Add EPUB import for data bank

public/scripts/chats.js

@@ -27,6 +27,7 @@ import {
     extractTextFromHTML,
     extractTextFromMarkdown,
     extractTextFromPDF,
+    extractTextFromEpub,
     getBase64Async,
     getStringHash,
     humanFileSize,
@@ -56,6 +57,7 @@ const converters = {
     'application/pdf': extractTextFromPDF,
     'text/html': extractTextFromHTML,
     'text/markdown': extractTextFromMarkdown,
+    'application/epub+zip': extractTextFromEpub,
 };
 
 /**

public/scripts/extensions/attachments/manager.html

@@ -7,8 +7,8 @@
     <div data-i18n="These files will be available for extensions that support attachments (e.g. Vector Storage).">
         These files will be available for extensions that support attachments (e.g. Vector Storage).
     </div>
-    <div data-i18n="Supported file types: Plain Text, PDF, Markdown, HTML." class="marginTopBot5">
-        Supported file types: Plain Text, PDF, Markdown, HTML.
+    <div data-i18n="Supported file types: Plain Text, PDF, Markdown, HTML, EPUB." class="marginTopBot5">
+        Supported file types: Plain Text, PDF, Markdown, HTML, EPUB.
     </div>
     <div class="flex-container marginTopBot5">
         <input type="search" id="attachmentSearch" class="attachmentSearch text_pole margin0 flex1" placeholder="Search...">

public/scripts/scrapers.js

@@ -179,7 +179,7 @@ class FileScraper {
         return new Promise(resolve => {
             const fileInput = document.createElement('input');
             fileInput.type = 'file';
-            fileInput.accept = '.txt, .md, .pdf, .html, .htm';
+            fileInput.accept = '.txt, .md, .pdf, .html, .htm, .epub';
             fileInput.multiple = true;
             fileInput.onchange = () => resolve(Array.from(fileInput.files));
             fileInput.click();

public/scripts/utils.js

@@ -695,7 +695,7 @@ export function splitRecursive(input, length, delimiters = ['\n\n', '\n', ' ', '
 
     const flatParts = parts.flatMap(p => {
         if (p.length < length) return p;
-        return splitRecursive(input, length, delimiters.slice(1));
+        return splitRecursive(p, length, delimiters.slice(1));
     });
 
     // Merge short chunks
@@ -1300,6 +1300,54 @@ export async function extractTextFromMarkdown(blob) {
     return text;
 }
 
+export async function extractTextFromEpub(blob) {
+    async function initEpubJs() {
+        const epubScript = new Promise((resolve, reject) => {
+            const epubScript = document.createElement('script');
+            epubScript.async = true;
+            epubScript.src = 'lib/epub.min.js';
+            epubScript.onload = resolve;
+            epubScript.onerror = reject;
+            document.head.appendChild(epubScript);
+        });
+
+        const jszipScript = new Promise((resolve, reject) => {
+            const jszipScript = document.createElement('script');
+            jszipScript.async = true;
+            jszipScript.src = 'lib/jszip.min.js';
+            jszipScript.onload = resolve;
+            jszipScript.onerror = reject;
+            document.head.appendChild(jszipScript);
+        });
+
+        return Promise.all([epubScript, jszipScript]);
+    }
+
+    if (!('ePub' in window)) {
+        await initEpubJs();
+    }
+
+    const book = ePub(blob);
+    await book.ready;
+    const sectionPromises = [];
+
+    book.spine.each((section) => {
+        const sectionPromise = (async () => {
+            const chapter = await book.load(section.href);
+            if (!(chapter instanceof Document) || !chapter.body?.textContent) {
+                return '';
+            }
+            return chapter.body.textContent.trim();
+        })();
+
+        sectionPromises.push(sectionPromise);
+    });
+
+    const content = await Promise.all(sectionPromises);
+    const text = content.filter(text => text);
+    return postProcessText(text.join('\n'), false);
+}
+
 /**
  * Sets a value in an object by a path.
  * @param {object} obj Object to set value in