Fix internal links and images replace.

longbridge · Nov 13, 2023 · e3bafb2 · e3bafb2
1 parent 0bc9b5d
commit e3bafb2
Show file tree

Hide file tree

Showing 7 changed files with 172 additions and 94 deletions.
diff --git a/feishu-pages/package.json b/feishu-pages/package.json
@@ -24,7 +24,7 @@
     "@types/node": "^20.5.7",
     "axios": "^1.5.0",
     "dotenv": "^16.3.1",
-    "feishu-docx": "0.5.2",
+    "feishu-docx": "0.5.4",
     "mime-types": "^2.1.35",
     "typescript": "^5.2.2"
   },
@@ -33,4 +33,4 @@
     "jest": "^29.6.4",
     "ts-jest": "^29.1.1"
   }
-}
+}
diff --git a/feishu-pages/src/doc.ts b/feishu-pages/src/doc.ts
@@ -82,16 +82,19 @@ export const generateFileMeta = (
     sidebar_position: position,
   };
 
-  // Replace double quote to avoid YAML parse error
-  meta.title = meta.title.replace(/"/g, '\\"');
-
   let output = `---\n`;
   for (const key in meta) {
-    const val = meta[key];
+    let val = meta[key];
     if (val === null || val === undefined) {
       continue;
     }
-    output += `${key}: "${val}"\n`;
+
+    // Replace double quote to avoid YAML parse error
+    if (typeof val === 'string') {
+      val = val.replace(/"/g, '\\"');
+      val = `"${val}"`;
+    }
+    output += `${key}: ${val}\n`;
   }
   output += `---\n`;
 

diff --git a/feishu-pages/src/feishu.ts b/feishu-pages/src/feishu.ts
@@ -218,7 +218,9 @@ export const feishuDownload = async (fileToken: string, localPath: string) => {
   fs.mkdirSync(CACHE_DIR, { recursive: true });
 
   let res: any = {};
+  let hasCache = false;
   if (fs.existsSync(cacheFilePath) && fs.existsSync(cacheFileMetaPath)) {
+    hasCache = true;
     res.data = fs.readFileSync(cacheFilePath);
     res.headers = JSON.parse(fs.readFileSync(cacheFileMetaPath, 'utf-8'));
     console.info(' -> Cache hit:', fileToken);
@@ -261,18 +263,22 @@ export const feishuDownload = async (fileToken: string, localPath: string) => {
 
   if (res.data) {
     let extension = mime.extension(res.headers['content-type']);
-    console.info(
-      ' =>',
-      res.headers['content-type'],
-      humanizeFileSize(res.data.length)
-    );
+    if (!hasCache) {
+      console.info(
+        ' =>',
+        res.headers['content-type'],
+        humanizeFileSize(res.data.length)
+      );
+    }
 
     if (extension) {
       localPath = localPath + '.' + extension;
     }
     const dir = path.dirname(localPath);
     fs.mkdirSync(dir, { recursive: true });
-    console.info(' -> Writing file:', localPath);
+    if (!hasCache) {
+      console.info(' -> Writing file:', localPath);
+    }
     fs.writeFileSync(localPath, res.data);
   }
 

diff --git a/feishu-pages/src/index.ts b/feishu-pages/src/index.ts
@@ -13,7 +13,7 @@ import {
   fetchTenantAccessToken,
 } from './feishu';
 import { FileDoc, generateSummary, prepareDocSlugs } from './summary';
-import { cleanupDocsForJSON, humanizeFileSize } from './utils';
+import { cleanupDocsForJSON, humanizeFileSize, replaceLinks } from './utils';
 import { fetchAllDocs } from './wiki';
 
 // App entry
@@ -88,12 +88,15 @@ const fetchDocAndWriteFile = async (
 
     let { content, fileTokens } = doc;
 
