[DRAFT] Sitemaps for reports and FECForms - working prototypes

fec/fec/urls.py

@@ -14,6 +14,13 @@
 from home import views as home_views
 from search import views as search_views
 
+# ############# NEW SITEMAP IMPORTS ############
+import re
+from django.contrib.sitemaps import Sitemap
+from wagtail.documents.models import Document
+from home.models import DocumentPage, DocumentFeedPage
+
+# ############# /END NEW SITEMAP IMPORTS ############
 
 urlpatterns = [
     re_path(
@@ -55,6 +62,99 @@
 ]
 
 
+# ########### NEW SITEMAPS ################
+# SHOULD MOVE THESE CLASSES TO A views.py FILE OR ITS OWN FILE AND IMPORT IT HERE
+
+# ######## FORMS SITEMAP #########
+
+class FormsSitemap(Sitemap):
+    changefreq = "never"
+    priority = 0.5
+    protocol = 'https'
+
+    def items(self):
+        return Document.objects.filter(file__icontains="fecfrm")
+
+    def location(self, obj):
+
+        # WORKS: RETURNS JUST THE FILE THEN CONCATENATES THE PATH TO IT
+        # return '/resources/cms_content/'+str(obj.file)
+        # return str(obj.file).replace('documents/', '/resources/cms_content/documents/')
+
+        # `obj.file.url` RETURNS THIS ON LOCAL:
+        #  https://127.0.0.1:8000/media/documents/fecfrm13.pdf (make sure to change to http to test)
+        # `obj.file.url` RETURNS THIS ON DEV:
+        # `https://dev.fec.govhttps://fec-dev-proxy.app.cloud.gov/resources/cms-content/documents/fecfrm2sf.pdf`
+
+        # WORKS: THIS ONE REMOVES THE `https://fec-dev-proxy.app.cloud.gov` , TESTED ON DEV
+        loc = re.sub(r'^[^:]+:\/\/[^/?#]+', '', obj.file.url)
+        return loc
+
+    def lastmod(self, obj):
+        return obj.created_at
+
+
+urlpatterns += [
+
+    re_path(
+        r'^sitemap-forms\.xml/$',
+        sitemap,
+        {"sitemaps": {'forms': FormsSitemap()}},
+        name="django.contrib.sitemaps.views.sitemap",
+    ),
+]
+
+# ######## REPORT SITEMAP #########
+
+
+class ReportsSitemap(Sitemap):
+    changefreq = "never"
+    priority = 0.5
+    protocol = 'https'
+
+    def items(self):
+        # could just return all DocumentPages, but instead return all DocumentPages \
+        # that are descendants of DocumentFeedPages
+        # return DocumentPage.objects.live()
+
+        docfeeds = DocumentFeedPage.objects.live()
+
+        reports = []
+        for pg in docfeeds:
+            rpts = DocumentPage.objects.live().descendant_of(pg, inclusive=False)  # or child_of()
+            reports.extend(list(rpts))
+
+        return reports
+
+    def location(self, obj):
+        if obj.file_url:
+            loc = re.sub(r'https:\/\/(www|beta)\.fec\.gov', '', obj.file_url)
+        else:
+            loc = obj.url_path.replace('/home', '')
+
+        return loc
+
+    def lastmod(self, obj):
+        return obj.latest_revision_created_at
+
+    # Not a thing
+    def description(self, obj):
+        return obj.page_title
+
+
+urlpatterns += [
+
+    re_path(
+        r'^sitemap-reports\.xml/$',
+        sitemap,
+        {"sitemaps": {'reports': ReportsSitemap()}},
+        name="django.contrib.sitemaps.views.sitemap",
+    ),
+]
+
+# ################# /END NEW SITEMAPS ################
+
+
 if settings.FEC_CMS_ENVIRONMENT != 'LOCAL':
     # admin/login always must come before admin/, so place at beginning of list
     urlpatterns.insert(0, re_path(r'^admin/login', uaa_views.login, name='login'))

fec/home/models.py

@@ -724,7 +724,8 @@ def display_date(self):
     @property
     def extension(self):
         # Return the file extension of file_url
-        return self.file_url.rsplit('.', 1)[1].upper()
+        if self.file_url:
+            return self.file_url.rsplit('.', 1)[1].upper()
 
