sitemap: sitemapindex template addition

* Adds support for sitemap pages.  (closes #3)

Signed-off-by: Jiri Kuncar <jiri.kuncar@cern.ch>

flask_sitemap/__init__.py

@@ -26,11 +26,15 @@
 
 from __future__ import absolute_import
 
+import gzip
 import sys
 
 from collections import Mapping
-from flask import current_app, request, Blueprint, render_template, url_for
+from flask import current_app, request, Blueprint, render_template, url_for, \
+    Response
+from flask.signals import Namespace
 from functools import wraps
+from itertools import islice
 from werkzeug.utils import import_string
 
 
@@ -39,9 +43,22 @@
 
 # PY2/3 compatibility
 if sys.version_info[0] == 3:
+    import io
+    StringIO = io.StringIO
     string_types = str,
+    from itertools import izip_longest as zip_longest
 else:
+    from cStringIO import StringIO
     string_types = basestring,
+    from itertolls import zip_longest
+
+
+# Signals
+_signals = Namespace()
+
+#: Sent when a sitemap index is generated and given page will need to be
+#: generated in the future from already calculated url set.
+sitemap_page_needed = _signals.signal('sitemap-page-needed')
 
 
 class Sitemap(object):
@@ -88,6 +105,11 @@ def init_app(self, app):
                 'sitemap',
                 self._decorate(self.sitemap)
             )
+            self.blueprint.add_url_rule(
+                app.config.get('SITEMAP_ENDPOINT_PAGE_URL'),
+                'page',
+                self._decorate(self.page)
+            )
             app.register_blueprint(
                 self.blueprint,
                 url_prefix=app.config.get('SITEMAP_BLUEPRINT_URL_PREFIX')
@@ -105,8 +127,34 @@ def wrapper(*args, **kwargs):
 
     def sitemap(self):
         """Generate sitemap.xml."""
-        return render_template('flask_sitemap/sitemap.xml',
-                               urlset=self._generate_all_urls())
+        size = self.app.config['SITEMAP_MAX_URL_COUNT']
+        args = [iter(self._generate_all_urls())] * size
+        run = zip_longest(*args)
+        urlset = next(run)
+
+        if urlset[-1] is None:
+            return render_template('flask_sitemap/sitemap.xml',
+                                   urlset=filter(None, urlset))
+
+        def pages():
+            yield {'loc': url_for('flask_sitemap.page', page=1)}
+            sitemap_page_needed.send(current_app._get_current_object(),
+                                     page=1, urlset=urlset)
+            for page, urlset_ in enumerate(run):
+                yield {'loc': url_for('flask_sitemap.page', page=page+2)}
+                sitemap_page_needed.send(current_app._get_current_object(),
+                                         page=page+2, urlset=urlset_)
+
+        return render_template('flask_sitemap/sitemapindex.xml',
+                               sitemaps=pages())
+
+    def page(self, page):
+        """Generate sitemap for given range of urls."""
+        size = self.app.config['SITEMAP_MAX_URL_COUNT']
+        urlset = islice(self._generate_all_urls(), (page-1)*size, page*size)
+        return self.gzip_response(
+            render_template('flask_sitemap/sitemap.xml', urlset=urlset)
+        )
 
     def register_generator(self, generator):
         """Register an URL generator.
@@ -166,5 +214,19 @@ def _generate_all_urls(self):
                         result['loc'] = url_for(endpoint, **values)
                     yield result
 
+    def gzip_response(self, data):
+        """GZip response data and create new Response instance."""
+        gzip_buffer = StringIO()
+        gzip_file = gzip.GzipFile(mode='wb', compresslevel=6,
+                                  fileobj=gzip_buffer)
+        gzip_file.write(data)
+        gzip_file.close()
+        response = Response()
+        response.data = gzip_buffer.getvalue()
+        response.headers['Content-Encoding'] = 'gzip'
+        response.headers['Content-Length'] = len(response.data)
+
+        return response
+
 
 __all__ = ('Sitemap', '__version__')

flask_sitemap/config.py

@@ -30,7 +30,19 @@
 SITEMAP_ENDPOINT_URL
 --------------------
 
-Default: ``/sitemap.xml``.
+Return sitemap index or sitemap for pages with less than
+``SITEMAP_MAX_URL_COUNT`` urls.
+
+Default: ``sitemap.xml``.
+
+SITEMAP_ENDPOINT_PAGE_URL
+-------------------------
+
+Return GZipped sitemap for given page range of urls.
+
+.. note:: It is strongly recommended to provide caching decorator.
+
+Default: ``sitemap<int:page>.xml.gz``
 
 SITEMAP_INCLUDE_RULES_WITHOUT_PARAMS
 ------------------------------------
@@ -46,6 +58,15 @@
 ----------------------
 
 Default: ``[]``.
+
+SITEMAP_MAX_URL_COUNT
+---------------------
+
+The maximum number of urls per one sitemap file can be up to 50000, however
+there is 10MB limitation for the file.
+
+Default: ``10000``.
+
 """
 
 SITEMAP_BLUEPRINT = 'flask_sitemap'
@@ -54,10 +75,14 @@
 
 SITEMAP_ENDPOINT_URL = 'sitemap.xml'
 
+SITEMAP_ENDPOINT_PAGE_URL = 'sitemap<int:page>.xml.gz'
+
 SITEMAP_URL_SCHEME = 'http'
 
 SITEMAP_INCLUDE_RULES_WITHOUT_PARAMS = False
 
 SITEMAP_IGNORE_ENDPOINTS = None
 
 SITEMAP_VIEW_DECORATORS = []
+
+SITEMAP_MAX_URL_COUNT = 10000

flask_sitemap/templates/flask_sitemap/sitemapindex.xml

@@ -0,0 +1,11 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
+  {%- for sitemap in sitemaps %}
+  <sitemap>
+    <loc>{{ sitemap.loc }}</loc>
+    {%- if sitemap.lastmod %}
+    <lastmod>{{ sitemap.lastmod }}</lastmod>
+    {%- endif %}
+  </sitemap>
+  {%- endfor %}
+</sitemapindex>

tests/test_ext.py

@@ -187,3 +187,36 @@ def third(dummy):
 
         with self.app.test_client() as c:
             assert b('third') == c.get('/sitemap.xml').data
+
+    def test_pagination(self):
+        self.app.config['SERVER_NAME'] = 'www.example.com'
+        self.app.config['SITEMAP_INCLUDE_RULES_WITHOUT_PARAMS'] = True
+        self.app.config['SITEMAP_MAX_URL_COUNT'] = 10
+        sitemap = Sitemap(app=self.app)
+        now = datetime.now().isoformat()
+
+        @self.app.route('/')
+        def index():
+            pass
+
+        @self.app.route('/first')
+        def first():
+            pass
+
+        @self.app.route('/second')
+        def second():
+            pass
+
+        @self.app.route('/<username>')
+        def user(username):
+            pass
+
+        @sitemap.register_generator
+        def user():
+            for number in range(20):
+                yield 'user', {'username': 'test{0}'.format(number)}
+
+        with self.app.test_client() as c:
+            assert b('sitemapindex') in c.get('/sitemap.xml').data
+            assert b('gzip') in c.get(
+                '/sitemap1.xml.gz').headers['Content-Encoding']