@kedataindo/docflow-core 0.0.49 → 0.0.50

package/dist/index.cjs

@@ -49,11 +49,13 @@ __export(index_exports, {
   createEditor: () => createEditor,
   createPerformanceMonitor: () => createPerformanceMonitor,
   definePlugin: () => definePlugin,
+  detectClipboardSource: () => detectClipboardSource,
   findMatches: () => findMatches,
   getSearchState: () => getSearchState,
   httpKeyStorage: () => httpKeyStorage,
   localStorageKeyStorage: () => localStorageKeyStorage,
   memoryKeyStorage: () => memoryKeyStorage,
+  normalizeClipboardHTML: () => normalizeClipboardHTML,
   openaiCompatibleProvider: () => openaiCompatibleProvider,
   replaceAll: () => replaceAll,
   replaceCurrent: () => replaceCurrent,
@@ -460,13 +462,13 @@ function replaceAll(editor, replacement) {
   return s.matches.length;
 }
 
-// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_@tiptap+core@2.27.2_@tiptap+pm@2.27.2__@tiptap+pm@2.27.2/node_modules/tiptap-pagination-plus/dist/PaginationPlus.js
+// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_patch_hash=4857789ba5672a185fb96bdd854e03e0c6ec8ed26abfeea_16730b2b361cd3f71113c52459ed62fd/node_modules/tiptap-pagination-plus/dist/PaginationPlus.js
 var import_core4 = require("@tiptap/core");
 var import_state3 = require("@tiptap/pm/state");
 var import_transform = require("@tiptap/pm/transform");
 var import_view3 = require("@tiptap/pm/view");
 
-// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_@tiptap+core@2.27.2_@tiptap+pm@2.27.2__@tiptap+pm@2.27.2/node_modules/tiptap-pagination-plus/dist/utils.js
+// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_patch_hash=4857789ba5672a185fb96bdd854e03e0c6ec8ed26abfeea_16730b2b361cd3f71113c52459ed62fd/node_modules/tiptap-pagination-plus/dist/utils.js
 var updateCssVariables = (targetNode, config) => {
   const cssVariables = {
     "rm-page-height": `${config.pageHeight}px`,
@@ -640,7 +642,7 @@ var footerClickEvent = (pageNumber, onFooterClick) => {
   };
 };
 
-// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_@tiptap+core@2.27.2_@tiptap+pm@2.27.2__@tiptap+pm@2.27.2/node_modules/tiptap-pagination-plus/dist/PaginationPlus.js
+// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_patch_hash=4857789ba5672a185fb96bdd854e03e0c6ec8ed26abfeea_16730b2b361cd3f71113c52459ed62fd/node_modules/tiptap-pagination-plus/dist/PaginationPlus.js
 var page_count_meta_key = "PAGE_COUNT_META_KEY";
 var key = new import_state3.PluginKey("brDecoration");
 function buildDecorations(doc) {
@@ -807,11 +809,6 @@ var PaginationPlus = import_core4.Extension.create({
         width: 100%;
         display: contents;
       }
-      .rm-with-pagination table tbody{
-        display: table;
-        max-height: 300px;
-        overflow-y: auto;
-      }
       .rm-with-pagination table tbody > tr{
         display: table-row !important;
       }
@@ -1134,6 +1131,7 @@ var getExistingPageCount = (view) => {
   }
   return 0;
 };
+var pageGapTracker = /* @__PURE__ */ new WeakMap();
 var calculatePageCount = (view, pageOptions, headerHeight = 0, footerHeight = 0) => {
   var _a;
   const editorDom = view.dom;
@@ -1151,8 +1149,36 @@ var calculatePageCount = (view, pageOptions, headerHeight = 0, footerHeight = 0)
       const lastPageGap = lastElementRect.bottom - lastPageBreakRect.bottom;
       if (lastPageGap > 0) {
         const addPage = Math.ceil(lastPageGap / pageContentAreaHeight);
+        if (lastElementRect.height > pageContentAreaHeight) {
+          const isSplittableTable = (lastElementOfEditor === null || lastElementOfEditor === void 0 ? void 0 : lastElementOfEditor.tagName) === "TABLE" && lastElementOfEditor.hasAttribute("data-tps-splittable");
+          if (!isSplittableTable) {
+            return currentPageCount;
+          }
+        }
+        const prev = pageGapTracker.get(editorDom);
+        if (prev && prev.pages < currentPageCount && lastPageGap > prev.gap - 2) {
+          return currentPageCount;
+        }
+        pageGapTracker.set(editorDom, { gap: lastPageGap, pages: currentPageCount });
+        let totalEditorContentHeight = 0;
+        for (const child of editorDom.children) {
+          if (!(child instanceof HTMLElement)) continue;
+          if (child.hasAttribute("data-rm-pagination")) continue;
+          if (child.classList.contains("rm-page-break")) continue;
+          totalEditorContentHeight += child.getBoundingClientRect().height;
+        }
+        const maxPagesByContent = Math.ceil(totalEditorContentHeight / pageContentAreaHeight);
+        const contentCap = Math.max(currentPageCount, maxPagesByContent + 1);
+        if (currentPageCount + addPage > contentCap) {
+          return contentCap;
+        }
+        const MAX_PAGES = 1e3;
+        if (currentPageCount + addPage > MAX_PAGES) {
+          return currentPageCount;
+        }
         return currentPageCount + addPage;
       } else {
+        pageGapTracker.delete(editorDom);
         const allBreaksAfterLastElement = Array.from(paginationElement.querySelectorAll(".breaker"));
         const allBreaksAfterLastElementRect = allBreaksAfterLastElement.filter((element) => element.getBoundingClientRect().top > lastElementRect.bottom);
         const removePage = allBreaksAfterLastElementRect.length;
@@ -1279,7 +1305,7 @@ function createDecoration(pageOptions, headerHeightMap, footerHeightMap) {
   return [pageWidget, firstHeaderWidget];
 }
 
-// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_@tiptap+core@2.27.2_@tiptap+pm@2.27.2__@tiptap+pm@2.27.2/node_modules/tiptap-pagination-plus/dist/constants.js
+// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_patch_hash=4857789ba5672a185fb96bdd854e03e0c6ec8ed26abfeea_16730b2b361cd3f71113c52459ed62fd/node_modules/tiptap-pagination-plus/dist/constants.js
 var A4_PAGE_SIZE = getPageSize(1123, 794, 95, 95, 76, 76);
 var A3_PAGE_SIZE = getPageSize(1591, 1123, 95, 95, 76, 76);
 var A5_PAGE_SIZE = getPageSize(794, 419, 76, 76, 57, 57);
@@ -1477,41 +1503,150 @@ function createPerformanceMonitor(options) {
   return monitor;
 }
 
-// src/Editor.ts
-function sanitizePastedHTML(html) {
-  let cleaned = html.replace(/<meta[^>]*>/gi, "");
-  cleaned = cleaned.replace(/<style[^>]*>[\s\S]*?<\/style>/gi, "");
-  cleaned = cleaned.replace(/<!--[\s\S]*?-->/g, "");
-  cleaned = cleaned.replace(/<link[^>]*>/gi, "");
-  cleaned = cleaned.replace(/<base[^>]*>/gi, "");
-  cleaned = cleaned.replace(/<title[^>]*>[\s\S]*?<\/title>/gi, "");
-  {
-    for (let pass = 0; pass < 10; pass++) {
-      const before = cleaned;
-      cleaned = cleaned.replace(
-        /<span\s([^>]*style\s*=\s*["'][^"']*(?:background-color|background)[^"']*["'][^>]*)>([^<]*(?:<(?!\/?span)[^>]*>[^<]*)*)<\/span>/gi,
-        (_full, attrs, content) => {
-          if (!/background(?:-color)?\s*:/.test(attrs)) return _full;
-          const colorMatch = attrs.match(/(?<!background-)color\s*:\s*([^;"]+)/i);
-          const colorVal = colorMatch ? colorMatch[1].trim() : null;
-          const markAttrs = colorVal ? attrs.replace(/(?<!background-)color\s*:\s*[^;"]+;?\s*/gi, "").trim() : attrs;
-          if (colorVal) {
-            return `<mark ${markAttrs}><span style="color: ${colorVal};">${content}</span></mark>`;
-          }
-          return `<mark ${markAttrs}>${content}</mark>`;
-        }
-      );
-      if (cleaned === before) break;
+// src/pasteNormalization.ts
+var KEEP_STYLE_PROPS = /* @__PURE__ */ new Set(["color", "background-color"]);
+var REMOVE_TAGS = /* @__PURE__ */ new Set([
+  "meta",
+  "style",
+  "link",
+  "base",
+  "title",
+  "script",
+  "iframe",
+  "object",
+  "embed",
+  "form",
+  "input",
+  "button",
+  "select",
+  "textarea",
+  "noscript",
+  // Word Office namespace dump paragraph
+  "o:p"
+]);
+var KEEP_ATTRS = {
+  a: /* @__PURE__ */ new Set(["href", "style"]),
+  img: /* @__PURE__ */ new Set(["src", "alt", "style"]),
+  td: /* @__PURE__ */ new Set(["colspan", "rowspan", "style"]),
+  th: /* @__PURE__ */ new Set(["colspan", "rowspan", "style"]),
+  col: /* @__PURE__ */ new Set(["span"]),
+  "*": /* @__PURE__ */ new Set(["style"])
+};
+function detectClipboardSource(html) {
+  if (/id\s*=\s*["']docs-internal/i.test(html)) return "google-docs";
+  if (/<o:p[\s>]|xmlns(:w|:o)?=|class\s*=\s*["']?Mso|mso-/i.test(html)) return "word";
+  return "generic";
+}
+function parseStyleDeclarations(style) {
+  const map = /* @__PURE__ */ new Map();
+  for (const decl of style.split(";")) {
+    const idx = decl.indexOf(":");
+    if (idx < 0) continue;
+    const prop = decl.slice(0, idx).trim().toLowerCase();
+    const val = decl.slice(idx + 1).trim();
+    if (prop && val) map.set(prop, val);
+  }
+  return map;
+}
+function cleanStyleAttr(el) {
+  const style = el.getAttribute("style");
+  if (!style) return;
+  const kept = [];
+  parseStyleDeclarations(style).forEach((val, prop) => {
+    if (KEEP_STYLE_PROPS.has(prop)) kept.push(`${prop}: ${val}`);
+  });
+  if (kept.length) el.setAttribute("style", kept.join("; "));
+  else el.removeAttribute("style");
+}
+function cleanAttrs(el) {
