Changelog

This update introduces automated Facebook sharing cleanup utilities, fixes Graph API deprecations, aligns auto-share paths, and consolidates the core pipeline documentation.

📲 Facebook Sharing & Ledgers

• Automated Cleaner Tool: Introduced clean_fb.py to recursively match active website articles, clean up orphaned records in local JSON ledgers (shared.json, pending_shares.json, articles.json), and automatically delete matching posts from the Facebook Page feed and schedule queue.
• Graph API Deprecation Fix: Resolved deprecate_post_aggregated_fields_for_attachement errors by querying the Graph API’s attachments nested fields instead of the deprecated link parameter.

️ Auto-Share Refactoring

• Partitioned News Paths: Aligned facebook_publisher.py and scrape_news.py to read and write articles dynamically from the new partitioned directory structures (src/content/news/[scope]/) instead of the deprecated src/content/blog/ path.
• Trigger Alignment: Updated .github/workflows/facebook_share.yml to trigger deployments and staggered sharing workflows on changes under src/content/news/**.

📚 Documentation & Orchestrator Sync

• Engine Doc Consolidation: Moved detailed phase design guides (phase_1.md through phase_5.md) and ARCHITECTURE.md to the core infra engine repository under docs/ and integrated links in the engine’s main README.
• Worker Scope Isolation: Rewrote SOURCES.md in each worker repository (theHmars-local, theHmars-national, theHmars-global) to exclusively document feed registries belonging to that worker’s active geographic tier.
• Run Schedule Exposure: Documented the central pipeline orchestrator cron run schedule (executing daily at 05:53, 13:53, and 21:53 IST) in both the website and engine README files.

This update introduces critical stability fixes to the automated crawling, deduplication, and production pipelines, and resolves content duplication issues.

Content & Deduplication

• Duplicate Clean Up: Inspected and removed 13 duplicate markdown posts from the news archive to maintain high editorial quality.
• Improved Deduplication Sorting: Updated the deduplication agent logic to sort posts alphabetically before checking historical content.
• Log Persistence: Updated .gitignore to track url logs in data/1/3 across scopes to persist crawled URL histories across workflows.

Workflow & Pipeline Optimization

• Actions Execution Order: Reordered GitHub workflows to checkout the website repository before running the curation engine, ensuring deduplication runs against accurate live state.
• Production Limiter: Modified the writer agent with a --limit option to allow controlled production targets (e.g. 1 article limit).
• Preflight Tooling Fixes: Added automatic start-time reset to prevent timeouts on delayed runs and resolved missing assembler.py file copy errors.

The project has transitioned from a personal journal to a high-volume, professional news outlet named thehmars. This update introduces a hierarchical news architecture, automated content pipelines, and a complete design overhaul.

🌿 Rebranding & Identity

• Identity Shift: Rebranded from BatchNode to thehmars, a news portal dedicated to global and Northeast Indian reporting.
• Visual Identity: Deployed a “White & Red” editorial theme.
• Custom Assets: Created a new “TH” SVG favicon and social sharing assets.
• Branding Logic: Unified all components to use the the<span class="text-red-600">hmars</span> branding pattern.

News Portal Architecture

• Hierarchical Taxonomy: Reorganized content into International, National, Local, and Editorial categories.
• Regional Desks: Implemented a segmented Local news system covering 8 Northeast India states (Manipur, Mizoram, Assam, etc.) with dedicated hubs.
• Flat Routing: Optimized SEO by removing the /blog prefix; articles now reside at /[id] and hubs at /[category].
• Headline Mosaic: Deployed a dynamic, non-duplicating grid system for homepage and category hubs.
• Topic Discovery: Introduced a new /topics system for cross-category reporting.

Content Intelligence

• Article Sidebar: Added a sticky dual-column layout for reports featuring Source Desk identification and Related Content discovery.
• Dynamic TOC: Implemented an automatic Table of Contents for long-form news reports.
• SEO & Schema: Integrated JSON-LD NewsArticle structured data and optimized robots.txt.
• Archive System: Unified historical records under /archives with advanced filtering by category, region, and tag.

Automation & Backend

• News Scraper Pipeline: Migrated 130+ high-quality reports from major outlets (BBC, The Hindu, EastMojo) to seed the portal.
• GitHub Actions Roadmap: Laid groundwork for .github/workflows/scrape.yml to automate 4x daily updates.
• Supabase Sync: Confirmed integration with Supabase Free Tier for supporter tracking and dynamic metadata.
• Razorpay Live: Finalized technical configuration for live payment processing.

🎨 UI & Design System

• Editorial Crimson Theme: Applied a sophisticated red accent system (#DC2626) across the interface.
• UI Symmetry: Achieved vertical and horizontal symmetry between navigation elements and content feeds.
• Rounded UI Patterns: Updated Sub-Header and card components with rounded-pill and ultra-tight alignment styles.
• Contact Refinement: Updated the contact system with Formspree AJAX integration and custom success modals.