epub2md 1.2.1 → 1.3.0

package/lib/bin/cli.cjs

@@ -9,7 +9,7 @@ var _args = _interopRequireDefault(require("args"));
 var _chalk = _interopRequireDefault(require("chalk"));
 var _nodeProcess = _interopRequireDefault(require("node:process"));
 var _parseEpub = _interopRequireDefault(require("../parseEpub.cjs"));
-var _convert = _interopRequireDefault(require("./convert.cjs"));
+var _convert = require("./convert.cjs");
 var _beautyJson = require("beauty-json");
 function _interopRequireDefault(e) { return e && e.__esModule ? e : { default: e }; }
 const name = "epub2md";
@@ -47,7 +47,7 @@ function run(cmd) {
   const epubPath = flags["md" /* markdown */] || flags["ma" /* autocorrect */] || flags["unzip" /* unzip */];
   if (epubPath) {
     console.log(_chalk.default.blueBright(`[${name}]: converting${cmd === "ma" /* autocorrect */ ? " with AutoCorrect" : ""}...`));
-    new _convert.default({
+    new _convert.Converter({
       eubPath: epubPath,
       cmd
     }).run(flags["unzip" /* unzip */]).then(outDir => {

package/lib/bin/convert.cjs

@@ -3,7 +3,7 @@
 Object.defineProperty(exports, "__esModule", {
   value: true
 });
-
+exports.Converter = void 0;
 var _nodePath = require("node:path");
 var _nodeFs = require("node:fs");
 var _writeFileSafe = require("write-file-safe");
@@ -155,6 +155,7 @@ class Converter {
       filterPool = {};
     const padding = Math.floor(Math.log10(this.structure.length));
     this.structure.forEach(s => {
+      const numLabel = ("0".repeat(padding) + num).slice(-(padding + 1));
       const {
         outFilePath,
         content
@@ -165,7 +166,7 @@ class Converter {
         const parsedPath = (0, _nodePath.parse)(outFilePath);
         numberedOutFilePath = (0, _nodePath.format)({
           ...parsedPath,
-          base: `${("0".repeat(padding) + num).slice(-(padding + 1))}-${parsedPath.base}`
+          base: `${numLabel}-${parsedPath.base}`
         });
         console.log(_chalk.default.yellow(`${num++}: [${(0, _nodePath.basename)(numberedOutFilePath)}]`));
       }
@@ -177,4 +178,4 @@ class Converter {
     return this.outDir;
   }
 }
-module.exports = Converter;
+exports.Converter = Converter;

