Merge branch 'main' into ai-vector

2025-11-19 05:41:08 +10:00 · 2025-09-16 01:06:34 +01:00
parent 8816134c7c 04ae1d7270
commit e6bb54b278
55 changed files with 1390 additions and 432 deletions
--- a/apps/server/src/integrations/import/import.module.ts
+++ b/apps/server/src/integrations/import/import.module.ts
@ -2,7 +2,7 @@ import { Module } from '@nestjs/common';
 import { ImportService } from './services/import.service';
 import { ImportController } from './import.controller';
 import { StorageModule } from '../storage/storage.module';
-import { FileTaskService } from './services/file-task.service';
+import { FileImportTaskService } from './services/file-import-task.service';
 import { FileTaskProcessor } from './processors/file-task.processor';
 import { ImportAttachmentService } from './services/import-attachment.service';
 import { FileTaskController } from './file-task.controller';
@ -11,7 +11,7 @@ import { PageModule } from '../../core/page/page.module';
@Module({
  providers: [
    ImportService,
-    FileTaskService,
+    FileImportTaskService,
    FileTaskProcessor,
    ImportAttachmentService,
  ],
--- a/apps/server/src/integrations/import/processors/file-task.processor.ts
+++ b/apps/server/src/integrations/import/processors/file-task.processor.ts
@ -2,7 +2,7 @@ import { Logger, OnModuleDestroy } from '@nestjs/common';
 import { OnWorkerEvent, Processor, WorkerHost } from '@nestjs/bullmq';
 import { Job } from 'bullmq';
 import { QueueJob, QueueName } from 'src/integrations/queue/constants';
-import { FileTaskService } from '../services/file-task.service';
+import { FileImportTaskService } from '../services/file-import-task.service';
 import { FileTaskStatus } from '../utils/file.utils';
 import { StorageService } from '../../storage/storage.service';

@ -11,7 +11,7 @@ export class FileTaskProcessor extends WorkerHost implements OnModuleDestroy {
  private readonly logger = new Logger(FileTaskProcessor.name);

  constructor(
-    private readonly fileTaskService: FileTaskService,
+    private readonly fileTaskService: FileImportTaskService,
    private readonly storageService: StorageService,
  ) {
    super();
@ -41,15 +41,32 @@ export class FileTaskProcessor extends WorkerHost implements OnModuleDestroy {
  @OnWorkerEvent('failed')
  async onFailed(job: Job) {
    this.logger.error(
-      `Error processing ${job.name} job. Reason: ${job.failedReason}`,
+      `Error processing ${job.name} job. Import Task ID: ${job.data.fileTaskId}. Reason: ${job.failedReason}`,
    );

+    await this.handleFailedJob(job);
+  }
+
+  @OnWorkerEvent('stalled')
+  async onStalled(job: Job) {
+    this.logger.error(
+      `Job ${job.name} stalled. . Import Task ID: ${job.data.fileTaskId}.. Job ID: ${job.id}`,
+    );
+
+    // Set failedReason for stalled jobs since it's not automatically set
+    job.failedReason = 'Job stalled and was marked as failed';
+    await this.handleFailedJob(job);
+  }
+
+  private async handleFailedJob(job: Job) {
    try {
      const fileTaskId = job.data.fileTaskId;
+      const reason = job.failedReason || 'Unknown error';
+
      await this.fileTaskService.updateTaskStatus(
        fileTaskId,
        FileTaskStatus.Failed,
-        job.failedReason,
+        reason,
      );

      const fileTask = await this.fileTaskService.getFileTask(fileTaskId);
@ -62,10 +79,22 @@ export class FileTaskProcessor extends WorkerHost implements OnModuleDestroy {
  }

