comments

app/item/models.py

@@ -1,6 +1,8 @@
 from django.utils.timezone import datetime, timedelta
 from django.utils import timezone
 import json
+import requests
+import lxml.html
 
 from django.conf import settings
 from django.contrib.auth import get_user_model
@@ -31,6 +33,10 @@ class Item(models.Model):
     announced = models.DateTimeField(null=True, default=None, blank=True)
     data = models.JSONField(default=dict, editable=False)
 
+    def save(self, *args, **kwargs):
+        if self.url and not self.data:
+            self.update_data()
+        super().save(*args, **kwargs)
     def __str__(self):
         return '%s (%s)' % (self.title, self.url)
 
@@ -70,6 +76,26 @@ class Item(models.Model):
     def get_absolute_url(self):
         return reverse('item', kwargs={'id': self.id})
 
+    def update_data(self):
+        self.data.update(self.parse_url())
+
+    def parse_url(self):
+        content = requests.get(self.url).text
+        doc = lxml.html.fromstring(content)
+        data = {}
+        for meta in doc.cssselect('meta'):
+            key = meta.attrib.get('name')
+            if not key:
+                key = meta.attrib.get('property')
+            value = meta.attrib.get('content')
+            if key and value:
+                if key in ('viewport', ):
+                    continue
+                key = key.replace('og:', '')
+                data[key] = value
+        return data
+
+
 class Comment(models.Model):
     created = models.DateTimeField(auto_now_add=True)
     modified = models.DateTimeField(auto_now=True)
@@ -83,6 +109,10 @@ class Comment(models.Model):
     data = models.JSONField(default=dict, editable=False)
     published = models.DateTimeField(null=True, default=None)
 
+    @property
+    def is_published(self):
+        return bool(self.published)
+
     def __str__(self):
         return '%s: %s' % (self.item, self.user)