@shenghuabi/knowledge 1.0.22 → 1.0.24

package/ocr.mjs

@@ -242,106 +242,106 @@ var ModelConfig = [
   {
     label: "简体中文",
     key: "ch_mobile",
-    det: "det/ch_PP-OCRv4_det_infer.onnx",
-    rec: "rec/ch_PP-OCRv4_rec_infer.onnx",
-    dict: "rec/ch_PP-OCRv4_rec_infer/ppocr_keys_v1.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/ch_PP-OCRv4_det_mobile.onnx",
+    rec: "rec/ch_PP-OCRv4_rec_mobile.onnx",
+    dict: "rec/ch_PP-OCRv4_rec_mobile/ppocr_keys_v1.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "简体中文(服务器)",
     key: "ch_server",
-    det: "det/ch_PP-OCRv4_det_server_infer.onnx",
-    rec: "rec/ch_PP-OCRv4_rec_server_infer.onnx",
-    dict: "rec/ch_PP-OCRv4_rec_server_infer/ppocr_keys_v1.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/ch_PP-OCRv4_det_server.onnx",
+    rec: "rec/ch_PP-OCRv4_rec_server.onnx",
+    dict: "rec/ch_PP-OCRv4_rec_server/ppocr_keys_v1.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "繁體中文",
     key: "chinese_cht",
-    det: "det/ch_PP-OCRv4_det_infer.onnx",
-    rec: "rec/chinese_cht_PP-OCRv3_rec_infer.onnx",
-    dict: "rec/chinese_cht_PP-OCRv3_rec_infer/chinese_cht_dict.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/ch_PP-OCRv4_det_server.onnx",
+    rec: "rec/chinese_cht_PP-OCRv3_rec_mobile.onnx",
+    dict: "rec/chinese_cht_PP-OCRv3_rec_mobile/chinese_cht_dict.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "英文",
     key: "en_mobile",
-    det: "det/en_PP-OCRv3_det_infer.onnx",
-    rec: "rec/en_PP-OCRv4_rec_infer.onnx",
-    dict: "rec/en_PP-OCRv4_rec_infer/en_dict.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/en_PP-OCRv3_det_mobile.onnx",
+    rec: "rec/en_PP-OCRv4_rec_mobile.onnx",
+    dict: "rec/en_PP-OCRv4_rec_mobile/en_dict.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "阿拉伯文",
     key: "ar_mobile",
-    det: "det/Multilingual_PP-OCRv3_det_infer.onnx",
-    rec: "rec/arabic_PP-OCRv4_rec_infer.onnx",
-    dict: "rec/arabic_PP-OCRv4_rec_infer/arabic_dict.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/multi_PP-OCRv3_det_mobile.onnx",
+    rec: "rec/arabic_PP-OCRv4_rec_mobile.onnx",
+    dict: "rec/arabic_PP-OCRv4_rec_mobile/arabic_dict.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "塞尔维亚文",
     key: "cyrillic_mobile",
-    det: "det/Multilingual_PP-OCRv3_det_infer.onnx",
-    rec: "rec/cyrillic_PP-OCRv3_rec_infer.onnx",
-    dict: "rec/cyrillic_PP-OCRv3_rec_infer/cyrillic_dict.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/multi_PP-OCRv3_det_mobile.onnx",
+    rec: "rec/cyrillic_PP-OCRv3_rec_mobile.onnx",
+    dict: "rec/cyrillic_PP-OCRv3_rec_mobile/cyrillic_dict.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "梵文",
     key: "devanagari_mobile",
-    det: "det/Multilingual_PP-OCRv3_det_infer.onnx",
-    rec: "rec/devanagari_PP-OCRv4_rec_infer.onnx",
-    dict: "rec/devanagari_PP-OCRv4_rec_infer/devanagari_dict.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/multi_PP-OCRv3_det_mobile.onnx",
+    rec: "rec/devanagari_PP-OCRv4_rec_mobile.onnx",
+    dict: "rec/devanagari_PP-OCRv4_rec_mobile/devanagari_dict.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "日文",
     key: "japan_mobile",
-    det: "det/Multilingual_PP-OCRv3_det_infer.onnx",
-    rec: "rec/japan_PP-OCRv4_rec_infer.onnx",
-    dict: "rec/japan_PP-OCRv4_rec_infer/japan_dict.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/multi_PP-OCRv3_det_mobile.onnx",
+    rec: "rec/japan_PP-OCRv4_rec_mobile.onnx",
+    dict: "rec/japan_PP-OCRv4_rec_mobile/japan_dict.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "卡纳达语",
     key: "ka_mobile",
-    det: "det/Multilingual_PP-OCRv3_det_infer.onnx",
-    rec: "rec/ka_PP-OCRv4_rec_infer.onnx",
-    dict: "rec/ka_PP-OCRv4_rec_infer/ka_dict.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/multi_PP-OCRv3_det_mobile.onnx",
+    rec: "rec/ka_PP-OCRv4_rec_mobile.onnx",
+    dict: "rec/ka_PP-OCRv4_rec_mobile/ka_dict.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "韩文",
     key: "korean_mobile",
-    det: "det/Multilingual_PP-OCRv3_det_infer.onnx",
-    rec: "rec/korean_PP-OCRv4_rec_infer.onnx",
-    dict: "rec/korean_PP-OCRv4_rec_infer/korean_dict.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/multi_PP-OCRv3_det_mobile.onnx",
+    rec: "rec/korean_PP-OCRv4_rec_mobile.onnx",
+    dict: "rec/korean_PP-OCRv4_rec_mobile/korean_dict.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "拉丁文",
     key: "latin_mobile",
-    det: "det/Multilingual_PP-OCRv3_det_infer.onnx",
-    rec: "rec/latin_PP-OCRv3_rec_infer.onnx",
-    dict: "rec/latin_PP-OCRv3_rec_infer/latin_dict.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/multi_PP-OCRv3_det_mobile.onnx",
+    rec: "rec/latin_PP-OCRv3_rec_mobile.onnx",
+    dict: "rec/latin_PP-OCRv3_rec_mobile/latin_dict.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "泰米尔文",
     key: "ta_mobile",
-    det: "det/Multilingual_PP-OCRv3_det_infer.onnx",
-    rec: "rec/ta_PP-OCRv4_rec_infer.onnx",
-    dict: "rec/ta_PP-OCRv4_rec_infer/ta_dict.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/multi_PP-OCRv3_det_mobile.onnx",
+    rec: "rec/ta_PP-OCRv4_rec_mobile.onnx",
+    dict: "rec/ta_PP-OCRv4_rec_mobile/ta_dict.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   },
   {
     label: "泰卢固文",
     key: "te_mobile",
-    det: "det/Multilingual_PP-OCRv3_det_infer.onnx",
-    rec: "rec/te_PP-OCRv4_rec_infer.onnx",
-    dict: "rec/te_PP-OCRv4_rec_infer/te_dict.txt",
-    cls: "cls/ch_ppocr_mobile_v2.0_cls_infer.onnx"
+    det: "det/multi_PP-OCRv3_det_mobile.onnx",
+    rec: "rec/te_PP-OCRv4_rec_mobile.onnx",
+    dict: "rec/te_PP-OCRv4_rec_mobile/te_dict.txt",
+    cls: "cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx"
   }
 ];
 export {