  @OnWorkerEvent('completed')
-  onCompleted(job: Job) {
+  async onCompleted(job: Job) {
    this.logger.log(
      `Completed ${job.name} job for File task ID ${job.data.fileTaskId}`,
    );
+
+    try {
+      const fileTask = await this.fileTaskService.getFileTask(
+        job.data.fileTaskId,
+      );
+      if (fileTask) {
+        await this.storageService.delete(fileTask.filePath);
+        this.logger.debug(`Deleted imported zip file: ${fileTask.filePath}`);
+      }
+    } catch (err) {
+      this.logger.error(`Failed to delete imported zip file:`, err);
+    }
  }

  async onModuleDestroy(): Promise<void> {
--- a/apps/server/src/integrations/import/services/file-import-task.service.ts
+++ b/apps/server/src/integrations/import/services/file-import-task.service.ts
@ -33,8 +33,8 @@ import { PageService } from '../../../core/page/services/page.service';
 import { ImportPageNode } from '../dto/file-task-dto';

@Injectable()
-export class FileTaskService {
-  private readonly logger = new Logger(FileTaskService.name);
+export class FileImportTaskService {
+  private readonly logger = new Logger(FileImportTaskService.name);

  constructor(
    private readonly storageService: StorageService,
@ -266,7 +266,7 @@ export class FileTaskService {
            attachmentCandidates,
          });

-        const { html, backlinks } = await formatImportHtml({
+        const { html, backlinks, pageIcon } = await formatImportHtml({
          html: htmlContent,
          currentFilePath: page.filePath,
          filePathToPageMetaMap: filePathToPageMetaMap,
@ -286,6 +286,7 @@ export class FileTaskService {
          id: page.id,
          slugId: page.slugId,
          title: title || page.name,
+          icon: pageIcon || null,
          content: prosemirrorJson,
          textContent: jsonToText(prosemirrorJson),
          ydoc: await this.importService.createYdoc(prosemirrorJson),
--- a/apps/server/src/integrations/import/services/import-attachment.service.ts
+++ b/apps/server/src/integrations/import/services/import-attachment.service.ts
@ -35,7 +35,7 @@ interface DrawioPair {
@Injectable()
 export class ImportAttachmentService {
  private readonly logger = new Logger(ImportAttachmentService.name);
-  private readonly CONCURRENT_UPLOADS = 3;
+  private readonly CONCURRENT_UPLOADS = 1;
  private readonly MAX_RETRIES = 2;
  private readonly RETRY_DELAY = 2000;

@ -53,6 +53,7 @@ export class ImportAttachmentService {
    fileTask: FileTask;
    attachmentCandidates: Map<string, string>;
    pageAttachments?: AttachmentInfo[];
+    isConfluenceImport?: boolean;
  }): Promise<string> {
    const {
      html,
@ -62,6 +63,7 @@ export class ImportAttachmentService {
      fileTask,
      attachmentCandidates,
      pageAttachments = [],
+      isConfluenceImport,
    } = opts;

    const attachmentTasks: (() => Promise<void>)[] = [];
@ -90,7 +92,10 @@ export class ImportAttachmentService {
    >();

    // Analyze attachments to identify Draw.io pairs
-    const { drawioPairs, skipFiles } = this.analyzeAttachments(pageAttachments);
+    const { drawioPairs, skipFiles } = this.analyzeAttachments(
+      pageAttachments,
+      isConfluenceImport,
+    );

    // Map to store processed Draw.io SVGs
    const drawioSvgMap = new Map<
@ -134,7 +139,9 @@ export class ImportAttachmentService {
            const stream = Readable.from(svgBuffer);

            // Upload to storage
-            await this.storageService.uploadStream(storageFilePath, stream);
+            await this.storageService.uploadStream(storageFilePath, stream, {
+              recreateClient: true,
+            });

            // Insert into database
            await this.db
@ -235,202 +242,197 @@ export class ImportAttachmentService {
    const pageDir = path.dirname(pageRelativePath);
    const $ = load(html);

