From 22b17d08fafb0b554276adf7470f2bec7aa30298 Mon Sep 17 00:00:00 2001
From: eric <eric@hellman.net>
Date: Sat, 9 Mar 2013 11:52:42 -0500
Subject: [PATCH 1/6] this was causing googlebot crawl errors

---
 core/feeds.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/core/feeds.py b/core/feeds.py
index 64abeb7b..6df720a1 100644
--- a/core/feeds.py
+++ b/core/feeds.py
@@ -15,7 +15,7 @@ class SupporterWishlistFeed(Feed):
         return "Latest wishbooks for %s on unglue.it" % obj.username
         
     def link(self, obj):
-        return "/%s/feed/" % obj.username
+        return "/supporter/%s/feed/" % obj.username
 
     def item_title(self, item):
         return "%s" % item.title

From d3077c988bc819dbe4a28261e097af57ca31e8da Mon Sep 17 00:00:00 2001
From: eric <eric@hellman.net>
Date: Sat, 9 Mar 2013 13:29:03 -0500
Subject: [PATCH 2/6] links should not point to http from https

---
 core/signals.py               | 6 +++---
 frontend/templates/about.html | 2 +-
 frontend/views.py             | 6 +++---
 payment/manager.py            | 6 +++---
 settings/common.py            | 2 +-
 5 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/core/signals.py b/core/signals.py
index b1ddda8a..b109f500 100644
--- a/core/signals.py
+++ b/core/signals.py
@@ -257,7 +257,7 @@ def handle_wishlist_added(supporter, work, **kwargs):
         notification.queue([claim[0].user], "new_wisher", {
             'supporter': supporter,
             'work': work,
-            'base_url': settings.BASE_URL,
+            'base_url': settings.BASE_URL_SECURE,
         }, True)
         
         from regluit.core.tasks import emit_notifications
@@ -277,13 +277,13 @@ def handle_wishlist_near_deadline(campaign, **kwargs):
     
     notification.queue(pledgers, "wishlist_near_deadline", {
             'campaign': campaign,
-            'domain': settings.BASE_URL,
+            'domain': settings.BASE_URL_SECURE,
             'pledged': True,
     }, True)
     
     notification.queue(nonpledgers, "wishlist_near_deadline", {
             'campaign': campaign,
-            'domain': settings.BASE_URL,
+            'domain': settings.BASE_URL_SECURE,
             'pledged': False,
     }, True)
 
diff --git a/frontend/templates/about.html b/frontend/templates/about.html
index 155e2e23..84bdcd85 100644
--- a/frontend/templates/about.html
+++ b/frontend/templates/about.html
@@ -2,7 +2,7 @@
 {% block title %} Everything You Always Wanted to Know {% endblock %}
 {% block doccontent %}
 <h2>About</h2>
-<p><a href="http://unglue.it">Unglue.it</a> is a service provided by <a href="http://gluejar.com">Gluejar, Inc.</a>  It's a place for individuals and institutions to join together to liberate specific ebooks and other types of digital content by paying rights holders to relicense their works under <a href="http://creativecommons.org">Creative Commons</a> licenses.</p>
+<p><a href="https://unglue.it">Unglue.it</a> is a service provided by <a href="http://gluejar.com">Gluejar, Inc.</a>  It's a place for individuals and institutions to join together to liberate specific ebooks and other types of digital content by paying rights holders to relicense their works under <a href="http://creativecommons.org">Creative Commons</a> licenses.</p>
 
 <p>What does this mean?</p>
 <ul>
diff --git a/frontend/views.py b/frontend/views.py
index 7970cc7f..f6f6ba59 100755
--- a/frontend/views.py
+++ b/frontend/views.py
@@ -912,7 +912,7 @@ class NonprofitCampaign(FormView):
         forward['amount']= int(amount)
         forward['sent']= Sent.objects.create(user=username,amount=form.cleaned_data['preapproval_amount']).pk
         token=signing.dumps(forward)
-        return HttpResponseRedirect(settings.BASE_URL + reverse('donation_credit',kwargs={'token':token}))
+        return HttpResponseRedirect(settings.BASE_URL_SECURE + reverse('donation_credit',kwargs={'token':token}))
 
 class DonationCredit(TemplateView):
     template_name="donation_credit.html"
@@ -2025,7 +2025,7 @@ def emailshare(request, action):
                 next = form.cleaned_data['next']
             except:
                 # if we totally failed to have a next value, we should still redirect somewhere useful
-                next = 'http://unglue.it'
+                next = 'https://unglue.it'
             return HttpResponseRedirect(next)
             
