@mdaemon/html-editor 1.1.0 → 1.2.0

package/README.md

@@ -103,6 +103,7 @@ const editor = new HTMLEditor(container, {
 | `entity_encoding` | 'raw' \| 'named' \| 'numeric' | 'raw' | HTML entity encoding mode |
 | `convert_unsafe_embeds` | boolean | true | Sanitize embedded content |
 | `format_empty_lines` | boolean | true | Format empty lines |
+| `paste_from_office` | boolean | true | Clean and preserve formatting when pasting from Microsoft Word and Excel |
 | `setup` | (editor) => void | - | Callback invoked before init — use to register custom buttons |
 
 ### Toolbar Toggle

package/dist/index.d.ts

@@ -138,6 +138,7 @@ export declare interface EditorConfig {
     quickbars_selection_toolbar?: string;
     quickbars_image_toolbar?: boolean;
     quickbars_insert_toolbar?: boolean;
+    paste_from_office?: boolean;
     browser_spellcheck?: boolean;
     elementpath?: boolean;
     entity_encoding?: 'raw' | 'named' | 'numeric';
@@ -373,6 +374,16 @@ export declare interface MDHTMLEditor {
 
 export declare const Mention: Node_2<any, any>;
 
+export declare const PasteFromOffice: Extension<PasteFromOfficeOptions, any>;
+
+export declare interface PasteFromOfficeOptions {
+    /**
+     * Enable/disable paste-from-office cleaning.
+     * When false, pasted content passes through unchanged.
+     */
+    enabled: boolean;
+}
+
 /**
  * Reset the global translation function to the default identity function.
  * This also clears the customized flag so that the next editor created

package/dist/index.js

@@ -44033,6 +44033,384 @@ const Mention = Node3.create({
     ];
   }
 });
+const WORD_MARKERS = [
+  'xmlns:o="urn:schemas-microsoft-com:office:office"',
+  'xmlns:w="urn:schemas-microsoft-com:office:word"',
+  'class="Mso',
+  "class=Mso",
+  'style="mso-',
+  "style='mso-"
+];
+const EXCEL_MARKERS = [
+  'xmlns:x="urn:schemas-microsoft-com:office:excel"',
+  "ProgId content=Excel.Sheet",
+  'ProgId content="Excel.Sheet"'
+];
+function isWordContent(html) {
+  return WORD_MARKERS.some((marker) => html.includes(marker));
+}
+function isExcelContent(html) {
+  return EXCEL_MARKERS.some((marker) => html.includes(marker));
+}
+function isOfficeContent(html) {
+  return isWordContent(html) || isExcelContent(html);
+}
+function removeConditionalComments(html) {
+  return html.replace(/<!--\[if[^\]]*\]>[\s\S]*?<!\[endif\]-->/gi, "");
+}
+const OFFICE_NS_TAG_PATTERN = /(?:<\/?)(?:o|w|v|m|x|st\d):/i;
+function removeOfficeElements(doc2) {
+  const allElements = Array.from(doc2.body.querySelectorAll("*"));
+  for (const el2 of allElements) {
+    const tagName = el2.tagName.toLowerCase();
+    if (OFFICE_NS_TAG_PATTERN.test(`<${tagName}`)) {
+      const parent = el2.parentNode;
+      if (parent) {
+        while (el2.firstChild) {
+          parent.insertBefore(el2.firstChild, el2);
+        }
+        parent.removeChild(el2);
+      }
+    }
+  }
+  const remaining = Array.from(doc2.body.querySelectorAll("*"));
+  for (const el2 of remaining) {
+    const attrs = Array.from(el2.attributes);
+    for (const attr of attrs) {
+      if (attr.name.startsWith("xmlns:") || attr.name === "xmlns") {
+        el2.removeAttribute(attr.name);
+      }
+    }
+  }
+}
+function parseListStyles(doc2) {
+  const styleMap = /* @__PURE__ */ new Map();
+  const styleEls = doc2.querySelectorAll("style");
+  for (const styleEl of Array.from(styleEls)) {
+    const css2 = styleEl.textContent ?? "";
+    const listRulePattern = /@list\s+(l\d+):level(\d+)\s*\{([^}]*)\}/gi;
+    let match;
+    while ((match = listRulePattern.exec(css2)) !== null) {
+      const listId = match[1];
+      const level = match[2];
+      const body = match[3];
+      const key = `${listId}:level${level}`;