-    // Cache for resolved paths to avoid repeated lookups
-    const resolvedPathCache = new Map<string, string | null>();
+    // image
+    for (const imgEl of $('img').toArray()) {
+      const $img = $(imgEl);
+      const src = cleanUrlString($img.attr('src') ?? '')!;
+      if (!src || src.startsWith('http')) continue;

-    const getCachedResolvedPath = (rawPath: string): string | null => {
-      if (resolvedPathCache.has(rawPath)) {
-        return resolvedPathCache.get(rawPath)!;
-      }
-      const resolved = resolveRelativeAttachmentPath(
-        rawPath,
+      const relPath = resolveRelativeAttachmentPath(
+        src,
        pageDir,
        attachmentCandidates,
      );
-      resolvedPathCache.set(rawPath, resolved);
-      return resolved;
-    };
+      if (!relPath) continue;

-    // Cache for file stats to avoid repeated file system calls
-    const statCache = new Map<string, any>();
+      // Check if this image is part of a Draw.io pair
+      const drawioSvg = drawioSvgMap.get(relPath);
+      if (drawioSvg) {
+        const $drawio = $('<div>')
+          .attr('data-type', 'drawio')
+          .attr('data-src', drawioSvg.apiFilePath)
+          .attr('data-title', 'diagram')
+          .attr('data-width', '100%')
+          .attr('data-align', 'center')
+          .attr('data-attachment-id', drawioSvg.attachmentId);

-    const getCachedStat = async (absPath: string) => {
-      if (statCache.has(absPath)) {
-        return statCache.get(absPath);
+        $img.replaceWith($drawio);
+        unwrapFromParagraph($, $drawio);
+        continue;
      }
-      const stat = await fs.stat(absPath);
-      statCache.set(absPath, stat);
-      return stat;
-    };

-    // Single DOM traversal for all attachment elements
-    const selector =
-      'img, video, div[data-type="attachment"], a, div[data-type="excalidraw"], div[data-type="drawio"]';
-    const elements = $(selector).toArray();
+      const { attachmentId, apiFilePath } = processFile(relPath);

-    for (const element of elements) {
-      const $el = $(element);
-      const tagName = element.tagName.toLowerCase();
+      const width = $img.attr('width') ?? '100%';
+      const align = $img.attr('data-align') ?? 'center';

-      // Process based on element type
-      if (tagName === 'img') {
-        const src = cleanUrlString($el.attr('src') ?? '');
-        if (!src || src.startsWith('http')) continue;
+      $img
+        .attr('src', apiFilePath)
+        .attr('data-attachment-id', attachmentId)
+        .attr('width', width)
+        .attr('data-align', align);

-        const relPath = getCachedResolvedPath(src);
-        if (!relPath) continue;
+      unwrapFromParagraph($, $img);
+    }

-        // Check if this image is part of a Draw.io pair
-        const drawioSvg = drawioSvgMap.get(relPath);
-        if (drawioSvg) {
-          const $drawio = $('<div>')
-            .attr('data-type', 'drawio')
-            .attr('data-src', drawioSvg.apiFilePath)
-            .attr('data-title', 'diagram')
-            .attr('data-width', '100%')
-            .attr('data-align', 'center')
-            .attr('data-attachment-id', drawioSvg.attachmentId);
+    // video
+    for (const vidEl of $('video').toArray()) {
+      const $vid = $(vidEl);
+      const src = cleanUrlString($vid.attr('src') ?? '')!;
+      if (!src || src.startsWith('http')) continue;

-          $el.replaceWith($drawio);
-          unwrapFromParagraph($, $drawio);
-          continue;
-        }
+      const relPath = resolveRelativeAttachmentPath(
+        src,
+        pageDir,
+        attachmentCandidates,
+      );
+      if (!relPath) continue;

-        const { attachmentId, apiFilePath, abs } = processFile(relPath);
-        const stat = await getCachedStat(abs);
+      const { attachmentId, apiFilePath } = processFile(relPath);

