From e60378f84bbc7762b935fcae66c91e1d0bf0b9a8 Mon Sep 17 00:00:00 2001
From: Sarah German <sgerman@gitlab.com>
Date: Tue, 8 Nov 2022 19:08:20 +0000
Subject: Add a script to report on pages not in the global nav

---
 scripts/pages_not_in_nav.js | 72 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 72 insertions(+)
 create mode 100755 scripts/pages_not_in_nav.js

(limited to 'scripts')

diff --git a/scripts/pages_not_in_nav.js b/scripts/pages_not_in_nav.js
new file mode 100755
index 00000000..87e15fe0
--- /dev/null
+++ b/scripts/pages_not_in_nav.js
@@ -0,0 +1,72 @@
+#!/usr/bin/env node
+
+/**
+ * @file pages_not_in_nav.js
+ * Generates a report of pages which are not included in navigation.yaml.
+ */
+
+/* eslint-disable no-console */
+
+const fs = require('fs');
+const glob = require('glob');
+const yaml = require('js-yaml');
+const fm = require('front-matter');
+
+// Load site data sources from nanoc config.
+const nanocConfig = yaml.load(fs.readFileSync('nanoc.yaml', 'utf8'));
+const dataSources = nanocConfig.data_sources.filter((source) => source.items_root !== '/');
+
+// Load the global navigation data file.
+const navYaml = yaml.load(fs.readFileSync('content/_data/navigation.yaml', 'utf8'));
+const nav = JSON.stringify(navYaml);
+
+// Read the markdown file and extract the fields we need.
+const getPageData = (filename) => {
+  const contents = fs.readFileSync(filename, 'utf-8');
+  return {
+    filename,
+    isRedirect: contents.includes('redirect_to'),
+    stage: fm(contents).attributes.stage,
+    group: fm(contents).attributes.group,
+  };
+};
+
+// Loop through each data source's markdown files.
+const lostPages = [];
+dataSources.forEach((source) => {
+  glob.sync(`${source.content_dir}/**/*.md`).forEach((filename) => {
+    const pageData = getPageData(filename);
+    if (pageData.isRedirect) {
+      return;
+    }
+
+    // Convert the markdown filepath into a string that matches the URL path on the website.
+    const path =
+      source.items_root.replaceAll('/', '') +
+      filename
+        .replace(source.content_dir, '')
+        .replace(source, '')
+        .replace('index.md', '')
+        .replace('.md', '.html');
+
+    if (
+      // Include pages that are not in the nav.
+      !nav.includes(path) &&
+      // Exclude sections that are intentionally not in the nav.
+      !path.includes('/architecture/blueprints') &&
+      !path.includes('/user/application_security/dast/checks/') &&
+      !path.includes('/legal/') &&
+      !path.includes('/drawers/') &&
+      !path.includes('/adr/')
+    ) {
+      lostPages.push({
+        url: `https://docs.gitlab.com/${path}`,
+        stage: pageData.stage,
+        group: pageData.group,
+      });
+    }
+  });
+});
+
+// Return results as JSON.
+console.log(JSON.stringify(lostPages));
-- 
cgit v1.2.3