feat: add meta tags endpoint to html to pdf (#789)

Signed-off-by: hxtree <[email protected]>

common/config/rush/browser-approved-packages.json

@@ -450,6 +450,10 @@
       "name": "bson",
       "allowedCategories": [ "apis", "libraries" ]
     },
+    {
+      "name": "cheerio",
+      "allowedCategories": [ "apis" ]
+    },
     {
       "name": "chokidar",
       "allowedCategories": [ "platform", "rigs" ]

common/config/rush/repo-state.json

@@ -1,5 +1,5 @@
 // DO NOT MODIFY THIS FILE MANUALLY BUT DO COMMIT IT. It is generated and used by Rush.
 {
-  "pnpmShrinkwrapHash": "ef99d2be9bb5fde2af843e6f081457fe57fa679f",
+  "pnpmShrinkwrapHash": "0f26d344f9b8b25f5850336a9f316858200a1459",
   "preferredVersionsHash": "a48003cf229dd47d077bcf6301ac15a6f90e1c34"
 }

services/html-to-pdf/package.json

@@ -61,7 +61,8 @@
     "@sparticuz/chromium-min": "112.0.0",
     "puppeteer-core": "19.8.0",
     "uuid": "~9.0.1",
-    "axios": "^0.21.1"
+    "axios": "^0.21.1",
+    "cheerio": "~1.0.0-rc.12"
   },
   "devDependencies": {
     "@cats-cradle/eslint-config": "1.0.11",

services/html-to-pdf/src/app.module.ts

@@ -1,9 +1,10 @@
 import { Module } from '@nestjs/common';
 import { HealthModule } from './module/health/health.module';
 import { PdfModule } from './module/pdf/pdf.module';
+import { MetaTagsModule } from './module/meta-tags/meta-tags.module';
 
 @Module({
-  imports: [HealthModule, PdfModule],
+  imports: [HealthModule, PdfModule, MetaTagsModule],
   providers: [],
   exports: [],
 })

services/html-to-pdf/src/module/meta-tags/meta-tags.controller.ts

@@ -0,0 +1,16 @@
+import {
+  Controller, Get, Param, Query, VERSION_NEUTRAL,
+} from '@nestjs/common';
+import { MetaTagsService } from './meta-tags.service';
+
+@Controller({ path: 'meta-tags', version: ['1', VERSION_NEUTRAL] })
+export class MetaTagsController {
+  constructor(private readonly metaTagsService: MetaTagsService) {}
+
+  @Get()
+  async getMetaTags(
+    @Query('url') url: string,
+  ): Promise<{ [key: string]: string }> {
+    return this.metaTagsService.getMetaTags(url);
+  }
+}

services/html-to-pdf/src/module/meta-tags/meta-tags.module.ts

@@ -0,0 +1,9 @@
+import { Module } from '@nestjs/common';
+import { MetaTagsController } from './meta-tags.controller';
+import { MetaTagsService } from './meta-tags.service';
+
+@Module({
+  controllers: [MetaTagsController],
+  providers: [MetaTagsService],
+})
+export class MetaTagsModule {}

services/html-to-pdf/src/module/meta-tags/meta-tags.service.ts

@@ -0,0 +1,36 @@
+import { Injectable } from '@nestjs/common';
+import axios from 'axios';
+import * as cheerio from 'cheerio';
+
+@Injectable()
+export class MetaTagsService {
+  async getMetaTags(url: string): Promise<{ [key: string]: string }> {
+    try {
+      const response = await axios.get(url);
+      const metaTags = this.extractMetaTags(response.data);
+      return metaTags;
+    } catch (err) {
+      // Handle errors (e.g., network issues, invalid URLs)
+      const error = err as Error;
+      console.error('Error fetching or parsing the page:', error.message);
+      throw new Error('Unable to fetch or parse the page');
+    }
+  }
+
+  private extractMetaTags(html: string): { [key: string]: string } {
+    const $ = cheerio.load(html);
+    const metaTags: { [key: string]: string } = {};
+
+    $('meta').each((_, element) => {
+      const tag = $(element);
+      const name = tag.attr('name') || tag.attr('property');
+      const content = tag.attr('content');
+
+      if (name && content) {
+        metaTags[name] = content;
+      }
+    });
+
+    return metaTags;
+  }
+}