+  const keep = /* @__PURE__ */ new Set([...KEEP_ATTRS["*"] ?? [], ...KEEP_ATTRS[el.tagName.toLowerCase()] ?? []]);
+  for (const attr of Array.from(el.attributes)) {
+    if (!keep.has(attr.name)) el.removeAttribute(attr.name);
+  }
+}
+function unwrap(el) {
+  const parent = el.parentNode;
+  if (!parent) return;
+  while (el.firstChild) parent.insertBefore(el.firstChild, el);
+  parent.removeChild(el);
+}
+function convertHighlightSpan(span, styles) {
+  const bg = styles.get("background-color") ?? styles.get("background");
+  if (!bg) return;
+  const color = styles.get("color");
+  const mark = span.ownerDocument.createElement("mark");
+  mark.setAttribute("style", `background-color: ${bg}`);
+  if (color) {
+    const inner = span.ownerDocument.createElement("span");
+    inner.setAttribute("style", `color: ${color}`);
+    while (span.firstChild) inner.appendChild(span.firstChild);
+    mark.appendChild(inner);
+  } else {
+    while (span.firstChild) mark.appendChild(span.firstChild);
+  }
+  span.replaceWith(mark);
+}
+function normalizeElement(el, source) {
+  if (el.tagName === "SPAN") {
+    const styles = parseStyleDeclarations(el.getAttribute("style") ?? "");
+    if (styles.has("background-color") || styles.has("background")) {
+      convertHighlightSpan(el, styles);
+      return;
+    }
+    cleanStyleAttr(el);
+    cleanAttrs(el);
+    if (el.attributes.length === 0) unwrap(el);
+    return;
+  }
+  if (el.tagName === "B") {
+    const styles = parseStyleDeclarations(el.getAttribute("style") ?? "");
+    if (styles.get("font-weight")?.replace(/\s/g, "") === "normal") {
+      unwrap(el);
+      return;
     }
   }
-  cleaned = cleaned.replace(
-    /<b\s[^>]*style\s*=\s*["'][^"']*font-weight:\s*normal[^"']*["'][^>]*>/gi,
-    ""
+  cleanStyleAttr(el);
+  cleanAttrs(el);
+  if (source === "word") {
+    el.removeAttribute("class");
+  }
+}
+function removeComments(root) {
+  const walker = root.ownerDocument.createTreeWalker(
+    root,
+    128
+    /* NodeFilter.SHOW_COMMENT */
   );
-  cleaned = cleaned.replace(/<\/b>/gi, "");
-  cleaned = cleaned.replace(/\sid\s*=\s*["']docs-internal[^"']*["']/gi, "");
-  return cleaned;
+  const comments = [];
+  let node = walker.nextNode();
+  while (node) {
+    comments.push(node);
+    node = walker.nextNode();
+  }
+  for (const c of comments) c.parentNode?.removeChild(c);
 }
+function normalizeClipboardHTML(html) {
+  if (!html || typeof html !== "string") return html ?? "";
+  if (!/[<][a-zA-Z]/.test(html)) return html;
+  const source = detectClipboardSource(html);
+  const doc = new DOMParser().parseFromString(html, "text/html");
+  removeComments(doc.body);
+  for (const tag of REMOVE_TAGS) {
+    for (const el of Array.from(doc.body.getElementsByTagName(tag))) {
+      el.parentNode?.removeChild(el);
+    }
+  }
+  const walk = (node) => {
+    if (node.nodeType !== 1) return;
+    for (const child of Array.from(node.childNodes)) walk(child);
+    normalizeElement(node, source);
+  };
+  for (const child of Array.from(doc.body.childNodes)) walk(child);
+  return doc.body.innerHTML;
+}
+var sanitizePastedHTML = normalizeClipboardHTML;
+
+// src/Editor.ts
 function migrateContent(content) {
   if (!content) return content;
   if (typeof content === "string") {
@@ -2473,11 +2608,13 @@ function openaiCompatibleProvider(config) {
   createEditor,
   createPerformanceMonitor,
   definePlugin,
+  detectClipboardSource,
   findMatches,
   getSearchState,
   httpKeyStorage,
   localStorageKeyStorage,
   memoryKeyStorage,
+  normalizeClipboardHTML,
   openaiCompatibleProvider,
   replaceAll,
   replaceCurrent,

