@hamster-note/document-parser 0.1.0 → 0.2.1

package/dist/index.d.ts

@@ -0,0 +1,53 @@
+import { IntermediateDocument, Number2 } from "@hamster-note/types";
+
+//#region src/types/DocumentAnchor.d.ts
+interface DocumentAnchorWithTextId {
+  pageId: string;
+  textId?: string;
+}
+interface DocumentAnchorWithPosition {
+  pageId: string;
+  position: Number2;
+}
+type DocumentAnchor = DocumentAnchorWithPosition | DocumentAnchorWithTextId;
+//#endregion
+//#region src/Parser/index.d.ts
+type ParserInput = ArrayBuffer | ArrayBufferView | Blob;
+declare abstract class DocumentParser {
+  static readonly exts: readonly string[];
+  /**
+   * 将原始输入解析为中间文档结构。
+   * - 输入必须是内存态数据（ArrayBuffer / ArrayBufferView / Blob）
+   * - 失败时抛出异常；不要返回 undefined。
+   */
+  abstract encode(input: ParserInput): Promise<IntermediateDocument>;
+  /**
+   * 可选：从中间文档逆序列化回原始文件数据。
+   * 默认不支持，需要的子类自行实现。
+   */
+  decode(_intermediateDocument: IntermediateDocument): Promise<ParserInput>;
+  protected static toArrayBuffer(input: ParserInput): Promise<ArrayBuffer>;
+  protected static toUint8Array(input: ParserInput): Promise<Uint8Array>;
+}
+//#endregion
+//#region src/register/index.d.ts
+interface DocumentParserConstructor<T extends DocumentParser = DocumentParser> {
+  new (): T;
+  readonly exts: readonly string[];
+}
+declare const registerParser: (Parser: DocumentParserConstructor) => void;
+declare function unregisterParser(Parser: DocumentParserConstructor): void;
+declare function getParserByExt(ext?: string): DocumentParserConstructor | undefined;
+interface ParseOptions {
+  filename?: string;
+  ext?: string;
+  parser?: DocumentParserConstructor;
+}
+declare function parse(input: ParserInput, options?: ParseOptions): Promise<IntermediateDocument>;
+declare function listRegisteredParsers(): {
+  ext: string;
+  parser: DocumentParserConstructor<DocumentParser>;
+}[];
+//#endregion
+export { DocumentAnchor, DocumentAnchorWithPosition, DocumentAnchorWithTextId, DocumentParser, DocumentParserConstructor, ParseOptions, ParserInput, getParserByExt, listRegisteredParsers, parse, registerParser, unregisterParser };
+//# sourceMappingURL=index.d.ts.map