package/lib/parseEpub.cjs

@@ -13,21 +13,21 @@ var _parseLink = _interopRequireDefault(require("./parseLink.cjs"));
 var _parseSection = _interopRequireDefault(require("./parseSection.cjs"));
 var _utils = require("./utils.cjs");
 function _interopRequireDefault(e) { return e && e.__esModule ? e : { default: e }; }
-const parseMetadata = metadata => {
-  const meta = metadata[0];
+const parseMetadata = (metadata = {}) => {
+  const meta = metadata;
   const info = {};
   ["title", "author", "description", "language", "publisher", "rights"].forEach(item => {
     if (item === "author") {
-      info.author = _lodash.default.get(meta, ["dc:creator", 0]);
-      if (_lodash.default.isString(info.author)) {
-        info.author = [info.author];
+      const author = _lodash.default.get(meta, ["dc:creator"]);
+      if (_lodash.default.isArray(author)) {
+        info.author = author.map(a => a["#text"]);
       } else {
-        info.author = [_lodash.default.get(info.author, ["_"])];
+        info.author = [author["#text"]];
       }
     } else if (item === "description") {
-      info.description = _lodash.default.get(meta, ["description", 0, "_"]);
+      info.description = _lodash.default.get(meta, [item, "_"]);
     } else {
-      info[item] = _lodash.default.get(meta, ["dc:" + item, 0]);
+      info[item] = _lodash.default.get(meta, ["dc:" + item]);
     }
   });
   return _lodash.default.pickBy(info, v => {
@@ -83,14 +83,15 @@ class Epub {
       throw new Error(`${_path} not found!`);
     }
   }
-  async _resolveXMLAsJsObject(path) {
+  _resolveXMLAsJsObject(path) {
     const xml = this.resolve(path).asText();
-    return (0, _utils.xmlToJs)(xml);
+    return (0, _utils.xmlToJson)(xml);
   }
-  async _getOpfPath() {
-    const container = await this._resolveXMLAsJsObject("/META-INF/container.xml");
-    const opfPath = container.container.rootfiles[0].rootfile[0]["$"]["full-path"];
-    return opfPath;
+  /**
+   * 获取EPUB文件中的OPF（Open Packaging Format）文件的路径。
+   */
+  _getOpfPath() {
+    return this._resolveXMLAsJsObject("/META-INF/container.xml").container.rootfiles.rootfile["@full-path"];
   }
   _resolveIdFromLink(href) {
     const {
@@ -105,13 +106,16 @@ class Epub {
     return _lodash.default.get(tarItem, "id");
   }
   getManifest(content) {
-    return this._manifest || _lodash.default.get(content, ["package", "manifest", 0, "item"], []).map(item => item.$);
+    return this._manifest || _lodash.default.get(content, ["package", "manifest", "item"], []).map(item => ({
+      href: item["@href"],
+      id: item["@id"]
+    }));
   }
   getSpine() {
     const spine = {};
     this.getManifest();
-    _lodash.default.get(this._content, ["package", "spine", 0, "itemref"], []).map((item, i) => {
-      return spine[item.$.idref] = i;
+    _lodash.default.get(this._content, ["package", "spine", "itemref"], []).map((item, i) => {
+      return spine[item["@idref"]] = i;
     });
     return spine;
   }
@@ -155,11 +159,11 @@ class Epub {
     if (tocObj.html) {
       return this._genStructureForHTML(tocObj);
     }
-    const rootNavPoints = _lodash.default.get(tocObj, ["ncx", "navMap", "0", "navPoint"], []);
+    const rootNavPoints = _lodash.default.get(tocObj, ["ncx", "navMap", "navPoint"], []);
     const parseNavPoint = navPoint => {
-      const path = _lodash.default.get(navPoint, ["content", "0", "$", "src"], "");
-      const name = _lodash.default.get(navPoint, ["navLabel", "0", "text", "0"]);
-      const playOrder = _lodash.default.get(navPoint, ["$", "playOrder"]);
+      const path = _lodash.default.get(navPoint, ["content", "@src"], "");
+      const name = _lodash.default.get(navPoint, ["navLabel", "text"]);
+      const playOrder = _lodash.default.get(navPoint, ["@playOrder"]);
       const {
         hash
       } = (0, _parseLink.default)(path);
@@ -171,14 +175,14 @@ class Epub {
       return {
         name,
         sectionId,
-        nodeId: hash || _lodash.default.get(navPoint, ["$", "id"]),
+        nodeId: hash || navPoint["@id"],
         path,
         playOrder,
         children
       };
     };
     const parseNavPoints = navPoints => {
-      return navPoints.map(point => {
+      return (_lodash.default.isArray(navPoints) ? navPoints : [navPoints]).map(point => {
         return parseNavPoint(point);
       });
     };
@@ -216,16 +220,16 @@ class Epub {
     return this.sections ? sectionIndex != -1 ? this.sections[sectionIndex] : null : null;
   }
   async parse() {
-    this._opfPath = await this._getOpfPath();
-    this._content = await this._resolveXMLAsJsObject("/" + this._opfPath);
+    this._opfPath = this._getOpfPath();
+    this._content = this._resolveXMLAsJsObject("/" + this._opfPath);
     this._root = (0, _utils.determineRoot)(this._opfPath);
     this._manifest = this.getManifest(this._content);
-    this._metadata = _lodash.default.get(this._content, ["package", "metadata"], []);
+    this._metadata = _lodash.default.get(this._content, ["package", "metadata"], {});
     this.tocFile = (_lodash.default.find(this._manifest, {
       id: "ncx"
     }) || {}).href;
     if (this.tocFile) {
-      const toc = await this._resolveXMLAsJsObject(this.tocFile);
+      const toc = this._resolveXMLAsJsObject(this.tocFile);
       this._toc = toc;
       this.structure = this._genStructure(toc);
     }

