balanced author selection logic

2025-05-07 09:39:32 +10:00
parent 4da83f1d4b
commit f6ab7e78d3
4 changed files with 122 additions and 52 deletions
@@ -314,14 +314,36 @@ def curate_from_google_trends(geo_list=['US']):

            final_summary = insert_link_naturally(final_summary, source_name, link)

-            post_data, author, category, image_url, image_source, uploader, page_url = prepare_post_data(final_summary, title, main_topic)
-            if not post_data:
-                attempts += 1
-                continue
+            # Balanced author selection
+            x_post_counts = load_json_file('/home/shane/foodie_automator/x_post_counts.json', expiration_hours=24*30)
+            monthly_counts = {entry['username']: entry['monthly_count'] for entry in x_post_counts}
+            low_post_authors = [u for u, c in monthly_counts.items() if c < 3]

-            image_url, image_source, uploader, page_url = get_flickr_image(image_query, relevance_keywords, main_topic)
-            if not image_url:
-                image_url, image_source, uploader, page_url = get_image(image_query)
+            if low_post_authors:
+                author_username = random.choice(low_post_authors)
+                author = next(a for a in AUTHORS if a['username'] == author_username)
+                logging.info(f"Prioritizing low-post author: {author_username}")
+                post_data = {
+                    "title": generate_title_from_summary(final_summary),
+                    "content": final_summary,
+                    "status": "publish",
+                    "author": author_username,
+                    "categories": [generate_category_from_summary(final_summary)]
+                }
+                category = post_data["categories"][0]
+                image_url, image_source, uploader, page_url = get_flickr_image(image_query, relevance_keywords, main_topic)
+                if not image_url:
+                    image_url, image_source, uploader, page_url = get_image(image_query)
+                    if not image_url:
+                        logging.warning(f"All image uploads failed for '{title}' - posting without image")
+                        image_source = None
+                        uploader = None
+                        page_url = None
+            else:
+                post_data, author, category, image_url, image_source, uploader, page_url = prepare_post_data(final_summary, title, main_topic)
+                if not post_data:
+                    attempts += 1
+                    continue

            hook = get_dynamic_hook(post_data["title"]).strip()