package/dist/index.js

@@ -0,0 +1,66 @@
+//#region src/Parser/index.ts
+var DocumentParser = class DocumentParser {
+	static exts = [];
+	/**
+	* 可选：从中间文档逆序列化回原始文件数据。
+	* 默认不支持，需要的子类自行实现。
+	*/
+	decode(_intermediateDocument) {
+		return Promise.reject(/* @__PURE__ */ new Error("decode is not implemented"));
+	}
+	static async toArrayBuffer(input) {
+		if (input instanceof ArrayBuffer) return input;
+		if (ArrayBuffer.isView(input)) {
+			const buffer = new ArrayBuffer(input.byteLength);
+			new Uint8Array(buffer).set(new Uint8Array(input.buffer, input.byteOffset, input.byteLength));
+			return buffer;
+		}
+		if (typeof Blob !== "undefined" && input instanceof Blob) return input.arrayBuffer();
+		throw new TypeError("Unsupported parser input");
+	}
+	static async toUint8Array(input) {
+		const buffer = await DocumentParser.toArrayBuffer(input);
+		return new Uint8Array(buffer);
+	}
+};
+
+//#endregion
+//#region src/register/index.ts
+const parserMap = /* @__PURE__ */ new Map();
+const normalizeExt = (ext) => ext.replace(/^\./, "").toLowerCase();
+const registerParser = (Parser) => {
+	if (!Array.isArray(Parser.exts) || Parser.exts.length === 0) throw new Error("Parser.exts 必须是非空数组");
+	Parser.exts.forEach((ext) => {
+		const key = normalizeExt(ext);
+		const exists = parserMap.get(key);
+		if (exists && exists !== Parser) throw new Error(`扩展名 ${key} 已被其他解析器占用`);
+		parserMap.set(key, Parser);
+	});
+};
+function unregisterParser(Parser) {
+	Parser.exts.forEach((ext) => {
+		const key = normalizeExt(ext);
+		if (parserMap.get(key) === Parser) parserMap.delete(key);
+	});
+}
+function getParserByExt(ext) {
+	if (!ext) return void 0;
+	return parserMap.get(normalizeExt(ext));
+}
+async function parse(input, options = {}) {
+	const extFromName = options.filename?.split(".").pop();
+	const ext = options.ext ?? extFromName;
+	const Parser = options.parser ?? getParserByExt(ext);
+	if (!Parser) throw new Error(`未找到匹配的解析器，扩展名: ${ext ?? "未知"}，已注册: ${[...parserMap.keys()].join(", ")}`);
+	return new Parser().encode(input);
+}
+function listRegisteredParsers() {
+	return [...parserMap.entries()].map(([ext, ctor]) => ({
+		ext,
+		parser: ctor
+	}));
+}
+
+//#endregion
+export { DocumentParser, getParserByExt, listRegisteredParsers, parse, registerParser, unregisterParser };
+//# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","names":[],"sources":["../src/Parser/index.ts","../src/register/index.ts"],"sourcesContent":["import type { IntermediateDocument } from '@hamster-note/types'\n\n// 解析器通用输入，兼容 Browser / Worker / Node（Buffer 属于 ArrayBufferView）\nexport type ParserInput = ArrayBuffer | ArrayBufferView | Blob\n\nexport abstract class DocumentParser {\n  // 子类必须声明支持的扩展名列表（不带点），用于注册分发\n  static readonly exts: readonly string[] = []\n\n  /**\n   * 将原始输入解析为中间文档结构。\n   * - 输入必须是内存态数据（ArrayBuffer / ArrayBufferView / Blob）\n   * - 失败时抛出异常；不要返回 undefined。\n   */\n  abstract encode(input: ParserInput): Promise<IntermediateDocument>\n\n  /**\n   * 可选：从中间文档逆序列化回原始文件数据。\n   * 默认不支持，需要的子类自行实现。\n   */\n  decode(_intermediateDocument: IntermediateDocument): Promise<ParserInput> {\n    return Promise.reject(new Error('decode is not implemented'))\n  }\n\n  // 将各种输入转换为 ArrayBuffer，方便子类直接消费二进制\n  protected static async toArrayBuffer(\n    input: ParserInput\n  ): Promise<ArrayBuffer> {\n    if (input instanceof ArrayBuffer) {\n      return input\n    }\n    if (ArrayBuffer.isView(input)) {\n      // 拷贝为独立的 ArrayBuffer，避免 SharedArrayBuffer 兼容性问题\n      const buffer = new ArrayBuffer(input.byteLength)\n      new Uint8Array(buffer).set(\n        new Uint8Array(input.buffer, input.byteOffset, input.byteLength)\n      )\n      return buffer\n    }\n    if (typeof Blob !