News sitemap

README.md

@@ -10,18 +10,51 @@ First install the module:
 $ npm install -s amphora-sitemaps
 ```
 
-Then pass the module into Amphora as an item for the `plugins` array property.
+Then, require the module and pass all the options you need to start the plugin:
+```javascript
+const amphoraSitemaps = require('amphora-sitemaps'),
+  amphoraSitemapsPlugin = amphoraSitemaps({ _news: { component: 'article' }});
+```
+
+After that, pass the module into Amphora as an item for the `plugins` array property.
 
 ```javascript
 amphora({
   ...
   plugins: [
     ...
-    require('amphora-sitemaps'),
+    amphoraSitemapsPlugin,
     ...
   ],
   ...
 })
 ```
 
-At startup time the module will create and expose endpoints that you can hit in your browser to get the all the published pages in an XML response for that site. `eg. yoursite.com/_sitemap`
+## Options
+
+The options object is used to initialize some endpoints like `_news`.
+
+The object should have the following format:
+
+```javascript
+var options = {
+  _news: { // Endpoint name
+    component: 'article' // The component used to get the data for the sitemap
+  }
+}
+```
+
+## Endpoints
+At startup time, the module will create the following XML endpoints:
+
+### _sitemaps
+Gets the all the published pages.
+
+`eg. yoursite.com/_sitemap`
+
+### _news
+Gets pages based on the [Google News Sitemap](https://support.google.com/news/publisher-center/answer/74288?hl=en) guidelines.
+
+In order to meet these guidelines, you must pass a component name with the properties `canonicalUrl` and `date` because those fields are required.
+
+`eg. yoursite.com/_news`

lib/controllers/news.js

@@ -0,0 +1,60 @@
+'use strict';
+
+const h = require('highland'),
+  db = require('../services/db'),
+  xml = require('xml'),
+  { setup, logError } = require('../services/logger'),
+  log = setup({ file: __filename }),
+  errorLogger = logError(log);
+
+/**
+ * Gets the published pages of a site.
+ * @param {string} uri
+ * @param {object} options
+ * @return {Promise<T | void>}
+ */
+function getPublishedContent({ uri }, options) {
+  const elem = xml.element({
+      _attr: {
+        xmlns: 'http://www.sitemaps.org/schemas/sitemap/0.9',
+        'xmlns:news': 'http://www.google.com/schemas/sitemap-news/0.9'
+      }
+    }),
+    stream = xml({ urlset: elem }, { stream: true, declaration: true }),
+    site = uri.replace('/_news', '');
+
+  h(db.selectPublishedContent(site, options))
+    .filter(content => options.componentFilter(content.data))
+    .map(({ data, title }) => {
+      return options.getKeywords(data)
+        .then(keywords => ({ data, title, keywords }));
+    })
+    .flatMap(h)
+    .map(({ data, title , keywords }) => {
+      const { date, canonicalUrl } = data;
+
+      elem.push({
+        url: [
+          { loc: canonicalUrl },
+          { 'news:news': [
+            { 'news:publication': [
+              { 'news:name': site },
+              { 'news:language': 'en' }
+            ] },
+            { 'news:publication_date': date },
+            { 'news:title': title },
+            { 'news:language': 'en' },
+            { 'news:keywords': keywords }
+          ] }
+        ]
+      });
+    })
+    .errors(errorLogger)
+    .done(() => {
+      elem.close();
+    });
+
+  return stream;
+}
+
+module.exports.getPublishedContent = getPublishedContent;

lib/services/sitemaps.js → lib/controllers/sitemap.js

@@ -1,9 +1,9 @@
 'use strict';
 
 const _ = require('highland'),
-  db = require('./db'),
+  db = require('../services/db'),
   xml = require('xml'),
-  { setup, logError } = require('./logger'),
+  { setup, logError } = require('../services/logger'),
   log = setup({ file: __filename }),
   errorLogger = logError(log);
 
@@ -14,11 +14,11 @@ const _ = require('highland'),
  */
 function getPublishedPages({ uri, query }) {
   const elem = xml.element({ _attr: { xmlns: 'http://www.sitemaps.org/schemas/sitemap/0.9' } }),
-    stream =  xml({ urlset: elem }, { stream: true });
+    stream = xml({ urlset: elem }, { stream: true , declaration: true });
 
   _(db.selectPublishedPages(uri, query))
     .map(item => {
-      var arr = [];
+      const arr = [];
 
       for (const key in item) {
         if (item.hasOwnProperty(key)) {

lib/services/sitemaps.test.js → lib/controllers/sitemap.test.js

