posts: added updating feature to automatically update post pub dates

from wired RSS feed.

5 files changed, 50 insertions, 22 deletions

diff --git a/app/posts/management/__init__.py b/app/posts/management/__init__.py
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/app/posts/management/__init__.py
diff --git a/app/posts/management/commands/__init__.py b/app/posts/management/commands/__init__.py
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/app/posts/management/commands/__init__.py
diff --git a/app/posts/management/commands/rss_updater.py b/app/posts/management/commands/rss_updater.py
new file mode 100644
index 0000000..7ac26e6
--- /dev/null
+++ b/app/posts/management/commands/rss_updater.py
@@ -0,0 +1,42 @@
+from django.core.management.base import BaseCommand, CommandError
+
+import datetime
+import feedparser
+from urllib.parse import urlparse
+
+from posts.models import PostStatus, Post
+
+"""
+run from a cronscript that looks line this:
+
+*/1 * * * * cd /home/lxf/sites/wired.luxagraf.net && source /home/lxf/sites/wired.luxagraf.net/venv/bin/activate && /home/lxf/sites/wired.luxagraf.net/venv/bin/python /home/lxf/sites/wired.luxagraf.net/manage.py rss_updater --settings=config.settings
+"""
+
+class Command(BaseCommand):
+    help = "Update all published posts"
+
+    def is_deal(tags):
+        for tag in tags:
+            if tag['term'] == "Deals":
+                return True 
+        return False
+    
+    def handle(self, *args, **options):
+        feed = feedparser.parse("https://www.wired.com/feed/tag/commerce/latest/rss")
+        for item in feed.entries:
+            url = urlparse(item.link)
+            story_type = url.path.split('/')[1]
+            if story_type == "story" or "gallery":
+                if not self.is_deal(item.tags):
+                    try:
+                        post = Post.objects.get(url=item.link)
+                        post.date_last_pub = datetime.datetime.strptime(item.published, '%a, %d %b %Y %H:%M:%S %z').date()
+                        post.post_status = PostStatus.PUBLISHED
+                        post.save()
+                        self.stdout.write(
+                            self.style.SUCCESS('Successfully updated post "%s"' % post.title)
+                        )
+                    except:
+                        continue
+
+
diff --git a/app/posts/models.py b/app/posts/models.py
index f42956e..b518554 100644
--- a/app/posts/models.py
+++ b/app/posts/models.py
@@ -6,28 +6,6 @@ import settings
 
 from products.models import ProductLink
 
-"""
-class Feed(models.Model):
-    name = models.CharField(max_length=255)
-    feed_url = models.CharField(max_length=512)
-    slug = models.CharField(max_length=50)
-    last_polled = models.DateTimeField(blank=True, null=True)
-    due_poll = models.DateTimeField(default=datetime.datetime(1900, 1, 1)) # default to distant past to put new sources to front of queue
-    etag = models.CharField(max_length=255, blank=True, null=True)
-    last_modified = models.CharField(max_length=255, blank=True, null=True) # just pass this back and forward between server and me , no need to parse
-    last_result = models.CharField(max_length=255,blank=True,null=True)
-    interval = models.PositiveIntegerField(default=400)
-    last_success = models.DateTimeField(blank=True, null=True)
-    last_change = models.DateTimeField(blank=True, null=True)
-    live = models.BooleanField(default=True)
-    status_code = models.PositiveIntegerField(default=0)
-    last_302_url = models.CharField(max_length=512, null=True, blank=True)
-    last_302_start = models.DateTimeField(null=True, blank=True)
-    
-    def __str__(self):
-        return self.name
-"""
-
 
 class PostType(models.IntegerChoices):
     REVIEW = 0, ('review')
diff --git a/app/posts/views.py b/app/posts/views.py
index 25a8074..29796c4 100644
--- a/app/posts/views.py
+++ b/app/posts/views.py
@@ -8,6 +8,14 @@ from .models import Post
 from .forms import PostUpdateForm
 from notes.models import Note
 
+"""
+TODO:
+    crawl the RSS feed and compare url to stored url 
+    if it matches update the pub_date.
+    Then scrape the HTML.
+    Then parse the scraped html for aff links and create products from them
+"""
+
 class PostListView(LoginRequiredMixin, ListView):
     model = Post 
     template_name = 'posts/post_table.html'