== 'undefined' && input instanceof Blob) {\n      return input.arrayBuffer()\n    }\n    throw new TypeError('Unsupported parser input')\n  }\n\n  // 子类若需要 Uint8Array，可直接复用\n  protected static async toUint8Array(input: ParserInput): Promise<Uint8Array> {\n    const buffer = await DocumentParser.toArrayBuffer(input)\n    return new Uint8Array(buffer)\n  }\n}\n","import type { DocumentParser, ParserInput } from '../Parser/index'\nimport type { IntermediateDocument } from '@hamster-note/types'\n\nexport interface DocumentParserConstructor<\n  T extends DocumentParser = DocumentParser\n> {\n  new (): T\n  readonly exts: readonly string[]\n}\n\nconst parserMap: Map<string, DocumentParserConstructor> = new Map()\n\nconst normalizeExt = (ext: string) => ext.replace(/^\\./, '').toLowerCase()\n\n// 注册解析器：防止重复注册同扩展名的不同类\nexport const registerParser = (Parser: DocumentParserConstructor) => {\n  if (!Array.isArray(Parser.exts) || Parser.exts.length === 0) {\n    throw new Error('Parser.exts 必须是非空数组')\n  }\n  Parser.exts.forEach((ext) => {\n    const key = normalizeExt(ext)\n    const exists = parserMap.get(key)\n    if (exists && exists !== Parser) {\n      throw new Error(`扩展名 ${key} 已被其他解析器占用`)\n    }\n    parserMap.set(key, Parser)\n  })\n}\n\nexport function unregisterParser(Parser: DocumentParserConstructor) {\n  Parser.exts.forEach((ext) => {\n    const key = normalizeExt(ext)\n    const exists = parserMap.get(key)\n    if (exists === Parser) {\n      parserMap.delete(key)\n    }\n  })\n}\n\nexport function getParserByExt(\n  ext?: string\n): DocumentParserConstructor | undefined {\n  if (!ext) return undefined\n  return parserMap.get(normalizeExt(ext))\n}\n\nexport interface ParseOptions {\n  filename?: string\n  ext?: string\n  parser?: DocumentParserConstructor\n}\n\nexport async function parse(\n  input: ParserInput,\n  options: ParseOptions = {}\n): Promise<IntermediateDocument> {\n  const extFromName = options.filename?.split('.').pop()\n  const ext = options.ext ?? extFromName\n  const Parser = options.parser ?? getParserByExt(ext)\n\n  if (!Parser) {\n    throw new Error(\n      `未找到匹配的解析器，扩展名: ${ext ?? '未知'}，已注册: ${[\n        ...parserMap.keys()\n      ].join(', ')}`\n    )\n  }\n\n  const parser = new Parser()\n  return parser.encode(input)\n}\n\n// 便于测试或运行时可观测性\nexport function listRegisteredParsers() {\n  return [...parserMap.entries()].map(([ext, ctor]) => ({ ext, parser: ctor }))\n}\n"],"mappings":";AAKA,IAAsB,iBAAtB,MAAsB,eAAe;CAEnC,OAAgB,OAA0B,EAAE;;;;;CAa5C,OAAO,uBAAmE;AACxE,SAAO,QAAQ,uBAAO,IAAI,MAAM,4BAA4B,CAAC;;CAI/D,aAAuB,cACrB,OACsB;AACtB,MAAI,iBAAiB,YACnB,QAAO;AAET,MAAI,YAAY,OAAO,MAAM,EAAE;GAE7B,MAAM,SAAS,IAAI,YAAY,MAAM,WAAW;AAChD,OAAI,WAAW,OAAO,CAAC,IACrB,IAAI,WAAW,MAAM,QAAQ,MAAM,YAAY,MAAM,WAAW,CACjE;AACD,UAAO;;AAET,MAAI,OAAO,SAAS,eAAe,iBAAiB,KAClD,QAAO,MAAM,aAAa;AAE5B,QAAM,IAAI,UAAU,2BAA2B;;CAIjD,aAAuB,aAAa,OAAyC;EAC3E,MAAM,SAAS,MAAM,eAAe,cAAc,MAAM;AACxD,SAAO,IAAI,WAAW,OAAO;;;;;;ACtCjC,MAAM,4BAAoD,IAAI,KAAK;AAEnE,MAAM,gBAAgB,QAAgB,IAAI,QAAQ,OAAO,GAAG,CAAC,aAAa;AAG1E,MAAa,kBAAkB,WAAsC;AACnE,KAAI,CAAC,MAAM,QAAQ,OAAO,KAAK,IAAI,OAAO,KAAK,WAAW,EACxD,OAAM,IAAI,MAAM,sBAAsB;AAExC,QAAO,KAAK,SAAS,QAAQ;EAC3B,MAAM,MAAM,aAAa,IAAI;EAC7B,MAAM,SAAS,UAAU,IAAI,IAAI;AACjC,MAAI,UAAU,WAAW,OACvB,OAAM,IAAI,MAAM,OAAO,IAAI,YAAY;AAEzC,YAAU,IAAI,KAAK,OAAO;GAC1B;;AAGJ,SAAgB,iBAAiB,QAAmC;AAClE,QAAO,KAAK,SAAS,QAAQ;EAC3B,MAAM,MAAM,aAAa,IAAI;AAE7B,MADe,UAAU,IAAI,IAAI,KAClB,OACb,WAAU,OAAO,IAAI;GAEvB;;AAGJ,SAAgB,eACd,KACuC;AACvC,KAAI,CAAC,IAAK,QAAO;AACjB,QAAO,UAAU,IAAI,aAAa,IAAI,CAAC;;AASzC,eAAsB,MACpB,OACA,UAAwB,EAAE,EACK;CAC/B,MAAM,cAAc,QAAQ,UAAU,MAAM,IAAI,CAAC,KAAK;CACtD,MAAM,MAAM,QAAQ,OAAO;CAC3B,MAAM,SAAS,QAAQ,UAAU,eAAe,IAAI;AAEpD,KAAI,CAAC,OACH,OAAM,IAAI,MACR,kBAAkB,OAAO,KAAK,QAAQ,CACpC,GAAG,UAAU,MAAM,CACpB,CAAC,KAAK,KAAK,GACb;AAIH,QADe,IAAI,QAAQ,CACb,OAAO,MAAM;;AAI7B,SAAgB,wBAAwB;AACtC,QAAO,CAAC,GAAG,UAAU,SAAS,CAAC,CAAC,KAAK,CAAC,KAAK,WAAW;EAAE;EAAK,QAAQ;EAAM,EAAE"}