+      const formatMatch = /mso-level-number-format:\s*([^;]+)/i.exec(body);
+      if (formatMatch) {
+        const format = formatMatch[1].trim().toLowerCase();
+        styleMap.set(key, format !== "bullet");
+      } else {
+        styleMap.set(key, true);
+      }
+    }
+  }
+  return styleMap;
+}
+function parseListStyle(style2) {
+  const match = /mso-list:\s*(l\d+)\s+level(\d+)\s+/i.exec(style2);
+  if (!match) return null;
+  return {
+    listId: match[1],
+    level: parseInt(match[2], 10),
+    isOrdered: false
+    // Will be resolved from style rules
+  };
+}
+function convertWordLists(doc2) {
+  const listStyles = parseListStyles(doc2);
+  const listParagraphs = Array.from(doc2.body.querySelectorAll('[class*="MsoList"]'));
+  if (listParagraphs.length === 0) return;
+  const groups = [];
+  let currentGroup = [];
+  for (const p of listParagraphs) {
+    const el2 = p;
+    const style2 = el2.getAttribute("style") ?? "";
+    if (!parseListStyle(style2)) continue;
+    if (currentGroup.length > 0) {
+      const lastEl = currentGroup[currentGroup.length - 1];
+      let nextSibling = lastEl.nextElementSibling;
+      while (nextSibling && nextSibling !== el2 && nextSibling.textContent?.trim() === "") {
+        nextSibling = nextSibling.nextElementSibling;
+      }
+      if (nextSibling !== el2) {
+        groups.push(currentGroup);
+        currentGroup = [];
+      }
+    }
+    currentGroup.push(el2);
+  }
+  if (currentGroup.length > 0) {
+    groups.push(currentGroup);
+  }
+  for (const group of groups) {
+    convertListGroup(doc2, group, listStyles);
+  }
+}
+function convertListGroup(doc2, paragraphs, listStyles) {
+  if (paragraphs.length === 0) return;
+  const rootInfo = parseListStyle(paragraphs[0].getAttribute("style") ?? "");
+  if (!rootInfo) return;
+  const parent = paragraphs[0].parentNode;
+  if (!parent) return;
+  const rootKey = `${rootInfo.listId}:level${rootInfo.level}`;
+  const isRootOrdered = listStyles.get(rootKey) ?? false;
+  const rootList = doc2.createElement(isRootOrdered ? "ol" : "ul");
+  const stack = [{ element: rootList, level: rootInfo.level }];
+  for (const p of paragraphs) {
+    const style2 = p.getAttribute("style") ?? "";
+    const info = parseListStyle(style2);
+    if (!info) continue;
+    const key = `${info.listId}:level${info.level}`;
+    const isOrdered = listStyles.get(key) ?? false;
+    while (stack.length > 1 && stack[stack.length - 1].level >= info.level) {
+      stack.pop();
+    }
+    if (info.level > stack[stack.length - 1].level) {
+      const currentList = stack[stack.length - 1].element;
+      let lastLi = currentList.lastElementChild;
+      if (!lastLi || lastLi.tagName.toLowerCase() !== "li") {
+        lastLi = doc2.createElement("li");
+        currentList.appendChild(lastLi);
+      }
+      const subList = doc2.createElement(isOrdered ? "ol" : "ul");
+      lastLi.appendChild(subList);
+      stack.push({ element: subList, level: info.level });
+    }
+    const li = doc2.createElement("li");
+    removeListMarkers(p);
+    while (p.firstChild) {
+      li.appendChild(p.firstChild);
+    }
+    const cleanedStyle = cleanMsoStyles(style2);
+    if (cleanedStyle) {
+      li.setAttribute("style", cleanedStyle);
+    }
+    stack[stack.length - 1].element.appendChild(li);
+  }
+  parent.insertBefore(rootList, paragraphs[0]);
+  for (const p of paragraphs) {
+    p.remove();
+  }
+}
+function removeListMarkers(el2) {
+  const spans = Array.from(el2.querySelectorAll("span"));
+  for (const span of spans) {
+    const style2 = span.getAttribute("style") ?? "";
+    if (/mso-list:\s*Ignore/i.test(style2)) {
+      span.remove();
+    }