package/ocr.mjs.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../packages/ocr/ocr.ts", "../packages/ocr/ImageRaw.ts", "../packages/ocr/FileUtils.ts", "../packages/image/convert.ts", "../packages/image/extract.ts", "../packages/image/image-metadata.ts", "../packages/ocr/model-config.ts"],
-  "sourcesContent": ["import BaseOcr, {\n  ModelCreateOptions,\n  registerBackend,\n} from '@gutenye/ocr-common';\nimport { splitIntoLineImages } from '@gutenye/ocr-common/splitIntoLineImages';\nimport { ImageRaw } from './ImageRaw';\nimport { FileUtils } from './FileUtils';\nimport { InferenceSession } from 'onnxruntime-node';\nimport fs from 'fs/promises';\nimport { path } from '@cyia/vfs2';\nimport { convertToRaw } from '../image';\nimport * as v from 'valibot';\nconst ImageAdjustDefine = v.object({\n  padding: v.pipe(\n    v.optional(\n      v.union([\n        v.pipe(\n          v.number(),\n          v.transform((item) => ({\n            top: item,\n            left: item,\n            right: item,\n            bottom: item,\n          })),\n        ),\n        v.object({\n          left: v.number(),\n          right: v.number(),\n          top: v.number(),\n          bottom: v.number(),\n        }),\n      ]),\n      { top: 50, right: 50, left: 50, bottom: 50 },\n    ),\n  ),\n  maxSideLen: v.optional(v.union([v.pipe(v.number())]), 1920),\n  // threshold: v.optional(v.union([v.pipe(v.number())]), 0.3),\n});\nexport type ImageAdjustType = v.InferInput<typeof ImageAdjustDefine>;\nregisterBackend({\n  FileUtils,\n  ImageRaw,\n  InferenceSession,\n  splitIntoLineImages,\n  defaultModels: undefined,\n});\n\nasync function convert(\n  this: BaseOcr,\n  input: string | Uint8Array,\n  options: ImageAdjustType = {},\n) {\n  const resolveOptions = v.parse(ImageAdjustDefine, options);\n  //100 80 0.8\n  // 50 40\n  let { raw } = await convertToRaw(input);\n  const metadata = await raw.metadata();\n  const maxSize = Math.max(metadata.width!, metadata.height!);\n  if (maxSize > resolveOptions.maxSideLen) {\n    let ratio = metadata.width! / metadata.height!;\n    ratio = ratio > 1 ? 1 / ratio : ratio;\n    raw = raw.resize({\n      width: Math.round(\n        maxSize === metadata.width!\n          ? resolveOptions.maxSideLen\n          : ratio * resolveOptions.maxSideLen,\n      ),\n      height: Math.round(\n        maxSize === metadata.height!\n          ? resolveOptions.maxSideLen\n          : ratio * resolveOptions.maxSideLen,\n      ),\n    });\n  }\n  raw = raw.extend({ ...resolveOptions.padding, background: '#fff' });\n  raw = raw.ensureAlpha(1);\n  return this.detect(raw as any);\n}\nexport class Ocr extends BaseOcr {\n  static override async create(options: ModelCreateOptions = {}) {\n    const ocr = await BaseOcr.create(options);\n    if (options.debugOutputDir) {\n      await fs.mkdir(path.normalize(options.debugOutputDir), {\n        recursive: true,\n      });\n    }\n    (ocr as any).convert = convert.bind(ocr);\n    return ocr as BaseOcr & { convert: typeof convert };\n  }\n}\n", "import filePath from 'node:path';\nimport { ImageRawBase } from '@gutenye/ocr-common';\nimport type { ImageRawData, LineImage, SizeOption } from '@gutenye/ocr-common';\nimport sharp from 'sharp';\nexport class ImageRaw extends ImageRawBase {\n  #sharp!: sharp.Sharp;\n\n  static async open(path: string): Promise<ImageRaw> {\n    // let { raw } = await convertToRaw(path);\n    return new ImageRaw(await toImageRaw(path as any));\n  }\n\n  constructor(imageRawData: ImageRawData) {\n    super(imageRawData);\n    this.#sharp = toSharp(imageRawData);\n  }\n\n  async write(path: string) {\n    const ext = filePath.extname(path).slice(1);\n    return this.#sharp.toFormat(ext as keyof sharp.FormatEnum).toFile(path);\n  }\n\n  async resize(size: SizeOption) {\n    return this.#apply(\n      this.#sharp.resize({\n        width: size.width,\n        height: size.height,\n        fit: 'contain',\n      }),\n    );\n  }\n\n  async drawBox(lineImages: LineImage[]) {\n    const svg = `\n      <svg width=\"${this.width}\" height=\"${this.height}\">\n        ${lineImages\n          .map((lineImage) => {\n            const [p1, p2, p3, p4] = lineImage.box;\n            return `<polygon points=\"${p1[0]},${p1[1]} ${p2[0]},${p2[1]} ${p3[0]},${p3[1]} ${p4[0]},${p4[1]}\"  fill=\"none\" stroke=\"red\" />`;\n          })\n          .join('\\n')}\n      </svg>\n    `;\n    return this.#apply(\n      this.#sharp.composite([{ input: Buffer.from(svg), left: 0, top: 0 }]),\n    );\n  }\n\n  async #apply(sharp: sharp.Sharp) {\n    this.#sharp = sharp;\n    const result = await toImageRaw(sharp);\n    this.data = result.data;\n    this.width = result.width;\n    this.height = result.height;\n    return this;\n  }\n}\n\nasync function toImageRaw(sharp: sharp.Sharp) {\n  const result = await sharp.raw().toBuffer({ resolveWithObject: true });\n  return {\n    data: result.data,\n    width: result.info.width,\n    height: result.info.height,\n  };\n}\nfunction toSharp(imageRawData: ImageRawData) {\n  return sharp(imageRawData.data, {\n    raw: {\n      width: imageRawData.width,\n      height: imageRawData.height,\n      channels: 4,\n    },\n  });\n}\n", "import fs from 'node:fs/promises';\nimport { FileUtilsBase } from '@gutenye/ocr-common';\nimport { path } from '@cyia/vfs2';\nexport class FileUtils extends FileUtilsBase {\n  static override async read(filePath: string) {\n    return await fs.readFile(path.normalize(filePath), 'utf8');\n  }\n}\n", "import * as v from 'valibot';\nimport * as fs from 'fs/promises';\nimport { path } from '@cyia/vfs2';\nimport { fileTypeFromBuffer } from 'file-type';\nimport sharp from 'sharp';\nimport heicdecode from 'heic-decode';\nimport { decode } from 'bmp-js';\n\nconst BASE64_HEAD_REPLACE_REG = /^data:image\\/[\\w]+;base64,/;\n\nconst InputDefine = v.union([\n  v.pipe(\n    v.string(),\n    v.check((input) => BASE64_HEAD_REPLACE_REG.test(input)),\n    v.transform((base64) => {\n      const result = base64.match(BASE64_HEAD_REPLACE_REG)!;\n      return new Uint8Array(\n        Buffer.from(base64.slice(result[0].length), 'base64'),\n      );\n    }),\n  ),\n  v.pipe(\n    v.string(),\n    v.transform((filePath) =>\n      fs\n        .readFile(path.normalize(filePath))\n        .then((buffer) => new Uint8Array(buffer)),\n    ),\n  ),\n  v.pipe(v.custom<Uint8Array>((input) => input instanceof Uint8Array)),\n]);\nexport async function decodeToBuffer(input: string | Uint8Array) {\n  const buffer = await v.parse(InputDefine, input);\n  return buffer;\n}\n/**\n * ocr处理用\n * 支持路径,base64,uint8array\n */\nexport async function convertToRaw(input: string | Uint8Array) {\n  const buffer = await decodeToBuffer(input);\n  const type = await fileTypeFromBuffer(buffer);\n  if (!type) {\n    throw new Error(`不支持的图片类型`);\n  }\n  if (type.mime === 'image/bmp') {\n    const data = decode(Buffer.from(buffer));\n    const resolvedBuffer = data.data;\n    //ABGR =>RGBA\n    for (let i = 0; i < resolvedBuffer.length; i += 4) {\n      const alpha = resolvedBuffer[i];\n      const blue = resolvedBuffer[i + 1];\n      const green = resolvedBuffer[i + 2];\n      const red = resolvedBuffer[i + 3];\n      resolvedBuffer[i] = red;\n      resolvedBuffer[i + 1] = green;\n      resolvedBuffer[i + 2] = blue;\n      resolvedBuffer[i + 3] = (data as any).is_with_alpha ? alpha : 0xff;\n    }\n    const result = sharp(resolvedBuffer, {\n      raw: {\n        width: data.width,\n        height: data.height,\n        channels: 4,\n      },\n    }).ensureAlpha(1);\n    return { type: 'image/png', raw: result };\n  } else if (type?.mime === 'image/heic' || type?.mime === 'image/heif') {\n    const data = await heicdecode({\n      buffer: buffer as any,\n    });\n    const result = sharp(data.data, {\n      raw: {\n        width: data.width,\n        height: data.height,\n        channels: 4,\n      },\n    });\n    return { type: 'image/png', raw: result };\n  } else {\n    const result = sharp(buffer);\n    return { type: type.mime, raw: result };\n  }\n}\n// todo 未来其实应该直接是Buffer转通道颜色\n/**\n * 转换为兼容的图片格式\n */\nexport async function convertToCompatibleBuffer(input: string | Uint8Array) {\n  const result2 = await convertToRaw(input);\n\n  return {\n    type: result2.type,\n    buffer: new Uint8Array(await result2.raw.png().toBuffer()),\n  };\n}\n\nexport function bufferToImageBase64(input: {\n  type: string;\n  buffer: Uint8Array;\n}) {\n  return `data:${input.type};base64,${Buffer.from(input.buffer).toString('base64')}`;\n}\nexport function bufferToFileBase64(input: {\n  type: string;\n  buffer: Uint8Array;\n}) {\n  return Buffer.from(input.buffer).toString('base64');\n}\n", "import sharp from 'sharp';\nimport { getImageMetadata } from './image-metadata';\nexport async function imageExtract(\n  buffer: Buffer,\n  position: sharp.Region,\n  padding: number = 0,\n) {\n  let metadata =await getImageMetadata(buffer);\n  let left = Math.min(\n    Math.max(Math.round(position.left - padding), 0),\n    metadata.width,\n  );\n  let top = Math.min(\n    Math.max(Math.round(position.top - padding), 0),\n    metadata.height,\n  );\n  return sharp(buffer)\n    .extract({\n      left,\n      top,\n      width: Math.min(\n        Math.max(Math.round(position.width + padding * 2), 0),\n        metadata.width - left,\n      ),\n      height: Math.min(\n        Math.max(Math.round(position.height + padding * 2), 0),\n        metadata.height - top,\n      ),\n    })\n    .toBuffer();\n}\n", "import sharp from 'sharp';\n\nexport function getImageMetadata(buffer: Buffer) {\n  let instance = sharp(buffer);\n  return instance.metadata();\n}\n", "export const ModelConfig = [\n  {\n    label: '简体中文',\n    key: 'ch_mobile',\n    det: 'det/ch_PP-OCRv4_det_infer.onnx',\n    rec: 'rec/ch_PP-OCRv4_rec_infer.onnx',\n    dict: 'rec/ch_PP-OCRv4_rec_infer/ppocr_keys_v1.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '简体中文(服务器)',\n    key: 'ch_server',\n    det: 'det/ch_PP-OCRv4_det_server_infer.onnx',\n    rec: 'rec/ch_PP-OCRv4_rec_server_infer.onnx',\n    dict: 'rec/ch_PP-OCRv4_rec_server_infer/ppocr_keys_v1.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '繁體中文',\n    key: 'chinese_cht',\n    det: 'det/ch_PP-OCRv4_det_infer.onnx',\n    rec: 'rec/chinese_cht_PP-OCRv3_rec_infer.onnx',\n    dict: 'rec/chinese_cht_PP-OCRv3_rec_infer/chinese_cht_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '英文',\n    key: 'en_mobile',\n    det: 'det/en_PP-OCRv3_det_infer.onnx',\n    rec: 'rec/en_PP-OCRv4_rec_infer.onnx',\n    dict: 'rec/en_PP-OCRv4_rec_infer/en_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '阿拉伯文',\n    key: 'ar_mobile',\n    det: 'det/Multilingual_PP-OCRv3_det_infer.onnx',\n    rec: 'rec/arabic_PP-OCRv4_rec_infer.onnx',\n    dict: 'rec/arabic_PP-OCRv4_rec_infer/arabic_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '塞尔维亚文',\n    key: 'cyrillic_mobile',\n    det: 'det/Multilingual_PP-OCRv3_det_infer.onnx',\n    rec: 'rec/cyrillic_PP-OCRv3_rec_infer.onnx',\n    dict: 'rec/cyrillic_PP-OCRv3_rec_infer/cyrillic_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '梵文',\n    key: 'devanagari_mobile',\n    det: 'det/Multilingual_PP-OCRv3_det_infer.onnx',\n    rec: 'rec/devanagari_PP-OCRv4_rec_infer.onnx',\n    dict: 'rec/devanagari_PP-OCRv4_rec_infer/devanagari_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '日文',\n    key: 'japan_mobile',\n    det: 'det/Multilingual_PP-OCRv3_det_infer.onnx',\n    rec: 'rec/japan_PP-OCRv4_rec_infer.onnx',\n    dict: 'rec/japan_PP-OCRv4_rec_infer/japan_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '卡纳达语',\n    key: 'ka_mobile',\n    det: 'det/Multilingual_PP-OCRv3_det_infer.onnx',\n    rec: 'rec/ka_PP-OCRv4_rec_infer.onnx',\n    dict: 'rec/ka_PP-OCRv4_rec_infer/ka_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '韩文',\n    key: 'korean_mobile',\n    det: 'det/Multilingual_PP-OCRv3_det_infer.onnx',\n    rec: 'rec/korean_PP-OCRv4_rec_infer.onnx',\n    dict: 'rec/korean_PP-OCRv4_rec_infer/korean_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '拉丁文',\n    key: 'latin_mobile',\n    det: 'det/Multilingual_PP-OCRv3_det_infer.onnx',\n    rec: 'rec/latin_PP-OCRv3_rec_infer.onnx',\n    dict: 'rec/latin_PP-OCRv3_rec_infer/latin_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '泰米尔文',\n    key: 'ta_mobile',\n    det: 'det/Multilingual_PP-OCRv3_det_infer.onnx',\n    rec: 'rec/ta_PP-OCRv4_rec_infer.onnx',\n    dict: 'rec/ta_PP-OCRv4_rec_infer/ta_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n  {\n    label: '泰卢固文',\n    key: 'te_mobile',\n    det: 'det/Multilingual_PP-OCRv3_det_infer.onnx',\n    rec: 'rec/te_PP-OCRv4_rec_infer.onnx',\n    dict: 'rec/te_PP-OCRv4_rec_infer/te_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_infer.onnx',\n  },\n];\n"],
+  "sourcesContent": ["import BaseOcr, {\n  ModelCreateOptions,\n  registerBackend,\n} from '@gutenye/ocr-common';\nimport { splitIntoLineImages } from '@gutenye/ocr-common/splitIntoLineImages';\nimport { ImageRaw } from './ImageRaw';\nimport { FileUtils } from './FileUtils';\nimport { InferenceSession } from 'onnxruntime-node';\nimport fs from 'fs/promises';\nimport { path } from '@cyia/vfs2';\nimport { convertToRaw } from '../image';\nimport * as v from 'valibot';\nconst ImageAdjustDefine = v.object({\n  padding: v.pipe(\n    v.optional(\n      v.union([\n        v.pipe(\n          v.number(),\n          v.transform((item) => ({\n            top: item,\n            left: item,\n            right: item,\n            bottom: item,\n          })),\n        ),\n        v.object({\n          left: v.number(),\n          right: v.number(),\n          top: v.number(),\n          bottom: v.number(),\n        }),\n      ]),\n      { top: 50, right: 50, left: 50, bottom: 50 },\n    ),\n  ),\n  maxSideLen: v.optional(v.union([v.pipe(v.number())]), 1920),\n  // threshold: v.optional(v.union([v.pipe(v.number())]), 0.3),\n});\nexport type ImageAdjustType = v.InferInput<typeof ImageAdjustDefine>;\nregisterBackend({\n  FileUtils,\n  ImageRaw,\n  InferenceSession,\n  splitIntoLineImages,\n  defaultModels: undefined,\n});\n\nasync function convert(\n  this: BaseOcr,\n  input: string | Uint8Array,\n  options: ImageAdjustType = {},\n) {\n  const resolveOptions = v.parse(ImageAdjustDefine, options);\n  //100 80 0.8\n  // 50 40\n  let { raw } = await convertToRaw(input);\n  const metadata = await raw.metadata();\n  const maxSize = Math.max(metadata.width!, metadata.height!);\n  if (maxSize > resolveOptions.maxSideLen) {\n    let ratio = metadata.width! / metadata.height!;\n    ratio = ratio > 1 ? 1 / ratio : ratio;\n    raw = raw.resize({\n      width: Math.round(\n        maxSize === metadata.width!\n          ? resolveOptions.maxSideLen\n          : ratio * resolveOptions.maxSideLen,\n      ),\n      height: Math.round(\n        maxSize === metadata.height!\n          ? resolveOptions.maxSideLen\n          : ratio * resolveOptions.maxSideLen,\n      ),\n    });\n  }\n  raw = raw.extend({ ...resolveOptions.padding, background: '#fff' });\n  raw = raw.ensureAlpha(1);\n  return this.detect(raw as any);\n}\nexport class Ocr extends BaseOcr {\n  static override async create(options: ModelCreateOptions = {}) {\n    const ocr = await BaseOcr.create(options);\n    if (options.debugOutputDir) {\n      await fs.mkdir(path.normalize(options.debugOutputDir), {\n        recursive: true,\n      });\n    }\n    (ocr as any).convert = convert.bind(ocr);\n    return ocr as BaseOcr & { convert: typeof convert };\n  }\n}\n", "import filePath from 'node:path';\nimport { ImageRawBase } from '@gutenye/ocr-common';\nimport type { ImageRawData, LineImage, SizeOption } from '@gutenye/ocr-common';\nimport sharp from 'sharp';\nexport class ImageRaw extends ImageRawBase {\n  #sharp!: sharp.Sharp;\n\n  static async open(path: string): Promise<ImageRaw> {\n    // let { raw } = await convertToRaw(path);\n    return new ImageRaw(await toImageRaw(path as any));\n  }\n\n  constructor(imageRawData: ImageRawData) {\n    super(imageRawData);\n    this.#sharp = toSharp(imageRawData);\n  }\n\n  async write(path: string) {\n    const ext = filePath.extname(path).slice(1);\n    return this.#sharp.toFormat(ext as keyof sharp.FormatEnum).toFile(path);\n  }\n\n  async resize(size: SizeOption) {\n    return this.#apply(\n      this.#sharp.resize({\n        width: size.width,\n        height: size.height,\n        fit: 'contain',\n      }),\n    );\n  }\n\n  async drawBox(lineImages: LineImage[]) {\n    const svg = `\n      <svg width=\"${this.width}\" height=\"${this.height}\">\n        ${lineImages\n          .map((lineImage) => {\n            const [p1, p2, p3, p4] = lineImage.box;\n            return `<polygon points=\"${p1[0]},${p1[1]} ${p2[0]},${p2[1]} ${p3[0]},${p3[1]} ${p4[0]},${p4[1]}\"  fill=\"none\" stroke=\"red\" />`;\n          })\n          .join('\\n')}\n      </svg>\n    `;\n    return this.#apply(\n      this.#sharp.composite([{ input: Buffer.from(svg), left: 0, top: 0 }]),\n    );\n  }\n\n  async #apply(sharp: sharp.Sharp) {\n    this.#sharp = sharp;\n    const result = await toImageRaw(sharp);\n    this.data = result.data;\n    this.width = result.width;\n    this.height = result.height;\n    return this;\n  }\n}\n\nasync function toImageRaw(sharp: sharp.Sharp) {\n  const result = await sharp.raw().toBuffer({ resolveWithObject: true });\n  return {\n    data: result.data,\n    width: result.info.width,\n    height: result.info.height,\n  };\n}\nfunction toSharp(imageRawData: ImageRawData) {\n  return sharp(imageRawData.data, {\n    raw: {\n      width: imageRawData.width,\n      height: imageRawData.height,\n      channels: 4,\n    },\n  });\n}\n", "import fs from 'node:fs/promises';\nimport { FileUtilsBase } from '@gutenye/ocr-common';\nimport { path } from '@cyia/vfs2';\nexport class FileUtils extends FileUtilsBase {\n  static override async read(filePath: string) {\n    return await fs.readFile(path.normalize(filePath), 'utf8');\n  }\n}\n", "import * as v from 'valibot';\nimport * as fs from 'fs/promises';\nimport { path } from '@cyia/vfs2';\nimport { fileTypeFromBuffer } from 'file-type';\nimport sharp from 'sharp';\nimport heicdecode from 'heic-decode';\nimport { decode } from 'bmp-js';\n\nconst BASE64_HEAD_REPLACE_REG = /^data:image\\/[\\w]+;base64,/;\n\nconst InputDefine = v.union([\n  v.pipe(\n    v.string(),\n    v.check((input) => BASE64_HEAD_REPLACE_REG.test(input)),\n    v.transform((base64) => {\n      const result = base64.match(BASE64_HEAD_REPLACE_REG)!;\n      return new Uint8Array(\n        Buffer.from(base64.slice(result[0].length), 'base64'),\n      );\n    }),\n  ),\n  v.pipe(\n    v.string(),\n    v.transform((filePath) =>\n      fs\n        .readFile(path.normalize(filePath))\n        .then((buffer) => new Uint8Array(buffer)),\n    ),\n  ),\n  v.pipe(v.custom<Uint8Array>((input) => input instanceof Uint8Array)),\n]);\nexport async function decodeToBuffer(input: string | Uint8Array) {\n  const buffer = await v.parse(InputDefine, input);\n  return buffer;\n}\n/**\n * ocr处理用\n * 支持路径,base64,uint8array\n */\nexport async function convertToRaw(input: string | Uint8Array) {\n  const buffer = await decodeToBuffer(input);\n  const type = await fileTypeFromBuffer(buffer);\n  if (!type) {\n    throw new Error(`不支持的图片类型`);\n  }\n  if (type.mime === 'image/bmp') {\n    const data = decode(Buffer.from(buffer));\n    const resolvedBuffer = data.data;\n    //ABGR =>RGBA\n    for (let i = 0; i < resolvedBuffer.length; i += 4) {\n      const alpha = resolvedBuffer[i];\n      const blue = resolvedBuffer[i + 1];\n      const green = resolvedBuffer[i + 2];\n      const red = resolvedBuffer[i + 3];\n      resolvedBuffer[i] = red;\n      resolvedBuffer[i + 1] = green;\n      resolvedBuffer[i + 2] = blue;\n      resolvedBuffer[i + 3] = (data as any).is_with_alpha ? alpha : 0xff;\n    }\n    const result = sharp(resolvedBuffer, {\n      raw: {\n        width: data.width,\n        height: data.height,\n        channels: 4,\n      },\n    }).ensureAlpha(1);\n    return { type: 'image/png', raw: result };\n  } else if (type?.mime === 'image/heic' || type?.mime === 'image/heif') {\n    const data = await heicdecode({\n      buffer: buffer as any,\n    });\n    const result = sharp(data.data, {\n      raw: {\n        width: data.width,\n        height: data.height,\n        channels: 4,\n      },\n    });\n    return { type: 'image/png', raw: result };\n  } else {\n    const result = sharp(buffer);\n    return { type: type.mime, raw: result };\n  }\n}\n// todo 未来其实应该直接是Buffer转通道颜色\n/**\n * 转换为兼容的图片格式\n */\nexport async function convertToCompatibleBuffer(input: string | Uint8Array) {\n  const result2 = await convertToRaw(input);\n\n  return {\n    type: result2.type,\n    buffer: new Uint8Array(await result2.raw.png().toBuffer()),\n  };\n}\n\nexport function bufferToImageBase64(input: {\n  type: string;\n  buffer: Uint8Array;\n}) {\n  return `data:${input.type};base64,${Buffer.from(input.buffer).toString('base64')}`;\n}\nexport function bufferToFileBase64(input: {\n  type: string;\n  buffer: Uint8Array;\n}) {\n  return Buffer.from(input.buffer).toString('base64');\n}\n", "import sharp from 'sharp';\nimport { getImageMetadata } from './image-metadata';\nexport async function imageExtract(\n  buffer: Buffer,\n  position: sharp.Region,\n  padding: number = 0,\n) {\n  let metadata =await getImageMetadata(buffer);\n  let left = Math.min(\n    Math.max(Math.round(position.left - padding), 0),\n    metadata.width,\n  );\n  let top = Math.min(\n    Math.max(Math.round(position.top - padding), 0),\n    metadata.height,\n  );\n  return sharp(buffer)\n    .extract({\n      left,\n      top,\n      width: Math.min(\n        Math.max(Math.round(position.width + padding * 2), 0),\n        metadata.width - left,\n      ),\n      height: Math.min(\n        Math.max(Math.round(position.height + padding * 2), 0),\n        metadata.height - top,\n      ),\n    })\n    .toBuffer();\n}\n", "import sharp from 'sharp';\n\nexport function getImageMetadata(buffer: Buffer) {\n  let instance = sharp(buffer);\n  return instance.metadata();\n}\n", "export const ModelConfig = [\n  {\n    label: '简体中文',\n    key: 'ch_mobile',\n    det: 'det/ch_PP-OCRv4_det_mobile.onnx',\n    rec: 'rec/ch_PP-OCRv4_rec_mobile.onnx',\n    dict: 'rec/ch_PP-OCRv4_rec_mobile/ppocr_keys_v1.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '简体中文(服务器)',\n    key: 'ch_server',\n    det: 'det/ch_PP-OCRv4_det_server.onnx',\n    rec: 'rec/ch_PP-OCRv4_rec_server.onnx',\n    dict: 'rec/ch_PP-OCRv4_rec_server/ppocr_keys_v1.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '繁體中文',\n    key: 'chinese_cht',\n    det: 'det/ch_PP-OCRv4_det_server.onnx',\n    rec: 'rec/chinese_cht_PP-OCRv3_rec_mobile.onnx',\n    dict: 'rec/chinese_cht_PP-OCRv3_rec_mobile/chinese_cht_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '英文',\n    key: 'en_mobile',\n    det: 'det/en_PP-OCRv3_det_mobile.onnx',\n    rec: 'rec/en_PP-OCRv4_rec_mobile.onnx',\n    dict: 'rec/en_PP-OCRv4_rec_mobile/en_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '阿拉伯文',\n    key: 'ar_mobile',\n    det: 'det/multi_PP-OCRv3_det_mobile.onnx',\n    rec: 'rec/arabic_PP-OCRv4_rec_mobile.onnx',\n    dict: 'rec/arabic_PP-OCRv4_rec_mobile/arabic_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '塞尔维亚文',\n    key: 'cyrillic_mobile',\n    det: 'det/multi_PP-OCRv3_det_mobile.onnx',\n    rec: 'rec/cyrillic_PP-OCRv3_rec_mobile.onnx',\n    dict: 'rec/cyrillic_PP-OCRv3_rec_mobile/cyrillic_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '梵文',\n    key: 'devanagari_mobile',\n    det: 'det/multi_PP-OCRv3_det_mobile.onnx',\n    rec: 'rec/devanagari_PP-OCRv4_rec_mobile.onnx',\n    dict: 'rec/devanagari_PP-OCRv4_rec_mobile/devanagari_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '日文',\n    key: 'japan_mobile',\n    det: 'det/multi_PP-OCRv3_det_mobile.onnx',\n    rec: 'rec/japan_PP-OCRv4_rec_mobile.onnx',\n    dict: 'rec/japan_PP-OCRv4_rec_mobile/japan_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '卡纳达语',\n    key: 'ka_mobile',\n    det: 'det/multi_PP-OCRv3_det_mobile.onnx',\n    rec: 'rec/ka_PP-OCRv4_rec_mobile.onnx',\n    dict: 'rec/ka_PP-OCRv4_rec_mobile/ka_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '韩文',\n    key: 'korean_mobile',\n    det: 'det/multi_PP-OCRv3_det_mobile.onnx',\n    rec: 'rec/korean_PP-OCRv4_rec_mobile.onnx',\n    dict: 'rec/korean_PP-OCRv4_rec_mobile/korean_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '拉丁文',\n    key: 'latin_mobile',\n    det: 'det/multi_PP-OCRv3_det_mobile.onnx',\n    rec: 'rec/latin_PP-OCRv3_rec_mobile.onnx',\n    dict: 'rec/latin_PP-OCRv3_rec_mobile/latin_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '泰米尔文',\n    key: 'ta_mobile',\n    det: 'det/multi_PP-OCRv3_det_mobile.onnx',\n    rec: 'rec/ta_PP-OCRv4_rec_mobile.onnx',\n    dict: 'rec/ta_PP-OCRv4_rec_mobile/ta_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n  {\n    label: '泰卢固文',\n    key: 'te_mobile',\n    det: 'det/multi_PP-OCRv3_det_mobile.onnx',\n    rec: 'rec/te_PP-OCRv4_rec_mobile.onnx',\n    dict: 'rec/te_PP-OCRv4_rec_mobile/te_dict.txt',\n    cls: 'cls/ch_ppocr_mobile_v2.0_cls_mobile.onnx',\n  },\n];\n"],
   "mappings": ";AAAA,OAAO;AAAA,EAEL;AAAA,OACK;AACP,SAAS,2BAA2B;;;ACJpC,OAAO,cAAc;AACrB,SAAS,oBAAoB;AAE7B,OAAO,WAAW;AACX,IAAM,WAAN,MAAM,kBAAiB,aAAa;AAAA,EACzC;AAAA,EAEA,aAAa,KAAKA,OAAiC;AAEjD,WAAO,IAAI,UAAS,MAAM,WAAWA,KAAW,CAAC;AAAA,EACnD;AAAA,EAEA,YAAY,cAA4B;AACtC,UAAM,YAAY;AAClB,SAAK,SAAS,QAAQ,YAAY;AAAA,EACpC;AAAA,EAEA,MAAM,MAAMA,OAAc;AACxB,UAAM,MAAM,SAAS,QAAQA,KAAI,EAAE,MAAM,CAAC;AAC1C,WAAO,KAAK,OAAO,SAAS,GAA6B,EAAE,OAAOA,KAAI;AAAA,EACxE;AAAA,EAEA,MAAM,OAAO,MAAkB;AAC7B,WAAO,KAAK;AAAA,MACV,KAAK,OAAO,OAAO;AAAA,QACjB,OAAO,KAAK;AAAA,QACZ,QAAQ,KAAK;AAAA,QACb,KAAK;AAAA,MACP,CAAC;AAAA,IACH;AAAA,EACF;AAAA,EAEA,MAAM,QAAQ,YAAyB;AACrC,UAAM,MAAM;AAAA,oBACI,KAAK,KAAK,aAAa,KAAK,MAAM;AAAA,UAC5C,WACC,IAAI,CAAC,cAAc;AAClB,YAAM,CAAC,IAAI,IAAI,IAAI,EAAE,IAAI,UAAU;AACnC,aAAO,oBAAoB,GAAG,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC;AAAA,IACjG,CAAC,EACA,KAAK,IAAI,CAAC;AAAA;AAAA;AAGjB,WAAO,KAAK;AAAA,MACV,KAAK,OAAO,UAAU,CAAC,EAAE,OAAO,OAAO,KAAK,GAAG,GAAG,MAAM,GAAG,KAAK,EAAE,CAAC,CAAC;AAAA,IACtE;AAAA,EACF;AAAA,EAEA,MAAM,OAAOC,QAAoB;AAC/B,SAAK,SAASA;AACd,UAAM,SAAS,MAAM,WAAWA,MAAK;AACrC,SAAK,OAAO,OAAO;AACnB,SAAK,QAAQ,OAAO;AACpB,SAAK,SAAS,OAAO;AACrB,WAAO;AAAA,EACT;AACF;AAEA,eAAe,WAAWA,QAAoB;AAC5C,QAAM,SAAS,MAAMA,OAAM,IAAI,EAAE,SAAS,EAAE,mBAAmB,KAAK,CAAC;AACrE,SAAO;AAAA,IACL,MAAM,OAAO;AAAA,IACb,OAAO,OAAO,KAAK;AAAA,IACnB,QAAQ,OAAO,KAAK;AAAA,EACtB;AACF;AACA,SAAS,QAAQ,cAA4B;AAC3C,SAAO,MAAM,aAAa,MAAM;AAAA,IAC9B,KAAK;AAAA,MACH,OAAO,aAAa;AAAA,MACpB,QAAQ,aAAa;AAAA,MACrB,UAAU;AAAA,IACZ;AAAA,EACF,CAAC;AACH;;;AC1EA,OAAO,QAAQ;AACf,SAAS,qBAAqB;AAC9B,SAAS,YAAY;AACd,IAAM,YAAN,cAAwB,cAAc;AAAA,EAC3C,aAAsB,KAAKC,WAAkB;AAC3C,WAAO,MAAM,GAAG,SAAS,KAAK,UAAUA,SAAQ,GAAG,MAAM;AAAA,EAC3D;AACF;;;AFAA,SAAS,wBAAwB;AACjC,OAAOC,SAAQ;AACf,SAAS,QAAAC,aAAY;;;AGTrB,YAAY,OAAO;AACnB,YAAYC,SAAQ;AACpB,SAAS,QAAAC,aAAY;AACrB,SAAS,0BAA0B;AACnC,OAAOC,YAAW;AAClB,OAAO,gBAAgB;AACvB,SAAS,cAAc;AAEvB,IAAM,0BAA0B;AAEhC,IAAM,cAAgB,QAAM;AAAA,EACxB;AAAA,IACE,SAAO;AAAA,IACP,QAAM,CAAC,UAAU,wBAAwB,KAAK,KAAK,CAAC;AAAA,IACpD,YAAU,CAAC,WAAW;AACtB,YAAM,SAAS,OAAO,MAAM,uBAAuB;AACnD,aAAO,IAAI;AAAA,QACT,OAAO,KAAK,OAAO,MAAM,OAAO,CAAC,EAAE,MAAM,GAAG,QAAQ;AAAA,MACtD;AAAA,IACF,CAAC;AAAA,EACH;AAAA,EACE;AAAA,IACE,SAAO;AAAA,IACP;AAAA,MAAU,CAACC,cAER,aAASF,MAAK,UAAUE,SAAQ,CAAC,EACjC,KAAK,CAAC,WAAW,IAAI,WAAW,MAAM,CAAC;AAAA,IAC5C;AAAA,EACF;AAAA,EACE,OAAO,SAAmB,CAAC,UAAU,iBAAiB,UAAU,CAAC;AACrE,CAAC;AACD,eAAsB,eAAe,OAA4B;AAC/D,QAAM,SAAS,MAAQ,QAAM,aAAa,KAAK;AAC/C,SAAO;AACT;AAKA,eAAsB,aAAa,OAA4B;AAC7D,QAAM,SAAS,MAAM,eAAe,KAAK;AACzC,QAAM,OAAO,MAAM,mBAAmB,MAAM;AAC5C,MAAI,CAAC,MAAM;AACT,UAAM,IAAI,MAAM,UAAU;AAAA,EAC5B;AACA,MAAI,KAAK,SAAS,aAAa;AAC7B,UAAM,OAAO,OAAO,OAAO,KAAK,MAAM,CAAC;AACvC,UAAM,iBAAiB,KAAK;AAE5B,aAAS,IAAI,GAAG,IAAI,eAAe,QAAQ,KAAK,GAAG;AACjD,YAAM,QAAQ,eAAe,CAAC;AAC9B,YAAM,OAAO,eAAe,IAAI,CAAC;AACjC,YAAM,QAAQ,eAAe,IAAI,CAAC;AAClC,YAAM,MAAM,eAAe,IAAI,CAAC;AAChC,qBAAe,CAAC,IAAI;AACpB,qBAAe,IAAI,CAAC,IAAI;AACxB,qBAAe,IAAI,CAAC,IAAI;AACxB,qBAAe,IAAI,CAAC,IAAK,KAAa,gBAAgB,QAAQ;AAAA,IAChE;AACA,UAAM,SAASD,OAAM,gBAAgB;AAAA,MACnC,KAAK;AAAA,QACH,OAAO,KAAK;AAAA,QACZ,QAAQ,KAAK;AAAA,QACb,UAAU;AAAA,MACZ;AAAA,IACF,CAAC,EAAE,YAAY,CAAC;AAChB,WAAO,EAAE,MAAM,aAAa,KAAK,OAAO;AAAA,EAC1C,WAAW,MAAM,SAAS,gBAAgB,MAAM,SAAS,cAAc;AACrE,UAAM,OAAO,MAAM,WAAW;AAAA,MAC5B;AAAA,IACF,CAAC;AACD,UAAM,SAASA,OAAM,KAAK,MAAM;AAAA,MAC9B,KAAK;AAAA,QACH,OAAO,KAAK;AAAA,QACZ,QAAQ,KAAK;AAAA,QACb,UAAU;AAAA,MACZ;AAAA,IACF,CAAC;AACD,WAAO,EAAE,MAAM,aAAa,KAAK,OAAO;AAAA,EAC1C,OAAO;AACL,UAAM,SAASA,OAAM,MAAM;AAC3B,WAAO,EAAE,MAAM,KAAK,MAAM,KAAK,OAAO;AAAA,EACxC;AACF;;;ACnFA,OAAOE,YAAW;;;ACAlB,OAAOC,YAAW;;;ALWlB,YAAYC,QAAO;AACnB,IAAM,oBAAsB,UAAO;AAAA,EACjC,SAAW;AAAA,IACP;AAAA,MACE,SAAM;AAAA,QACJ;AAAA,UACE,UAAO;AAAA,UACP,aAAU,CAAC,UAAU;AAAA,YACrB,KAAK;AAAA,YACL,MAAM;AAAA,YACN,OAAO;AAAA,YACP,QAAQ;AAAA,UACV,EAAE;AAAA,QACJ;AAAA,QACE,UAAO;AAAA,UACP,MAAQ,UAAO;AAAA,UACf,OAAS,UAAO;AAAA,UAChB,KAAO,UAAO;AAAA,UACd,QAAU,UAAO;AAAA,QACnB,CAAC;AAAA,MACH,CAAC;AAAA,MACD,EAAE,KAAK,IAAI,OAAO,IAAI,MAAM,IAAI,QAAQ,GAAG;AAAA,IAC7C;AAAA,EACF;AAAA,EACA,YAAc,YAAW,SAAM,CAAG,QAAO,UAAO,CAAC,CAAC,CAAC,GAAG,IAAI;AAAA;AAE5D,CAAC;AAED,gBAAgB;AAAA,EACd;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA,eAAe;AACjB,CAAC;AAED,eAAe,QAEb,OACA,UAA2B,CAAC,GAC5B;AACA,QAAM,iBAAmB,SAAM,mBAAmB,OAAO;AAGzD,MAAI,EAAE,IAAI,IAAI,MAAM,aAAa,KAAK;AACtC,QAAM,WAAW,MAAM,IAAI,SAAS;AACpC,QAAM,UAAU,KAAK,IAAI,SAAS,OAAQ,SAAS,MAAO;AAC1D,MAAI,UAAU,eAAe,YAAY;AACvC,QAAI,QAAQ,SAAS,QAAS,SAAS;AACvC,YAAQ,QAAQ,IAAI,IAAI,QAAQ;AAChC,UAAM,IAAI,OAAO;AAAA,MACf,OAAO,KAAK;AAAA,QACV,YAAY,SAAS,QACjB,eAAe,aACf,QAAQ,eAAe;AAAA,MAC7B;AAAA,MACA,QAAQ,KAAK;AAAA,QACX,YAAY,SAAS,SACjB,eAAe,aACf,QAAQ,eAAe;AAAA,MAC7B;AAAA,IACF,CAAC;AAAA,EACH;AACA,QAAM,IAAI,OAAO,EAAE,GAAG,eAAe,SAAS,YAAY,OAAO,CAAC;AAClE,QAAM,IAAI,YAAY,CAAC;AACvB,SAAO,KAAK,OAAO,GAAU;AAC/B;AACO,IAAM,MAAN,cAAkB,QAAQ;AAAA,EAC/B,aAAsB,OAAO,UAA8B,CAAC,GAAG;AAC7D,UAAM,MAAM,MAAM,QAAQ,OAAO,OAAO;AACxC,QAAI,QAAQ,gBAAgB;AAC1B,YAAMC,IAAG,MAAMC,MAAK,UAAU,QAAQ,cAAc,GAAG;AAAA,QACrD,WAAW;AAAA,MACb,CAAC;AAAA,IACH;AACA,IAAC,IAAY,UAAU,QAAQ,KAAK,GAAG;AACvC,WAAO;AAAA,EACT;AACF;;;AMzFO,IAAM,cAAc;AAAA,EACzB;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AAAA,EACA;AAAA,IACE,OAAO;AAAA,IACP,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AAAA,IACL,MAAM;AAAA,IACN,KAAK;AAAA,EACP;AACF;",
   "names": ["path", "sharp", "filePath", "fs", "path", "fs", "path", "sharp", "filePath", "sharp", "sharp", "v", "fs", "path"]
 }