package/lib/parseEpub.d.ts

@@ -45,7 +45,10 @@ export declare class Epub {
         asText: () => string;
         asNodeBuffer: () => Buffer;
     };
-    _resolveXMLAsJsObject(path: string): Promise<GeneralObject>;
+    _resolveXMLAsJsObject(path: string): GeneralObject;
+    /**
+     * 获取EPUB文件中的OPF（Open Packaging Format）文件的路径。
+     */
     private _getOpfPath;
     _resolveIdFromLink(href: string): string;
     getManifest(content?: GeneralObject): Manifest[];

package/lib/parseEpub.mjs

@@ -3,22 +3,22 @@ import _ from "lodash";
 import nodeZip from "node-zip";
 import parseLink from "./parseLink.mjs";
 import parseSection from "./parseSection.mjs";
-import { xmlToJs, determineRoot } from "./utils.mjs";
-const parseMetadata = (metadata) => {
-  const meta = metadata[0];
+import { xmlToJson, determineRoot } from "./utils.mjs";
+const parseMetadata = (metadata = {}) => {
+  const meta = metadata;
   const info = {};
   ["title", "author", "description", "language", "publisher", "rights"].forEach((item) => {
     if (item === "author") {
-      info.author = _.get(meta, ["dc:creator", 0]);
-      if (_.isString(info.author)) {
-        info.author = [info.author];
+      const author = _.get(meta, ["dc:creator"]);
+      if (_.isArray(author)) {
+        info.author = author.map((a) => a["#text"]);
       } else {
-        info.author = [_.get(info.author, ["_"])];
+        info.author = [author["#text"]];
       }
     } else if (item === "description") {
-      info.description = _.get(meta, ["description", 0, "_"]);
+      info.description = _.get(meta, [item, "_"]);
     } else {
-      info[item] = _.get(meta, ["dc:" + item, 0]);
+      info[item] = _.get(meta, ["dc:" + item]);
     }
   });
   return _.pickBy(info, (v) => {
@@ -64,14 +64,15 @@ export class Epub {
       throw new Error(`${_path} not found!`);
     }
   }
-  async _resolveXMLAsJsObject(path) {
+  _resolveXMLAsJsObject(path) {
     const xml = this.resolve(path).asText();
-    return xmlToJs(xml);
+    return xmlToJson(xml);
   }
-  async _getOpfPath() {
-    const container = await this._resolveXMLAsJsObject("/META-INF/container.xml");
-    const opfPath = container.container.rootfiles[0].rootfile[0]["$"]["full-path"];
-    return opfPath;
+  /**
+   * 获取EPUB文件中的OPF（Open Packaging Format）文件的路径。
+   */
+  _getOpfPath() {
+    return this._resolveXMLAsJsObject("/META-INF/container.xml").container.rootfiles.rootfile["@full-path"];
   }
   _resolveIdFromLink(href) {
     const { name: tarName } = parseLink(href);
@@ -82,14 +83,19 @@ export class Epub {
     return _.get(tarItem, "id");
   }
   getManifest(content) {
-    return this._manifest || _.get(content, ["package", "manifest", 0, "item"], []).map((item) => item.$);
+    return this._manifest || _.get(content, ["package", "manifest", "item"], []).map(
+      (item) => ({
+        href: item["@href"],
+        id: item["@id"]
+      })
+    );
   }
   getSpine() {
     const spine = {};
     this.getManifest();
-    _.get(this._content, ["package", "spine", 0, "itemref"], []).map(
+    _.get(this._content, ["package", "spine", "itemref"], []).map(
       (item, i) => {
-        return spine[item.$.idref] = i;
+        return spine[item["@idref"]] = i;
       }
     );
     return spine;
@@ -132,11 +138,11 @@ export class Epub {
     if (tocObj.html) {
       return this._genStructureForHTML(tocObj);
     }
-    const rootNavPoints = _.get(tocObj, ["ncx", "navMap", "0", "navPoint"], []);
+    const rootNavPoints = _.get(tocObj, ["ncx", "navMap", "navPoint"], []);
     const parseNavPoint = (navPoint) => {
-      const path = _.get(navPoint, ["content", "0", "$", "src"], "");
-      const name = _.get(navPoint, ["navLabel", "0", "text", "0"]);
-      const playOrder = _.get(navPoint, ["$", "playOrder"]);
+      const path = _.get(navPoint, ["content", "@src"], "");
+      const name = _.get(navPoint, ["navLabel", "text"]);
+      const playOrder = _.get(navPoint, ["@playOrder"]);
       const { hash } = parseLink(path);
       let children = navPoint.navPoint;
       if (children) {
@@ -146,14 +152,14 @@ export class Epub {
       return {
         name,
         sectionId,
-        nodeId: hash || _.get(navPoint, ["$", "id"]),
+        nodeId: hash || navPoint["@id"],
         path,
         playOrder,
         children
       };
     };
     const parseNavPoints = (navPoints) => {
-      return navPoints.map((point) => {
+      return (_.isArray(navPoints) ? navPoints : [navPoints]).map((point) => {
         return parseNavPoint(point);
       });
     };
@@ -189,14 +195,14 @@ export class Epub {
     return this.sections ? sectionIndex != -1 ? this.sections[sectionIndex] : null : null;
   }
   async parse() {
-    this._opfPath = await this._getOpfPath();
-    this._content = await this._resolveXMLAsJsObject("/" + this._opfPath);
+    this._opfPath = this._getOpfPath();
+    this._content = this._resolveXMLAsJsObject("/" + this._opfPath);
     this._root = determineRoot(this._opfPath);
     this._manifest = this.getManifest(this._content);
-    this._metadata = _.get(this._content, ["package", "metadata"], []);
+    this._metadata = _.get(this._content, ["package", "metadata"], {});
     this.tocFile = (_.find(this._manifest, { id: "ncx" }) || {}).href;
     if (this.tocFile) {
-      const toc = await this._resolveXMLAsJsObject(this.tocFile);
+      const toc = this._resolveXMLAsJsObject(this.tocFile);
       this._toc = toc;
       this.structure = this._genStructure(toc);
     }