     else: 
@@ -2180,7 +2180,7 @@ def download(request, work_id):
         'unglued_ebooks': unglued_ebooks,
         'other_ebooks': other_ebooks,
         'readmill_epub_url': readmill_epub_url,
-        'base_url': settings.BASE_URL
+        'base_url': settings.BASE_URL_SECURE
     })
     
     return render(request, "download.html", context)
diff --git a/payment/manager.py b/payment/manager.py
index c89c2057..478a66ed 100644
--- a/payment/manager.py
+++ b/payment/manager.py
@@ -565,7 +565,7 @@ class PaymentManager( object ):
         if return_url is None:
             return_path = "{0}?{1}".format(reverse('pledge_complete'), 
                                 urllib.urlencode({'tid':transaction.id})) 
-            return_url = urlparse.urljoin(settings.BASE_URL, return_path)
+            return_url = urlparse.urljoin(settings.BASE_URL_SECURE, return_path)
         
         p = transaction.get_payment_class().Preapproval(transaction, transaction.max_amount, expiry, return_url=return_url, paymentReason=paymentReason) 
        
@@ -657,7 +657,7 @@ class PaymentManager( object ):
             credit.pledge_transaction(t,user,amount)
             return_path = "{0}?{1}".format(reverse('pledge_complete'), 
                                 urllib.urlencode({'tid':t.id})) 
-            return_url = urlparse.urljoin(settings.BASE_URL, return_path)
+            return_url = urlparse.urljoin(settings.BASE_URL_SECURE, return_path)
             pledge_created.send(sender=self, transaction=t)
             return t, return_url
         else:
@@ -762,7 +762,7 @@ class PaymentManager( object ):
                 credit.pledge_transaction(transaction,transaction.user,amount)
                 return_path = "{0}?{1}".format(reverse('pledge_complete'), 
                                     urllib.urlencode({'tid':transaction.id})) 
-                return_url = urlparse.urljoin(settings.BASE_URL, return_path)
+                return_url = urlparse.urljoin(settings.BASE_URL_SECURE, return_path)
                 
                 logger.info("Updated amount of transaction to %f" % amount)
                 pledge_modified.send(sender=self, transaction=transaction,up_or_down=up_or_down)
diff --git a/settings/common.py b/settings/common.py
index 4b9d391f..6c20731f 100644
--- a/settings/common.py
+++ b/settings/common.py
@@ -216,7 +216,7 @@ LOGIN_URL = "/accounts/login/"
 LOGIN_REDIRECT_URL = "/"
 LOGOUT_URL = "/accounts/logout/"
 
-USER_AGENT = "unglue.it.bot v0.0.1 <http://unglue.it>"
+USER_AGENT = "unglue.it.bot v0.0.1 <https://unglue.it>"
 
 SOUTH_TESTS_MIGRATE = True
 

From c037764289eca46b4efabdf12e36948291e4b94f Mon Sep 17 00:00:00 2001
From: eric <eric@hellman.net>
Date: Sat, 9 Mar 2013 17:37:33 -0500
Subject: [PATCH 3/6] sitemaps! [#25581139]

---
 core/models.py     |  5 +++++
 core/sitemaps.py   | 36 ++++++++++++++++++++++++++++++++++++
 settings/common.py |  1 +
 urls.py            | 11 +++++++++++
 4 files changed, 53 insertions(+)
 create mode 100644 core/sitemaps.py

diff --git a/core/models.py b/core/models.py
index d2fc1b56..1dab0bb9 100755
--- a/core/models.py
+++ b/core/models.py
@@ -13,9 +13,11 @@ from django.db import models
 from django.db.models import Q, get_model
 from django.contrib.auth.models import User
 from django.contrib.sites.models import Site
+from django.core.urlresolvers import reverse
 from django.conf import settings
 from django.utils.translation import ugettext_lazy as _
 
+
 import regluit
 import regluit.core.isbn
 from regluit.core.signals import successful_campaign, unsuccessful_campaign, wishlist_added
@@ -808,6 +810,9 @@ class Work(models.Model):
         except:
             return False
 