+  }
+}
+const MSO_PROPERTY_PATTERN = /\bmso-[^:]+:[^;]+;?\s*/gi;
+const MSO_CLASS_PATTERN = /\bMso\w*|\bxl\d+/g;
+function cleanMsoStyles(style2) {
+  let cleaned = style2.replace(MSO_PROPERTY_PATTERN, "");
+  cleaned = cleaned.replace(/;\s*;/g, ";").replace(/^\s*;\s*/, "").replace(/\s*;\s*$/, "").trim();
+  return cleaned || "";
+}
+function cleanFontFamily(value) {
+  const first2 = value.split(",")[0].trim();
+  return first2.replace(/^["']|["']$/g, "");
+}
+function cleanElementStyles(doc2) {
+  const allElements = Array.from(doc2.body.querySelectorAll("*"));
+  for (const el2 of allElements) {
+    const className = el2.getAttribute("class");
+    if (className) {
+      const cleaned = className.replace(MSO_CLASS_PATTERN, "").trim();
+      if (cleaned) {
+        el2.setAttribute("class", cleaned);
+      } else {
+        el2.removeAttribute("class");
+      }
+    }
+    const style2 = el2.getAttribute("style");
+    if (style2) {
+      let cleaned = cleanMsoStyles(style2);
+      const fontFamilyMatch = /font-family:\s*([^;]+)/i.exec(cleaned);
+      if (fontFamilyMatch) {
+        const cleanedFont = cleanFontFamily(fontFamilyMatch[1]);
+        cleaned = cleaned.replace(fontFamilyMatch[0], `font-family: ${cleanedFont}`);
+      }
+      if (cleaned) {
+        el2.setAttribute("style", cleaned);
+      } else {
+        el2.removeAttribute("style");
+      }
+    }
+    const attrsToRemove = Array.from(el2.attributes).filter(
+      (attr) => attr.name.startsWith("v:") || attr.name.startsWith("o:") || attr.name === "lang"
+    );
+    for (const attr of attrsToRemove) {
+      el2.removeAttribute(attr.name);
+    }
+  }
+}
+function removeEmptyWrappers(doc2) {
+  let changed = true;
+  while (changed) {
+    changed = false;
+    const spans = Array.from(doc2.body.querySelectorAll("span"));
+    for (const span of spans) {
+      if (span.attributes.length === 0) {
+        const parent = span.parentNode;
+        if (parent) {
+          while (span.firstChild) {
+            parent.insertBefore(span.firstChild, span);
+          }
+          parent.removeChild(span);
+          changed = true;
+        }
+      }
+    }
+  }
+  const emptyParas = Array.from(doc2.body.querySelectorAll("p"));
+  for (const p of emptyParas) {
+    if (p.innerHTML.trim() === "" && doc2.body.children.length > 1) {
+      p.remove();
+    }
+  }
+}
+function normalizeExcelTables(doc2) {
+  const tables = Array.from(doc2.body.querySelectorAll("table"));
+  for (const table of tables) {
+    removeExcelAttributes(table);
+    const cells = Array.from(table.querySelectorAll("td, th"));
+    for (const cell of cells) {
+      removeExcelAttributes(cell);
+      const style2 = cell.getAttribute("style");
+      if (style2) {
+        const cleaned = cleanMsoStyles(style2);
+        if (cleaned) {
+          cell.setAttribute("style", cleaned);
+        } else {
+          cell.removeAttribute("style");
+        }
+      }
+    }
+    const rows = Array.from(table.querySelectorAll("tr"));
+    for (const row of rows) {
+      removeExcelAttributes(row);
+    }
+  }
+}
+const EXCEL_ATTRS = [
+  "x:num",
+  "x:str",
+  "x:fmla",
+  "x:autofilter",
+  "mso-number-format"
+];
+function removeExcelAttributes(el2) {
+  for (const attr of EXCEL_ATTRS) {
+    el2.removeAttribute(attr);
+  }
+  const className = el2.getAttribute("class");
+  if (className) {
+    const cleaned = className.replace(MSO_CLASS_PATTERN, "").trim();
+    if (cleaned) {
+      el2.setAttribute("class", cleaned);
+    } else {
+      el2.removeAttribute("class");
+    }
+  }
+}
+const DANGEROUS_TAGS = [
+  "script",
+  "iframe",
+  "object",
+  "embed",
+  "applet",
+  "form",
+  "input",