-        $el
+      const width = $vid.attr('width') ?? '100%';
+      const align = $vid.attr('data-align') ?? 'center';
+
+      $vid
+        .attr('src', apiFilePath)
+        .attr('data-attachment-id', attachmentId)
+        .attr('width', width)
+        .attr('data-align', align);
+
+      unwrapFromParagraph($, $vid);
+    }
+
+    // <div data-type="attachment">
+    for (const el of $('div[data-type="attachment"]').toArray()) {
+      const $oldDiv = $(el);
+      const rawUrl = cleanUrlString($oldDiv.attr('data-attachment-url') ?? '')!;
+      if (!rawUrl || rawUrl.startsWith('http')) continue;
+
+      const relPath = resolveRelativeAttachmentPath(
+        rawUrl,
+        pageDir,
+        attachmentCandidates,
+      );
+      if (!relPath) continue;
+
+      const { attachmentId, apiFilePath, abs } = processFile(relPath);
+      const fileName = path.basename(abs);
+      const mime = getMimeType(abs);
+
+      const $newDiv = $('<div>')
+        .attr('data-type', 'attachment')
+        .attr('data-attachment-url', apiFilePath)
+        .attr('data-attachment-name', fileName)
+        .attr('data-attachment-mime', mime)
+        .attr('data-attachment-id', attachmentId);
+
+      $oldDiv.replaceWith($newDiv);
+      unwrapFromParagraph($, $newDiv);
+    }
+
+    // rewrite other attachments via <a>
+    for (const aEl of $('a').toArray()) {
+      const $a = $(aEl);
+      const href = cleanUrlString($a.attr('href') ?? '')!;
+      if (!href || href.startsWith('http')) continue;
+
+      const relPath = resolveRelativeAttachmentPath(
+        href,
+        pageDir,
+        attachmentCandidates,
+      );
+      if (!relPath) continue;
+
+      // Check if this is a Draw.io file
+      const drawioSvg = drawioSvgMap.get(relPath);
+      if (drawioSvg) {
+        const $drawio = $('<div>')
+          .attr('data-type', 'drawio')
+          .attr('data-src', drawioSvg.apiFilePath)
+          .attr('data-title', 'diagram')
+          .attr('data-width', '100%')
+          .attr('data-align', 'center')
+          .attr('data-attachment-id', drawioSvg.attachmentId);
+
+        $a.replaceWith($drawio);
+        unwrapFromParagraph($, $drawio);
+        continue;
+      }
+
+      // Skip files that should be ignored
+      if (skipFiles.has(relPath)) {
+        $a.remove();
+        continue;
+      }
+
+      const { attachmentId, apiFilePath, abs } = processFile(relPath);
+      const ext = path.extname(relPath).toLowerCase();
+
+      if (ext === '.mp4') {
+        const $video = $('<video>')
          .attr('src', apiFilePath)
          .attr('data-attachment-id', attachmentId)
-          .attr('data-size', stat.size.toString())
-          .attr('width', $el.attr('width') ?? '100%')
-          .attr('data-align', $el.attr('data-align') ?? 'center');
+          .attr('width', '100%')
+          .attr('data-align', 'center');
+        $a.replaceWith($video);
+        unwrapFromParagraph($, $video);
+      } else {
+        const confAliasName = $a.attr('data-linked-resource-default-alias');
+        let attachmentName = path.basename(abs);
+        if (confAliasName) attachmentName = confAliasName;

-        unwrapFromParagraph($, $el);
-      } else if (tagName === 'video') {
-        const src = cleanUrlString($el.attr('src') ?? '');
-        if (!src || src.startsWith('http')) continue;
+        const $div = $('<div>')
+          .attr('data-type', 'attachment')
+          .attr('data-attachment-url', apiFilePath)
+          .attr('data-attachment-name', attachmentName)
+          .attr('data-attachment-mime', getMimeType(abs))
+          .attr('data-attachment-id', attachmentId);