+    def get_absolute_url(self):
+        return reverse('work', args=[str(self.id)])
+        
 class Author(models.Model):
     created = models.DateTimeField(auto_now_add=True)
     name = models.CharField(max_length=500)
diff --git a/core/sitemaps.py b/core/sitemaps.py
new file mode 100644
index 00000000..64606a76
--- /dev/null
+++ b/core/sitemaps.py
@@ -0,0 +1,36 @@
+from django.contrib.sitemaps import Sitemap
+from django.core.urlresolvers import reverse
+from regluit.core.models import Work, Edition
+
+class WorkSitemap(Sitemap):
+    protocol = 'https'
+
+    def items(self):
+        return Work.objects.all()
+        
+    def priority(self,work):
+        if work.last_campaign():
+            return '1.0'
+        if work.num_wishes>1000:
+            return '0.8'
+        if work.num_wishes>100:
+            return '0.6'
+        if work.num_wishes>10:
+            return '0.4'
+        if work.num_wishes>1:
+            return '0.3'
+        if work.num_wishes==1:
+            return '0.2'
+        if work.num_wishes==0:
+            return '0.1'
+        return '0.1'
+
+class PublisherSitemap(Sitemap):
+    priority = 0.2
+    protocol = 'https'
+            
+    def items(self):
+        return Edition.objects.exclude(publisher__isnull=True).exclude(publisher="").order_by('publisher').values('publisher').distinct()
+    
+    def location(self, pub):
+        return reverse("bypub_list",args=[pub['publisher']])
diff --git a/settings/common.py b/settings/common.py
index 6c20731f..efe35af3 100644
--- a/settings/common.py
+++ b/settings/common.py
@@ -119,6 +119,7 @@ INSTALLED_APPS = (
     'django.contrib.contenttypes',  
     'django.contrib.sessions',
     'django.contrib.sites',
+    'django.contrib.sitemaps',
     'django.contrib.messages',
     'django.contrib.staticfiles',
     'django.contrib.comments',
diff --git a/urls.py b/urls.py
index 39f96633..e9650444 100755
--- a/urls.py
+++ b/urls.py
@@ -3,6 +3,12 @@ from frontend.forms import ProfileForm
 from django.views.generic.simple import direct_to_template
 from regluit.admin import admin_site
 import notification.urls
+from regluit.core.sitemaps import WorkSitemap, PublisherSitemap
+
+sitemaps = {
+        'works': WorkSitemap,
+        'publishers': PublisherSitemap,
+    }
 
 urlpatterns = patterns('',
     url(r'^accounts/activate/complete/$','django.contrib.auth.views.login',
@@ -23,3 +29,8 @@ urlpatterns = patterns('',
     (r'^notification/', include(notification.urls)),
     (r'^ckeditor/', include('ckeditor.urls')),
 )
+
+urlpatterns += patterns('django.contrib.sitemaps.views',
+    (r'^sitemap\.xml$', 'index', {'sitemaps': sitemaps}),
+    (r'^sitemap-(?P<section>.+)\.xml$', 'sitemap', {'sitemaps': sitemaps}),
+)

From f3543bd91427c6bc0821ff47d659e6e4aa2006b0 Mon Sep 17 00:00:00 2001
From: eric <eric@hellman.net>
Date: Sat, 9 Mar 2013 17:38:47 -0500
Subject: [PATCH 4/6] googlebot likes the wishlist feeds- give it more

---
 core/feeds.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/core/feeds.py b/core/feeds.py
index 6df720a1..002ba72b 100644
--- a/core/feeds.py
+++ b/core/feeds.py
@@ -24,4 +24,4 @@ class SupporterWishlistFeed(Feed):
         return "/work/%s" % item.id
 
     def items(self, obj):
-        return obj.wishlist.works.all().order_by('-id')[:5]    
\ No newline at end of file
+        return obj.wishlist.works.all().order_by('-id')[:100]    
\ No newline at end of file

From 55b30f494ed0eb03259b9846412b64632254413f Mon Sep 17 00:00:00 2001
From: eric <eric@hellman.net>
Date: Sat, 9 Mar 2013 20:25:47 -0500
Subject: [PATCH 5/6] schema.org metadata!

