });
      text = fallbackResult.text;
      model = `${fallbackResult.model}-dataurl`;
    }

    const parsedJson = parseJsonText(text || "");
    let items: OcrItem[] = [];

    if (parsedJson && Array.isArray(parsedJson.items)) {
      items = sanitizeItems(parsedJson.items);
    }

    if (items.length === 0 && text) {
      items = fallbackItemsFromRaw(text);
    }

    return NextResponse.json({
      success: true,
      data: {
        items,
        model,
      },
    });
  } catch (error) {
    logger.error("text_detect_failed", error);
    return NextResponse.json(
      {
        success: false,
        error: error instanceof Error ? error.message : "文字识别失败",
      },
      { status: 500 },
    );
  }
}