package/dist/index.d.cts

@@ -186,15 +186,30 @@ declare class PerformanceMonitor {
 declare function createPerformanceMonitor(options?: PerformanceMonitorOptions): PerformanceMonitor;
 
 /**
- * Sanitize pasted HTML content (e.g. from Google Docs) to prevent crashes
- * during ProseMirror parsing. Strips non-content tags like <meta>, <style>,
- * and HTML comments that are common in rich clipboard data but not valid
- * in the ProseMirror schema.
+ * DOM-based clipboard HTML normalization (Stage 2 of the paste pipeline —
+ * docs/plans/clipboard-paste-pipeline-plan.md).
  *
- * Also converts Google Docs-specific markup into standard TipTap-compatible
- * HTML so that text color and background-color survive the paste.
+ * Replaces the old regex-based sanitizePastedHTML: pasted HTML from Google
+ * Docs, Word, Notion, or generic web pages is parsed into an inert Document,
+ * normalized against a whitelist, and serialized for ProseMirror's parser.
+ *
+ * Design decisions (see plan §3):
+ * - Typography normalizes to destination styles: font-family, font-size,
+ *   line-height, margins and padding are stripped; only `color` and
+ *   `background-color` survive as inline styles.
+ * - One pipeline for all HTML sources; detectClipboardSource only tunes
+ *   which wrapper-stripping rules run.
+ * - Idempotent by construction: normalizing twice changes nothing.
+ */
+type ClipboardSource = 'google-docs' | 'word' | 'generic';
+declare function detectClipboardSource(html: string): ClipboardSource;
+declare function normalizeClipboardHTML(html: string): string;
+/**
+ * Backward-compatible name kept for existing callers
+ * (transformPastedHTML in Editor.ts, handlePaste in DocsEditor.vue).
  */
-declare function sanitizePastedHTML(html: string): string;
+declare const sanitizePastedHTML: typeof normalizeClipboardHTML;
+
 interface EditorOptions {
     target?: HTMLElement;
     content?: object | string;
@@ -380,4 +395,4 @@ declare class SubdocumentProvider {
     destroy(): void;
 }
 
-export { AIDraftFn, AIStreamFn, type AwarenessState, BlockAttributesExtension, CitationPort, type CollaborationOptions, type CollaborationSetup, type DocsEditor, type DocsEditorPlugin, EditorContextExtension, type EditorContextOptions, type EditorOptions, FontSizeExtension, ImageUploadHandler, type MonitorMetric, PerformanceMonitor, type PerformanceMonitorOptions, SearchAndReplaceExtension, type SearchMatch, type SearchState, type SlashCommand, type SubdocState, SubdocumentProvider, type SubdocumentProviderOptions, type ToolbarItem, clearSearch, collaborationExtensions, collectExtensions, createActionMap, createCollaboration, createEditor, createPerformanceMonitor, definePlugin, findMatches, getSearchState, replaceAll, replaceCurrent, resolveAction, sanitizePastedHTML, searchAndReplaceKey, searchNext, searchPrev, setSearchQuery };
+export { AIDraftFn, AIStreamFn, type AwarenessState, BlockAttributesExtension, CitationPort, type ClipboardSource, type CollaborationOptions, type CollaborationSetup, type DocsEditor, type DocsEditorPlugin, EditorContextExtension, type EditorContextOptions, type EditorOptions, FontSizeExtension, ImageUploadHandler, type MonitorMetric, PerformanceMonitor, type PerformanceMonitorOptions, SearchAndReplaceExtension, type SearchMatch, type SearchState, type SlashCommand, type SubdocState, SubdocumentProvider, type SubdocumentProviderOptions, type ToolbarItem, clearSearch, collaborationExtensions, collectExtensions, createActionMap, createCollaboration, createEditor, createPerformanceMonitor, definePlugin, detectClipboardSource, findMatches, getSearchState, normalizeClipboardHTML, replaceAll, replaceCurrent, resolveAction, sanitizePastedHTML, searchAndReplaceKey, searchNext, searchPrev, setSearchQuery };

package/dist/index.d.ts

@@ -186,15 +186,30 @@ declare class PerformanceMonitor {
 declare function createPerformanceMonitor(options?: PerformanceMonitorOptions): PerformanceMonitor;
 