-    // TODO: Replace link's node_token into slug
+    // Replace node_token to slug
     for (const node_token in slugMap) {
-      // Replace Markdown link and HTML link
-      const re = new RegExp(`${node_token}`, 'gm');
-      const newLink = `${BASE_URL}${slugMap[node_token]}`;
-      content = content.replace(re, newLink);
+      if (slugMap[node_token]) {
+        content = replaceLinks(
+          content,
+          node_token,
+          `${BASE_URL}${slugMap[node_token]}`
+        );
+      }
     }
 
     const metaInfo = generateFileMeta(doc, doc.slug, doc.position);
@@ -141,8 +144,10 @@ const downloadFiles = async (
 
     const extension = path.extname(filePath);
 
-    const re = new RegExp(`${fileToken}`, 'gm');
-    content = content.replace(re, './assets/' + fileToken + extension);
+    let assetURL = `./assets/${fileToken}${extension}`;
+
+    // Replase Markdown image
+    content = replaceLinks(content, fileToken, assetURL);
   }
 
   return content;

diff --git a/feishu-pages/src/utils.ts b/feishu-pages/src/utils.ts
@@ -1,4 +1,4 @@
-import { FileDoc } from "./summary";
+import { FileDoc } from './summary';
 
 export const normalizeSlug = (slug: string | number) => {
   // force convert slug into string
@@ -37,19 +37,19 @@ export const humanizeFileSize = (bytes, dp = 1) => {
 };
 
 const allowKeys = [
-  "depth",
-  "title",
-  "slug",
-  "filename",
-  "node_token",
-  "parent_node_token",
-  "children",
-  "obj_create_time",
-  "obj_edit_time",
-  "obj_token",
-  "has_child",
-  "meta",
-  "position"
+  'depth',
+  'title',
+  'slug',
+  'filename',
+  'node_token',
+  'parent_node_token',
+  'children',
+  'obj_create_time',
+  'obj_edit_time',
+  'obj_token',
+  'has_child',
+  'meta',
+  'position',
 ];
 
 export function cleanupDocsForJSON(docs: FileDoc[]) {
@@ -58,7 +58,7 @@ export function cleanupDocsForJSON(docs: FileDoc[]) {
   for (let idx = 0; idx < docs.length; idx++) {
     const doc = docs[idx];
 
-    Object.keys(doc).forEach(key => {
+    Object.keys(doc).forEach((key) => {
       if (!allowKeys.includes(key)) {
         delete doc[key];
       }
@@ -78,3 +78,27 @@ export function cleanupDocsForJSON(docs: FileDoc[]) {
     docs.splice(nodesToDelete[i], 1);
   }
 }
+
+export function replaceLinks(
+  content: string,
+  node_token: string,
+  newLink?: string
+): string {
+  if (!newLink) {
+    return content;
+  }
+
+  /*
+    match all
+
+    1 - ]( or src=" or href="
+    2 - https://ywh1bkansf.feishu.cn/wiki/aabbdd
+    3 - node_token
+    4 - ) or "
+  */
+  const re = new RegExp(
+    `(]\\(|src="|href=")(http[s]?:\\\/\\\/[\\w]+\\.(feishu\\.cn|larksuite\.com)\\\/.*)?(${node_token}.*)(\\)|")`,
+    'gm'
+  );
+  return content.replace(re, `$1${newLink}$5`);
+}
diff --git a/feishu-pages/tests/doc.test.ts b/feishu-pages/tests/doc.test.ts
@@ -5,15 +5,15 @@ import { generateFileMeta } from '../src/doc';
 describe('Doc', () => {
   test('generateFileMeta', () => {
     let doc: any = {
-      title: 'hello world',
+      title: 'Docs: "hello world"',
     };
 
     const urlPath = '/hello/world';
     const position = 1;
 
     let expected = `---
-title: hello world
-slug: /hello/world
+title: "Docs: \\"hello world\\""
+slug: "/hello/world"
 sidebar_position: 1
 ---
 `;
@@ -24,7 +24,7 @@ sidebar_position: 1
       title: null,
     };
     expected = `---
-slug: /hello/world
+slug: "/hello/world"
 sidebar_position: 1
 ---
 `;