add bypass paywall links for several sites #205

app/post/routes.py

@@ -15,7 +15,8 @@ from app.community.util import save_post, send_to_remote_instance
 from app.inoculation import inoculation
 from app.post.forms import NewReplyForm, ReportPostForm, MeaCulpaForm
 from app.community.forms import CreateLinkForm, CreateImageForm, CreateDiscussionForm, CreateVideoForm, CreatePollForm
-from app.post.util import post_replies, get_comment_branch, post_reply_count, tags_to_string
+from app.post.util import post_replies, get_comment_branch, post_reply_count, tags_to_string, url_has_paywall, \
+    generate_paywall_bypass_link
 from app.constants import SUBSCRIPTION_MEMBER, SUBSCRIPTION_OWNER, SUBSCRIPTION_MODERATOR, POST_TYPE_LINK, \
     POST_TYPE_IMAGE, \
     POST_TYPE_ARTICLE, POST_TYPE_VIDEO, NOTIF_REPLY, NOTIF_POST, POST_TYPE_POLL
@@ -299,6 +300,11 @@ def show_post(post_id: int):
             else:
                 poll_form = True
 
+    # Bypass paywalls link
+    bypass_paywall_link = None
+    if post.type == POST_TYPE_LINK and 'https://archive.' not in post.body_html and url_has_paywall(post.url):
+        bypass_paywall_link = generate_paywall_bypass_link(post.url)
+
     response = render_template('post/post.html', title=post.title, post=post, is_moderator=is_moderator, community=post.community,
                            breadcrumbs=breadcrumbs, related_communities=related_communities, mods=mod_list,
                            poll_form=poll_form, poll_results=poll_results, poll_data=poll_data, poll_choices=poll_choices, poll_total_votes=poll_total_votes,
@@ -306,7 +312,7 @@ def show_post(post_id: int):
                            description=description, og_image=og_image, POST_TYPE_IMAGE=constants.POST_TYPE_IMAGE,
                            POST_TYPE_LINK=constants.POST_TYPE_LINK, POST_TYPE_ARTICLE=constants.POST_TYPE_ARTICLE,
                            POST_TYPE_VIDEO=constants.POST_TYPE_VIDEO, POST_TYPE_POLL=constants.POST_TYPE_POLL,
-                           autoplay=request.args.get('autoplay', False),
+                           autoplay=request.args.get('autoplay', False), bypass_paywall_link=bypass_paywall_link,
                            noindex=not post.author.indexable, preconnect=post.url if post.url else None,
                            recently_upvoted=recently_upvoted, recently_downvoted=recently_downvoted,
                            recently_upvoted_replies=recently_upvoted_replies, recently_downvoted_replies=recently_downvoted_replies,

app/post/util.py

@@ -1,4 +1,5 @@
 from typing import List
+from urllib.parse import urlparse
 
 from flask_login import current_user
 from sqlalchemy import desc, text, or_
@@ -78,3 +79,22 @@ def post_reply_count(post_id) -> int:
 def tags_to_string(post: Post) -> str:
     if post.tags.count() > 0:
         return ', '.join([tag.name for tag in post.tags])
+
+
+def url_has_paywall(url) -> bool:
+    paywalled_sites = ['washingtonpost.com', 'wapo.st', 'nytimes.com', 'wsj.com', 'economist.com', 'ft.com', 'telegraph.co.uk',
+                       'bild.de', 'theatlantic.com', 'lemonde.fr']
+    if url:
+        try:
+            parsed_url = urlparse(url.replace('www.', ''))
+            hostname = parsed_url.hostname.lower()
+        except:
+            return False
+        return hostname in paywalled_sites
+    else:
+        return False
+
+
+def generate_paywall_bypass_link(url) -> bool:
+    url_without_protocol = url.replace('https://', '').replace('http://', '')
+    return 'https://archive.ph/' + url

app/templates/post/_post_full.html

@@ -144,6 +144,9 @@
             {% endif -%}
             <div class="post_body">
                 {{ post.body_html|community_links|safe if post.body_html else '' }}
+                {% if bypass_paywall_link -%}
+                    <p><a href="{{ bypass_paywall_link }}">{{ _('Bypass paywall') }}</a></p>
+                {% endif -%}
             </div>
             {% if post.type == POST_TYPE_POLL -%}
                 <div class="post_poll">