-        const relPath = getCachedResolvedPath(src);
+        $a.replaceWith($div);
+        unwrapFromParagraph($, $div);
+      }
+    }
+
+    // excalidraw and drawio
+    for (const type of ['excalidraw', 'drawio'] as const) {
+      for (const el of $(`div[data-type="${type}"]`).toArray()) {
+        const $oldDiv = $(el);
+        const rawSrc = cleanUrlString($oldDiv.attr('data-src') ?? '')!;
+        if (!rawSrc || rawSrc.startsWith('http')) continue;
+
+        const relPath = resolveRelativeAttachmentPath(
+          rawSrc,
+          pageDir,
+          attachmentCandidates,
+        );
        if (!relPath) continue;

        const { attachmentId, apiFilePath, abs } = processFile(relPath);
-        const stat = await getCachedStat(abs);
+        const fileName = path.basename(abs);

-        $el
-          .attr('src', apiFilePath)
-          .attr('data-attachment-id', attachmentId)
-          .attr('data-size', stat.size.toString())
-          .attr('width', $el.attr('width') ?? '100%')
-          .attr('data-align', $el.attr('data-align') ?? 'center');
+        const width = $oldDiv.attr('data-width') || '100%';
+        const align = $oldDiv.attr('data-align') || 'center';

-        unwrapFromParagraph($, $el);
-      } else if (tagName === 'div') {
-        const dataType = $el.attr('data-type');
+        const $newDiv = $('<div>')
+          .attr('data-type', type)
+          .attr('data-src', apiFilePath)
+          .attr('data-title', fileName)
+          .attr('data-width', width)
+          .attr('data-align', align)
+          .attr('data-attachment-id', attachmentId);

-        if (dataType === 'attachment') {
-          const rawUrl = cleanUrlString($el.attr('data-attachment-url') ?? '');
-          if (!rawUrl || rawUrl.startsWith('http')) continue;
-
-          const relPath = getCachedResolvedPath(rawUrl);
-          if (!relPath) continue;
-
-          const { attachmentId, apiFilePath, abs } = processFile(relPath);
-          const stat = await getCachedStat(abs);
-          const fileName = path.basename(abs);
-          const mime = getMimeType(abs);
-
-          const $newDiv = $('<div>')
-            .attr('data-type', 'attachment')
-            .attr('data-attachment-url', apiFilePath)
-            .attr('data-attachment-name', fileName)
-            .attr('data-attachment-mime', mime)
-            .attr('data-attachment-size', stat.size.toString())
-            .attr('data-attachment-id', attachmentId);
-
-          $el.replaceWith($newDiv);
-          unwrapFromParagraph($, $newDiv);
-        } else if (dataType === 'excalidraw' || dataType === 'drawio') {
-          const rawSrc = cleanUrlString($el.attr('data-src') ?? '');
-          if (!rawSrc || rawSrc.startsWith('http')) continue;
-
-          const relPath = getCachedResolvedPath(rawSrc);
-          if (!relPath) continue;
-
-          const { attachmentId, apiFilePath, abs } = processFile(relPath);
-          const stat = await getCachedStat(abs);
-          const fileName = path.basename(abs);
-
-          const $newDiv = $('<div>')
-            .attr('data-type', dataType)
-            .attr('data-src', apiFilePath)
-            .attr('data-title', fileName)
-            .attr('data-width', $el.attr('data-width') || '100%')
-            .attr('data-size', stat.size.toString())
-            .attr('data-align', $el.attr('data-align') || 'center')
-            .attr('data-attachment-id', attachmentId);
-
-          $el.replaceWith($newDiv);
-          unwrapFromParagraph($, $newDiv);
-        }
-      } else if (tagName === 'a') {
-        const href = cleanUrlString($el.attr('href') ?? '');
-        if (!href || href.startsWith('http')) continue;
-
-        const relPath = getCachedResolvedPath(href);
-        if (!relPath) continue;
-
-        // Check if this is a Draw.io file
-        const drawioSvg = drawioSvgMap.get(relPath);
-        if (drawioSvg) {
-          const $drawio = $('<div>')
-            .attr('data-type', 'drawio')
-            .attr('data-src', drawioSvg.apiFilePath)
-            .attr('data-title', 'diagram')
-            .attr('data-width', '100%')
-            .attr('data-align', 'center')
-            .attr('data-attachment-id', drawioSvg.attachmentId);
-
-          $el.replaceWith($drawio);
-          unwrapFromParagraph($, $drawio);
-          continue;
-        }
-
-        // Skip files that should be ignored
-        if (skipFiles.has(relPath)) {
-          $el.remove();
-          continue;
-        }
-
-        const { attachmentId, apiFilePath, abs } = processFile(relPath);
-        const stat = await getCachedStat(abs);
-        const ext = path.extname(relPath).toLowerCase();
-
-        if (ext === '.mp4') {
-          const $video = $('<video>')
-            .attr('src', apiFilePath)
-            .attr('data-attachment-id', attachmentId)
-            .attr('data-size', stat.size.toString())
-            .attr('width', '100%')
-            .attr('data-align', 'center');
-          $el.replaceWith($video);
-          unwrapFromParagraph($, $video);
-        } else {
-          const confAliasName = $el.attr('data-linked-resource-default-alias');
-          let attachmentName = path.basename(abs);
-          if (confAliasName) attachmentName = confAliasName;
-
-          const $div = $('<div>')
-            .attr('data-type', 'attachment')
-            .attr('data-attachment-url', apiFilePath)
-            .attr('data-attachment-name', attachmentName)
-            .attr('data-attachment-mime', getMimeType(abs))
-            .attr('data-attachment-size', stat.size.toString())
-            .attr('data-attachment-id', attachmentId);
-
-          $el.replaceWith($div);
-          unwrapFromParagraph($, $div);
-        }
+        $oldDiv.replaceWith($newDiv);
+        unwrapFromParagraph($, $newDiv);
      }
    }