+  "textarea",
+  "select",
+  "button"
+];
+const EVENT_ATTR_PATTERN = /^on/i;
+const DANGEROUS_URL_PATTERN = /^\s*(javascript|vbscript|data\s*:(?!image))/i;
+function sanitize(doc2) {
+  for (const tag of DANGEROUS_TAGS) {
+    const elements = Array.from(doc2.body.querySelectorAll(tag));
+    for (const el2 of elements) {
+      el2.remove();
+    }
+  }
+  const allElements = Array.from(doc2.body.querySelectorAll("*"));
+  for (const el2 of allElements) {
+    const attrs = Array.from(el2.attributes);
+    for (const attr of attrs) {
+      if (EVENT_ATTR_PATTERN.test(attr.name)) {
+        el2.removeAttribute(attr.name);
+        continue;
+      }
+      if ((attr.name === "href" || attr.name === "src" || attr.name === "action") && DANGEROUS_URL_PATTERN.test(attr.value)) {
+        el2.removeAttribute(attr.name);
+      }
+    }
+    const src = el2.getAttribute("src");
+    if (src && src.startsWith("file://")) {
+      el2.removeAttribute("src");
+    }
+    const href = el2.getAttribute("href");
+    if (href && href.startsWith("file://")) {
+      el2.removeAttribute("href");
+    }
+  }
+}
+function removeStyleBlocks(doc2) {
+  const styles = Array.from(doc2.querySelectorAll("style"));
+  for (const s of styles) {
+    s.remove();
+  }
+}
+function removeHeadElements(doc2) {
+  const selectors = ["meta", "link", "title", "xml"];
+  for (const selector of selectors) {
+    const elements = Array.from(doc2.body.querySelectorAll(selector));
+    for (const el2 of elements) {
+      el2.remove();
+    }
+  }
+}
+function transformOfficeHTML(html) {
+  if (!isOfficeContent(html)) {
+    return html;
+  }
+  let cleaned = removeConditionalComments(html);
+  const parser = new DOMParser();
+  const doc2 = parser.parseFromString(cleaned, "text/html");
+  removeHeadElements(doc2);
+  if (isWordContent(html)) {
+    convertWordLists(doc2);
+  }
+  removeOfficeElements(doc2);
+  if (isExcelContent(html)) {
+    normalizeExcelTables(doc2);
+  }
+  cleanElementStyles(doc2);
+  removeStyleBlocks(doc2);
+  removeEmptyWrappers(doc2);
+  sanitize(doc2);
+  return doc2.body.innerHTML;
+}
+const PasteFromOffice = Extension.create({
+  name: "pasteFromOffice",
+  addOptions() {
+    return {
+      enabled: true
+    };
+  },
+  addStorage() {
+    return {};
+  },
+  transformPastedHTML(html) {
+    if (!this.options.enabled) {
+      return html;
+    }
+    return transformOfficeHTML(html);
+  }
+});
 const en = {
   "Bold": "Bold",
   "Italic": "Italic",
@@ -46403,7 +46781,8 @@ class HTMLEditor {
       convert_unsafe_embeds: config.convert_unsafe_embeds ?? true,
       format_empty_lines: config.format_empty_lines ?? true,
       toolbar_narrow_breakpoint: config.toolbar_narrow_breakpoint,
-      toolbar_priority: config.toolbar_priority
+      toolbar_priority: config.toolbar_priority,
+      paste_from_office: config.paste_from_office ?? true
     };
   }
   createEditor() {
@@ -46606,6 +46985,9 @@ class HTMLEditor {
       }),
       TextDirection
     ];
+    if (this.config.paste_from_office !== false) {
+      extensions.push(PasteFromOffice);
+    }
     if (!this.config.basicEditor) {
       extensions.push(
         Image.configure({
@@ -46830,6 +47212,7 @@ exports.HTMLEditor = HTMLEditor;
 exports.LineHeight = LineHeight;
 exports.LinkEditor = LinkEditor;
 exports.Mention = Mention;
+exports.PasteFromOffice = PasteFromOffice;
 exports.SearchReplace = SearchReplace;
 exports.SignatureBlock = SignatureBlock;
 exports.SourceEditor = SourceEditor;

package/dist/index.mjs

@@ -44031,6 +44031,384 @@ const Mention = Node3.create({
     ];
   }
 });
+const WORD_MARKERS = [
+  'xmlns:o="urn:schemas-microsoft-com:office:office"',