package/package.json

@@ -1,11 +1,11 @@
 {
   "name": "@shenghuabi/knowledge",
-  "version": "1.0.22",
+  "version": "1.0.24",
   "description": "知识库",
   "author": "wszgrcy",
   "sideEffects": false,
   "peerDependencies": {
-    "@cyia/vfs2": "^1.1.26",
+    "@cyia/vfs2": "^1.5.3",
     "handlebars": "^4.7.8",
     "lru-cache": "^11.2.1",
     "rfdc": "^1.4.1",
@@ -19,14 +19,14 @@
     "html-to-text": "^9.0.5",
     "fastq": "^1.19.1",
     "sharp": "0.34.2",
-    "@cyia/dl": "^1.0.15",
-    "@cyia/external-call": "^1.0.35"
+    "@cyia/dl": "^1.5.3",
+    "@cyia/external-call": "^1.5.3"
   },
   "dependencies": {
     "html-entities": "^2.6.0",
     "@qdrant/qdrant-js": "1.15.1",
     "@cyia/mdict-reader": "^1.0.9",
-    "@langchain/community": "^1.1.1",
+    "@langchain/community": "1.1.1",
     "@langchain/core": "1.1.8",
     "@langchain/textsplitters": "^1.0.1",
     "@xhmikosr/decompress-tarbz2": "^8.0.2",
@@ -47,7 +47,7 @@
     "@gutenye/ocr-common": "^1.4.8",
     "bmp-js": "^0.1.0",
     "onnxruntime-node": "1.20.1",
-    "@huggingface/transformers": "3.7.2",
+    "@huggingface/transformers": "4.2.0",
     "xlsx": "^0.18.5",
     "pdfjs-dist": "^5.4.449"
   },