package/lib/utils.cjs

@@ -5,11 +5,15 @@ Object.defineProperty(exports, "__esModule", {
 });
 exports.determineRoot = void 0;
 exports.findRealPath = findRealPath;
-exports.xmlToJs = exports.traverseNestedObject = exports.sanitizeFileName = void 0;
+exports.xmlToJson = exports.traverseNestedObject = exports.sanitizeFileName = void 0;
 var _lodash = _interopRequireDefault(require("lodash"));
-var _xml2js = _interopRequireDefault(require("xml2js"));
+var _fastXmlParser = require("fast-xml-parser");
 function _interopRequireDefault(e) { return e && e.__esModule ? e : { default: e }; }
-const xmlParser = new _xml2js.default.Parser();
+const xmlParser = new _fastXmlParser.XMLParser({
+  attributeNamePrefix: "@",
+  ignoreAttributes: false,
+  parseAttributeValue: true
+});
 const cacheNavPool = {};
 function findRealPath(filePath, navs) {
   if (!navs) return;
@@ -37,18 +41,10 @@ const sanitizeFileName = (fileName, replacementChar = "_") => {
   return fileName.replace(invalidCharsPattern, replacementChar);
 };
 exports.sanitizeFileName = sanitizeFileName;
-const xmlToJs = xml => {
-  return new Promise((resolve, reject) => {
-    xmlParser.parseString(xml, (err, object) => {
-      if (err) {
-        reject(err);
-      } else {
-        resolve(object);
-      }
-    });
-  });
+const xmlToJson = xml => {
+  return xmlParser.parse(xml);
 };
-exports.xmlToJs = xmlToJs;
+exports.xmlToJson = xmlToJson;
 const determineRoot = opfPath => {
   let root = "";
   if (opfPath.match(/\//)) {

package/lib/utils.d.ts

@@ -12,7 +12,7 @@ export interface TraverseNestedObject {
  */
 export declare function findRealPath(filePath: string, navs?: TOCItem[]): TOCItem | undefined;
 export declare const sanitizeFileName: (fileName: string, replacementChar?: string) => string;
-export declare const xmlToJs: (xml: string) => Promise<any>;
+export declare const xmlToJson: (xml: string) => any;
 export declare const determineRoot: (opfPath: string) => string;
 /**
  * traverseNestedObject