+  'xmlns:w="urn:schemas-microsoft-com:office:word"',
+  'class="Mso',
+  "class=Mso",
+  'style="mso-',
+  "style='mso-"
+];
+const EXCEL_MARKERS = [
+  'xmlns:x="urn:schemas-microsoft-com:office:excel"',
+  "ProgId content=Excel.Sheet",
+  'ProgId content="Excel.Sheet"'
+];
+function isWordContent(html) {
+  return WORD_MARKERS.some((marker) => html.includes(marker));
+}
+function isExcelContent(html) {
+  return EXCEL_MARKERS.some((marker) => html.includes(marker));
+}
+function isOfficeContent(html) {
+  return isWordContent(html) || isExcelContent(html);
+}
+function removeConditionalComments(html) {
+  return html.replace(/<!--\[if[^\]]*\]>[\s\S]*?<!\[endif\]-->/gi, "");
+}
+const OFFICE_NS_TAG_PATTERN = /(?:<\/?)(?:o|w|v|m|x|st\d):/i;
+function removeOfficeElements(doc2) {
+  const allElements = Array.from(doc2.body.querySelectorAll("*"));
+  for (const el2 of allElements) {
+    const tagName = el2.tagName.toLowerCase();
+    if (OFFICE_NS_TAG_PATTERN.test(`<${tagName}`)) {
+      const parent = el2.parentNode;
+      if (parent) {
+        while (el2.firstChild) {
+          parent.insertBefore(el2.firstChild, el2);
+        }
+        parent.removeChild(el2);
+      }
+    }
+  }
+  const remaining = Array.from(doc2.body.querySelectorAll("*"));
+  for (const el2 of remaining) {
+    const attrs = Array.from(el2.attributes);
+    for (const attr of attrs) {
+      if (attr.name.startsWith("xmlns:") || attr.name === "xmlns") {
+        el2.removeAttribute(attr.name);
+      }
+    }
+  }
+}
+function parseListStyles(doc2) {
+  const styleMap = /* @__PURE__ */ new Map();
+  const styleEls = doc2.querySelectorAll("style");
+  for (const styleEl of Array.from(styleEls)) {
+    const css2 = styleEl.textContent ?? "";
+    const listRulePattern = /@list\s+(l\d+):level(\d+)\s*\{([^}]*)\}/gi;
+    let match;
+    while ((match = listRulePattern.exec(css2)) !== null) {
+      const listId = match[1];
+      const level = match[2];
+      const body = match[3];
+      const key = `${listId}:level${level}`;
+      const formatMatch = /mso-level-number-format:\s*([^;]+)/i.exec(body);
+      if (formatMatch) {
+        const format = formatMatch[1].trim().toLowerCase();
+        styleMap.set(key, format !== "bullet");
+      } else {
+        styleMap.set(key, true);
+      }
+    }
+  }
+  return styleMap;
+}
+function parseListStyle(style2) {
+  const match = /mso-list:\s*(l\d+)\s+level(\d+)\s+/i.exec(style2);
+  if (!match) return null;
+  return {
+    listId: match[1],
+    level: parseInt(match[2], 10),
+    isOrdered: false
+    // Will be resolved from style rules
+  };
+}
+function convertWordLists(doc2) {
+  const listStyles = parseListStyles(doc2);
+  const listParagraphs = Array.from(doc2.body.querySelectorAll('[class*="MsoList"]'));
+  if (listParagraphs.length === 0) return;
+  const groups = [];
+  let currentGroup = [];
+  for (const p of listParagraphs) {
+    const el2 = p;
+    const style2 = el2.getAttribute("style") ?? "";
+    if (!parseListStyle(style2)) continue;
+    if (currentGroup.length > 0) {
+      const lastEl = currentGroup[currentGroup.length - 1];
+      let nextSibling = lastEl.nextElementSibling;
+      while (nextSibling && nextSibling !== el2 && nextSibling.textContent?.trim() === "") {
+        nextSibling = nextSibling.nextElementSibling;
+      }
+      if (nextSibling !== el2) {
+        groups.push(currentGroup);
+        currentGroup = [];
+      }
+    }
+    currentGroup.push(el2);
+  }
+  if (currentGroup.length > 0) {
+    groups.push(currentGroup);
+  }
+  for (const group of groups) {
+    convertListGroup(doc2, group, listStyles);
+  }
+}
+function convertListGroup(doc2, paragraphs, listStyles) {