@ -492,24 +494,17 @@ export class ImportAttachmentService {

      // This attachment was in the list but not referenced in HTML - add it
      const { attachmentId, apiFilePath, abs } = processFile(href);
+      const mime = mimeType || getMimeType(abs);

-      try {
-        const stat = await fs.stat(abs);
-        const mime = mimeType || getMimeType(abs);
+      // Add as attachment node at the end
+      const $attachmentDiv = $('<div>')
+        .attr('data-type', 'attachment')
+        .attr('data-attachment-url', apiFilePath)
+        .attr('data-attachment-name', fileName)
+        .attr('data-attachment-mime', mime)
+        .attr('data-attachment-id', attachmentId);

-        // Add as attachment node at the end
-        const $attachmentDiv = $('<div>')
-          .attr('data-type', 'attachment')
-          .attr('data-attachment-url', apiFilePath)
-          .attr('data-attachment-name', fileName)
-          .attr('data-attachment-mime', mime)
-          .attr('data-attachment-size', stat.size.toString())
-          .attr('data-attachment-id', attachmentId);
-
-        $.root().append($attachmentDiv);
-      } catch (error) {
-        this.logger.error(`Failed to process attachment ${fileName}:`, error);
-      }
+      $.root().append($attachmentDiv);
    }

    // wait for all uploads & DB inserts
@ -534,16 +529,49 @@ export class ImportAttachmentService {
      }
    }