package/worker/custom-cache.d.ts

@@ -1,4 +1,3 @@
-import fs from 'fs';
 import { InitOptions } from './set-transformers-config';
 export interface NodeProxy {
     match: (request: string) => Promise<ArrayBuffer | undefined>;
@@ -7,22 +6,6 @@ export interface NodeProxy {
 export declare class FileProxyCache {
     #private;
     constructor(initOptions: InitOptions);
-    match(request: string): Promise<FileResponse | undefined>;
-    put(request: string, response: Response | FileResponse): Promise<void>;
+    match(request: string): Promise<Response | undefined>;
+    put(request: string, response: Response): Promise<void>;
 }
-declare class FileResponse {
-    filePath: string;
-    headers: import("undici-types").Headers;
-    exists: boolean;
-    status: number;
-    statusText: string;
-    body: fs.ReadStream;
-    constructor(filePath: string);
-    updateContentType(): void;
-    clone(): FileResponse;
-    arrayBuffer(): Promise<ArrayBuffer>;
-    blob(): Promise<Blob>;
-    text(): Promise<string>;
-    json(): Promise<object>;
-}
-export {};

package/worker/ocr/index.d.ts

@@ -3,7 +3,7 @@ import { MessagePort } from 'worker_threads';
 declare function init(ocrConfig: {
     key: string;
     modelDir: string;
-    port: MessagePort;
+    port?: MessagePort;
 }): Promise<void>;
 declare function convert(input: {
     filePath: string | Uint8Array;

package/worker/ocr.mjs

@@ -9,7 +9,7 @@ var BaseUrl = "https://www.modelscope.cn/models/RapidAI/RapidOCR/resolve/master/
 var DictUrl = "https://www.modelscope.cn/models/RapidAI/RapidOCR/resolve/master/paddle/PP-OCRv4";
 async function init(ocrConfig) {
   const messageCb = (message) => {
-    ocrConfig.port.postMessage({ type: "progress", message });
+    ocrConfig.port?.postMessage({ type: "progress", message });
   };
   const modelConfig = ModelConfig.find((item) => item.key === ocrConfig.key);
   const fs = createNormalizeVfs({ dir: ocrConfig.modelDir });

package/worker/ocr.mjs.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../packages/worker/ocr/index.ts"],
-  "sourcesContent": ["import { ImageAdjustType, Ocr, ModelConfig } from '@shenghuabi/knowledge/ocr';\nimport { createNormalizeVfs, path } from '@cyia/vfs2';\n// import * as ort from 'onnxruntime-node';\nimport { MessagePort } from 'worker_threads';\nimport { getUniqueObjectKey } from '@shenghuabi/knowledge/util';\nimport { downloadFile } from '@cyia/dl';\nlet key!: string;\nlet ocrInstance: ReturnType<(typeof Ocr)['create']>;\nconst BaseUrl =\n  'https://www.modelscope.cn/models/RapidAI/RapidOCR/resolve/master/onnx/PP-OCRv4';\n\nconst DictUrl =\n  'https://www.modelscope.cn/models/RapidAI/RapidOCR/resolve/master/paddle/PP-OCRv4';\nasync function init(ocrConfig: {\n  key: string;\n  modelDir: string;\n  port: MessagePort;\n}) {\n  const messageCb = (message: any) => {\n    ocrConfig.port.postMessage({ type: 'progress', message });\n  };\n  const modelConfig = ModelConfig.find((item) => item.key === ocrConfig.key)!;\n  const fs = createNormalizeVfs({ dir: ocrConfig.modelDir });\n  // 自动下载模型\n  const absDetectionPath = path.join(ocrConfig.modelDir, modelConfig.det);\n  if (!(await fs.exists(modelConfig.det))) {\n    await downloadFile(`${BaseUrl}/${modelConfig.det}`, {\n      savePath: absDetectionPath,\n      message: messageCb,\n    });\n  }\n  const absRecognitionPath = path.join(ocrConfig.modelDir, modelConfig.rec);\n  if (!(await fs.exists(modelConfig.rec))) {\n    await downloadFile(`${BaseUrl}/${modelConfig.rec}`, {\n      savePath: absRecognitionPath,\n      message: messageCb,\n    });\n  }\n  const absDictionaryPath = path.join(ocrConfig.modelDir, modelConfig.dict);\n  if (!(await fs.exists(modelConfig.dict))) {\n    await downloadFile(`${DictUrl}/${modelConfig.dict}`, {\n      savePath: absDictionaryPath,\n      message: messageCb,\n    });\n  }\n}\n// 改为init和convert\nasync function convert(input: {\n  filePath: string | Uint8Array;\n  ocrConfig: { key: string; modelDir: string; device?: 'dml' | 'cuda' | 'cpu' };\n  options?: ImageAdjustType;\n}) {\n  const inputKey = getUniqueObjectKey(input.ocrConfig);\n  if (key !== inputKey) {\n    const modelConfig = ModelConfig.find(\n      (item) => item.key === input.ocrConfig.key,\n    )!;\n    // 自动下载模型\n    const absDetectionPath = path.join(\n      input.ocrConfig.modelDir,\n      modelConfig.det,\n    );\n    const absRecognitionPath = path.join(\n      input.ocrConfig.modelDir,\n      modelConfig.rec,\n    );\n    const absDictionaryPath = path.join(\n      input.ocrConfig.modelDir,\n      modelConfig.dict,\n    );\n    ocrInstance = Ocr.create({\n      onnxOptions: {\n        executionProviders: input.ocrConfig.device\n          ? [input.ocrConfig.device]\n          : ['dml', 'cuda', 'cpu'],\n        executionMode: 'parallel',\n      },\n      models: {\n        detectionPath: absDetectionPath,\n        recognitionPath: absRecognitionPath,\n        dictionaryPath: absDictionaryPath,\n      },\n    });\n    key = inputKey;\n  }\n  return (await ocrInstance).convert(input.filePath, input.options);\n}\n\nexport { init, convert };\n"],
-  "mappings": ";AAAA,SAA0B,KAAK,mBAAmB;AAClD,SAAS,oBAAoB,YAAY;AAGzC,SAAS,0BAA0B;AACnC,SAAS,oBAAoB;AAC7B,IAAI;AACJ,IAAI;AACJ,IAAM,UACJ;AAEF,IAAM,UACJ;AACF,eAAe,KAAK,WAIjB;AACD,QAAM,YAAY,CAAC,YAAiB;AAClC,cAAU,KAAK,YAAY,EAAE,MAAM,YAAY,QAAQ,CAAC;AAAA,EAC1D;AACA,QAAM,cAAc,YAAY,KAAK,CAAC,SAAS,KAAK,QAAQ,UAAU,GAAG;AACzE,QAAM,KAAK,mBAAmB,EAAE,KAAK,UAAU,SAAS,CAAC;AAEzD,QAAM,mBAAmB,KAAK,KAAK,UAAU,UAAU,YAAY,GAAG;AACtE,MAAI,CAAE,MAAM,GAAG,OAAO,YAAY,GAAG,GAAI;AACvC,UAAM,aAAa,GAAG,OAAO,IAAI,YAAY,GAAG,IAAI;AAAA,MAClD,UAAU;AAAA,MACV,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AACA,QAAM,qBAAqB,KAAK,KAAK,UAAU,UAAU,YAAY,GAAG;AACxE,MAAI,CAAE,MAAM,GAAG,OAAO,YAAY,GAAG,GAAI;AACvC,UAAM,aAAa,GAAG,OAAO,IAAI,YAAY,GAAG,IAAI;AAAA,MAClD,UAAU;AAAA,MACV,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AACA,QAAM,oBAAoB,KAAK,KAAK,UAAU,UAAU,YAAY,IAAI;AACxE,MAAI,CAAE,MAAM,GAAG,OAAO,YAAY,IAAI,GAAI;AACxC,UAAM,aAAa,GAAG,OAAO,IAAI,YAAY,IAAI,IAAI;AAAA,MACnD,UAAU;AAAA,MACV,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AACF;AAEA,eAAe,QAAQ,OAIpB;AACD,QAAM,WAAW,mBAAmB,MAAM,SAAS;AACnD,MAAI,QAAQ,UAAU;AACpB,UAAM,cAAc,YAAY;AAAA,MAC9B,CAAC,SAAS,KAAK,QAAQ,MAAM,UAAU;AAAA,IACzC;AAEA,UAAM,mBAAmB,KAAK;AAAA,MAC5B,MAAM,UAAU;AAAA,MAChB,YAAY;AAAA,IACd;AACA,UAAM,qBAAqB,KAAK;AAAA,MAC9B,MAAM,UAAU;AAAA,MAChB,YAAY;AAAA,IACd;AACA,UAAM,oBAAoB,KAAK;AAAA,MAC7B,MAAM,UAAU;AAAA,MAChB,YAAY;AAAA,IACd;AACA,kBAAc,IAAI,OAAO;AAAA,MACvB,aAAa;AAAA,QACX,oBAAoB,MAAM,UAAU,SAChC,CAAC,MAAM,UAAU,MAAM,IACvB,CAAC,OAAO,QAAQ,KAAK;AAAA,QACzB,eAAe;AAAA,MACjB;AAAA,MACA,QAAQ;AAAA,QACN,eAAe;AAAA,QACf,iBAAiB;AAAA,QACjB,gBAAgB;AAAA,MAClB;AAAA,IACF,CAAC;AACD,UAAM;AAAA,EACR;AACA,UAAQ,MAAM,aAAa,QAAQ,MAAM,UAAU,MAAM,OAAO;AAClE;",
+  "sourcesContent": ["import { ImageAdjustType, Ocr, ModelConfig } from '@shenghuabi/knowledge/ocr';\nimport { createNormalizeVfs, path } from '@cyia/vfs2';\n// import * as ort from 'onnxruntime-node';\nimport { MessagePort } from 'worker_threads';\nimport { getUniqueObjectKey } from '@shenghuabi/knowledge/util';\nimport { downloadFile } from '@cyia/dl';\nlet key!: string;\nlet ocrInstance: ReturnType<(typeof Ocr)['create']>;\nconst BaseUrl =\n  'https://www.modelscope.cn/models/RapidAI/RapidOCR/resolve/master/onnx/PP-OCRv4';\n\nconst DictUrl =\n  'https://www.modelscope.cn/models/RapidAI/RapidOCR/resolve/master/paddle/PP-OCRv4';\nasync function init(ocrConfig: {\n  key: string;\n  modelDir: string;\n  port?: MessagePort;\n}) {\n  const messageCb = (message: any) => {\n    ocrConfig.port?.postMessage({ type: 'progress', message });\n  };\n  const modelConfig = ModelConfig.find((item) => item.key === ocrConfig.key)!;\n  const fs = createNormalizeVfs({ dir: ocrConfig.modelDir });\n  // 自动下载模型\n  const absDetectionPath = path.join(ocrConfig.modelDir, modelConfig.det);\n  if (!(await fs.exists(modelConfig.det))) {\n    await downloadFile(`${BaseUrl}/${modelConfig.det}`, {\n      savePath: absDetectionPath,\n      message: messageCb,\n    });\n  }\n  const absRecognitionPath = path.join(ocrConfig.modelDir, modelConfig.rec);\n  if (!(await fs.exists(modelConfig.rec))) {\n    await downloadFile(`${BaseUrl}/${modelConfig.rec}`, {\n      savePath: absRecognitionPath,\n      message: messageCb,\n    });\n  }\n  const absDictionaryPath = path.join(ocrConfig.modelDir, modelConfig.dict);\n  if (!(await fs.exists(modelConfig.dict))) {\n    await downloadFile(`${DictUrl}/${modelConfig.dict}`, {\n      savePath: absDictionaryPath,\n      message: messageCb,\n    });\n  }\n}\n// 改为init和convert\nasync function convert(input: {\n  filePath: string | Uint8Array;\n  ocrConfig: { key: string; modelDir: string; device?: 'dml' | 'cuda' | 'cpu' };\n  options?: ImageAdjustType;\n}) {\n  const inputKey = getUniqueObjectKey(input.ocrConfig);\n  if (key !== inputKey) {\n    const modelConfig = ModelConfig.find(\n      (item) => item.key === input.ocrConfig.key,\n    )!;\n    // 自动下载模型\n    const absDetectionPath = path.join(\n      input.ocrConfig.modelDir,\n      modelConfig.det,\n    );\n    const absRecognitionPath = path.join(\n      input.ocrConfig.modelDir,\n      modelConfig.rec,\n    );\n    const absDictionaryPath = path.join(\n      input.ocrConfig.modelDir,\n      modelConfig.dict,\n    );\n    ocrInstance = Ocr.create({\n      onnxOptions: {\n        executionProviders: input.ocrConfig.device\n          ? [input.ocrConfig.device]\n          : ['dml', 'cuda', 'cpu'],\n        executionMode: 'parallel',\n      },\n      models: {\n        detectionPath: absDetectionPath,\n        recognitionPath: absRecognitionPath,\n        dictionaryPath: absDictionaryPath,\n      },\n    });\n    key = inputKey;\n  }\n  return (await ocrInstance).convert(input.filePath, input.options);\n}\n\nexport { init, convert };\n"],
+  "mappings": ";AAAA,SAA0B,KAAK,mBAAmB;AAClD,SAAS,oBAAoB,YAAY;AAGzC,SAAS,0BAA0B;AACnC,SAAS,oBAAoB;AAC7B,IAAI;AACJ,IAAI;AACJ,IAAM,UACJ;AAEF,IAAM,UACJ;AACF,eAAe,KAAK,WAIjB;AACD,QAAM,YAAY,CAAC,YAAiB;AAClC,cAAU,MAAM,YAAY,EAAE,MAAM,YAAY,QAAQ,CAAC;AAAA,EAC3D;AACA,QAAM,cAAc,YAAY,KAAK,CAAC,SAAS,KAAK,QAAQ,UAAU,GAAG;AACzE,QAAM,KAAK,mBAAmB,EAAE,KAAK,UAAU,SAAS,CAAC;AAEzD,QAAM,mBAAmB,KAAK,KAAK,UAAU,UAAU,YAAY,GAAG;AACtE,MAAI,CAAE,MAAM,GAAG,OAAO,YAAY,GAAG,GAAI;AACvC,UAAM,aAAa,GAAG,OAAO,IAAI,YAAY,GAAG,IAAI;AAAA,MAClD,UAAU;AAAA,MACV,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AACA,QAAM,qBAAqB,KAAK,KAAK,UAAU,UAAU,YAAY,GAAG;AACxE,MAAI,CAAE,MAAM,GAAG,OAAO,YAAY,GAAG,GAAI;AACvC,UAAM,aAAa,GAAG,OAAO,IAAI,YAAY,GAAG,IAAI;AAAA,MAClD,UAAU;AAAA,MACV,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AACA,QAAM,oBAAoB,KAAK,KAAK,UAAU,UAAU,YAAY,IAAI;AACxE,MAAI,CAAE,MAAM,GAAG,OAAO,YAAY,IAAI,GAAI;AACxC,UAAM,aAAa,GAAG,OAAO,IAAI,YAAY,IAAI,IAAI;AAAA,MACnD,UAAU;AAAA,MACV,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AACF;AAEA,eAAe,QAAQ,OAIpB;AACD,QAAM,WAAW,mBAAmB,MAAM,SAAS;AACnD,MAAI,QAAQ,UAAU;AACpB,UAAM,cAAc,YAAY;AAAA,MAC9B,CAAC,SAAS,KAAK,QAAQ,MAAM,UAAU;AAAA,IACzC;AAEA,UAAM,mBAAmB,KAAK;AAAA,MAC5B,MAAM,UAAU;AAAA,MAChB,YAAY;AAAA,IACd;AACA,UAAM,qBAAqB,KAAK;AAAA,MAC9B,MAAM,UAAU;AAAA,MAChB,YAAY;AAAA,IACd;AACA,UAAM,oBAAoB,KAAK;AAAA,MAC7B,MAAM,UAAU;AAAA,MAChB,YAAY;AAAA,IACd;AACA,kBAAc,IAAI,OAAO;AAAA,MACvB,aAAa;AAAA,QACX,oBAAoB,MAAM,UAAU,SAChC,CAAC,MAAM,UAAU,MAAM,IACvB,CAAC,OAAO,QAAQ,KAAK;AAAA,QACzB,eAAe;AAAA,MACjB;AAAA,MACA,QAAQ;AAAA,QACN,eAAe;AAAA,QACf,iBAAiB;AAAA,QACjB,gBAAgB;AAAA,MAClB;AAAA,IACF,CAAC;AACD,UAAM;AAAA,EACR;AACA,UAAQ,MAAM,aAAa,QAAQ,MAAM,UAAU,MAAM,OAAO;AAClE;",
   "names": []
 }