package/lib/utils.mjs

@@ -1,6 +1,10 @@
 import _ from "lodash";
-import xml2js from "xml2js";
-const xmlParser = new xml2js.Parser();
+import { XMLParser } from "fast-xml-parser";
+const xmlParser = new XMLParser({
+  attributeNamePrefix: "@",
+  ignoreAttributes: false,
+  parseAttributeValue: true
+});
 const cacheNavPool = {};
 export function findRealPath(filePath, navs) {
   if (!navs) return;
@@ -24,16 +28,8 @@ export const sanitizeFileName = (fileName, replacementChar = "_") => {
   const invalidCharsPattern = /[\\/:*?"<>|]/g;
   return fileName.replace(invalidCharsPattern, replacementChar);
 };
-export const xmlToJs = (xml) => {
-  return new Promise((resolve, reject) => {
-    xmlParser.parseString(xml, (err, object) => {
-      if (err) {
-        reject(err);
-      } else {
-        resolve(object);
-      }
-    });
-  });
+export const xmlToJson = (xml) => {
+  return xmlParser.parse(xml);
 };
 export const determineRoot = (opfPath) => {
   let root = "";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "epub2md",
-  "version": "1.2.1",
+  "version": "1.3.0",
   "description": "A epub parser that also can convert epub to markdown using the command line",
   "type": "module",
   "main": "lib/index.cjs",
@@ -19,18 +19,6 @@
     }
   },
   "types": "./lib/index.d.ts",
-  "scripts": {
-    "prebuild": "pnpm clean",
-    "build": "unbuild && pnpm run build:deno",
-    "build:deno": "node ./deno-build.mjs",
-    "watch": "tsc --watch",
-    "clean": "rimraf lib dist",
-    "format": "prettier --write \"src/**/*.{js,jsx,ts,tsx,json,md,css,scss}\"",
-    "test": "jest",
-    "test-debug": "node --inspect-brk -r ts-node/register node_modules/.bin/jest --runInBand",
-    "v": "pnpm dlx commit-and-tag-version --preset angular --release-as",
-    "postv": "git push --follow-tags origin main"
-  },
   "repository": {
     "type": "git",
     "url": "git+https://github.com/uxiew/epub2MD.git"
@@ -55,20 +43,20 @@
     "args": "^5.0.3",
     "autocorrect-node": "^2.6.2",
     "beauty-json": "^1.0.0",
+    "fast-xml-parser": "^5.2.0",
     "jsdom": "^24.0.0",
     "lodash": "^4.17.15",
     "node-html-markdown": "^1.3.0",
     "node-zip": "^1.1.1",
-    "write-file-safe": "^1.3.1",
-    "xml2js": "^0.5.0"
+    "write-file-safe": "^1.3.1"
   },
   "devDependencies": {
+    "@jsdevtools/npm-publish": "^3.1.1",
     "@types/args": "^5.0.0",
     "@types/jest": "^24.0.18",
     "@types/jsdom": "^21.1.1",
     "@types/lodash": "^4.14.137",
     "@types/node": "^20.2.3",
-    "@types/xml2js": "^0.4.4",
     "chalk": "^4.1.2",
     "cross-env": "^5.2.0",
     "dotenv": "^8.1.0",
@@ -83,5 +71,17 @@
     "typescript": "^5.5.4",
     "unbuild": "3.0.0-rc.7",
     "vrsource-tslint-rules": "^6.0.0"
+  },
+  "scripts": {
+    "prebuild": "pnpm clean",
+    "build": "unbuild && pnpm run build:deno",
+    "build:deno": "node ./deno-build.mjs",
+    "watch": "tsc --watch",
+    "clean": "rimraf lib dist",
+    "format": "prettier --write \"src/**/*.{js,jsx,ts,tsx,json,md,css,scss}\"",
+    "test": "jest",
+    "test-debug": "node --inspect-brk -r ts-node/register node_modules/.bin/jest --runInBand",
+    "v": "pnpm dlx commit-and-tag-version --preset angular --release-as",
+    "postv": "git push --follow-tags origin main"
   }
-}
+}