feat: support get dynamic page meta data (#2632)

apps/electron/layers/main/src/ui/get-meta-data/get-html.ts

@@ -0,0 +1,49 @@
+import { BrowserWindow } from 'electron';
+
+import type { GetHTMLOptions } from './types';
+
+async function getHTMLFromWindow(win: BrowserWindow): Promise<string> {
+  return win.webContents
+    .executeJavaScript(`document.documentElement.outerHTML;`)
+    .then(html => html);
+}
+
+// For normal web pages, obtaining html can be done directly,
+// but for some dynamic web pages, obtaining html should wait for the complete loading of web pages. shouldReGetHTML should be used to judge whether to obtain html again
+export async function getHTMLByURL(
+  url: string,
+  options: GetHTMLOptions
+): Promise<string> {
+  return new Promise(resolve => {
+    const { timeout = 10000, shouldReGetHTML } = options;
+    const window = new BrowserWindow({
+      show: false,
+    });
+    let html = '';
+    window.loadURL(url);
+
+    const timer = setTimeout(() => {
+      resolve(html);
+      window.close();
+    }, timeout);
+
+    async function loopHandle() {
+      html = await getHTMLFromWindow(window);
+      if (!shouldReGetHTML) {
+        return html;
+      }
+
+      if (await shouldReGetHTML(html)) {
+        setTimeout(loopHandle, 1000);
+      } else {
+        window.close();
+        clearTimeout(timer);
+        resolve(html);
+      }
+    }
+
+    window.webContents.on('did-finish-load', async () => {
+      loopHandle();
+    });
+  });
+}

apps/electron/layers/main/src/ui/get-meta-data/index.ts

@@ -1,23 +1,15 @@
 import type { CheerioAPI, Element } from 'cheerio';
 import { load } from 'cheerio';
-import got from 'got';
 
 import type { Context, MetaData, Options, RuleSet } from './types';
 
 export * from './types';
 
+import { getHTMLByURL } from './get-html';
 import { metaDataRules } from './rules';
+import type { GetMetaDataOptions } from './types';
 
-const defaultOptions = {
-  maxRedirects: 5,
-  ua: 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/113.0.0.0 Safari/537.36',
-  lang: '*',
-  timeout: 10000,
-  forceImageHttps: true,
-  customRules: {},
-};
-
-const runRule = function (ruleSet: RuleSet, $: CheerioAPI, context: Context) {
+function runRule(ruleSet: RuleSet, $: CheerioAPI, context: Context) {
   let maxScore = 0;
   let value;
 
@@ -58,61 +50,31 @@ const runRule = function (ruleSet: RuleSet, $: CheerioAPI, context: Context) {
   }
 
   return undefined;
-};
+}
 
-const getMetaData = async function (
-  input: string | Partial<Options>,
-  inputOptions: Partial<Options> = {}
+async function getMetaDataByHTML(
+  html: string,
+  url: string,
+  options: GetMetaDataOptions
 ) {
-  let url;
-  if (typeof input === 'object') {
-    inputOptions = input;
-    url = input.url || '';
-  } else {
-    url = input;
-  }
-
-  const options = Object.assign({}, defaultOptions, inputOptions);
-
+  const { customRules = {} } = options;
   const rules: Record<string, RuleSet> = { ...metaDataRules };
-  Object.keys(options.customRules).forEach((key: string) => {
+  Object.keys(customRules).forEach((key: string) => {
     rules[key] = {
-      rules: [...metaDataRules[key].rules, ...options.customRules[key].rules],
+      rules: [...metaDataRules[key].rules, ...customRules[key].rules],
       defaultValue:
-        options.customRules[key].defaultValue ||
-        metaDataRules[key].defaultValue,
-      processor:
-        options.customRules[key].processor || metaDataRules[key].processor,
+        customRules[key].defaultValue || metaDataRules[key].defaultValue,
+      processor: customRules[key].processor || metaDataRules[key].processor,
     };
   });
 
-  let html;
-  if (!options.html) {
-    const response = await got(url, {
-      headers: {
-        'User-Agent': options.ua,
-        'Accept-Language': options.lang,
-      },
-      timeout: options.timeout,
-      ...(options.maxRedirects === 0
-        ? { followRedirect: false }
-        : { maxRedirects: options.maxRedirects }),
-    });
-    html = response.body;
-  } else {
-    html = options.html;
-  }
-
   const metadata: MetaData = {};
   const context: Context = {
     url,
-    options,
+    ...options,
   };
 
   const $ = load(html);
-  // console.log('===============================');
-  // console.log('html');
-  // console.log(doc);
 