 
 class DocumentFeedPage(ContentPage):
@@ -1372,7 +1373,7 @@ class ReportingDatesTable(Page):
         ('html', blocks.RawHTMLBlock()),
         ('internal_button', InternalButtonBlock()),
         ('external_button', ExternalButtonBlock()),
-        ('dates_table', ReportingTableBlock(blank=True, required=False,form_classname='title')),
+        ('dates_table', ReportingTableBlock(blank=True, required=False, form_classname='title')),
     ], blank=True, null=True, use_json_field=True, collapsed=False)
 
     footnotes = StreamField([

fec/home/templates/home/document_page.html

@@ -4,23 +4,41 @@
 {% load static %}
 {% block body_class %}template-{{ self.get_verbose_name | slugify }}{% endblock %}
 
+
+
 {% block content %}
 
+<!--{% if "reports-about-fec" in request.path and page.content_type.model == "documentpage" and self.file_url %}
+    <meta http-equiv="refresh" content="0; url={% include_block self.file_url %}">
+{% endif %} -->
+
+
+
+
 {% include 'partials/breadcrumbs.html' with page=self links=self.get_ancestors style='secondary' %}
 
 <article class="main">
   <div class="container">
     <header class="heading--main heading--with-date">
       <ul class="tags">
         <li class="tag tag--secondary t-upper">
-          {{ self.category }}
+     <a href="{{self.get_parent.url}}?category={{ self.category | cms_query_syntax }}">{{ self.category }}</a>
         </li>
       </ul>
       <div class="row">
         {% spaceless %}{# for inline blocks #}
-        <h1 class="heading__left">{% formatted_title self %}</h1>
+      <h1 class="heading__left">{% if self.file_url %}<a href="{{ self.file_url }}">{% formatted_title self %}</a>
+      {% else %}
+       {% formatted_title self %}
+      {% endif %}</h1>
         <div class="heading__right">
-          <span class="t-sans">{{ self.display_date }}</span>
+          <span class="t-sans">{{ self.display_date }} |
+        {% if self.file_url %} <a href="{{ self.file_url }}">{{ self.extension }}</a>
+        {% else %}
+          {{ self.extension }}
+        {% endif %}
+        {% if self.size %} | ({{ self.size }}){% endif %}</span>
+
         </div>
         {% endspaceless %}
       </div>
@@ -31,5 +49,15 @@ <h1 class="heading__left">{% formatted_title self %}</h1>
 </article>
 
 {% include 'partials/disclaimer.html' %}
+<script>
+// document.getElementsByClassName('main')[0].insertAdjacentHTML("beforebegin", "<p>You will be directed</p>")
+// setTimeout(() => {
+// var meta = document.createElement('meta');
+// meta.httpEquiv = "refresh" ;
+// meta.content = "3; url={% include_block self.file_url %}";
+// document.getElementsByTagName('head')[0].appendChild(meta);
+// }, "0");
+
+</script>
 
 {% endblock %}

fec/home/templates/partials/document.html

@@ -29,6 +29,6 @@ <h3>
     </h3>
   {% endif %}
     <div class="post__meta">
-      <a class="tag tag--secondary" href="">{{ document.category }}</a>
+      <a class="tag tag--secondary" href="{{self.get_parent.url}}?category={{ document.category | cms_query_syntax }}">{{ document.category }}</a>
     </div>
 </article>

fec/home/templatetags/filters.py

@@ -189,3 +189,10 @@ def get_file_type(value):
     file_type = "EXCEL" if xl else file_extension
 
     return file_type
+
+
+@register.filter(name='cms_query_syntax')
+def cms_query_syntax(string):
+    query = string.replace(' ', '+').lower()
+
+    return query

tasks.py

@@ -74,7 +74,7 @@ def _detect_space(repo, branch=None, yes=False):
 DEPLOY_RULES = (
     ('prod', _detect_prod),
     ('stage', lambda _, branch: branch.startswith('release')),
-    ('dev', lambda _, branch: branch == 'develop'),
+    ('dev', lambda _, branch: branch == 'feature/innovation-document-form-sitemaps'),
     # Uncomment below and adjust branch name to deploy desired feature branch to the feature space
     # ('feature', lambda _, branch: branch == '[BRANCH NAME]'),
 )