create endpoint to get articles by url

app.js

@@ -83,6 +83,7 @@ app.get("/version", (req, res) => {
 
 //BLOG DATA APIs
 app.get("/blogs", verifyToken, routes.api.browse_data);
+app.get("/articles", verifyToken, routes.api.get_articles);
 app.get("/blogs/stats", verifyToken, routes.api.get_blog_stats);
 
 app.post("/get-webpage-content", verifyToken, routes.api.getWebContent);

controllers/blog/api/get_articles.js

@@ -0,0 +1,44 @@
+const { DB } = include("db/");
+
+exports.get_articles = async (req, res) => {
+  const { id, url } = req.query;
+
+  // Ensure at least one of `id` or `url` is provided
+  if (
+    (!id || (Array.isArray(id) && id.length === 0)) &&
+    (!url || (Array.isArray(url) && url.length === 0))
+  ) {
+    return res
+      .status(400)
+      .json({ error: "At least one 'id' or 'url' must be provided." });
+  }
+
+  // Normalize `id` and `url` into arrays
+  const idList = id ? (Array.isArray(id) ? id : [id]).map(Number) : [];
+  const urlList = url ? (Array.isArray(url) ? url : [url]) : [];
+
+  try {
+    const results = await DB.blog.any(
+      `
+      SELECT a.id, a.url, a.article_type, a.title, a.iso3, a.posted_date, a.posted_date_str, a.parsed_date, a.language, a.created_at, c.html_content,
+          regexp_replace(
+              regexp_replace(COALESCE(b.content, c.html_content), E'\\n', ' ', 'g'),
+              E'<iframe[^>]*>.*?</iframe>',
+              '',
+              'gi'
+          ) AS content
+      FROM articles a
+      JOIN article_content b ON b.article_id = a.id 
+      JOIN article_html_content c ON c.article_id = a.id
+      WHERE (array_length($1::int[], 1) > 0 AND a.id = ANY ($1::int[]))
+         OR (array_length($2::text[], 1) > 0 AND a.url = ANY ($2::text[]))
+      `,
+      [idList, urlList]
+    );
+
+    return res.status(200).json(results);
+  } catch (err) {
+    console.error("Database query failed:", err);
+    return res.status(500).json({ error: "An unexpected error occurred." });
+  }
+};

controllers/blog/api/query.js

@@ -80,7 +80,7 @@ exports.searchBlogQuery = (
   return {
     text: `
     WITH search_results AS (
-      SELECT a.url, a.article_type, a.title, a.iso3, a.posted_date, a.posted_date_str, a.parsed_date, a.language, a.created_at, c.html_content,
+      SELECT a.id, a.url, a.article_type, a.title, a.iso3, a.posted_date, a.posted_date_str, a.parsed_date, a.language, a.created_at, c.html_content,
           regexp_replace(
               regexp_replace(${textColumn}, E'\\n', ' ', 'g'),
               E'<iframe[^>]*>.*?</iframe>',

controllers/blog/index.js

@@ -1,7 +1,7 @@
 const { browse_data } = require("./api/data");
 const { main } = require("./api/stats");
 const { DB } = include("db");
-const cleanup = require('./scrapper/clean-up')
+const cleanup = require("./scrapper/clean-up");
 
 exports.browse_data = async (req, res) => {
   const data = await browse_data(DB.blog, req, res);
@@ -19,5 +19,7 @@ exports.cleanup = async (req, res) => {
   cleanup(DB.blog, req, res);
 };
 
-exports.getWebContent = require('../blog/api/scap')
-exports.scrap_medium_posts = require('./blog_type/medium').scrap_medium_posts
+exports.getWebContent = require("../blog/api/scap");
+exports.scrap_medium_posts = require("./blog_type/medium").scrap_medium_posts;
+
+exports.get_articles = require("./api/get_articles").get_articles;

controllers/index.js

@@ -3,7 +3,8 @@ const {
   get_stats,
   cleanup,
   getWebContent,
-  scrap_medium_posts
+  scrap_medium_posts,
+  get_articles,
 } = require("./blog");
 const { get_toolkit_data, toolkit_scrapper } = require("./toolkits");
 
@@ -18,4 +19,4 @@ exports.cleanup = cleanup;
 exports.scrap_medium_posts = scrap_medium_posts;
 
 exports.getWebContent = getWebContent;
-
+exports.get_articles = get_articles;

routes/index.js

@@ -7,6 +7,7 @@ const {
   scrap_medium_posts,
   getWebContent,
   get_ce_rave,
+  get_articles,
 } = include("/controllers");
 if (!exports.api) {
   exports.api = {};
@@ -25,7 +26,7 @@ exports.api.get_toolkit_data = get_toolkit_data;
 exports.api.cleanup = cleanup;
 
 exports.api.getWebContent = getWebContent;
-
+exports.api.get_articles = get_articles;
 
 exports.cron.scrap_medium_posts = scrap_medium_posts;
 exports.cron.toolkit_scrapper = toolkit_scrapper;