lumbunglib/feed.py

@@ -85,6 +85,11 @@ def create_frontmatter(entry):
         for t in entry.tags:
             tags.append(t['term'])
 
+    if "featured_image" in entry:
+        featured_image = entry.featured_image
+    else:
+        featured_image = ''
+
     card_type = "network"
     if entry.feed_name == "pen.lumbung.space":
         card_type = "pen"
@@ -110,7 +115,8 @@ def create_frontmatter(entry):
         'original_link': entry.link,
         'feed_name': entry['feed_name'],
         'tags': str(tags),
-        'card_type': card_type
+        'card_type': card_type,
+        'featured_image': featured_image
         }
 
     return frontmatter
@@ -136,11 +142,33 @@ def sanitize_yaml (frontmatter):
 
     return frontmatter
 
+def parse_enclosures(post_dir, entry):
+    """
+    Parses feed enclosures which are featured media
+    Can be featured image but also podcast entries
+    https://pythonhosted.org/feedparser/reference-entry-enclosures.html
+    """
+    #TODO parse more than images
+    #TODO handle the fact it could be multiple items
+
+    for e in entry.enclosures:
+        if "type" in e:
+            print("found enclosed media", e.type)
+            if "image/" in e.type:
+                featured_image = grab_media(post_dir, e.href)
+                entry["featured_image"] = featured_image
+            else:
+                print("FIXME:ignoring enclosed", e.type)
+    return entry
+
 
 def create_post(post_dir, entry):
     """
     write hugo post based on RSS entry
     """
+    if "enclosures" in entry:
+        entry = parse_enclosures(post_dir, entry)
+
     frontmatter = create_frontmatter(entry)
 
     if not os.path.exists(post_dir):
@@ -169,18 +197,25 @@ def grab_media(post_directory, url, prefered_name=None):
     """
     media_item = urlparse(url).path.split('/')[-1]
 
+    headers = {
+    'User-Agent': 'https://git.autonomic.zone/ruangrupa/lumbunglib',
+    'From': 'info@lumbung.space'  # This is another valid field
+    }
     if prefered_name:
         media_item = prefered_name
 
     try:
         if not os.path.exists(os.path.join(post_directory, media_item)):
             #TODO: stream is true is a conditional so we could check the headers for things, mimetype etc
-            response = requests.get(url, stream=True)
+            response = requests.get(url, headers=headers, stream=True)
             if response.ok:
                 with open(os.path.join(post_directory, media_item), 'wb') as media_file:
                     shutil.copyfileobj(response.raw, media_file)
                     print('Downloaded media item', media_item)
                     return media_item
+            else:
+                print("Download failed", response.status_code)
+                return url
             return media_item
         elif os.path.exists(os.path.join(post_directory, media_item)):
             return media_item
@@ -235,6 +270,7 @@ def grab_feed(feed_url):
         print(e)
         return False
 
+    if "status" in data:
         print(data.status, feed_url)
         if data.status == 200:
             # 304 means the feed has not been modified since we last checked

lumbunglib/templates/feed.md

@@ -8,6 +8,7 @@ original_link: "{{ frontmatter.original_link }}"
 feed_name: "{{ frontmatter.feed_name}}"
 categories: ["{{ frontmatter.card_type }}", "{{ frontmatter.feed_name}}"]
 tags: {{ frontmatter.tags }}
+{% if frontmatter.featured_image %}featured_image: "{{frontmatter.featured_image}}"{% endif %}
 ---
 
 {{ content }}