package/package.json

@@ -1,16 +1,20 @@
 {
   "name": "@hamster-note/document-parser",
-  "version": "0.1.0",
+  "version": "0.2.1",
   "description": "",
   "scripts": {
-    "build:all": "echo 'no-build'",
-    "test": "echo 'no-test'"
+    "build:all": "rolldown --config rolldown.config.ts",
+    "prepublishOnly": "yarn build:all",
+    "test": "echo 'no-test'",
+    "lint": "echo 'no-lint'"
   },
   "dependencies": {
-    "@hamster-note/types": "^0.1.0"
+    "@hamster-note/types": "^0.5.1"
   },
   "devDependencies": {
     "@eslint/js": "^9.39.2",
+    "rolldown": "^1.0.0-beta.58",
+    "rolldown-plugin-dts": "^0.20.0",
     "@system-ui-js/development-base": "^0.1.2",
     "@typescript-eslint/eslint-plugin": "^8.50.0",
     "@typescript-eslint/parser": "^8.50.0",
@@ -27,11 +31,18 @@
     "typescript": "^5.0.2",
     "typescript-eslint": "^8.50.0"
   },
-  "main": "src/index.ts",
-  "module": "src/index.ts",
-  "types": "src/index.ts",
+  "type": "module",
+  "main": "./dist/index.js",
+  "module": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js"
+    }
+  },
   "files": [
-    "src"
+    "dist"
   ],
   "packageManager": "yarn@1.22.22+sha512.a6b2f7906b721bba3d67d4aff083df04dad64c399707841b7acf00f6b133b7ac24255f2652fa22ae3534329dc6180534e98d17432037ff6fd140556e2bb3137e"
-}
+}

