markitdown-ts 0.0.3 → 0.0.4

package/README.md

@@ -7,7 +7,6 @@
 It supports:
 
 - [x] PDF
-- [x] PowerPoint
 - [x] Word (.docx)
 - [x] Excel (.xlsx)
 - [x] Images (EXIF metadata extraction and optional LLM-based description)
@@ -17,6 +16,7 @@ It supports:
 - [x] Jupyter Notebooks (.ipynb)
 - [x] Bing Search Result Pages (SERP)
 - [x] ZIP files (recursively iterates over contents)
+- [ ] PowerPoint
 
 > [!NOTE]
 >

package/dist/index.cjs

@@ -16,7 +16,6 @@ const util = require('util');
 const fs$1 = require('fs/promises');
 const os = require('os');
 const ai = require('ai');
-const unzipper = require('unzipper');
 
 function _interopDefaultCompat (e) { return e && typeof e === 'object' && 'default' in e ? e.default : e; }
 
@@ -40,12 +39,11 @@ const fs__namespace = /*#__PURE__*/_interopNamespaceCompat(fs);
 const TurndownService__default = /*#__PURE__*/_interopDefaultCompat(TurndownService);
 const turndownPluginGfm__default = /*#__PURE__*/_interopDefaultCompat(turndownPluginGfm);
 const Mammoth__default = /*#__PURE__*/_interopDefaultCompat(Mammoth);
-const XLSX__default = /*#__PURE__*/_interopDefaultCompat(XLSX);
+const XLSX__namespace = /*#__PURE__*/_interopNamespaceCompat(XLSX);
 const childProcess__namespace = /*#__PURE__*/_interopNamespaceCompat(childProcess);
 const util__namespace = /*#__PURE__*/_interopNamespaceCompat(util);
 const fs__namespace$1 = /*#__PURE__*/_interopNamespaceCompat(fs$1);
 const os__namespace = /*#__PURE__*/_interopNamespaceCompat(os);
-const unzipper__namespace = /*#__PURE__*/_interopNamespaceCompat(unzipper);
 
 class PlainTextConverter {
   async convert(local_path, options = {}) {
@@ -707,12 +705,12 @@ class XlsxConverter extends HtmlConverter {
       if (!exists) {
         throw new Error("File does'nt exists");
       }
-      let workbook = XLSX__default.readFile(local_path);
+      let workbook = XLSX__namespace.readFile(local_path);
       let mdContent = "";
       for (const sheetName of workbook.SheetNames) {
         mdContent += `## ${sheetName}
 `;
-        let htmlContent = XLSX__default.utils.sheet_to_html(workbook.Sheets[sheetName]);
+        let htmlContent = XLSX__namespace.utils.sheet_to_html(workbook.Sheets[sheetName]);
         mdContent += (await this._convert(htmlContent))?.text_content.trim() + "\n\n";
       }
       return {
@@ -959,9 +957,18 @@ class ZipConverter {
         text_content: `[ERROR] Invalid zip file path: ${localPath}`
       };
     }
+    let unzipper;
+    try {
+      unzipper = await import('unzipper').then((mod) => mod.default);
+    } catch (error) {
+      console.error(
+        "Optional dependency 'unzipper' is not installed. Run `npm install unzipper` to enable this feature."
+      );
+      return null;
+    }
     try {
       await fs__namespace$1.mkdir(newFolder, { recursive: true });
-      const zip = await unzipper__namespace.Open.file(localPath);
+      const zip = await unzipper.Open.file(localPath);
       await zip.extract({ path: newFolder });
       const files = await this._walk(newFolder);
       for (const { root, name } of files) {

package/dist/index.mjs

@@ -10,13 +10,12 @@ import { DOMParser } from '@xmldom/xmldom';
 import { URL as URL$1 } from 'url';
 import { pdfToText } from 'pdf-ts';
 import Mammoth from 'mammoth';
-import XLSX from 'xlsx';
+import * as XLSX from 'xlsx';
 import * as childProcess from 'child_process';
 import * as util from 'util';
 import * as fs$1 from 'fs/promises';
 import * as os from 'os';
 import { generateText } from 'ai';
-import * as unzipper from 'unzipper';
 
 class PlainTextConverter {
   async convert(local_path, options = {}) {
@@ -930,6 +929,15 @@ class ZipConverter {
         text_content: `[ERROR] Invalid zip file path: ${localPath}`
       };
     }
+    let unzipper;
+    try {
+      unzipper = await import('unzipper').then((mod) => mod.default);
+    } catch (error) {
+      console.error(
+        "Optional dependency 'unzipper' is not installed. Run `npm install unzipper` to enable this feature."
+      );
+      return null;
+    }
     try {
       await fs$1.mkdir(newFolder, { recursive: true });
       const zip = await unzipper.Open.file(localPath);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "markitdown-ts",
-  "version": "0.0.3",
+  "version": "0.0.4",
   "description": "",
   "keywords": [],
   "homepage": "https://github.com/dead8309/markitdown-ts#readme",
@@ -50,10 +50,10 @@
     "mime-types": "^2.1.35",
     "pdf-ts": "^0.0.2",
     "turndown": "^7.2.0",
-    "xlsx": "^0.18.5"
+    "xlsx": "^0.18.5",
+    "ai": "^4.0.22"
   },
   "peerDependencies": {
-    "ai": "^4.0.22",
     "youtube-transcript": "^1.2.1",
     "unzipper": "^0.12.3"
   },
@@ -61,9 +61,6 @@
     "youtube-transcript": {
       "optional": true
     },
-    "ai": {
-      "optional": true
-    },
     "unzipper": {
       "optional": true
     }