+  if (paragraphs.length === 0) return;
+  const rootInfo = parseListStyle(paragraphs[0].getAttribute("style") ?? "");
+  if (!rootInfo) return;
+  const parent = paragraphs[0].parentNode;
+  if (!parent) return;
+  const rootKey = `${rootInfo.listId}:level${rootInfo.level}`;
+  const isRootOrdered = listStyles.get(rootKey) ?? false;
+  const rootList = doc2.createElement(isRootOrdered ? "ol" : "ul");
+  const stack = [{ element: rootList, level: rootInfo.level }];
+  for (const p of paragraphs) {
+    const style2 = p.getAttribute("style") ?? "";
+    const info = parseListStyle(style2);
+    if (!info) continue;
+    const key = `${info.listId}:level${info.level}`;
+    const isOrdered = listStyles.get(key) ?? false;
+    while (stack.length > 1 && stack[stack.length - 1].level >= info.level) {
+      stack.pop();
+    }
+    if (info.level > stack[stack.length - 1].level) {
+      const currentList = stack[stack.length - 1].element;
+      let lastLi = currentList.lastElementChild;
+      if (!lastLi || lastLi.tagName.toLowerCase() !== "li") {
+        lastLi = doc2.createElement("li");
+        currentList.appendChild(lastLi);
+      }
+      const subList = doc2.createElement(isOrdered ? "ol" : "ul");
+      lastLi.appendChild(subList);
+      stack.push({ element: subList, level: info.level });
+    }
+    const li = doc2.createElement("li");
+    removeListMarkers(p);
+    while (p.firstChild) {
+      li.appendChild(p.firstChild);
+    }
+    const cleanedStyle = cleanMsoStyles(style2);
+    if (cleanedStyle) {
+      li.setAttribute("style", cleanedStyle);
+    }
+    stack[stack.length - 1].element.appendChild(li);
+  }
+  parent.insertBefore(rootList, paragraphs[0]);
+  for (const p of paragraphs) {
+    p.remove();
+  }
+}
+function removeListMarkers(el2) {
+  const spans = Array.from(el2.querySelectorAll("span"));
+  for (const span of spans) {
+    const style2 = span.getAttribute("style") ?? "";
+    if (/mso-list:\s*Ignore/i.test(style2)) {
+      span.remove();
+    }
+  }
+}
+const MSO_PROPERTY_PATTERN = /\bmso-[^:]+:[^;]+;?\s*/gi;
+const MSO_CLASS_PATTERN = /\bMso\w*|\bxl\d+/g;
+function cleanMsoStyles(style2) {
+  let cleaned = style2.replace(MSO_PROPERTY_PATTERN, "");
+  cleaned = cleaned.replace(/;\s*;/g, ";").replace(/^\s*;\s*/, "").replace(/\s*;\s*$/, "").trim();
+  return cleaned || "";
+}
+function cleanFontFamily(value) {
+  const first2 = value.split(",")[0].trim();
+  return first2.replace(/^["']|["']$/g, "");
+}
+function cleanElementStyles(doc2) {
+  const allElements = Array.from(doc2.body.querySelectorAll("*"));
+  for (const el2 of allElements) {
+    const className = el2.getAttribute("class");
+    if (className) {
+      const cleaned = className.replace(MSO_CLASS_PATTERN, "").trim();
+      if (cleaned) {
+        el2.setAttribute("class", cleaned);
+      } else {
+        el2.removeAttribute("class");
+      }
+    }
+    const style2 = el2.getAttribute("style");
+    if (style2) {
+      let cleaned = cleanMsoStyles(style2);
+      const fontFamilyMatch = /font-family:\s*([^;]+)/i.exec(cleaned);
+      if (fontFamilyMatch) {
+        const cleanedFont = cleanFontFamily(fontFamilyMatch[1]);
+        cleaned = cleaned.replace(fontFamilyMatch[0], `font-family: ${cleanedFont}`);
+      }
+      if (cleaned) {
+        el2.setAttribute("style", cleaned);
+      } else {
+        el2.removeAttribute("style");
+      }
+    }
+    const attrsToRemove = Array.from(el2.attributes).filter(
+      (attr) => attr.name.startsWith("v:") || attr.name.startsWith("o:") || attr.name === "lang"
+    );
+    for (const attr of attrsToRemove) {
+      el2.removeAttribute(attr.name);
+    }
+  }
+}
+function removeEmptyWrappers(doc2) {