@@ -1,8 +1,8 @@
 'use strict';
 
-const sitemaps = require('./sitemaps'),
+const sitemaps = require('./sitemap'),
   h = require('highland'),
-  db = require('./db'),
+  db = require('../services/db'),
   data = [{
     loc: 'http://localhost.nymag.com/',
     lastmod: '2018-10-01T15:45:44.494Z'

lib/routes/_news.js

@@ -0,0 +1,24 @@
+'use strict';
+
+const responses = require('../services/responses'),
+  controller = require('../controllers/news'),
+  handlers = {
+    getPublishedContent: options => {
+      return (req, res) => {
+        // set response type
+        res.type('text/xml');
+        // stream from the db
+        controller.getPublishedContent(req, options).pipe(res);
+      };
+    },
+  };
+
+function routes(router, options) {
+  router.all('/', responses.allow({ allow: ['get'] }));
+  router.get('/', handlers.getPublishedContent(options));
+}
+
+module.exports = routes;
+
+// For testing
+module.exports.route = handlers;

lib/routes/_sitemap.js

@@ -1,13 +1,11 @@
 'use strict';
 
 const responses = require('../services/responses'),
-  controller = require('../services/sitemaps'),
+  controller = require('../controllers/sitemap'),
   route = {
     getList: function (req, res) {
       // set response type
       res.type('text/xml');
-      // We need to add in the xml version as first line
-      res.write('<?xml version="1.0" encoding="UTF-8"?>');
       // stream from the db
       controller.getPublishedPages(req).pipe(res);
     },

lib/routes/_sitemap.test.js

@@ -1,7 +1,7 @@
 'use strict';
 
 const sitemap = require('./_sitemap'),
-  controller = require('../services/sitemaps'),
+  controller = require('../controllers/sitemap'),
   h =  require('highland');
 
 jest.mock('../services/sitemaps');

lib/services/db.js

@@ -54,6 +54,23 @@ function selectPublishedPages(uri, { year = '' }) {
   return db.raw(sql, params);
 }
 
+function selectPublishedContent(site, options) {
+  const componentSql = options.components.map(component => {
+    return `(SELECT  pages.meta ->> 'title' as title, componentTable.data as data, componentTable.data->> 'date' as pubDate
+      FROM components."${component}" as componentTable, public.pages as pages, public.uris as uris
+      WHERE pages.meta->>'url' = componentTable.data->>'canonicalUrl'
+      AND pages.id = uris.data
+      AND (componentTable.data->>'date')::date between (now() - interval '2 days') AND now()
+      AND componentTable.id like '${site}'||'/_components%')`
+  });
+  let sql = componentSql.join(' UNION ');
+
+  sql += ` ORDER BY (pubDate) desc
+    LIMIT 1000;`;
+
+  return db.raw(sql);
+}
 
 module.exports = init;
 module.exports.selectPublishedPages = selectPublishedPages;
+module.exports.selectPublishedContent = selectPublishedContent;

lib/services/init.js

@@ -4,16 +4,18 @@ const db = require('./db'),
   routes = require('./routes');
 
 /**
- * Initializes plugin.
- * @param {Object} router
- * @param {Object} storage
- * @param {Object} _
- * @param {Object} sites
- * @return {Promise}
+ * Sets up the plugin with some options
+ *
+ * @param {Object} options
+ * @param {Object} options._news
+ * @returns {Function}
  */
-function onInit(router, storage) {
-  return db(storage)
-    .then(() => routes(router));
+function setup(options) {
+  // Initializes plugin
+  return (router, storage) => {
+    return db(storage)
+      .then(() => routes(router, options));
+  };
 }
 
-module.exports = onInit;
+module.exports = setup;

lib/services/routes.js

@@ -9,19 +9,20 @@ const express = require('express'),
 /**
 * Set up sitemaping routes for all sites. eg. nymag.com/_sitemap
 * @param {object} router
+* @param {object} options
 */
-function setupRoutes(router) {
+function setupRoutes(router, options) {
   const routesPath = 'routes';
 
   // load all controller routers
   files.getFiles([__dirname, '..', routesPath].join(path.sep)).filter((filename) => {
     const pathRouter = express.Router(),
       name = removeExtension(filename),
-      controller = files.tryRequire([__dirname, '..', routesPath, name].join(path.sep));
+      route = files.tryRequire([__dirname, '..', routesPath, name].join(path.sep));
 
     pathRouter.use(jsonBodyParser);
 
-    controller(pathRouter);
+    route(pathRouter, options[name]);
     router.use(`/${name}`, pathRouter);
   });
 }