package/worker/reranker.mjs

@@ -29,6 +29,20 @@ var FileProxyCache = class {
     this.#downloadConfig = initOptions.downloadConfig;
     this.#initOptions = initOptions;
   }
+  async #createResponse(filePath) {
+    const stats = await fs.promises.stat(filePath);
+    const extension = filePath.split(".").pop().toLowerCase();
+    const contentType = CONTENT_TYPE_MAP[extension] ?? "application/octet-stream";
+    const stream = fs.createReadStream(filePath);
+    return new Response(stream, {
+      status: 200,
+      statusText: "OK",
+      headers: {
+        "content-type": contentType,
+        "content-length": stats.size.toString()
+      }
+    });
+  }
   async match(request) {
     let filePath;
     if (request.startsWith("http")) {
@@ -54,7 +68,7 @@ var FileProxyCache = class {
     }
     const exists = await this.#vfs.exists(filePath);
     if (exists) {
-      return new FileResponse(filePath);
+      return this.#createResponse(filePath);
     }
     return void 0;
   }
@@ -62,7 +76,6 @@ var FileProxyCache = class {
     throw new Error("no put");
   }
 };
-var decoder = new TextDecoder("utf-8");
 var CONTENT_TYPE_MAP = {
   txt: "text/plain",
   html: "text/html",
@@ -74,51 +87,6 @@ var CONTENT_TYPE_MAP = {
   jpeg: "image/jpeg",
   gif: "image/gif"
 };
-var FileResponse = class _FileResponse {
-  filePath;
-  headers;
-  exists = true;
-  status = 200;
-  statusText = "OK";
-  body;
-  constructor(filePath) {
-    this.filePath = filePath;
-    this.headers = new Headers();
-    this.updateContentType();
-    this.body = fs.createReadStream(filePath);
-  }
-  updateContentType() {
-    const stats = fs.statSync(this.filePath);
-    this.headers.set("content-length", stats.size.toString());
-    const extension = this.filePath.toString().split(".").pop().toLowerCase();
-    this.headers.set(
-      "content-type",
-      CONTENT_TYPE_MAP[extension] ?? "application/octet-stream"
-    );
-  }
-  clone() {
-    const response = new _FileResponse(this.filePath);
-    response.exists = this.exists;
-    response.status = this.status;
-    response.statusText = this.statusText;
-    response.headers = new Headers(this.headers);
-    return response;
-  }
-  async arrayBuffer() {
-    return fs.promises.readFile(this.filePath).then((buffer) => buffer.buffer);
-  }
-  async blob() {
-    return new Blob([await this.arrayBuffer()], {
-      type: this.headers.get("content-type")
-    });
-  }
-  async text() {
-    return decoder.decode(await this.arrayBuffer());
-  }
-  async json() {
-    return JSON.parse(await this.text());
-  }
-};
 
 // packages/worker/set-transformers-config.ts
 function setTransformersConfig(options) {