+  let changed = true;
+  while (changed) {
+    changed = false;
+    const spans = Array.from(doc2.body.querySelectorAll("span"));
+    for (const span of spans) {
+      if (span.attributes.length === 0) {
+        const parent = span.parentNode;
+        if (parent) {
+          while (span.firstChild) {
+            parent.insertBefore(span.firstChild, span);
+          }
+          parent.removeChild(span);
+          changed = true;
+        }
+      }
+    }
+  }
+  const emptyParas = Array.from(doc2.body.querySelectorAll("p"));
+  for (const p of emptyParas) {
+    if (p.innerHTML.trim() === "" && doc2.body.children.length > 1) {
+      p.remove();
+    }
+  }
+}
+function normalizeExcelTables(doc2) {
+  const tables = Array.from(doc2.body.querySelectorAll("table"));
+  for (const table of tables) {
+    removeExcelAttributes(table);
+    const cells = Array.from(table.querySelectorAll("td, th"));
+    for (const cell of cells) {
+      removeExcelAttributes(cell);
+      const style2 = cell.getAttribute("style");
+      if (style2) {
+        const cleaned = cleanMsoStyles(style2);
+        if (cleaned) {
+          cell.setAttribute("style", cleaned);
+        } else {
+          cell.removeAttribute("style");
+        }
+      }
+    }
+    const rows = Array.from(table.querySelectorAll("tr"));
+    for (const row of rows) {
+      removeExcelAttributes(row);
+    }
+  }
+}
+const EXCEL_ATTRS = [
+  "x:num",
+  "x:str",
+  "x:fmla",
+  "x:autofilter",
+  "mso-number-format"
+];
+function removeExcelAttributes(el2) {
+  for (const attr of EXCEL_ATTRS) {
+    el2.removeAttribute(attr);
+  }
+  const className = el2.getAttribute("class");
+  if (className) {
+    const cleaned = className.replace(MSO_CLASS_PATTERN, "").trim();
+    if (cleaned) {
+      el2.setAttribute("class", cleaned);
+    } else {
+      el2.removeAttribute("class");
+    }
+  }
+}
+const DANGEROUS_TAGS = [
+  "script",
+  "iframe",
+  "object",
+  "embed",
+  "applet",
+  "form",
+  "input",
+  "textarea",
+  "select",
+  "button"
+];
+const EVENT_ATTR_PATTERN = /^on/i;
+const DANGEROUS_URL_PATTERN = /^\s*(javascript|vbscript|data\s*:(?!image))/i;
+function sanitize(doc2) {
+  for (const tag of DANGEROUS_TAGS) {
+    const elements = Array.from(doc2.body.querySelectorAll(tag));
+    for (const el2 of elements) {
+      el2.remove();
+    }
+  }
+  const allElements = Array.from(doc2.body.querySelectorAll("*"));
+  for (const el2 of allElements) {
+    const attrs = Array.from(el2.attributes);
+    for (const attr of attrs) {
+      if (EVENT_ATTR_PATTERN.test(attr.name)) {
+        el2.removeAttribute(attr.name);
+        continue;
+      }
+      if ((attr.name === "href" || attr.name === "src" || attr.name === "action") && DANGEROUS_URL_PATTERN.test(attr.value)) {
+        el2.removeAttribute(attr.name);
+      }
+    }
+    const src = el2.getAttribute("src");
+    if (src && src.startsWith("file://")) {
+      el2.removeAttribute("src");
+    }
+    const href = el2.getAttribute("href");
+    if (href && href.startsWith("file://")) {
+      el2.removeAttribute("href");
+    }
+  }
+}
+function removeStyleBlocks(doc2) {
+  const styles = Array.from(doc2.querySelectorAll("style"));
+  for (const s of styles) {
+    s.remove();
+  }
+}
+function removeHeadElements(doc2) {
+  const selectors = ["meta", "link", "title", "xml"];
+  for (const selector of selectors) {
+    const elements = Array.from(doc2.body.querySelectorAll(selector));
+    for (const el2 of elements) {
+      el2.remove();
+    }
+  }
+}
+function transformOfficeHTML(html) {
+  if (!isOfficeContent(html)) {
+    return html;
+  }
+  let cleaned = removeConditionalComments(html);
