Cloudflare Static Deployment

Relevant source files

The following files were used as context for generating this wiki page:

• frontend/README.md
• frontend/publish_snapshot.py
• frontend/chart_publication.py
• DO NOT DELETE EDL PIPELINE/src/edl_pipeline/runner.py
• DO NOT DELETE EDL PIPELINE/README.md

Cloudflare Static Deployment serves the Nexus Scanner frontend and its immutable data releases using Cloudflare Pages and R2 storage. The system builds a React-based research application and integrates it with a content-addressed data architecture generated by a Python-based Extraction, Delivery, and Loading (EDL) pipeline. Sources: frontend/README.md:1-10, frontend/README.md:465-470

The deployment utilizes a root wrangler.jsonc configuration to build and serve the static frontend for production environments and Pull Request (PR) previews. While the frontend logic runs in the browser, advanced historical queries depend on a separate, externally hosted Python backend. Sources: frontend/README.md:465-475

Architecture and Data Flow

The static deployment separates application logic from the underlying data snapshots. The frontend acts as a consumer of immutable JSON artifacts produced by the publish_snapshot.py script. Sources: frontend/publish_snapshot.py:1-10, frontend/README.md:210-215

Deployment Components

• Cloudflare Pages: Hosts the React 19 / Vite 8 frontend bundle. Sources: frontend/README.md:465-470, frontend/package.json:20-35
• Cloudflare R2: Stores immutable chart objects, historical candles, and archived releases. Sources: frontend/chart_publication.py:115-130, frontend/README.md:245-250
• Python EDL Pipeline: Runs periodically to generate the data snapshots and upload them to the static origin. Sources: DO NOT DELETE EDL PIPELINE/README.md:25-30, frontend/publish_snapshot.py:100-110

Logic Flow for Release Publication

The following diagram illustrates the process of promoting a new data release to the static deployment:

Immutable Data Structure

Data is deployed using a content-addressed directory structure. This ensures that browsers can cache artifacts indefinitely and that different scanner versions do not conflict. Sources: frontend/README.md:225-235, frontend/chart_publication.py:210-220

Directory Layout

Path	Purpose	Format
/data/current.json	Active-release pointer (published last)	JSON
/data/revisions/<hash>/stocks.json.gz	Mainboard stocks snapshot	Gzip JSON
/data/revisions/<hash>/ipos.json.gz	IPO catalogue for the release	Gzip JSON
/data/objects/<hash>.json.gz	Content-addressed chart/event objects	Gzip JSON
/data/revisions/<hash>/release.json	Revision metadata and URL templates	JSON

Sources: frontend/README.md:230-245, frontend/publish_snapshot.py:140-150, frontend/chart_publication.py:230-245

Deployment Configuration

Cloudflare deployment requires specific environment variables and build-time settings to link the static frontend to its data sources and optional backend. Sources: frontend/README.md:445-455

Environment Variables

• VITE_API_BASE_URL: Defines the base path for the historical evaluation service. The frontend appends /screens/run to this base. Sources: frontend/README.md:450-455, frontend/README.md:470-475
• EDL_CHART_STORAGE: Set to r2 to enable remote object storage; otherwise defaults to local. Sources: frontend/chart_publication.py:25-30
• R2_PUBLIC_BASE_URL: The HTTPS URL used by the browser to fetch chart objects from Cloudflare R2. Sources: frontend/chart_publication.py:135-140

Build and Deployment Procedure

1. Build Frontend: Run npm run build to generate the dist/ directory. Sources: frontend/README.md:95-100
2. Stage Data: The pipeline produces validated artifacts in frontend/public/data/revisions/. Sources: frontend/README.md:225-230
3. R2 Synchronization: rclone synchronizes local chart objects to the R2 bucket using the R2_ACCESS_KEY_ID and R2_SECRET_ACCESS_KEY. Sources: frontend/chart_publication.py:145-165
4. Promote Pointer: The pipeline updates current.json to point to the new revision hash. Sources: frontend/publish_snapshot.py:155-160

Limitations of Static Deployment

The Cloudflare Pages deployment is primarily static and does not execute Python code on the edge. Sources: frontend/README.md:105-110

• Browser-Based Evaluation: Only conditions supported by the Web Worker engine function without an external service. Sources: frontend/README.md:275-285
• External API Requirement: Expressions requiring historical lookbacks (e.g., specific year-over-year fundamental growth) must be proxied to a separate Python backend. Sources: frontend/README.md:480-485
• SPA Fallback: The deployment uses Single Page Application (SPA) fallback to support deep links, but this does not provide dynamic API endpoints. Sources: frontend/README.md:475-480

Sources: frontend/README.md:465-485, frontend/publish_snapshot.py:155-165

Cloudflare Static Deployment provides a scalable, cache-efficient hosting model for Nexus Scanner by separating application delivery from the data pipeline. While the core screener functions as a static site, it relies on strict immutable naming conventions and revision-pinned metadata to maintain data integrity across releases. Sources: frontend/README.md:265-275, frontend/chart_publication.py:195-205