   Object.keys(rules).forEach((key: string) => {
     const ruleSet = rules[key];
@@ -120,6 +82,26 @@ const getMetaData = async function (
   });
 
   return metadata;
-};
+}
 
-export { getMetaData };
+export async function getMetaData(url: string, options: Options = {}) {
+  const { customRules, forceImageHttps, shouldReGetHTML, ...other } = options;
+  const html = await getHTMLByURL(url, {
+    ...other,
+    shouldReGetHTML: async html => {
+      const meta = await getMetaDataByHTML(html, url, {
+        customRules,
+        forceImageHttps,
+      });
+      return shouldReGetHTML ? await shouldReGetHTML(meta) : false;
+    },
+  }).catch(() => {
+    //   TODO: report error
+    return '';
+  });
+
+  return await getMetaDataByHTML(html, url, {
+    customRules,
+    forceImageHttps,
+  });
+}

apps/electron/layers/main/src/ui/get-meta-data/rules.ts

@@ -591,7 +591,7 @@ export const metaDataRules: Record<string, RuleSet> = {
       ],
     ],
     processor: (imageUrl: any, context) =>
-      context.options.forceImageHttps === true
+      context.forceImageHttps === true
         ? makeUrlSecure(makeUrlAbsolute(context.url, imageUrl))
         : makeUrlAbsolute(context.url, imageUrl),
   },
@@ -625,7 +625,7 @@ export const metaDataRules: Record<string, RuleSet> = {
     },
     defaultValue: context => makeUrlAbsolute(context.url, '/favicon.ico'),
     processor: (iconUrl, context) =>
-      context.options.forceImageHttps === true
+      context.forceImageHttps === true
         ? makeUrlSecure(makeUrlAbsolute(context.url, iconUrl))
         : makeUrlAbsolute(context.url, iconUrl),
   },
@@ -654,7 +654,7 @@ export const metaDataRules: Record<string, RuleSet> = {
       ['meta[name="og:video"][content]', element => element.attribs['content']],
     ],
     processor: (imageUrl: any, context) =>
-      context.options.forceImageHttps === true
+      context.forceImageHttps === true
         ? makeUrlSecure(makeUrlAbsolute(context.url, imageUrl))
         : makeUrlAbsolute(context.url, imageUrl),
   },
@@ -683,7 +683,7 @@ export const metaDataRules: Record<string, RuleSet> = {
       ['meta[name="og:audio"][content]', element => element.attribs['content']],
     ],
     processor: (imageUrl: any, context) =>
-      context.options.forceImageHttps === true
+      context.forceImageHttps === true
         ? makeUrlSecure(makeUrlAbsolute(context.url, imageUrl))
         : makeUrlAbsolute(context.url, imageUrl),
   },

apps/electron/layers/main/src/ui/get-meta-data/types.ts

@@ -1,6 +1,6 @@
 import type { Element } from 'cheerio';
 
-export interface MetaData {
+export type MetaData = {
   title?: string;
   description?: string;
   icon?: string;
@@ -12,29 +12,32 @@ export interface MetaData {
   provider?: string;
 
   [x: string]: string | string[] | undefined;
-}
+};
 
 export type MetadataRule = [string, (el: Element) => string | null];
 
-export interface Context {
+export type Context = {
   url: string;
-  options: Options;
-}
+} & GetMetaDataOptions;
 
-export interface RuleSet {
+export type RuleSet = {
   rules: MetadataRule[];
   defaultValue?: (context: Context) => string | string[];
   scorer?: (el: Element, score: any) => any;
   processor?: (input: any, context: Context) => any;
-}
+};
 
-export interface Options {
-  maxRedirects?: number;
-  ua?: string;
-  lang?: string;
-  timeout?: number;
-  forceImageHttps?: boolean;
-  html?: string;
-  url?: string;
+export type GetMetaDataOptions = {
   customRules?: Record<string, RuleSet>;
-}
+  forceImageHttps?: boolean;
+};
+
+export type GetHTMLOptions = {
+  timeout?: number;
+  shouldReGetHTML?: (currentHTML: string) => boolean | Promise<boolean>;
+};
+
+export type Options = {
+  shouldReGetHTML?: (metaData: MetaData) => boolean | Promise<boolean>;
+} & GetMetaDataOptions &
+  Omit<GetHTMLOptions, 'shouldReGetHTML'>;

apps/electron/layers/main/src/ui/index.ts

@@ -1,4 +1,4 @@
-import { app, BrowserWindow, nativeTheme, session } from 'electron';
+import { app, BrowserWindow, nativeTheme } from 'electron';
 
 import type { NamespaceHandlers } from '../type';
 import { isMacOS } from '../utils';
@@ -42,7 +42,9 @@ export const uiHandlers = {
   },
   getBookmarkDataByLink: async (_, url: string) => {
     return getMetaData(url, {
-      ua: session.defaultSession.getUserAgent(),
+      shouldReGetHTML: metaData => {
+        return !metaData.title && !metaData.description;
+      },
     });
   },
 } satisfies NamespaceHandlers;