+  const parser = new DOMParser();
+  const doc2 = parser.parseFromString(cleaned, "text/html");
+  removeHeadElements(doc2);
+  if (isWordContent(html)) {
+    convertWordLists(doc2);
+  }
+  removeOfficeElements(doc2);
+  if (isExcelContent(html)) {
+    normalizeExcelTables(doc2);
+  }
+  cleanElementStyles(doc2);
+  removeStyleBlocks(doc2);
+  removeEmptyWrappers(doc2);
+  sanitize(doc2);
+  return doc2.body.innerHTML;
+}
+const PasteFromOffice = Extension.create({
+  name: "pasteFromOffice",
+  addOptions() {
+    return {
+      enabled: true
+    };
+  },
+  addStorage() {
+    return {};
+  },
+  transformPastedHTML(html) {
+    if (!this.options.enabled) {
+      return html;
+    }
+    return transformOfficeHTML(html);
+  }
+});
 const en = {
   "Bold": "Bold",
   "Italic": "Italic",
@@ -46401,7 +46779,8 @@ class HTMLEditor {
       convert_unsafe_embeds: config.convert_unsafe_embeds ?? true,
       format_empty_lines: config.format_empty_lines ?? true,
       toolbar_narrow_breakpoint: config.toolbar_narrow_breakpoint,
-      toolbar_priority: config.toolbar_priority
+      toolbar_priority: config.toolbar_priority,
+      paste_from_office: config.paste_from_office ?? true
     };
   }
   createEditor() {
@@ -46604,6 +46983,9 @@ class HTMLEditor {
       }),
       TextDirection
     ];
+    if (this.config.paste_from_office !== false) {
+      extensions.push(PasteFromOffice);
+    }
     if (!this.config.basicEditor) {
       extensions.push(
         Image.configure({
@@ -46829,6 +47211,7 @@ export {
   LineHeight,
   LinkEditor,
   Mention,
+  PasteFromOffice,
   SearchReplace,
   SignatureBlock,
   SourceEditor,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mdaemon/html-editor",
-  "version": "1.1.0",
+  "version": "1.2.0",
   "description": "A TinyMCE-compatible HTML editor built on TipTap",
   "homepage": "https://github.com/mdaemon-technologies/MDHTMLEditor",
   "repository": {
@@ -60,24 +60,24 @@
   },
   "license": "LGPL-3.0-or-later",
   "dependencies": {
-    "@tiptap/core": "^3.23.1",
-    "@tiptap/extension-character-count": "^3.23.1",
-    "@tiptap/extension-code-block-lowlight": "^3.23.1",
-    "@tiptap/extension-color": "^3.23.1",
-    "@tiptap/extension-font-family": "^3.23.1",
-    "@tiptap/extension-highlight": "^3.23.1",
-    "@tiptap/extension-image": "^3.23.1",
-    "@tiptap/extension-link": "^3.23.1",
-    "@tiptap/extension-placeholder": "^3.23.1",
-    "@tiptap/extension-table": "^3.23.1",
-    "@tiptap/extension-table-cell": "^3.23.1",
-    "@tiptap/extension-table-header": "^3.23.1",
-    "@tiptap/extension-table-row": "^3.23.1",
-    "@tiptap/extension-text-align": "^3.23.1",
-    "@tiptap/extension-text-style": "^3.23.1",
-    "@tiptap/extension-underline": "^3.23.1",
-    "@tiptap/pm": "^3.23.1",
-    "@tiptap/starter-kit": "^3.23.1",
+    "@tiptap/core": "^3.23.4",
+    "@tiptap/extension-character-count": "^3.23.4",
+    "@tiptap/extension-code-block-lowlight": "^3.23.4",
+    "@tiptap/extension-color": "^3.23.4",
+    "@tiptap/extension-font-family": "^3.23.4",
+    "@tiptap/extension-highlight": "^3.23.4",
+    "@tiptap/extension-image": "^3.23.4",
+    "@tiptap/extension-link": "^3.23.4",
+    "@tiptap/extension-placeholder": "^3.23.4",
+    "@tiptap/extension-table": "^3.23.4",
+    "@tiptap/extension-table-cell": "^3.23.4",
+    "@tiptap/extension-table-header": "^3.23.4",
+    "@tiptap/extension-table-row": "^3.23.4",
+    "@tiptap/extension-text-align": "^3.23.4",
+    "@tiptap/extension-text-style": "^3.23.4",
+    "@tiptap/extension-underline": "^3.23.4",
+    "@tiptap/pm": "^3.23.4",
+    "@tiptap/starter-kit": "^3.23.4",
     "lowlight": "^3.3.0"
   },
   "devDependencies": {