---
 frontend/templates/comments/list.html | 22 +++++++++++++---------
 frontend/templates/supporter.html     |  4 ++--
 frontend/templates/work.html          | 22 +++++++++++-----------
 3 files changed, 26 insertions(+), 22 deletions(-)

diff --git a/frontend/templates/comments/list.html b/frontend/templates/comments/list.html
index 879bca73..5fcd093e 100644
--- a/frontend/templates/comments/list.html
+++ b/frontend/templates/comments/list.html
@@ -1,16 +1,20 @@
 <div id="comments">
   {% for comment in comment_list reversed %}
   
-  <div class="work_supporter {% if comment.content_object.last_campaign and comment.user in comment.content_object.last_campaign.managers.all %}official{% endif %}">
-      <a href="{% url supporter supporter_username=comment.user.username %}">
-      <div class="work_supporter_avatar">
-        {% if comment.user.profile.pic_url %}
-            <img class="user-avatar" src="{{ comment.user.profile.pic_url }}" height="50" width="50" alt="Picture of {{ comment.user }}" title="{{ comment.user }}" />
-               {% else %}
-            <img class="user-avatar" src="/static/images/header/avatar.png" height="50" width="50" alt="Generic Ungluer Avatar" title="Ungluer" />
+  <div itemprop="comment" itemscope itemtype="http://schema.org/UserComments" class="work_supporter {% if comment.content_object.last_campaign and comment.user in comment.content_object.last_campaign.managers.all %}official{% endif %}">
+        <span itemprop="creator" itemscope itemtype="http://schema.org/Person">
+        <a itemprop="url" href="{% url supporter supporter_username=comment.user.username %}">
+            <div class="work_supporter_avatar">
+            {% if comment.user.profile.pic_url %}
+                <img class="user-avatar" src="{{ comment.user.profile.pic_url }}" height="50" width="50" alt="Picture of {{ comment.user }}" title="{{ comment.user }}" />
+            {% else %}
+                <img class="user-avatar" src="/static/images/header/avatar.png" height="50" width="50" alt="Generic Ungluer Avatar" title="Ungluer" />
             {% endif %}
-    </div>
-    <span class="comment_username">{{comment.user.username }}</span></a> <span>({{ comment.submit_date }})</span> <br /><span class="comment">{{ comment.comment|linebreaksbr }}<br /></span> 
+            </div>
+            <span class="comment_username">{{comment.user.username }}</span>
+        </a>
+        </span>
+      <span itemprop="commentTime">({{ comment.submit_date }})</span> <br /><span class="comment" itemProp="commentText">{{ comment.comment|linebreaksbr }}<br /></span> 
   </div>  
   {% endfor %}
 </div>
diff --git a/frontend/templates/supporter.html b/frontend/templates/supporter.html
index 13f863f7..fcbe2ade 100644
--- a/frontend/templates/supporter.html
+++ b/frontend/templates/supporter.html
@@ -96,7 +96,7 @@ there's no tab for seeing ALL my books, only the filters! huh.
             <div class="js-topnews1">
                 <div class="js-topnews2">
                     <div class="js-topnews3">
-                        <div class="user-block">
+                        <div class="user-block" itemscope itemtype="http://schema.org/Person">
                             <div id="user-block1">
                                 <div class="block-inner">
                                     {% if supporter.profile.pic_url %}
@@ -105,7 +105,7 @@ there's no tab for seeing ALL my books, only the filters! huh.
                                         <img class="user-avatar" src="/static/images/header/avatar.png" height="50" width="50" alt="Generic Ungluer Avatar" title="Ungluer" />
                                     {% endif %}
                                     <span class="user-name">
-                                        <a href="#">{{ supporter.username }}</a>
+                                        <a href="#"><span itemprop="name">{{ supporter.username }}</span></a>
                                     </span>
                                 </div>
                                 <span class="user-badges">
diff --git a/frontend/templates/work.html b/frontend/templates/work.html
index 291e7d46..88bec9e5 100644
--- a/frontend/templates/work.html
+++ b/frontend/templates/work.html
@@ -139,7 +139,7 @@ $j(document).ready(function(){
         </div>
         <div id="js-maincol">
         <div class="js-maincol-inner">
-            <div id="content-block">
+            <div id="content-block" itemscope itemtype="http://schema.org/Book">
                 <div class="book-detail">
                     {% if work.googlebooks_id %}
                     <div id="book-detail-img"><a href="{{ work.googlebooks_url }}">
@@ -147,16 +147,16 @@ $j(document).ready(function(){
                     </div>
                     {% else %}
                     <div id="book-detail-img">