package/worker/reranker.mjs.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../packages/worker/reranker/index.ts", "../../packages/worker/set-transformers-config.ts", "../../packages/worker/custom-cache.ts"],
-  "sourcesContent": ["import {\n  AutoModelForSequenceClassification,\n  AutoTokenizer,\n} from '@huggingface/transformers';\nimport type {\n  Tensor,\n  XLMRobertaModel,\n  XLMRobertaTokenizer,\n} from '@huggingface/transformers';\n\nimport { InitOptions, setTransformersConfig } from '../set-transformers-config';\n\nclass ReRanderService {\n  init = async (options: InitOptions) => {\n    if (!this.model || !this.tokenizer) {\n      await this.#downloadOnly(options);\n    }\n    return true;\n  };\n  convert = async (input: { value: string; docs: string[] }) => {\n    const inputs = this.tokenizer!(\n      new Array(input.docs.length).fill(input.value),\n      {\n        text_pair: input.docs,\n        padding: true,\n        truncation: true,\n      },\n    );\n    const { logits } = await this.model!(inputs);\n    return (logits as Tensor)\n      .sigmoid()\n      .tolist()\n      .map(([score], i: number) => ({\n        index: i,\n        score,\n      }))\n      .sort((a, b) => b.score - a.score);\n  };\n  tokenizer?: XLMRobertaTokenizer;\n  model?: XLMRobertaModel;\n  async #downloadOnly(options: InitOptions) {\n    setTransformersConfig(options);\n\n    this.tokenizer = await AutoTokenizer.from_pretrained(options.modelName);\n    this.model = await AutoModelForSequenceClassification.from_pretrained(\n      options.modelName,\n      {\n        ...options.options,\n      } as any,\n    );\n  }\n}\nconst instance = new ReRanderService();\nconst init = instance.init;\nconst convert = instance.convert;\nexport { init, convert };\n", "import { env, pipeline } from '@huggingface/transformers';\nimport { MessagePort } from 'worker_threads';\nimport { FileProxyCache } from './custom-cache';\nimport type { DownloadConfigType } from '@cyia/external-call';\ntype PipeLineOptions = Partial<NonNullable<Parameters<typeof pipeline>[2]>>;\nexport interface InitOptions {\n  /** 文件夹 */\n  dir: string;\n  /** 模型 */\n  modelName: string;\n  /** 模型参数 */\n  options: PipeLineOptions;\n  /**直接链接 */\n  remoteHost: string;\n  downloadConfig?: DownloadConfigType;\n  port?: MessagePort;\n  hfToken?: string;\n}\nexport function setTransformersConfig(options: InitOptions) {\n  env.useFS = false;\n  env.localModelPath = options.dir;\n  env.allowLocalModels = false;\n  env.allowRemoteModels = true;\n  env.cacheDir = options.dir;\n  env.customCache = new FileProxyCache(options);\n  env.useBrowserCache = false;\n  env.useFSCache = true;\n  env.useCustomCache = true;\n\n  env.remoteHost = `https://${options.remoteHost}`;\n}\n", "import { env } from '@huggingface/transformers';\nimport { createNormalizeVfs, path } from '@cyia/vfs2';\nimport { downloadFile } from '@cyia/dl';\nimport fs from 'fs';\nimport { InitOptions } from './set-transformers-config';\nexport interface NodeProxy {\n  match: (request: string) => Promise<ArrayBuffer | undefined>;\n  put: (request: string, arraybuffer: ArrayBuffer) => Promise<void>;\n}\nexport class FileProxyCache {\n  #path;\n  #vfs;\n  #sendMessage;\n  #modelName;\n  #downloadConfig;\n  #initOptions;\n  constructor(initOptions: InitOptions) {\n    this.#modelName = initOptions.modelName;\n    this.#sendMessage = (message: any) => {\n      initOptions.port?.postMessage({ type: 'progress', message });\n    };\n    this.#path = initOptions.dir;\n    this.#vfs = createNormalizeVfs({ dir: initOptions.dir });\n    this.#downloadConfig = initOptions.downloadConfig;\n    this.#initOptions = initOptions;\n  }\n  async match(request: string): Promise<FileResponse | undefined> {\n    let filePath;\n    if (request.startsWith('http')) {\n      const data = new URL(request);\n      filePath = path.join(\n        this.#path,\n        data.pathname.replace(\n          '/' +\n            env.remotePathTemplate\n              .replaceAll('{model}', this.#modelName)\n              .replaceAll('{revision}', encodeURIComponent('main')),\n          `/${this.#modelName}/`,\n        ),\n      );\n\n      await downloadFile(request, {\n        ...this.#downloadConfig,\n        savePath: filePath,\n        message: this.#sendMessage,\n        headers: {\n          token: this.#initOptions?.hfToken ?? '',\n          'software-bbs': 'bbs.shenghuabi.site',\n        },\n      });\n    } else {\n      filePath = request;\n    }\n    const exists = await this.#vfs.exists(filePath);\n    if (exists) {\n      return new FileResponse(filePath);\n    }\n    return undefined;\n  }\n\n  async put(request: string, response: Response | FileResponse): Promise<void> {\n    throw new Error('no put');\n  }\n}\nconst decoder = new TextDecoder('utf-8');\nconst CONTENT_TYPE_MAP = {\n  txt: 'text/plain',\n  html: 'text/html',\n  css: 'text/css',\n  js: 'text/javascript',\n  json: 'application/json',\n  png: 'image/png',\n  jpg: 'image/jpeg',\n  jpeg: 'image/jpeg',\n  gif: 'image/gif',\n};\nclass FileResponse {\n  filePath;\n  headers;\n  exists = true;\n  status = 200;\n  statusText = 'OK';\n  body;\n  constructor(filePath: string) {\n    this.filePath = filePath;\n    this.headers = new Headers();\n\n    this.updateContentType();\n\n    this.body = fs.createReadStream(filePath);\n  }\n\n  updateContentType() {\n    const stats = fs.statSync(this.filePath);\n    this.headers.set('content-length', stats.size.toString());\n\n    const extension = this.filePath.toString().split('.').pop()!.toLowerCase();\n    this.headers.set(\n      'content-type',\n      (CONTENT_TYPE_MAP as any)[extension] ?? 'application/octet-stream',\n    );\n  }\n\n  clone(): FileResponse {\n    const response = new FileResponse(this.filePath);\n    response.exists = this.exists;\n    response.status = this.status;\n    response.statusText = this.statusText;\n    response.headers = new Headers(this.headers);\n    return response;\n  }\n\n  async arrayBuffer(): Promise<ArrayBuffer> {\n    return fs.promises\n      .readFile(this.filePath)\n      .then((buffer) => buffer.buffer as ArrayBuffer);\n  }\n\n  async blob(): Promise<Blob> {\n    return new Blob([await this.arrayBuffer()], {\n      type: this.headers.get('content-type')!,\n    });\n  }\n\n  async text(): Promise<string> {\n    return decoder.decode(await this.arrayBuffer());\n  }\n\n  async json(): Promise<object> {\n    return JSON.parse(await this.text());\n  }\n}\n"],
-  "mappings": ";AAAA;AAAA,EACE;AAAA,EACA;AAAA,OACK;;;ACHP,SAAS,OAAAA,YAAqB;;;ACA9B,SAAS,WAAW;AACpB,SAAS,oBAAoB,YAAY;AACzC,SAAS,oBAAoB;AAC7B,OAAO,QAAQ;AAMR,IAAM,iBAAN,MAAqB;AAAA,EAC1B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA,YAAY,aAA0B;AACpC,SAAK,aAAa,YAAY;AAC9B,SAAK,eAAe,CAAC,YAAiB;AACpC,kBAAY,MAAM,YAAY,EAAE,MAAM,YAAY,QAAQ,CAAC;AAAA,IAC7D;AACA,SAAK,QAAQ,YAAY;AACzB,SAAK,OAAO,mBAAmB,EAAE,KAAK,YAAY,IAAI,CAAC;AACvD,SAAK,kBAAkB,YAAY;AACnC,SAAK,eAAe;AAAA,EACtB;AAAA,EACA,MAAM,MAAM,SAAoD;AAC9D,QAAI;AACJ,QAAI,QAAQ,WAAW,MAAM,GAAG;AAC9B,YAAM,OAAO,IAAI,IAAI,OAAO;AAC5B,iBAAW,KAAK;AAAA,QACd,KAAK;AAAA,QACL,KAAK,SAAS;AAAA,UACZ,MACE,IAAI,mBACD,WAAW,WAAW,KAAK,UAAU,EACrC,WAAW,cAAc,mBAAmB,MAAM,CAAC;AAAA,UACxD,IAAI,KAAK,UAAU;AAAA,QACrB;AAAA,MACF;AAEA,YAAM,aAAa,SAAS;AAAA,QAC1B,GAAG,KAAK;AAAA,QACR,UAAU;AAAA,QACV,SAAS,KAAK;AAAA,QACd,SAAS;AAAA,UACP,OAAO,KAAK,cAAc,WAAW;AAAA,UACrC,gBAAgB;AAAA,QAClB;AAAA,MACF,CAAC;AAAA,IACH,OAAO;AACL,iBAAW;AAAA,IACb;AACA,UAAM,SAAS,MAAM,KAAK,KAAK,OAAO,QAAQ;AAC9C,QAAI,QAAQ;AACV,aAAO,IAAI,aAAa,QAAQ;AAAA,IAClC;AACA,WAAO;AAAA,EACT;AAAA,EAEA,MAAM,IAAI,SAAiB,UAAkD;AAC3E,UAAM,IAAI,MAAM,QAAQ;AAAA,EAC1B;AACF;AACA,IAAM,UAAU,IAAI,YAAY,OAAO;AACvC,IAAM,mBAAmB;AAAA,EACvB,KAAK;AAAA,EACL,MAAM;AAAA,EACN,KAAK;AAAA,EACL,IAAI;AAAA,EACJ,MAAM;AAAA,EACN,KAAK;AAAA,EACL,KAAK;AAAA,EACL,MAAM;AAAA,EACN,KAAK;AACP;AACA,IAAM,eAAN,MAAM,cAAa;AAAA,EACjB;AAAA,EACA;AAAA,EACA,SAAS;AAAA,EACT,SAAS;AAAA,EACT,aAAa;AAAA,EACb;AAAA,EACA,YAAY,UAAkB;AAC5B,SAAK,WAAW;AAChB,SAAK,UAAU,IAAI,QAAQ;AAE3B,SAAK,kBAAkB;AAEvB,SAAK,OAAO,GAAG,iBAAiB,QAAQ;AAAA,EAC1C;AAAA,EAEA,oBAAoB;AAClB,UAAM,QAAQ,GAAG,SAAS,KAAK,QAAQ;AACvC,SAAK,QAAQ,IAAI,kBAAkB,MAAM,KAAK,SAAS,CAAC;AAExD,UAAM,YAAY,KAAK,SAAS,SAAS,EAAE,MAAM,GAAG,EAAE,IAAI,EAAG,YAAY;AACzE,SAAK,QAAQ;AAAA,MACX;AAAA,MACC,iBAAyB,SAAS,KAAK;AAAA,IAC1C;AAAA,EACF;AAAA,EAEA,QAAsB;AACpB,UAAM,WAAW,IAAI,cAAa,KAAK,QAAQ;AAC/C,aAAS,SAAS,KAAK;AACvB,aAAS,SAAS,KAAK;AACvB,aAAS,aAAa,KAAK;AAC3B,aAAS,UAAU,IAAI,QAAQ,KAAK,OAAO;AAC3C,WAAO;AAAA,EACT;AAAA,EAEA,MAAM,cAAoC;AACxC,WAAO,GAAG,SACP,SAAS,KAAK,QAAQ,EACtB,KAAK,CAAC,WAAW,OAAO,MAAqB;AAAA,EAClD;AAAA,EAEA,MAAM,OAAsB;AAC1B,WAAO,IAAI,KAAK,CAAC,MAAM,KAAK,YAAY,CAAC,GAAG;AAAA,MAC1C,MAAM,KAAK,QAAQ,IAAI,cAAc;AAAA,IACvC,CAAC;AAAA,EACH;AAAA,EAEA,MAAM,OAAwB;AAC5B,WAAO,QAAQ,OAAO,MAAM,KAAK,YAAY,CAAC;AAAA,EAChD;AAAA,EAEA,MAAM,OAAwB;AAC5B,WAAO,KAAK,MAAM,MAAM,KAAK,KAAK,CAAC;AAAA,EACrC;AACF;;;ADjHO,SAAS,sBAAsB,SAAsB;AAC1D,EAAAC,KAAI,QAAQ;AACZ,EAAAA,KAAI,iBAAiB,QAAQ;AAC7B,EAAAA,KAAI,mBAAmB;AACvB,EAAAA,KAAI,oBAAoB;AACxB,EAAAA,KAAI,WAAW,QAAQ;AACvB,EAAAA,KAAI,cAAc,IAAI,eAAe,OAAO;AAC5C,EAAAA,KAAI,kBAAkB;AACtB,EAAAA,KAAI,aAAa;AACjB,EAAAA,KAAI,iBAAiB;AAErB,EAAAA,KAAI,aAAa,WAAW,QAAQ,UAAU;AAChD;;;ADlBA,IAAM,kBAAN,MAAsB;AAAA,EACpB,OAAO,OAAO,YAAyB;AACrC,QAAI,CAAC,KAAK,SAAS,CAAC,KAAK,WAAW;AAClC,YAAM,KAAK,cAAc,OAAO;AAAA,IAClC;AACA,WAAO;AAAA,EACT;AAAA,EACA,UAAU,OAAO,UAA6C;AAC5D,UAAM,SAAS,KAAK;AAAA,MAClB,IAAI,MAAM,MAAM,KAAK,MAAM,EAAE,KAAK,MAAM,KAAK;AAAA,MAC7C;AAAA,QACE,WAAW,MAAM;AAAA,QACjB,SAAS;AAAA,QACT,YAAY;AAAA,MACd;AAAA,IACF;AACA,UAAM,EAAE,OAAO,IAAI,MAAM,KAAK,MAAO,MAAM;AAC3C,WAAQ,OACL,QAAQ,EACR,OAAO,EACP,IAAI,CAAC,CAAC,KAAK,GAAG,OAAe;AAAA,MAC5B,OAAO;AAAA,MACP;AAAA,IACF,EAAE,EACD,KAAK,CAAC,GAAG,MAAM,EAAE,QAAQ,EAAE,KAAK;AAAA,EACrC;AAAA,EACA;AAAA,EACA;AAAA,EACA,MAAM,cAAc,SAAsB;AACxC,0BAAsB,OAAO;AAE7B,SAAK,YAAY,MAAM,cAAc,gBAAgB,QAAQ,SAAS;AACtE,SAAK,QAAQ,MAAM,mCAAmC;AAAA,MACpD,QAAQ;AAAA,MACR;AAAA,QACE,GAAG,QAAQ;AAAA,MACb;AAAA,IACF;AAAA,EACF;AACF;AACA,IAAM,WAAW,IAAI,gBAAgB;AACrC,IAAM,OAAO,SAAS;AACtB,IAAM,UAAU,SAAS;",
+  "sourcesContent": ["import {\n  AutoModelForSequenceClassification,\n  AutoTokenizer,\n} from '@huggingface/transformers';\nimport type {\n  Tensor,\n  XLMRobertaModel,\n  XLMRobertaTokenizer,\n} from '@huggingface/transformers';\n\nimport { InitOptions, setTransformersConfig } from '../set-transformers-config';\n\nclass ReRanderService {\n  init = async (options: InitOptions) => {\n    if (!this.model || !this.tokenizer) {\n      await this.#downloadOnly(options);\n    }\n    return true;\n  };\n  convert = async (input: { value: string; docs: string[] }) => {\n    const inputs = this.tokenizer!(\n      new Array(input.docs.length).fill(input.value),\n      {\n        text_pair: input.docs,\n        padding: true,\n        truncation: true,\n      },\n    );\n    const { logits } = await this.model!(inputs);\n    return (logits as Tensor)\n      .sigmoid()\n      .tolist()\n      .map(([score], i: number) => ({\n        index: i,\n        score,\n      }))\n      .sort((a, b) => b.score - a.score);\n  };\n  tokenizer?: XLMRobertaTokenizer;\n  model?: XLMRobertaModel;\n  async #downloadOnly(options: InitOptions) {\n    setTransformersConfig(options);\n\n    this.tokenizer = await AutoTokenizer.from_pretrained(options.modelName);\n    this.model = await AutoModelForSequenceClassification.from_pretrained(\n      options.modelName,\n      {\n        ...options.options,\n      } as any,\n    );\n  }\n}\nconst instance = new ReRanderService();\nconst init = instance.init;\nconst convert = instance.convert;\nexport { init, convert };\n", "import { env, pipeline } from '@huggingface/transformers';\nimport { MessagePort } from 'worker_threads';\nimport { FileProxyCache } from './custom-cache';\nimport type { DownloadConfigType } from '@cyia/external-call';\ntype PipeLineOptions = Partial<NonNullable<Parameters<typeof pipeline>[2]>>;\nexport interface InitOptions {\n  /** 文件夹 */\n  dir: string;\n  /** 模型 */\n  modelName: string;\n  /** 模型参数 */\n  options: PipeLineOptions;\n  /**直接链接 */\n  remoteHost: string;\n  downloadConfig?: DownloadConfigType;\n  port?: MessagePort;\n  hfToken?: string;\n}\nexport function setTransformersConfig(options: InitOptions) {\n  env.useFS = false;\n  env.localModelPath = options.dir;\n  env.allowLocalModels = false;\n  env.allowRemoteModels = true;\n  env.cacheDir = options.dir;\n  env.customCache = new FileProxyCache(options);\n  env.useBrowserCache = false;\n  env.useFSCache = true;\n  env.useCustomCache = true;\n\n  env.remoteHost = `https://${options.remoteHost}`;\n}\n", "import { env } from '@huggingface/transformers';\nimport { createNormalizeVfs, path } from '@cyia/vfs2';\nimport { downloadFile } from '@cyia/dl';\nimport fs from 'fs';\nimport { InitOptions } from './set-transformers-config';\nexport interface NodeProxy {\n  match: (request: string) => Promise<ArrayBuffer | undefined>;\n  put: (request: string, arraybuffer: ArrayBuffer) => Promise<void>;\n}\nexport class FileProxyCache {\n  #path;\n  #vfs;\n  #sendMessage;\n  #modelName;\n  #downloadConfig;\n  #initOptions;\n  constructor(initOptions: InitOptions) {\n    this.#modelName = initOptions.modelName;\n    this.#sendMessage = (message: any) => {\n      initOptions.port?.postMessage({ type: 'progress', message });\n    };\n    this.#path = initOptions.dir;\n    this.#vfs = createNormalizeVfs({ dir: initOptions.dir });\n    this.#downloadConfig = initOptions.downloadConfig;\n    this.#initOptions = initOptions;\n  }\n  async #createResponse(filePath: string): Promise<Response> {\n    const stats = await fs.promises.stat(filePath);\n    const extension = filePath.split('.').pop()!.toLowerCase();\n    const contentType =\n      (CONTENT_TYPE_MAP as any)[extension] ?? 'application/octet-stream';\n    const stream = fs.createReadStream(filePath);\n    return new Response(stream, {\n      status: 200,\n      statusText: 'OK',\n      headers: {\n        'content-type': contentType,\n        'content-length': stats.size.toString(),\n      },\n    });\n  }\n  async match(request: string): Promise<Response | undefined> {\n    let filePath;\n    if (request.startsWith('http')) {\n      const data = new URL(request);\n      filePath = path.join(\n        this.#path,\n        data.pathname.replace(\n          '/' +\n            env.remotePathTemplate\n              .replaceAll('{model}', this.#modelName)\n              .replaceAll('{revision}', encodeURIComponent('main')),\n          `/${this.#modelName}/`,\n        ),\n      );\n\n      await downloadFile(request, {\n        ...this.#downloadConfig,\n        savePath: filePath,\n        message: this.#sendMessage,\n        headers: {\n          token: this.#initOptions?.hfToken ?? '',\n          'software-bbs': 'bbs.shenghuabi.site',\n        },\n      });\n    } else {\n      filePath = request;\n    }\n    const exists = await this.#vfs.exists(filePath);\n    if (exists) {\n      return this.#createResponse(filePath);\n    }\n    return undefined;\n  }\n\n  async put(request: string, response: Response): Promise<void> {\n    throw new Error('no put');\n  }\n}\nconst CONTENT_TYPE_MAP: Record<string, string> = {\n  txt: 'text/plain',\n  html: 'text/html',\n  css: 'text/css',\n  js: 'text/javascript',\n  json: 'application/json',\n  png: 'image/png',\n  jpg: 'image/jpeg',\n  jpeg: 'image/jpeg',\n  gif: 'image/gif',\n};\n"],
+  "mappings": ";AAAA;AAAA,EACE;AAAA,EACA;AAAA,OACK;;;ACHP,SAAS,OAAAA,YAAqB;;;ACA9B,SAAS,WAAW;AACpB,SAAS,oBAAoB,YAAY;AACzC,SAAS,oBAAoB;AAC7B,OAAO,QAAQ;AAMR,IAAM,iBAAN,MAAqB;AAAA,EAC1B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA,YAAY,aAA0B;AACpC,SAAK,aAAa,YAAY;AAC9B,SAAK,eAAe,CAAC,YAAiB;AACpC,kBAAY,MAAM,YAAY,EAAE,MAAM,YAAY,QAAQ,CAAC;AAAA,IAC7D;AACA,SAAK,QAAQ,YAAY;AACzB,SAAK,OAAO,mBAAmB,EAAE,KAAK,YAAY,IAAI,CAAC;AACvD,SAAK,kBAAkB,YAAY;AACnC,SAAK,eAAe;AAAA,EACtB;AAAA,EACA,MAAM,gBAAgB,UAAqC;AACzD,UAAM,QAAQ,MAAM,GAAG,SAAS,KAAK,QAAQ;AAC7C,UAAM,YAAY,SAAS,MAAM,GAAG,EAAE,IAAI,EAAG,YAAY;AACzD,UAAM,cACH,iBAAyB,SAAS,KAAK;AAC1C,UAAM,SAAS,GAAG,iBAAiB,QAAQ;AAC3C,WAAO,IAAI,SAAS,QAAQ;AAAA,MAC1B,QAAQ;AAAA,MACR,YAAY;AAAA,MACZ,SAAS;AAAA,QACP,gBAAgB;AAAA,QAChB,kBAAkB,MAAM,KAAK,SAAS;AAAA,MACxC;AAAA,IACF,CAAC;AAAA,EACH;AAAA,EACA,MAAM,MAAM,SAAgD;AAC1D,QAAI;AACJ,QAAI,QAAQ,WAAW,MAAM,GAAG;AAC9B,YAAM,OAAO,IAAI,IAAI,OAAO;AAC5B,iBAAW,KAAK;AAAA,QACd,KAAK;AAAA,QACL,KAAK,SAAS;AAAA,UACZ,MACE,IAAI,mBACD,WAAW,WAAW,KAAK,UAAU,EACrC,WAAW,cAAc,mBAAmB,MAAM,CAAC;AAAA,UACxD,IAAI,KAAK,UAAU;AAAA,QACrB;AAAA,MACF;AAEA,YAAM,aAAa,SAAS;AAAA,QAC1B,GAAG,KAAK;AAAA,QACR,UAAU;AAAA,QACV,SAAS,KAAK;AAAA,QACd,SAAS;AAAA,UACP,OAAO,KAAK,cAAc,WAAW;AAAA,UACrC,gBAAgB;AAAA,QAClB;AAAA,MACF,CAAC;AAAA,IACH,OAAO;AACL,iBAAW;AAAA,IACb;AACA,UAAM,SAAS,MAAM,KAAK,KAAK,OAAO,QAAQ;AAC9C,QAAI,QAAQ;AACV,aAAO,KAAK,gBAAgB,QAAQ;AAAA,IACtC;AACA,WAAO;AAAA,EACT;AAAA,EAEA,MAAM,IAAI,SAAiB,UAAmC;AAC5D,UAAM,IAAI,MAAM,QAAQ;AAAA,EAC1B;AACF;AACA,IAAM,mBAA2C;AAAA,EAC/C,KAAK;AAAA,EACL,MAAM;AAAA,EACN,KAAK;AAAA,EACL,IAAI;AAAA,EACJ,MAAM;AAAA,EACN,KAAK;AAAA,EACL,KAAK;AAAA,EACL,MAAM;AAAA,EACN,KAAK;AACP;;;ADvEO,SAAS,sBAAsB,SAAsB;AAC1D,EAAAC,KAAI,QAAQ;AACZ,EAAAA,KAAI,iBAAiB,QAAQ;AAC7B,EAAAA,KAAI,mBAAmB;AACvB,EAAAA,KAAI,oBAAoB;AACxB,EAAAA,KAAI,WAAW,QAAQ;AACvB,EAAAA,KAAI,cAAc,IAAI,eAAe,OAAO;AAC5C,EAAAA,KAAI,kBAAkB;AACtB,EAAAA,KAAI,aAAa;AACjB,EAAAA,KAAI,iBAAiB;AAErB,EAAAA,KAAI,aAAa,WAAW,QAAQ,UAAU;AAChD;;;ADlBA,IAAM,kBAAN,MAAsB;AAAA,EACpB,OAAO,OAAO,YAAyB;AACrC,QAAI,CAAC,KAAK,SAAS,CAAC,KAAK,WAAW;AAClC,YAAM,KAAK,cAAc,OAAO;AAAA,IAClC;AACA,WAAO;AAAA,EACT;AAAA,EACA,UAAU,OAAO,UAA6C;AAC5D,UAAM,SAAS,KAAK;AAAA,MAClB,IAAI,MAAM,MAAM,KAAK,MAAM,EAAE,KAAK,MAAM,KAAK;AAAA,MAC7C;AAAA,QACE,WAAW,MAAM;AAAA,QACjB,SAAS;AAAA,QACT,YAAY;AAAA,MACd;AAAA,IACF;AACA,UAAM,EAAE,OAAO,IAAI,MAAM,KAAK,MAAO,MAAM;AAC3C,WAAQ,OACL,QAAQ,EACR,OAAO,EACP,IAAI,CAAC,CAAC,KAAK,GAAG,OAAe;AAAA,MAC5B,OAAO;AAAA,MACP;AAAA,IACF,EAAE,EACD,KAAK,CAAC,GAAG,MAAM,EAAE,QAAQ,EAAE,KAAK;AAAA,EACrC;AAAA,EACA;AAAA,EACA;AAAA,EACA,MAAM,cAAc,SAAsB;AACxC,0BAAsB,OAAO;AAE7B,SAAK,YAAY,MAAM,cAAc,gBAAgB,QAAQ,SAAS;AACtE,SAAK,QAAQ,MAAM,mCAAmC;AAAA,MACpD,QAAQ;AAAA,MACR;AAAA,QACE,GAAG,QAAQ;AAAA,MACb;AAAA,IACF;AAAA,EACF;AACF;AACA,IAAM,WAAW,IAAI,gBAAgB;AACrC,IAAM,OAAO,SAAS;AACtB,IAAM,UAAU,SAAS;",
   "names": ["env", "env"]
 }