+    // Post-process DOM elements to add file sizes after uploads complete
+    // This avoids blocking file operations during initial DOM processing
+    const elementsNeedingSize = $('[data-attachment-id]:not([data-size])');
+    for (const element of elementsNeedingSize.toArray()) {
+      const $el = $(element);
+      const attachmentId = $el.attr('data-attachment-id');
+      if (!attachmentId) continue;
+
+      // Find the corresponding processed file info
+      const processedEntry = Array.from(processed.values()).find(
+        (entry) => entry.attachmentId === attachmentId,
+      );
+
+      if (processedEntry) {
+        try {
+          const stat = await fs.stat(processedEntry.abs);
+          $el.attr('data-size', stat.size.toString());
+        } catch (error) {
+          this.logger.debug(
+            `Could not get size for ${processedEntry.abs}:`,
+            error,
+          );
+        }
+      }
+    }
+
    return $.root().html() || '';
  }

-  private analyzeAttachments(attachments: AttachmentInfo[]): {
+  private analyzeAttachments(
+    attachments: AttachmentInfo[],
+    isConfluenceImport?: boolean,
+  ): {
    drawioPairs: Map<string, DrawioPair>;
    skipFiles: Set<string>;
  } {
    const drawioPairs = new Map<string, DrawioPair>();
    const skipFiles = new Set<string>();

+    if (!isConfluenceImport) {
+      return { drawioPairs, skipFiles };
+    }
+
    // Group attachments by type
    const drawioFiles: AttachmentInfo[] = [];
    const pngByBaseName = new Map<string, AttachmentInfo[]>();
@ -776,7 +804,10 @@ export class ImportAttachmentService {
    for (let attempt = 1; attempt <= this.MAX_RETRIES; attempt++) {
      try {
        const fileStream = createReadStream(abs);
-        await this.storageService.uploadStream(storageFilePath, fileStream);
+        await this.storageService.uploadStream(storageFilePath, fileStream, {
+          recreateClient: true,
+        });
+
        const stat = await fs.stat(abs);

        await this.db
@ -807,7 +838,7 @@ export class ImportAttachmentService {
                attempts: 1,
                backoff: {
                  type: 'exponential',
-                  delay: 30 * 1000,
+                  delay: 3 * 60 * 1000,
                },
                deduplication: {
                  id: attachmentId,
--- a/apps/server/src/integrations/import/utils/import-formatter.ts
+++ b/apps/server/src/integrations/import/utils/import-formatter.ts
@ -4,6 +4,11 @@ import { v7 } from 'uuid';
 import { InsertableBacklink } from '@docmost/db/types/entity.types';
 import { Cheerio, CheerioAPI, load } from 'cheerio';

+// Check if text contains Unicode characters (for emojis/icons)
+function isUnicodeCharacter(text: string): boolean {
+  return text.length > 0 && text.codePointAt(0)! > 127; // Non-ASCII characters
+}
+
 export async function formatImportHtml(opts: {
  html: string;
  currentFilePath: string;
@ -16,7 +21,11 @@ export async function formatImportHtml(opts: {
  workspaceId: string;
  pageDir?: string;
  attachmentCandidates?: string[];
-}): Promise<{ html: string; backlinks: InsertableBacklink[] }> {
+}): Promise<{
+  html: string;
+  backlinks: InsertableBacklink[];
+  pageIcon?: string;
+}> {
  const {
    html,
    currentFilePath,
@ -28,6 +37,17 @@ export async function formatImportHtml(opts: {
  const $: CheerioAPI = load(html);
  const $root: Cheerio<any> = $.root();

+  let pageIcon: string | null = null;
+  // extract notion page icon
+  const headerIconSpan = $root.find('header .page-header-icon .icon');
+
+  if (headerIconSpan.length > 0) {
+    const iconText = headerIconSpan.text().trim();
+    if (iconText && isUnicodeCharacter(iconText)) {
+      pageIcon = iconText;
+    }
+  }
+
  notionFormatter($, $root);
  defaultHtmlFormatter($, $root);

@ -44,6 +64,7 @@ export async function formatImportHtml(opts: {
  return {
    html: $root.html() || '',
    backlinks,
+    pageIcon: pageIcon || undefined,
  };
 }