-                        <img src="{% if work.cover_image_thumbnail %}{{ work.cover_image_thumbnail }}{% else %}/static/images/generic_cover_larger.png{% endif %}" alt="{{ work.title }}" title="{{ work.title }}" width="131" height="192" />
+                        <img itemprop="image" src="{% if work.cover_image_thumbnail %}{{ work.cover_image_thumbnail }}{% else %}/static/images/generic_cover_larger.png{% endif %}" alt="{{ work.title }}" title="{{ work.title }}" width="131" height="192" />
                     </div>
                     {% endif %}
                     <div class="book-detail-info">
                       <div class="layout">
-                        <h2 class="book-name">{{ work.title }}</h2>
+                        <h2 class="book-name" itemprop="name">{{ work.title }}</h2>
                         <div>
                           <div class="pubinfo">
-                            <h3 class="book-author">{{ work.author }}</h3>
-                            <h3 class="book-year">{{ work.publication_date_year }}</h3>
+                            <h3 class="book-author" itemprop="author">{{ work.author }}</h3>
+                            <h3 class="book-year" itemprop="datePublished">{{ work.publication_date_year }}</h3>
 					      </div>
                           {% if status == 'ACTIVE' %}
                           {% if pledged %}
@@ -249,10 +249,10 @@ $j(document).ready(function(){
                        <div id="tabs-1" class="tabs {% if activetab == '1' %}active{% endif %}">
                     <div class="tabs-content">
                         {% if status == 'ACTIVE' or status == 'SUCCESSFUL' %}
-                        {{ work.last_campaign.description|safe }}
+                        <span  itemprop="description">{{ work.last_campaign.description|safe }}</span>
                         {% else %}
                         <h3 class="tabcontent-title">{{work.title}}</h3>
-                        <p>{{ work.description|safe }}
+                        <p itemprop="description">{{ work.description|safe }}
                         {% endif %}
                         </p>
                     </div>
@@ -272,8 +272,8 @@ $j(document).ready(function(){
                     <div class="tabs-content">
                         {% for wish in work.wishes.all reversed %}
                         {% with wish.wishlist.user as supporter %}
-                        	<div class="work_supporter_nocomment">
-                        		<a href="{% url supporter supporter_username=supporter.username %}">
+                        	<div class="work_supporter_nocomment" itemscope itemtype="http://schema.org/Person">
+                        		<a itemprop="url" href="{% url supporter supporter_username=supporter.username %}">
                               		<div class="work_supporter_avatar">
                                 		{% if supporter.profile.pic_url %}
                                     		<img class="user-avatar" src="{{ supporter.profile.pic_url }}" height="50" width="50" alt="Picture of {{ supporter }}" title="{{ supporter }}" />
@@ -351,7 +351,7 @@ $j(document).ready(function(){
                         <h4>Subjects</h4>
                         <ul>
                         {% for subject in work.subjects.all %}
-                            <li>{{ subject.name }}</li>
+                            <li itemprop="keywords">{{ subject.name }}</li>
                         {% endfor %}
                         </ul>
                         {% endif %}
@@ -373,7 +373,7 @@ $j(document).ready(function(){
                                     <div class="metadata" id="edition_{{edition.id}}">{% if edition.publisher %}Publisher: <a href="{% url bypub_list edition.publisher %}">{{edition.publisher}}</a><br />{% endif %}
                                         {% if edition.publication_date %}Published: {{edition.publication_date}}<br />{% endif %}
                                             {% if edition.isbn_13 %}
-                                                ISBN: {{ edition.isbn_13 }}<br />
+                                                ISBN: <spam itemprop="isbn">{{ edition.isbn_13 }}</span><br />
                                             {% endif %}
                                             {% if edition.oclc %}
                                                 OCLC: <a href="http://www.worldcat.org/oclc/{{ edition.oclc }}">{{ edition.oclc }}</a><br />

From 9dbdafed7d20791d1cd63a6d48dc25be038e0a00 Mon Sep 17 00:00:00 2001
From: eric <eric@hellman.net>
Date: Fri, 15 Mar 2013 12:32:29 -0400
Subject: [PATCH 6/6] There IS a setting for the items per page!

---
 core/sitemaps.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/core/sitemaps.py b/core/sitemaps.py
index 64606a76..b676d79d 100644
--- a/core/sitemaps.py
+++ b/core/sitemaps.py
@@ -4,6 +4,7 @@ from regluit.core.models import Work, Edition
 
 class WorkSitemap(Sitemap):
     protocol = 'https'
+    limit = 10000
 
     def items(self):
         return Work.objects.all()