From 97412400ff50a8c1227733bb919f81607da365de Mon Sep 17 00:00:00 2001
From: Anna Khismatullina <anna.khismatullina@gmail.com>
Date: Fri, 8 Nov 2024 12:41:18 +0700
Subject: [PATCH] Fixes for Notion import (#7119)

Signed-off-by: Anna Khismatullina <anna.khismatullina@gmail.com>
---
 dev/import-tool/src/importer/uploader.ts | 10 ++++-
 dev/import-tool/src/index.ts             |  2 +-
 dev/import-tool/src/notion/notion.ts     | 47 +++++++++++++++++-------
 3 files changed, 44 insertions(+), 15 deletions(-)

diff --git a/dev/import-tool/src/importer/uploader.ts b/dev/import-tool/src/importer/uploader.ts
index 2edf7b2f25..64269f1176 100644
--- a/dev/import-tool/src/importer/uploader.ts
+++ b/dev/import-tool/src/importer/uploader.ts
@@ -24,6 +24,7 @@ import {
 export interface FileUploader {
   uploadFile: (id: Ref<Doc>, name: string, file: File, contentType?: string) => Promise<Response>
   uploadCollaborativeDoc: (id: Ref<Doc>, collabId: CollaborativeDoc, data: Buffer) => Promise<Response>
+  getFileUrl: (id: string) => string
 }
 
 export interface UploadResult {
@@ -34,8 +35,11 @@ export interface UploadResult {
 export class FrontFileUploader implements FileUploader {
   constructor (
     private readonly frontUrl: string,
+    private readonly workspaceId: string,
     private readonly token: string
-  ) {}
+  ) {
+    this.getFileUrl = this.getFileUrl.bind(this)
+  }
 
   public async uploadFile (id: Ref<Doc>, name: string, file: File, contentType?: string): Promise<Response> {
     const form = new FormData()
@@ -55,6 +59,10 @@ export class FrontFileUploader implements FileUploader {
     })
   }
 
+  public getFileUrl (id: string): string {
+    return concatLink(this.frontUrl, `/files/${this.workspaceId}/${id}?file=${id}&workspace=${this.workspaceId}`)
+  }
+
   public async uploadCollaborativeDoc (id: Ref<Doc>, collabId: CollaborativeDoc, data: Buffer): Promise<Response> {
     const file = new File([data], collabId)
     const { documentId } = collaborativeDocParse(collabId)
diff --git a/dev/import-tool/src/index.ts b/dev/import-tool/src/index.ts
index cf8ab5f456..df3759fa5d 100644
--- a/dev/import-tool/src/index.ts
+++ b/dev/import-tool/src/index.ts
@@ -78,7 +78,7 @@ export function importTool (): void {
       return
     }
     const client = new TxOperations(connection, acc._id)
-    const fileUploader = new FrontFileUploader(getFrontUrl(), selectedWs.token)
+    const fileUploader = new FrontFileUploader(getFrontUrl(), selectedWs.workspaceId, selectedWs.token)
     try {
       await f(client, fileUploader)
     } catch (err: any) {
diff --git a/dev/import-tool/src/notion/notion.ts b/dev/import-tool/src/notion/notion.ts
index 1ac2aedb0f..7af6e37082 100644
--- a/dev/import-tool/src/notion/notion.ts
+++ b/dev/import-tool/src/notion/notion.ts
@@ -409,7 +409,7 @@ async function importAttachment (
   }
 
   const file = new File([data], docMeta.name)
-  await fileUploader.uploadFile(docMeta.id as Ref<Doc>, docMeta.name, file)
+  await fileUploader.uploadFile(docMeta.id as Ref<Doc>, docMeta.id, file)
 
   const attachedData: AttachedData<Attachment> = {
     file: docMeta.id as Ref<Blob>,
@@ -442,7 +442,7 @@ async function importPageDocument (
   const md = data.toString() ?? ''
   const json = parseMessageMarkdown(md ?? '', 'image://')
   if (documentMetaMap !== undefined) {
-    preProcessMarkdown(json, documentMetaMap)
+    preProcessMarkdown(json, documentMetaMap, fileUploader)
   }
   const yDoc = jsonToYDocNoSchema(json, 'content')
   const buffer = yDocToBuffer(yDoc)
@@ -472,7 +472,11 @@ async function importPageDocument (
   await client.createDoc(document.class.Document, space, attachedData, id)
 }
 
-function preProcessMarkdown (json: MarkupNode, documentMetaMap: Map<string, DocumentMetadata>): void {
+function preProcessMarkdown (
+  json: MarkupNode,
+  documentMetaMap: Map<string, DocumentMetadata>,
+  fileUploader: FileUploader
+): void {
   traverseNode(json, (node) => {
     if (node.type === MarkupNodeType.image) {
       const src = node.attrs?.src
@@ -480,7 +484,7 @@ function preProcessMarkdown (json: MarkupNode, documentMetaMap: Map<string, Docu
         const notionId = getFileId('', src as string)
         const meta = documentMetaMap.get(notionId)
         if (meta !== undefined) {
-          alterImageNode(node, meta)
+          alterImageNode(node, meta, fileUploader.getFileUrl(meta.id))
         }
       }
     } else {
@@ -561,12 +565,21 @@ function alterInternalLinkNode (node: MarkupNode, targetMeta: DocumentMetadata):
   }
 }
 
-function alterImageNode (node: MarkupNode, meta: DocumentMetadata): void {
+function alterImageNode (node: MarkupNode, meta: DocumentMetadata, fileUrl: string): void {
   node.type = MarkupNodeType.image
   if (node.attrs !== undefined) {
-    node.attrs['file-id'] = meta.id
-    if (meta.mimeType !== undefined) {
-      node.attrs['data-file-type'] = meta.mimeType
+    node.attrs = {
+      'file-id': meta.id,
+      src: fileUrl,
+      width: node.attrs.width ?? null,
+      height: node.attrs.height ?? null,
+      align: node.attrs.align ?? null,
+      alt: meta.name,
+      title: meta.name
+    }
+    const mimeType = getContentType(meta.name)
+    if (mimeType !== undefined) {
+      node.attrs['data-file-type'] = mimeType
     }
   }
 }
@@ -580,19 +593,27 @@ function isExternalLink (href: any): boolean {
   return URL.canParse(href)
 }
 
+function safeDecodeURI (uri: string): string {
+  try {
+    return decodeURI(uri)
+  } catch (e) {
+    return uri
+  }
+}
+
 function extractNotionId (fileName: string): string | undefined {
-  const decoded = decodeURI(fileName).trimEnd()
+  const decoded = safeDecodeURI(fileName).trimEnd()
   const matched = decoded.match(/ ([\w\d]{32}(_all)?)(\.|$)/)
   return matched !== null && matched.length >= 2 ? matched[1] : undefined
 }
 
 function extractExtension (fileName: string): string {
-  const decoded = decodeURI(fileName)
+  const decoded = safeDecodeURI(fileName)
   return parse(decoded).ext.toLowerCase()
 }
 
 function extractNameWoExtension (fileName: string): string {
-  const decoded = decodeURI(fileName)
+  const decoded = safeDecodeURI(fileName)
   return parse(decoded).name
 }
 
@@ -608,8 +629,8 @@ function getFileId (filePath: string, fileName: string): string {
   if (notionId !== '' && notionId !== undefined) {
     return notionId
   }
-  const decodedPath = decodeURI(filePath)
-  const decodedName = decodeURI(fileName)
+  const decodedPath = safeDecodeURI(filePath)
+  const decodedName = safeDecodeURI(fileName)
   return join(basename(decodedPath), decodedName)
 }