package/worker/text2vec.mjs

@@ -26,6 +26,20 @@ var FileProxyCache = class {
     this.#downloadConfig = initOptions.downloadConfig;
     this.#initOptions = initOptions;
   }
+  async #createResponse(filePath) {
+    const stats = await fs.promises.stat(filePath);
+    const extension = filePath.split(".").pop().toLowerCase();
+    const contentType = CONTENT_TYPE_MAP[extension] ?? "application/octet-stream";
+    const stream = fs.createReadStream(filePath);
+    return new Response(stream, {
+      status: 200,
+      statusText: "OK",
+      headers: {
+        "content-type": contentType,
+        "content-length": stats.size.toString()
+      }
+    });
+  }
   async match(request) {
     let filePath;
     if (request.startsWith("http")) {
@@ -51,7 +65,7 @@ var FileProxyCache = class {
     }
     const exists = await this.#vfs.exists(filePath);
     if (exists) {
-      return new FileResponse(filePath);
+      return this.#createResponse(filePath);
     }
     return void 0;
   }
@@ -59,7 +73,6 @@ var FileProxyCache = class {
     throw new Error("no put");
   }
 };
-var decoder = new TextDecoder("utf-8");
 var CONTENT_TYPE_MAP = {
   txt: "text/plain",
   html: "text/html",
@@ -71,51 +84,6 @@ var CONTENT_TYPE_MAP = {
   jpeg: "image/jpeg",
   gif: "image/gif"
 };
-var FileResponse = class _FileResponse {
-  filePath;
-  headers;
-  exists = true;
-  status = 200;
-  statusText = "OK";
-  body;
-  constructor(filePath) {
-    this.filePath = filePath;
-    this.headers = new Headers();
-    this.updateContentType();
-    this.body = fs.createReadStream(filePath);
-  }
-  updateContentType() {
-    const stats = fs.statSync(this.filePath);
-    this.headers.set("content-length", stats.size.toString());
-    const extension = this.filePath.toString().split(".").pop().toLowerCase();
-    this.headers.set(
-      "content-type",
-      CONTENT_TYPE_MAP[extension] ?? "application/octet-stream"
-    );
-  }
-  clone() {
-    const response = new _FileResponse(this.filePath);
-    response.exists = this.exists;
-    response.status = this.status;
-    response.statusText = this.statusText;
-    response.headers = new Headers(this.headers);
-    return response;
-  }
-  async arrayBuffer() {
-    return fs.promises.readFile(this.filePath).then((buffer) => buffer.buffer);
-  }
-  async blob() {
-    return new Blob([await this.arrayBuffer()], {
-      type: this.headers.get("content-type")
-    });
-  }
-  async text() {
-    return decoder.decode(await this.arrayBuffer());
-  }
-  async json() {
-    return JSON.parse(await this.text());
-  }
-};
 
 // packages/worker/set-transformers-config.ts
 function setTransformersConfig(options) {