package/src/Document/index.ts

@@ -1,13 +0,0 @@
-import { HamsterPage } from '../Page'
-import { IntermediateOutline } from '@hamster-note/types'
-
-export abstract class HamsterDocument {
-  // 获取 Page 对象列表
-  abstract getPages(): Promise<HamsterPage[]>
-  // 获取 Page 对象
-  abstract getPage(pageNumber: number): Promise<HamsterPage | undefined>
-  // 获取大纲（PDF如果有内置的话）
-  abstract getOutline(): Promise<IntermediateOutline | undefined>
-  // 获取封面
-  abstract getCover(): Promise<HTMLCanvasElement | HTMLImageElement>
-}

package/src/Page/index.ts

@@ -1,22 +0,0 @@
-import { Number2 } from '@hamster-note/types'
-
-export enum RenderViews {
-  THUMBNAIL = 'thumbnail',
-  TEXT = 'text'
-}
-
-export interface RenderOptions {
-  views: RenderViews[]
-  scale: number
-}
-
-export abstract class HamsterPage {
-  abstract render(
-    container: HTMLDivElement,
-    options?: RenderOptions
-  ): Promise<void>
-  abstract getNumber(): number
-  abstract getSize(scale: number): Number2
-  abstract getPureText(): string
-  // abstract getTextDom
-}

package/src/Parser/index.ts

@@ -1,28 +0,0 @@
-import { HamsterDocument } from '../Document'
-import { IntermediateDocument } from '@hamster-note/types'
-
-export class DocumentParser {
-  static readonly ext: string
-  // Base static methods to be overridden by concrete parsers
-  static async encode(
-    _file: File | ArrayBuffer
-  ): Promise<IntermediateDocument | HamsterDocument | undefined> {
-    return Promise.resolve(undefined)
-  }
-  static async decode(
-    _intermediateDocument: IntermediateDocument
-  ): Promise<File | ArrayBuffer | undefined> {
-    return Promise.resolve(undefined)
-  }
-  protected static async toArrayBuffer(
-    fileOrBuffer: File | ArrayBuffer
-  ): Promise<ArrayBuffer> {
-    if (fileOrBuffer instanceof ArrayBuffer) return fileOrBuffer
-    return await new Promise<ArrayBuffer>((resolve, reject) => {
-      const reader = new FileReader()
-      reader.readAsArrayBuffer(fileOrBuffer)
-      reader.onload = () => resolve(reader.result as ArrayBuffer)
-      reader.onerror = (e) => reject(e)
-    })
-  }
-}

package/src/index.ts

@@ -1,5 +0,0 @@
-export * from './Document'
-export * from './Page'
-export * from './types/DocumentAnchor'
-export * from './Parser'
-export * from './register'

package/src/register/index.ts

@@ -1,30 +0,0 @@
-import { DocumentParser } from '../Parser'
-
-// Store constructors (classes), not instances
-const parserMap: Map<string, typeof DocumentParser> = new Map()
-
-// Register a parser class (constructor) which exposes a static `ext`
-export const registerParser = (parser: typeof DocumentParser) => {
-  console.log(parser.ext, parser)
-  parserMap.set(parser.ext, parser)
-}
-
-export function getParser(file: File): typeof DocumentParser | undefined {
-  const extension = file.name.split('.').pop()
-  if (!extension) {
-    return undefined
-  }
-  return parserMap.get(extension)
-}
-
-export function parse(file: File) {
-  const Parser = getParser(file)
-  console.log('Parse:', Parser)
-  if (!Parser) {
-    return undefined
-  }
-  // call static encode on the parser class
-  const res = Parser.encode(file)
-  // res.then((_doc) => { /* window.doc1 = _doc */ })
-  return res
-}

package/src/types/DocumentAnchor.ts

@@ -1,17 +0,0 @@
-// 文档锚（标记了文字id）
-import { Number2 } from '@hamster-note/types'
-
-export interface DocumentAnchorWithTextId {
-  pageId: string
-  textId?: string
-}
-
-// 文档锚（标记了位置）
-export interface DocumentAnchorWithPosition {
-  pageId: string
-  position: Number2
-}
-
-export type DocumentAnchor =
-  | DocumentAnchorWithPosition
-  | DocumentAnchorWithTextId