Add site map

src/.buildinfo.json

@@ -0,0 +1,6 @@
+{
+    "tag": "jc-ng-localtest:",
+    "date": "2025-06-15",
+    "host": "jake-e580",
+    "user": "jake"
+}

src/index.py

@@ -12,6 +12,7 @@ app = Flask(__name__)
 # These imports need to come after our app is defined as they add routes to it.
 import projects # pylint: disable=wrong-import-position,unused-import
 import contact  # pylint: disable=wrong-import-position,unused-import
+import sitemap  # pylint: disable=wrong-import-position,unused-import
 
 class DiscordLogger(logging.Handler):
     ''' Simple logging handler to send a message to Discord '''

src/projects.py

@@ -45,7 +45,7 @@ def to_html(content: str) -> str:
     ''' Jninja filter to wrap markdown '''
     return markdown(content)
 
-def get_all_posts(directory: str) -> list:
+def get_all_posts(directory: str = md_directory) -> list:
     ''' Get all posts in the posts directory '''
     abs_paths = [path.join(directory, x) for x in glob(f'{directory}/*.md')]
     return [frontmatter.load(x) for x in abs_paths]

src/projects.wsgi

@@ -4,3 +4,6 @@ import sys
 sys.path.append('/var/www/jc')
 
 from index import app as application
+
+if __name__ == '__main__':
+    application.run(debug=True)

src/sitemap.py

@@ -0,0 +1,42 @@
+#!/usr/bin/python3
+
+import xml.etree.ElementTree as ET
+import json
+from flask import url_for, request, Response
+from re import match
+from index import app
+from projects import get_all_posts
+
+def get_routes() -> list:
+    routes = []
+    for rule in app.url_map.iter_rules():
+        if 0 >= len(rule.arguments):
+            url = url_for(rule.endpoint, **(rule.defaults or {}))
+            routes.append(url)
+    return routes
+
+def get_build_date():
+    try:
+        with open('/var/www/jc/.buildinfo.json', encoding='utf8') as build:
+            build_json = json.load(build)
+            return build_json['date']
+    except:
+        return '1970-01-01'
+
+@app.route('/sitemap.xml')
+def sitemap():
+    date = get_build_date()
+    root = ET.Element('urlset', xmlns='http://www.sitemaps.org/schemas/sitemap/0.9')
+    base_url = match(r'^https?:\/\/.+:?\d*(?=\/)', request.base_url).group()
+    for route in get_routes():
+        url = ET.SubElement(root, 'url')
+        ET.SubElement(url, 'loc').text = base_url + route
+        ET.SubElement(url, 'lastmod').text = date
+    for article in get_all_posts():
+        if 'link' in article.metadata:
+            continue
+        url = ET.SubElement(root, 'url')
+        ET.SubElement(url, 'loc').text = f'{base_url}/projects/{article.metadata['id']}'
+        ET.SubElement(url, 'lastmod').text = article.metadata['date'].strftime('%Y-%m-%d')
+        
+    return Response(ET.tostring(root, encoding='utf-8'), 200, {'content-type': 'application/xml'})

src/static/robots.txt

@@ -0,0 +1,4 @@
+User-agent: *
+Allow: /
+
+Sitemap: https://jakecharman.co.uk/sitemap.xml