@ -69,6 +90,10 @@ export function defaultHtmlFormatter($: CheerioAPI, $root: Cheerio<any>) {
 }

 export function notionFormatter($: CheerioAPI, $root: Cheerio<any>) {
+  // remove page header icon and cover image
+  $root.find('.page-header-icon').remove();
+  $root.find('.page-cover-image').remove();
+
  // remove empty description paragraphs
  $root.find('p.page-description').each((_, el) => {
    if (!$(el).text().trim()) $(el).remove();
@ -189,6 +214,9 @@ export function notionFormatter($: CheerioAPI, $root: Cheerio<any>) {
      $fig.replaceWith($newAnchor);
    });

+  // remove user icons
+  $root.find('span.user img.user-icon').remove();
+
  // remove toc
  $root.find('nav.table_of_contents').remove();
 }
--- a/apps/server/src/integrations/storage/drivers/local.driver.ts
+++ b/apps/server/src/integrations/storage/drivers/local.driver.ts
@ -28,7 +28,7 @@ export class LocalDriver implements StorageDriver {
    }
  }

-  async uploadStream(filePath: string, file: Readable): Promise<void> {
+  async uploadStream(filePath: string, file: Readable, options?: { recreateClient?: boolean }): Promise<void> {
    try {
      const fullPath = this._fullPath(filePath);
      await fs.mkdir(dirname(fullPath), { recursive: true });
--- a/apps/server/src/integrations/storage/drivers/s3.driver.ts
+++ b/apps/server/src/integrations/storage/drivers/s3.driver.ts
@ -41,12 +41,26 @@ export class S3Driver implements StorageDriver {
    }
  }

-  async uploadStream(filePath: string, file: Readable): Promise<void> {
+  async uploadStream(
+    filePath: string,
+    file: Readable,
+    options?: { recreateClient?: boolean },
+  ): Promise<void> {
+    let clientToUse = this.s3Client;
+    let shouldDestroyClient = false;
+
+    // optionally recreate client to avoid socket hang errors
+    // (during multi-attachments imports)
+    if (options?.recreateClient) {
+      clientToUse = new S3Client(this.config as any);
+      shouldDestroyClient = true;
+    }
+
    try {
      const contentType = getMimeType(filePath);

      const upload = new Upload({
-        client: this.s3Client,
+        client: clientToUse,
        params: {
          Bucket: this.config.bucket,
          Key: filePath,
@ -58,6 +72,10 @@ export class S3Driver implements StorageDriver {
      await upload.done();
    } catch (err) {
      throw new Error(`Failed to upload file: ${(err as Error).message}`);
+    } finally {
+      if (shouldDestroyClient && clientToUse) {
+        clientToUse.destroy();
+      }
    }
  }

--- a/apps/server/src/integrations/storage/interfaces/storage-driver.interface.ts
+++ b/apps/server/src/integrations/storage/interfaces/storage-driver.interface.ts
@ -3,7 +3,7 @@ import { Readable } from 'stream';
 export interface StorageDriver {
  upload(filePath: string, file: Buffer): Promise<void>;

-  uploadStream(filePath: string, file: Readable): Promise<void>;
+  uploadStream(filePath: string, file: Readable, options?: { recreateClient?: boolean }): Promise<void>;

  copy(fromFilePath: string, toFilePath: string): Promise<void>;

--- a/apps/server/src/integrations/storage/storage.service.ts
+++ b/apps/server/src/integrations/storage/storage.service.ts
@ -15,8 +15,8 @@ export class StorageService {
    this.logger.debug(`File uploaded successfully. Path: ${filePath}`);
  }

-  async uploadStream(filePath: string, fileContent: Readable) {
-    await this.storageDriver.uploadStream(filePath, fileContent);
+  async uploadStream(filePath: string, fileContent: Readable, options?: { recreateClient?: boolean }) {
+    await this.storageDriver.uploadStream(filePath, fileContent, options);
    this.logger.debug(`File uploaded successfully. Path: ${filePath}`);
  }