 /**
- * Sanitize pasted HTML content (e.g. from Google Docs) to prevent crashes
- * during ProseMirror parsing. Strips non-content tags like <meta>, <style>,
- * and HTML comments that are common in rich clipboard data but not valid
- * in the ProseMirror schema.
+ * DOM-based clipboard HTML normalization (Stage 2 of the paste pipeline —
+ * docs/plans/clipboard-paste-pipeline-plan.md).
  *
- * Also converts Google Docs-specific markup into standard TipTap-compatible
- * HTML so that text color and background-color survive the paste.
+ * Replaces the old regex-based sanitizePastedHTML: pasted HTML from Google
+ * Docs, Word, Notion, or generic web pages is parsed into an inert Document,
+ * normalized against a whitelist, and serialized for ProseMirror's parser.
+ *
+ * Design decisions (see plan §3):
+ * - Typography normalizes to destination styles: font-family, font-size,
+ *   line-height, margins and padding are stripped; only `color` and
+ *   `background-color` survive as inline styles.
+ * - One pipeline for all HTML sources; detectClipboardSource only tunes
+ *   which wrapper-stripping rules run.
+ * - Idempotent by construction: normalizing twice changes nothing.
+ */
+type ClipboardSource = 'google-docs' | 'word' | 'generic';
+declare function detectClipboardSource(html: string): ClipboardSource;
+declare function normalizeClipboardHTML(html: string): string;
+/**
+ * Backward-compatible name kept for existing callers
+ * (transformPastedHTML in Editor.ts, handlePaste in DocsEditor.vue).
  */
-declare function sanitizePastedHTML(html: string): string;
+declare const sanitizePastedHTML: typeof normalizeClipboardHTML;
+
 interface EditorOptions {
     target?: HTMLElement;
     content?: object | string;
@@ -380,4 +395,4 @@ declare class SubdocumentProvider {
     destroy(): void;
 }
 
-export { AIDraftFn, AIStreamFn, type AwarenessState, BlockAttributesExtension, CitationPort, type CollaborationOptions, type CollaborationSetup, type DocsEditor, type DocsEditorPlugin, EditorContextExtension, type EditorContextOptions, type EditorOptions, FontSizeExtension, ImageUploadHandler, type MonitorMetric, PerformanceMonitor, type PerformanceMonitorOptions, SearchAndReplaceExtension, type SearchMatch, type SearchState, type SlashCommand, type SubdocState, SubdocumentProvider, type SubdocumentProviderOptions, type ToolbarItem, clearSearch, collaborationExtensions, collectExtensions, createActionMap, createCollaboration, createEditor, createPerformanceMonitor, definePlugin, findMatches, getSearchState, replaceAll, replaceCurrent, resolveAction, sanitizePastedHTML, searchAndReplaceKey, searchNext, searchPrev, setSearchQuery };
+export { AIDraftFn, AIStreamFn, type AwarenessState, BlockAttributesExtension, CitationPort, type ClipboardSource, type CollaborationOptions, type CollaborationSetup, type DocsEditor, type DocsEditorPlugin, EditorContextExtension, type EditorContextOptions, type EditorOptions, FontSizeExtension, ImageUploadHandler, type MonitorMetric, PerformanceMonitor, type PerformanceMonitorOptions, SearchAndReplaceExtension, type SearchMatch, type SearchState, type SlashCommand, type SubdocState, SubdocumentProvider, type SubdocumentProviderOptions, type ToolbarItem, clearSearch, collaborationExtensions, collectExtensions, createActionMap, createCollaboration, createEditor, createPerformanceMonitor, definePlugin, detectClipboardSource, findMatches, getSearchState, normalizeClipboardHTML, replaceAll, replaceCurrent, resolveAction, sanitizePastedHTML, searchAndReplaceKey, searchNext, searchPrev, setSearchQuery };

package/dist/index.js

@@ -402,13 +402,13 @@ function replaceAll(editor, replacement) {
   return s.matches.length;
 }
 
-// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_@tiptap+core@2.27.2_@tiptap+pm@2.27.2__@tiptap+pm@2.27.2/node_modules/tiptap-pagination-plus/dist/PaginationPlus.js
+// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_patch_hash=4857789ba5672a185fb96bdd854e03e0c6ec8ed26abfeea_16730b2b361cd3f71113c52459ed62fd/node_modules/tiptap-pagination-plus/dist/PaginationPlus.js
 import { Extension as Extension4 } from "@tiptap/core";
 import { Plugin as Plugin3, PluginKey as PluginKey3 } from "@tiptap/pm/state";
 import { ReplaceStep, ReplaceAroundStep, AddMarkStep, RemoveMarkStep, RemoveNodeMarkStep, AttrStep } from "@tiptap/pm/transform";
 import { Decoration as Decoration3, DecorationSet as DecorationSet3 } from "@tiptap/pm/view";
 
-// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_@tiptap+core@2.27.2_@tiptap+pm@2.27.2__@tiptap+pm@2.27.2/node_modules/tiptap-pagination-plus/dist/utils.js
+// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_patch_hash=4857789ba5672a185fb96bdd854e03e0c6ec8ed26abfeea_16730b2b361cd3f71113c52459ed62fd/node_modules/tiptap-pagination-plus/dist/utils.js
 var updateCssVariables = (targetNode, config) => {
   const cssVariables = {
     "rm-page-height": `${config.pageHeight}px`,
@@ -582,7 +582,7 @@ var footerClickEvent = (pageNumber, onFooterClick) => {
   };
 };
 
-// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_@tiptap+core@2.27.2_@tiptap+pm@2.27.2__@tiptap+pm@2.27.2/node_modules/tiptap-pagination-plus/dist/PaginationPlus.js
+// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_patch_hash=4857789ba5672a185fb96bdd854e03e0c6ec8ed26abfeea_16730b2b361cd3f71113c52459ed62fd/node_modules/tiptap-pagination-plus/dist/PaginationPlus.js
 var page_count_meta_key = "PAGE_COUNT_META_KEY";
 var key = new PluginKey3("brDecoration");
 function buildDecorations(doc) {
@@ -749,11 +749,6 @@ var PaginationPlus = Extension4.create({
         width: 100%;
         display: contents;
       }
-      .rm-with-pagination table tbody{
-        display: table;
-        max-height: 300px;
-        overflow-y: auto;
-      }
       .rm-with-pagination table tbody > tr{
         display: table-row !important;
       }
@@ -1076,6 +1071,7 @@ var getExistingPageCount = (view) => {
   }
   return 0;
 };
+var pageGapTracker = /* @__PURE__ */ new WeakMap();
 var calculatePageCount = (view, pageOptions, headerHeight = 0, footerHeight = 0) => {
   var _a;
   const editorDom = view.dom;
@@ -1093,8 +1089,36 @@ var calculatePageCount = (view, pageOptions, headerHeight = 0, footerHeight = 0)
       const lastPageGap = lastElementRect.bottom - lastPageBreakRect.bottom;
       if (lastPageGap > 0) {
         const addPage = Math.ceil(lastPageGap / pageContentAreaHeight);
+        if (lastElementRect.height > pageContentAreaHeight) {
+          const isSplittableTable = (lastElementOfEditor === null || lastElementOfEditor === void 0 ? void 0 : lastElementOfEditor.tagName) === "TABLE" && lastElementOfEditor.hasAttribute("data-tps-splittable");
+          if (!isSplittableTable) {
+            return currentPageCount;
+          }
+        }
+        const prev = pageGapTracker.get(editorDom);
+        if (prev && prev.pages < currentPageCount && lastPageGap > prev.gap - 2) {
+          return currentPageCount;
+        }
+        pageGapTracker.set(editorDom, { gap: lastPageGap, pages: currentPageCount });
+        let totalEditorContentHeight = 0;
+        for (const child of editorDom.children) {
+          if (!(child instanceof HTMLElement)) continue;
+          if (child.hasAttribute("data-rm-pagination")) continue;
+          if (child.classList.contains("rm-page-break")) continue;
+          totalEditorContentHeight += child.getBoundingClientRect().height;
+        }
+        const maxPagesByContent = Math.ceil(totalEditorContentHeight / pageContentAreaHeight);
+        const contentCap = Math.max(currentPageCount, maxPagesByContent + 1);
+        if (currentPageCount + addPage > contentCap) {
+          return contentCap;
+        }
+        const MAX_PAGES = 1e3;
+        if (currentPageCount + addPage > MAX_PAGES) {
+          return currentPageCount;
+        }
         return currentPageCount + addPage;
       } else {
+        pageGapTracker.delete(editorDom);
         const allBreaksAfterLastElement = Array.from(paginationElement.querySelectorAll(".breaker"));
         const allBreaksAfterLastElementRect = allBreaksAfterLastElement.filter((element) => element.getBoundingClientRect().top > lastElementRect.bottom);
         const removePage = allBreaksAfterLastElementRect.length;
@@ -1221,7 +1245,7 @@ function createDecoration(pageOptions, headerHeightMap, footerHeightMap) {
   return [pageWidget, firstHeaderWidget];
 }
 
-// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_@tiptap+core@2.27.2_@tiptap+pm@2.27.2__@tiptap+pm@2.27.2/node_modules/tiptap-pagination-plus/dist/constants.js
+// ../../node_modules/.pnpm/tiptap-pagination-plus@3.1.0_patch_hash=4857789ba5672a185fb96bdd854e03e0c6ec8ed26abfeea_16730b2b361cd3f71113c52459ed62fd/node_modules/tiptap-pagination-plus/dist/constants.js
 var A4_PAGE_SIZE = getPageSize(1123, 794, 95, 95, 76, 76);
 var A3_PAGE_SIZE = getPageSize(1591, 1123, 95, 95, 76, 76);
 var A5_PAGE_SIZE = getPageSize(794, 419, 76, 76, 57, 57);
@@ -1419,41 +1443,150 @@ function createPerformanceMonitor(options) {
   return monitor;
 }
 
-// src/Editor.ts
-function sanitizePastedHTML(html) {
-  let cleaned = html.replace(/<meta[^>]*>/gi, "");
-  cleaned = cleaned.replace(/<style[^>]*>[\s\S]*?<\/style>/gi, "");
-  cleaned = cleaned.replace(/<!--[\s\S]*?-->/g, "");
-  cleaned = cleaned.replace(/<link[^>]*>/gi, "");
-  cleaned = cleaned.replace(/<base[^>]*>/gi, "");
-  cleaned = cleaned.replace(/<title[^>]*>[\s\S]*?<\/title>/gi, "");
-  {
-    for (let pass = 0; pass < 10; pass++) {
-      const before = cleaned;
-      cleaned = cleaned.replace(
-        /<span\s([^>]*style\s*=\s*["'][^"']*(?:background-color|background)[^"']*["'][^>]*)>([^<]*(?:<(?!\/?span)[^>]*>[^<]*)*)<\/span>/gi,
-        (_full, attrs, content) => {
-          if (!/background(?:-color)?\s*:/.test(attrs)) return _full;
-          const colorMatch = attrs.match(/(?<!background-)color\s*:\s*([^;"]+)/i);
-          const colorVal = colorMatch ? colorMatch[1].trim() : null;
-          const markAttrs = colorVal ? attrs.replace(/(?<!background-)color\s*:\s*[^;"]+;?\s*/gi, "").trim() : attrs;
-          if (colorVal) {
-            return `<mark ${markAttrs}><span style="color: ${colorVal};">${content}</span></mark>`;
-          }
-          return `<mark ${markAttrs}>${content}</mark>`;
-        }
-      );
-      if (cleaned === before) break;
+// src/pasteNormalization.ts
+var KEEP_STYLE_PROPS = /* @__PURE__ */ new Set(["color", "background-color"]);
+var REMOVE_TAGS = /* @__PURE__ */ new Set([
+  "meta",
+  "style",
+  "link",
+  "base",
+  "title",
+  "script",
+  "iframe",
+  "object",
+  "embed",
+  "form",
+  "input",
+  "button",
+  "select",
+  "textarea",
+  "noscript",
+  // Word Office namespace dump paragraph
+  "o:p"
+]);
+var KEEP_ATTRS = {
+  a: /* @__PURE__ */ new Set(["href", "style"]),
+  img: /* @__PURE__ */ new Set(["src", "alt", "style"]),
+  td: /* @__PURE__ */ new Set(["colspan", "rowspan", "style"]),
+  th: /* @__PURE__ */ new Set(["colspan", "rowspan", "style"]),
+  col: /* @__PURE__ */ new Set(["span"]),
+  "*": /* @__PURE__ */ new Set(["style"])
+};
+function detectClipboardSource(html) {
+  if (/id\s*=\s*["']docs-internal/i.test(html)) return "google-docs";
+  if (/<o:p[\s>]|xmlns(:w|:o)?=|class\s*=\s*["']?Mso|mso-/i.test(html)) return "word";
+  return "generic";
+}
+function parseStyleDeclarations(style) {
+  const map = /* @__PURE__ */ new Map();
+  for (const decl of style.split(";")) {
+    const idx = decl.indexOf(":");
+    if (idx < 0) continue;
+    const prop = decl.slice(0, idx).trim().toLowerCase();
+    const val = decl.slice(idx + 1).trim();
+    if (prop && val) map.set(prop, val);
+  }
+  return map;
+}
+function cleanStyleAttr(el) {
+  const style = el.getAttribute("style");
+  if (!style) return;
+  const kept = [];
+  parseStyleDeclarations(style).forEach((val, prop) => {
+    if (KEEP_STYLE_PROPS.has(prop)) kept.push(`${prop}: ${val}`);
+  });
+  if (kept.length) el.setAttribute("style", kept.join("; "));
+  else el.removeAttribute("style");
+}
+function cleanAttrs(el) {
+  const keep = /* @__PURE__ */ new Set([...KEEP_ATTRS["*"] ?? [], ...KEEP_ATTRS[el.tagName.toLowerCase()] ?? []]);