package/worker/text2vec.mjs.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../packages/worker/text2vec/index.ts", "../../packages/worker/set-transformers-config.ts", "../../packages/worker/custom-cache.ts"],
-  "sourcesContent": ["import { pipeline } from '@huggingface/transformers';\nimport type { FeatureExtractionPipeline } from '@huggingface/transformers';\n\nimport { InitOptions, setTransformersConfig } from '../set-transformers-config';\nfunction qwen3ToVec(\n  extractor: FeatureExtractionPipeline,\n  query: string[],\n  description = 'Given a web search query, retrieve relevant passages that answer the query',\n) {\n  return extractor!(\n    query.map((item) => `Instruct: ${description}\\nQuery:${item}`),\n    {\n      pooling: 'last_token',\n      normalize: true,\n    },\n  );\n}\n\nclass Text2VecService {\n  #extractor!: FeatureExtractionPipeline | undefined;\n  constructor() {}\n  init = async (options: InitOptions) => {\n    if (this.#extractor) {\n      return true;\n    }\n    this.#extractor = await this.#downloadOnly(options);\n    return !!this.#extractor;\n  };\n  convert = async (\n    input: {\n      value: string | string[];\n      mode?: 'qwen3';\n      taskDescription?: string;\n    } & InitOptions,\n  ) => {\n    if (!this.#extractor) {\n      await this.init(input);\n    }\n    const inputList =\n      typeof input.value === 'string' ? [input.value] : input.value;\n    let result;\n    if (input.mode === 'qwen3') {\n      result = qwen3ToVec(this.#extractor!, inputList, input.taskDescription);\n    } else {\n      result = this.#extractor!(inputList, {\n        pooling: 'mean',\n        normalize: true,\n      });\n    }\n    return result.then((result) => {\n      const list = result.tolist();\n      return typeof input.value === 'string' ? list[0] : list;\n    });\n  };\n  async #downloadOnly(options: InitOptions) {\n    setTransformersConfig(options);\n\n    return await pipeline(\n      'feature-extraction',\n      options.modelName,\n      options.options,\n    );\n  }\n  getSize = () => (this.#extractor!.model.config as any).hidden_size;\n}\nconst instance = new Text2VecService();\nconst init = instance.init;\nconst getSize = instance.getSize;\nconst convert = instance.convert;\nexport { init, getSize, convert };\n", "import { env, pipeline } from '@huggingface/transformers';\nimport { MessagePort } from 'worker_threads';\nimport { FileProxyCache } from './custom-cache';\nimport type { DownloadConfigType } from '@cyia/external-call';\ntype PipeLineOptions = Partial<NonNullable<Parameters<typeof pipeline>[2]>>;\nexport interface InitOptions {\n  /** 文件夹 */\n  dir: string;\n  /** 模型 */\n  modelName: string;\n  /** 模型参数 */\n  options: PipeLineOptions;\n  /**直接链接 */\n  remoteHost: string;\n  downloadConfig?: DownloadConfigType;\n  port?: MessagePort;\n  hfToken?: string;\n}\nexport function setTransformersConfig(options: InitOptions) {\n  env.useFS = false;\n  env.localModelPath = options.dir;\n  env.allowLocalModels = false;\n  env.allowRemoteModels = true;\n  env.cacheDir = options.dir;\n  env.customCache = new FileProxyCache(options);\n  env.useBrowserCache = false;\n  env.useFSCache = true;\n  env.useCustomCache = true;\n\n  env.remoteHost = `https://${options.remoteHost}`;\n}\n", "import { env } from '@huggingface/transformers';\nimport { createNormalizeVfs, path } from '@cyia/vfs2';\nimport { downloadFile } from '@cyia/dl';\nimport fs from 'fs';\nimport { InitOptions } from './set-transformers-config';\nexport interface NodeProxy {\n  match: (request: string) => Promise<ArrayBuffer | undefined>;\n  put: (request: string, arraybuffer: ArrayBuffer) => Promise<void>;\n}\nexport class FileProxyCache {\n  #path;\n  #vfs;\n  #sendMessage;\n  #modelName;\n  #downloadConfig;\n  #initOptions;\n  constructor(initOptions: InitOptions) {\n    this.#modelName = initOptions.modelName;\n    this.#sendMessage = (message: any) => {\n      initOptions.port?.postMessage({ type: 'progress', message });\n    };\n    this.#path = initOptions.dir;\n    this.#vfs = createNormalizeVfs({ dir: initOptions.dir });\n    this.#downloadConfig = initOptions.downloadConfig;\n    this.#initOptions = initOptions;\n  }\n  async match(request: string): Promise<FileResponse | undefined> {\n    let filePath;\n    if (request.startsWith('http')) {\n      const data = new URL(request);\n      filePath = path.join(\n        this.#path,\n        data.pathname.replace(\n          '/' +\n            env.remotePathTemplate\n              .replaceAll('{model}', this.#modelName)\n              .replaceAll('{revision}', encodeURIComponent('main')),\n          `/${this.#modelName}/`,\n        ),\n      );\n\n      await downloadFile(request, {\n        ...this.#downloadConfig,\n        savePath: filePath,\n        message: this.#sendMessage,\n        headers: {\n          token: this.#initOptions?.hfToken ?? '',\n          'software-bbs': 'bbs.shenghuabi.site',\n        },\n      });\n    } else {\n      filePath = request;\n    }\n    const exists = await this.#vfs.exists(filePath);\n    if (exists) {\n      return new FileResponse(filePath);\n    }\n    return undefined;\n  }\n\n  async put(request: string, response: Response | FileResponse): Promise<void> {\n    throw new Error('no put');\n  }\n}\nconst decoder = new TextDecoder('utf-8');\nconst CONTENT_TYPE_MAP = {\n  txt: 'text/plain',\n  html: 'text/html',\n  css: 'text/css',\n  js: 'text/javascript',\n  json: 'application/json',\n  png: 'image/png',\n  jpg: 'image/jpeg',\n  jpeg: 'image/jpeg',\n  gif: 'image/gif',\n};\nclass FileResponse {\n  filePath;\n  headers;\n  exists = true;\n  status = 200;\n  statusText = 'OK';\n  body;\n  constructor(filePath: string) {\n    this.filePath = filePath;\n    this.headers = new Headers();\n\n    this.updateContentType();\n\n    this.body = fs.createReadStream(filePath);\n  }\n\n  updateContentType() {\n    const stats = fs.statSync(this.filePath);\n    this.headers.set('content-length', stats.size.toString());\n\n    const extension = this.filePath.toString().split('.').pop()!.toLowerCase();\n    this.headers.set(\n      'content-type',\n      (CONTENT_TYPE_MAP as any)[extension] ?? 'application/octet-stream',\n    );\n  }\n\n  clone(): FileResponse {\n    const response = new FileResponse(this.filePath);\n    response.exists = this.exists;\n    response.status = this.status;\n    response.statusText = this.statusText;\n    response.headers = new Headers(this.headers);\n    return response;\n  }\n\n  async arrayBuffer(): Promise<ArrayBuffer> {\n    return fs.promises\n      .readFile(this.filePath)\n      .then((buffer) => buffer.buffer as ArrayBuffer);\n  }\n\n  async blob(): Promise<Blob> {\n    return new Blob([await this.arrayBuffer()], {\n      type: this.headers.get('content-type')!,\n    });\n  }\n\n  async text(): Promise<string> {\n    return decoder.decode(await this.arrayBuffer());\n  }\n\n  async json(): Promise<object> {\n    return JSON.parse(await this.text());\n  }\n}\n"],
-  "mappings": ";AAAA,SAAS,YAAAA,iBAAgB;;;ACAzB,SAAS,OAAAC,YAAqB;;;ACA9B,SAAS,WAAW;AACpB,SAAS,oBAAoB,YAAY;AACzC,SAAS,oBAAoB;AAC7B,OAAO,QAAQ;AAMR,IAAM,iBAAN,MAAqB;AAAA,EAC1B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA,YAAY,aAA0B;AACpC,SAAK,aAAa,YAAY;AAC9B,SAAK,eAAe,CAAC,YAAiB;AACpC,kBAAY,MAAM,YAAY,EAAE,MAAM,YAAY,QAAQ,CAAC;AAAA,IAC7D;AACA,SAAK,QAAQ,YAAY;AACzB,SAAK,OAAO,mBAAmB,EAAE,KAAK,YAAY,IAAI,CAAC;AACvD,SAAK,kBAAkB,YAAY;AACnC,SAAK,eAAe;AAAA,EACtB;AAAA,EACA,MAAM,MAAM,SAAoD;AAC9D,QAAI;AACJ,QAAI,QAAQ,WAAW,MAAM,GAAG;AAC9B,YAAM,OAAO,IAAI,IAAI,OAAO;AAC5B,iBAAW,KAAK;AAAA,QACd,KAAK;AAAA,QACL,KAAK,SAAS;AAAA,UACZ,MACE,IAAI,mBACD,WAAW,WAAW,KAAK,UAAU,EACrC,WAAW,cAAc,mBAAmB,MAAM,CAAC;AAAA,UACxD,IAAI,KAAK,UAAU;AAAA,QACrB;AAAA,MACF;AAEA,YAAM,aAAa,SAAS;AAAA,QAC1B,GAAG,KAAK;AAAA,QACR,UAAU;AAAA,QACV,SAAS,KAAK;AAAA,QACd,SAAS;AAAA,UACP,OAAO,KAAK,cAAc,WAAW;AAAA,UACrC,gBAAgB;AAAA,QAClB;AAAA,MACF,CAAC;AAAA,IACH,OAAO;AACL,iBAAW;AAAA,IACb;AACA,UAAM,SAAS,MAAM,KAAK,KAAK,OAAO,QAAQ;AAC9C,QAAI,QAAQ;AACV,aAAO,IAAI,aAAa,QAAQ;AAAA,IAClC;AACA,WAAO;AAAA,EACT;AAAA,EAEA,MAAM,IAAI,SAAiB,UAAkD;AAC3E,UAAM,IAAI,MAAM,QAAQ;AAAA,EAC1B;AACF;AACA,IAAM,UAAU,IAAI,YAAY,OAAO;AACvC,IAAM,mBAAmB;AAAA,EACvB,KAAK;AAAA,EACL,MAAM;AAAA,EACN,KAAK;AAAA,EACL,IAAI;AAAA,EACJ,MAAM;AAAA,EACN,KAAK;AAAA,EACL,KAAK;AAAA,EACL,MAAM;AAAA,EACN,KAAK;AACP;AACA,IAAM,eAAN,MAAM,cAAa;AAAA,EACjB;AAAA,EACA;AAAA,EACA,SAAS;AAAA,EACT,SAAS;AAAA,EACT,aAAa;AAAA,EACb;AAAA,EACA,YAAY,UAAkB;AAC5B,SAAK,WAAW;AAChB,SAAK,UAAU,IAAI,QAAQ;AAE3B,SAAK,kBAAkB;AAEvB,SAAK,OAAO,GAAG,iBAAiB,QAAQ;AAAA,EAC1C;AAAA,EAEA,oBAAoB;AAClB,UAAM,QAAQ,GAAG,SAAS,KAAK,QAAQ;AACvC,SAAK,QAAQ,IAAI,kBAAkB,MAAM,KAAK,SAAS,CAAC;AAExD,UAAM,YAAY,KAAK,SAAS,SAAS,EAAE,MAAM,GAAG,EAAE,IAAI,EAAG,YAAY;AACzE,SAAK,QAAQ;AAAA,MACX;AAAA,MACC,iBAAyB,SAAS,KAAK;AAAA,IAC1C;AAAA,EACF;AAAA,EAEA,QAAsB;AACpB,UAAM,WAAW,IAAI,cAAa,KAAK,QAAQ;AAC/C,aAAS,SAAS,KAAK;AACvB,aAAS,SAAS,KAAK;AACvB,aAAS,aAAa,KAAK;AAC3B,aAAS,UAAU,IAAI,QAAQ,KAAK,OAAO;AAC3C,WAAO;AAAA,EACT;AAAA,EAEA,MAAM,cAAoC;AACxC,WAAO,GAAG,SACP,SAAS,KAAK,QAAQ,EACtB,KAAK,CAAC,WAAW,OAAO,MAAqB;AAAA,EAClD;AAAA,EAEA,MAAM,OAAsB;AAC1B,WAAO,IAAI,KAAK,CAAC,MAAM,KAAK,YAAY,CAAC,GAAG;AAAA,MAC1C,MAAM,KAAK,QAAQ,IAAI,cAAc;AAAA,IACvC,CAAC;AAAA,EACH;AAAA,EAEA,MAAM,OAAwB;AAC5B,WAAO,QAAQ,OAAO,MAAM,KAAK,YAAY,CAAC;AAAA,EAChD;AAAA,EAEA,MAAM,OAAwB;AAC5B,WAAO,KAAK,MAAM,MAAM,KAAK,KAAK,CAAC;AAAA,EACrC;AACF;;;ADjHO,SAAS,sBAAsB,SAAsB;AAC1D,EAAAC,KAAI,QAAQ;AACZ,EAAAA,KAAI,iBAAiB,QAAQ;AAC7B,EAAAA,KAAI,mBAAmB;AACvB,EAAAA,KAAI,oBAAoB;AACxB,EAAAA,KAAI,WAAW,QAAQ;AACvB,EAAAA,KAAI,cAAc,IAAI,eAAe,OAAO;AAC5C,EAAAA,KAAI,kBAAkB;AACtB,EAAAA,KAAI,aAAa;AACjB,EAAAA,KAAI,iBAAiB;AAErB,EAAAA,KAAI,aAAa,WAAW,QAAQ,UAAU;AAChD;;;AD1BA,SAAS,WACP,WACA,OACA,cAAc,8EACd;AACA,SAAO;AAAA,IACL,MAAM,IAAI,CAAC,SAAS,aAAa,WAAW;AAAA,QAAW,IAAI,EAAE;AAAA,IAC7D;AAAA,MACE,SAAS;AAAA,MACT,WAAW;AAAA,IACb;AAAA,EACF;AACF;AAEA,IAAM,kBAAN,MAAsB;AAAA,EACpB;AAAA,EACA,cAAc;AAAA,EAAC;AAAA,EACf,OAAO,OAAO,YAAyB;AACrC,QAAI,KAAK,YAAY;AACnB,aAAO;AAAA,IACT;AACA,SAAK,aAAa,MAAM,KAAK,cAAc,OAAO;AAClD,WAAO,CAAC,CAAC,KAAK;AAAA,EAChB;AAAA,EACA,UAAU,OACR,UAKG;AACH,QAAI,CAAC,KAAK,YAAY;AACpB,YAAM,KAAK,KAAK,KAAK;AAAA,IACvB;AACA,UAAM,YACJ,OAAO,MAAM,UAAU,WAAW,CAAC,MAAM,KAAK,IAAI,MAAM;AAC1D,QAAI;AACJ,QAAI,MAAM,SAAS,SAAS;AAC1B,eAAS,WAAW,KAAK,YAAa,WAAW,MAAM,eAAe;AAAA,IACxE,OAAO;AACL,eAAS,KAAK,WAAY,WAAW;AAAA,QACnC,SAAS;AAAA,QACT,WAAW;AAAA,MACb,CAAC;AAAA,IACH;AACA,WAAO,OAAO,KAAK,CAACC,YAAW;AAC7B,YAAM,OAAOA,QAAO,OAAO;AAC3B,aAAO,OAAO,MAAM,UAAU,WAAW,KAAK,CAAC,IAAI;AAAA,IACrD,CAAC;AAAA,EACH;AAAA,EACA,MAAM,cAAc,SAAsB;AACxC,0BAAsB,OAAO;AAE7B,WAAO,MAAMC;AAAA,MACX;AAAA,MACA,QAAQ;AAAA,MACR,QAAQ;AAAA,IACV;AAAA,EACF;AAAA,EACA,UAAU,MAAO,KAAK,WAAY,MAAM,OAAe;AACzD;AACA,IAAM,WAAW,IAAI,gBAAgB;AACrC,IAAM,OAAO,SAAS;AACtB,IAAM,UAAU,SAAS;AACzB,IAAM,UAAU,SAAS;",
+  "sourcesContent": ["import { pipeline } from '@huggingface/transformers';\nimport type { FeatureExtractionPipeline } from '@huggingface/transformers';\n\nimport { InitOptions, setTransformersConfig } from '../set-transformers-config';\nfunction qwen3ToVec(\n  extractor: FeatureExtractionPipeline,\n  query: string[],\n  description = 'Given a web search query, retrieve relevant passages that answer the query',\n) {\n  return extractor!(\n    query.map((item) => `Instruct: ${description}\\nQuery:${item}`),\n    {\n      pooling: 'last_token',\n      normalize: true,\n    },\n  );\n}\n\nclass Text2VecService {\n  #extractor!: FeatureExtractionPipeline | undefined;\n  constructor() {}\n  init = async (options: InitOptions) => {\n    if (this.#extractor) {\n      return true;\n    }\n    this.#extractor = await this.#downloadOnly(options);\n    return !!this.#extractor;\n  };\n  convert = async (\n    input: {\n      value: string | string[];\n      mode?: 'qwen3';\n      taskDescription?: string;\n    } & InitOptions,\n  ) => {\n    if (!this.#extractor) {\n      await this.init(input);\n    }\n    const inputList =\n      typeof input.value === 'string' ? [input.value] : input.value;\n    let result;\n    if (input.mode === 'qwen3') {\n      result = qwen3ToVec(this.#extractor!, inputList, input.taskDescription);\n    } else {\n      result = this.#extractor!(inputList, {\n        pooling: 'mean',\n        normalize: true,\n      });\n    }\n    return result.then((result) => {\n      const list = result.tolist();\n      return typeof input.value === 'string' ? list[0] : list;\n    });\n  };\n  async #downloadOnly(options: InitOptions) {\n    setTransformersConfig(options);\n\n    return await pipeline(\n      'feature-extraction',\n      options.modelName,\n      options.options,\n    );\n  }\n  getSize = () => (this.#extractor!.model.config as any).hidden_size;\n}\nconst instance = new Text2VecService();\nconst init = instance.init;\nconst getSize = instance.getSize;\nconst convert = instance.convert;\nexport { init, getSize, convert };\n", "import { env, pipeline } from '@huggingface/transformers';\nimport { MessagePort } from 'worker_threads';\nimport { FileProxyCache } from './custom-cache';\nimport type { DownloadConfigType } from '@cyia/external-call';\ntype PipeLineOptions = Partial<NonNullable<Parameters<typeof pipeline>[2]>>;\nexport interface InitOptions {\n  /** 文件夹 */\n  dir: string;\n  /** 模型 */\n  modelName: string;\n  /** 模型参数 */\n  options: PipeLineOptions;\n  /**直接链接 */\n  remoteHost: string;\n  downloadConfig?: DownloadConfigType;\n  port?: MessagePort;\n  hfToken?: string;\n}\nexport function setTransformersConfig(options: InitOptions) {\n  env.useFS = false;\n  env.localModelPath = options.dir;\n  env.allowLocalModels = false;\n  env.allowRemoteModels = true;\n  env.cacheDir = options.dir;\n  env.customCache = new FileProxyCache(options);\n  env.useBrowserCache = false;\n  env.useFSCache = true;\n  env.useCustomCache = true;\n\n  env.remoteHost = `https://${options.remoteHost}`;\n}\n", "import { env } from '@huggingface/transformers';\nimport { createNormalizeVfs, path } from '@cyia/vfs2';\nimport { downloadFile } from '@cyia/dl';\nimport fs from 'fs';\nimport { InitOptions } from './set-transformers-config';\nexport interface NodeProxy {\n  match: (request: string) => Promise<ArrayBuffer | undefined>;\n  put: (request: string, arraybuffer: ArrayBuffer) => Promise<void>;\n}\nexport class FileProxyCache {\n  #path;\n  #vfs;\n  #sendMessage;\n  #modelName;\n  #downloadConfig;\n  #initOptions;\n  constructor(initOptions: InitOptions) {\n    this.#modelName = initOptions.modelName;\n    this.#sendMessage = (message: any) => {\n      initOptions.port?.postMessage({ type: 'progress', message });\n    };\n    this.#path = initOptions.dir;\n    this.#vfs = createNormalizeVfs({ dir: initOptions.dir });\n    this.#downloadConfig = initOptions.downloadConfig;\n    this.#initOptions = initOptions;\n  }\n  async #createResponse(filePath: string): Promise<Response> {\n    const stats = await fs.promises.stat(filePath);\n    const extension = filePath.split('.').pop()!.toLowerCase();\n    const contentType =\n      (CONTENT_TYPE_MAP as any)[extension] ?? 'application/octet-stream';\n    const stream = fs.createReadStream(filePath);\n    return new Response(stream, {\n      status: 200,\n      statusText: 'OK',\n      headers: {\n        'content-type': contentType,\n        'content-length': stats.size.toString(),\n      },\n    });\n  }\n  async match(request: string): Promise<Response | undefined> {\n    let filePath;\n    if (request.startsWith('http')) {\n      const data = new URL(request);\n      filePath = path.join(\n        this.#path,\n        data.pathname.replace(\n          '/' +\n            env.remotePathTemplate\n              .replaceAll('{model}', this.#modelName)\n              .replaceAll('{revision}', encodeURIComponent('main')),\n          `/${this.#modelName}/`,\n        ),\n      );\n\n      await downloadFile(request, {\n        ...this.#downloadConfig,\n        savePath: filePath,\n        message: this.#sendMessage,\n        headers: {\n          token: this.#initOptions?.hfToken ?? '',\n          'software-bbs': 'bbs.shenghuabi.site',\n        },\n      });\n    } else {\n      filePath = request;\n    }\n    const exists = await this.#vfs.exists(filePath);\n    if (exists) {\n      return this.#createResponse(filePath);\n    }\n    return undefined;\n  }\n\n  async put(request: string, response: Response): Promise<void> {\n    throw new Error('no put');\n  }\n}\nconst CONTENT_TYPE_MAP: Record<string, string> = {\n  txt: 'text/plain',\n  html: 'text/html',\n  css: 'text/css',\n  js: 'text/javascript',\n  json: 'application/json',\n  png: 'image/png',\n  jpg: 'image/jpeg',\n  jpeg: 'image/jpeg',\n  gif: 'image/gif',\n};\n"],
+  "mappings": ";AAAA,SAAS,YAAAA,iBAAgB;;;ACAzB,SAAS,OAAAC,YAAqB;;;ACA9B,SAAS,WAAW;AACpB,SAAS,oBAAoB,YAAY;AACzC,SAAS,oBAAoB;AAC7B,OAAO,QAAQ;AAMR,IAAM,iBAAN,MAAqB;AAAA,EAC1B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA,YAAY,aAA0B;AACpC,SAAK,aAAa,YAAY;AAC9B,SAAK,eAAe,CAAC,YAAiB;AACpC,kBAAY,MAAM,YAAY,EAAE,MAAM,YAAY,QAAQ,CAAC;AAAA,IAC7D;AACA,SAAK,QAAQ,YAAY;AACzB,SAAK,OAAO,mBAAmB,EAAE,KAAK,YAAY,IAAI,CAAC;AACvD,SAAK,kBAAkB,YAAY;AACnC,SAAK,eAAe;AAAA,EACtB;AAAA,EACA,MAAM,gBAAgB,UAAqC;AACzD,UAAM,QAAQ,MAAM,GAAG,SAAS,KAAK,QAAQ;AAC7C,UAAM,YAAY,SAAS,MAAM,GAAG,EAAE,IAAI,EAAG,YAAY;AACzD,UAAM,cACH,iBAAyB,SAAS,KAAK;AAC1C,UAAM,SAAS,GAAG,iBAAiB,QAAQ;AAC3C,WAAO,IAAI,SAAS,QAAQ;AAAA,MAC1B,QAAQ;AAAA,MACR,YAAY;AAAA,MACZ,SAAS;AAAA,QACP,gBAAgB;AAAA,QAChB,kBAAkB,MAAM,KAAK,SAAS;AAAA,MACxC;AAAA,IACF,CAAC;AAAA,EACH;AAAA,EACA,MAAM,MAAM,SAAgD;AAC1D,QAAI;AACJ,QAAI,QAAQ,WAAW,MAAM,GAAG;AAC9B,YAAM,OAAO,IAAI,IAAI,OAAO;AAC5B,iBAAW,KAAK;AAAA,QACd,KAAK;AAAA,QACL,KAAK,SAAS;AAAA,UACZ,MACE,IAAI,mBACD,WAAW,WAAW,KAAK,UAAU,EACrC,WAAW,cAAc,mBAAmB,MAAM,CAAC;AAAA,UACxD,IAAI,KAAK,UAAU;AAAA,QACrB;AAAA,MACF;AAEA,YAAM,aAAa,SAAS;AAAA,QAC1B,GAAG,KAAK;AAAA,QACR,UAAU;AAAA,QACV,SAAS,KAAK;AAAA,QACd,SAAS;AAAA,UACP,OAAO,KAAK,cAAc,WAAW;AAAA,UACrC,gBAAgB;AAAA,QAClB;AAAA,MACF,CAAC;AAAA,IACH,OAAO;AACL,iBAAW;AAAA,IACb;AACA,UAAM,SAAS,MAAM,KAAK,KAAK,OAAO,QAAQ;AAC9C,QAAI,QAAQ;AACV,aAAO,KAAK,gBAAgB,QAAQ;AAAA,IACtC;AACA,WAAO;AAAA,EACT;AAAA,EAEA,MAAM,IAAI,SAAiB,UAAmC;AAC5D,UAAM,IAAI,MAAM,QAAQ;AAAA,EAC1B;AACF;AACA,IAAM,mBAA2C;AAAA,EAC/C,KAAK;AAAA,EACL,MAAM;AAAA,EACN,KAAK;AAAA,EACL,IAAI;AAAA,EACJ,MAAM;AAAA,EACN,KAAK;AAAA,EACL,KAAK;AAAA,EACL,MAAM;AAAA,EACN,KAAK;AACP;;;ADvEO,SAAS,sBAAsB,SAAsB;AAC1D,EAAAC,KAAI,QAAQ;AACZ,EAAAA,KAAI,iBAAiB,QAAQ;AAC7B,EAAAA,KAAI,mBAAmB;AACvB,EAAAA,KAAI,oBAAoB;AACxB,EAAAA,KAAI,WAAW,QAAQ;AACvB,EAAAA,KAAI,cAAc,IAAI,eAAe,OAAO;AAC5C,EAAAA,KAAI,kBAAkB;AACtB,EAAAA,KAAI,aAAa;AACjB,EAAAA,KAAI,iBAAiB;AAErB,EAAAA,KAAI,aAAa,WAAW,QAAQ,UAAU;AAChD;;;AD1BA,SAAS,WACP,WACA,OACA,cAAc,8EACd;AACA,SAAO;AAAA,IACL,MAAM,IAAI,CAAC,SAAS,aAAa,WAAW;AAAA,QAAW,IAAI,EAAE;AAAA,IAC7D;AAAA,MACE,SAAS;AAAA,MACT,WAAW;AAAA,IACb;AAAA,EACF;AACF;AAEA,IAAM,kBAAN,MAAsB;AAAA,EACpB;AAAA,EACA,cAAc;AAAA,EAAC;AAAA,EACf,OAAO,OAAO,YAAyB;AACrC,QAAI,KAAK,YAAY;AACnB,aAAO;AAAA,IACT;AACA,SAAK,aAAa,MAAM,KAAK,cAAc,OAAO;AAClD,WAAO,CAAC,CAAC,KAAK;AAAA,EAChB;AAAA,EACA,UAAU,OACR,UAKG;AACH,QAAI,CAAC,KAAK,YAAY;AACpB,YAAM,KAAK,KAAK,KAAK;AAAA,IACvB;AACA,UAAM,YACJ,OAAO,MAAM,UAAU,WAAW,CAAC,MAAM,KAAK,IAAI,MAAM;AAC1D,QAAI;AACJ,QAAI,MAAM,SAAS,SAAS;AAC1B,eAAS,WAAW,KAAK,YAAa,WAAW,MAAM,eAAe;AAAA,IACxE,OAAO;AACL,eAAS,KAAK,WAAY,WAAW;AAAA,QACnC,SAAS;AAAA,QACT,WAAW;AAAA,MACb,CAAC;AAAA,IACH;AACA,WAAO,OAAO,KAAK,CAACC,YAAW;AAC7B,YAAM,OAAOA,QAAO,OAAO;AAC3B,aAAO,OAAO,MAAM,UAAU,WAAW,KAAK,CAAC,IAAI;AAAA,IACrD,CAAC;AAAA,EACH;AAAA,EACA,MAAM,cAAc,SAAsB;AACxC,0BAAsB,OAAO;AAE7B,WAAO,MAAMC;AAAA,MACX;AAAA,MACA,QAAQ;AAAA,MACR,QAAQ;AAAA,IACV;AAAA,EACF;AAAA,EACA,UAAU,MAAO,KAAK,WAAY,MAAM,OAAe;AACzD;AACA,IAAM,WAAW,IAAI,gBAAgB;AACrC,IAAM,OAAO,SAAS;AACtB,IAAM,UAAU,SAAS;AACzB,IAAM,UAAU,SAAS;",
   "names": ["pipeline", "env", "env", "result", "pipeline"]
 }