+  for (const attr of Array.from(el.attributes)) {
+    if (!keep.has(attr.name)) el.removeAttribute(attr.name);
+  }
+}
+function unwrap(el) {
+  const parent = el.parentNode;
+  if (!parent) return;
+  while (el.firstChild) parent.insertBefore(el.firstChild, el);
+  parent.removeChild(el);
+}
+function convertHighlightSpan(span, styles) {
+  const bg = styles.get("background-color") ?? styles.get("background");
+  if (!bg) return;
+  const color = styles.get("color");
+  const mark = span.ownerDocument.createElement("mark");
+  mark.setAttribute("style", `background-color: ${bg}`);
+  if (color) {
+    const inner = span.ownerDocument.createElement("span");
+    inner.setAttribute("style", `color: ${color}`);
+    while (span.firstChild) inner.appendChild(span.firstChild);
+    mark.appendChild(inner);
+  } else {
+    while (span.firstChild) mark.appendChild(span.firstChild);
+  }
+  span.replaceWith(mark);
+}
+function normalizeElement(el, source) {
+  if (el.tagName === "SPAN") {
+    const styles = parseStyleDeclarations(el.getAttribute("style") ?? "");
+    if (styles.has("background-color") || styles.has("background")) {
+      convertHighlightSpan(el, styles);
+      return;
+    }
+    cleanStyleAttr(el);
+    cleanAttrs(el);
+    if (el.attributes.length === 0) unwrap(el);
+    return;
+  }
+  if (el.tagName === "B") {
+    const styles = parseStyleDeclarations(el.getAttribute("style") ?? "");
+    if (styles.get("font-weight")?.replace(/\s/g, "") === "normal") {
+      unwrap(el);
+      return;
     }
   }
-  cleaned = cleaned.replace(
-    /<b\s[^>]*style\s*=\s*["'][^"']*font-weight:\s*normal[^"']*["'][^>]*>/gi,
-    ""
+  cleanStyleAttr(el);
+  cleanAttrs(el);
+  if (source === "word") {
+    el.removeAttribute("class");
+  }
+}
+function removeComments(root) {
+  const walker = root.ownerDocument.createTreeWalker(
+    root,
+    128
+    /* NodeFilter.SHOW_COMMENT */
   );
-  cleaned = cleaned.replace(/<\/b>/gi, "");
-  cleaned = cleaned.replace(/\sid\s*=\s*["']docs-internal[^"']*["']/gi, "");
-  return cleaned;
+  const comments = [];
+  let node = walker.nextNode();
+  while (node) {
+    comments.push(node);
+    node = walker.nextNode();
+  }
+  for (const c of comments) c.parentNode?.removeChild(c);
 }
+function normalizeClipboardHTML(html) {
+  if (!html || typeof html !== "string") return html ?? "";
+  if (!/[<][a-zA-Z]/.test(html)) return html;
+  const source = detectClipboardSource(html);
+  const doc = new DOMParser().parseFromString(html, "text/html");
+  removeComments(doc.body);
+  for (const tag of REMOVE_TAGS) {
+    for (const el of Array.from(doc.body.getElementsByTagName(tag))) {
+      el.parentNode?.removeChild(el);
+    }
+  }
+  const walk = (node) => {
+    if (node.nodeType !== 1) return;
+    for (const child of Array.from(node.childNodes)) walk(child);
+    normalizeElement(node, source);
+  };
+  for (const child of Array.from(doc.body.childNodes)) walk(child);
+  return doc.body.innerHTML;
+}
+var sanitizePastedHTML = normalizeClipboardHTML;
+
+// src/Editor.ts
 function migrateContent(content) {
   if (!content) return content;
   if (typeof content === "string") {
@@ -2151,11 +2284,13 @@ export {
   createEditor,
   createPerformanceMonitor,
   definePlugin,
+  detectClipboardSource,
   findMatches,
   getSearchState,
   httpKeyStorage,
   localStorageKeyStorage,
   memoryKeyStorage,
+  normalizeClipboardHTML,
   openaiCompatibleProvider,
   replaceAll,
   replaceCurrent,

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@kedataindo/docflow-core",
   "license": "UNLICENSED",
-  "version": "0.0.49",
+  "version": "0.0.50",
   "type": "module",
   "main": "./dist/index.js",
   "module": "./dist/index.js",