Pinterest-to-WordPress content automation toolkit. Scrapes Pinterest Trends and PinClicks data, uses LLM analysis (DeepSeek) to identify high-signal keywords, generates SEO-optimized articles with AI images (Fal.ai), validates and repairs content quality, publishes to WordPress, and exports Pinterest bulk-upload CSVs.

Supports three blog brands and offers both a Streamlit UI and a CLI interface.

• Python >= 3.11
• Playwright browsers (for scraping)
• API keys: DeepSeek, Fal.ai, WordPress REST API credentials

pip install -e .
playwright install chromium

Copy .env.example to .env and fill in your credentials.

# Launch the Streamlit web app (main interface)
streamlit run src/automating_wf/ui/streamlit_app.py

# Or if streamlit isn't on PATH:
python -m streamlit run src/automating_wf/ui/streamlit_app.py

The app has two tabs:

• Single Article — generate an individual SEO blog draft and publish to WordPress
• Bulk Pipeline — run the full Trends → PinClicks → Generation → Publish workflow

# Pinterest bulk pipeline (headless)
pinterest-engine --blog THE_SUNDAY_PATIO [seed keywords...]

# Resume a previous run
pinterest-engine --blog THE_SUNDAY_PATIO --resume 20260224_143000

# WordPress site onboarding (categories, pages, theme, plugins)
wp-onboarding --help

# Build WordPress theme ZIP
python scripts/build_theme_zip.py

# Run tests
python -m pytest -q

Automating_WF/
├── src/automating_wf/          # Application package
│   ├── analysis/               # LLM keyword analysis, PinClicks ranking, Trends ranking
│   ├── config/                 # Blog profiles, category keywords, artifact paths
│   ├── content/                # Article generation, validation/repair, single-article flow
│   ├── design/                 # Pillow-based Pinterest pin image rendering
│   ├── engine/                 # Pipeline orchestration (config + 3-phase pipeline)
│   ├── export/                 # Pinterest bulk-upload CSV builder
│   ├── models/                 # Dataclasses (PinRecord, BrainOutput, CsvRow, etc.)
│   ├── prompts/                # LLM system prompts (analysis + validator)
│   ├── scrapers/               # Playwright scrapers (PinClicks, Trends, subprocess runner)
│   ├── ui/                     # Streamlit app + bulk pipeline wizard
│   └── wordpress/              # WP REST API uploader + onboarding
├── tests/                      # pytest suite
├── docs/                       # Documentation and reference
├── scripts/                    # Build utilities (theme ZIP builder)
├── assets/theme/               # WordPress theme source files (The Sunday Patio)
└── artifacts/                  # Runtime output (gitignored)

1. Configuration — Select blog, enter seed keywords, set filters (region, range, top-N), target article count
2. Trends Collection — Playwright scrapes Pinterest Trends → CSV parsing → hybrid ranking (trend index 50%, growth 30%, consistency 20%)
3. PinClicks Analysis — Playwright scrapes PinClicks top-pins per keyword → scoring by frequency (50%), engagement (35%), intent (15%)
4. Generation + Publishing — For each winner keyword:
   • LLM analysis → BrainOutput (primary keyword, image prompt, pin text)
   • DeepSeek article generation with SEO validation
   • LLM repair loop for content quality
   • Pillow renders 1000×1500 Pinterest pin image
   • Fal.ai generates hero image
   • Publishes to WordPress via REST API
   • Appends row to per-run Pinterest bulk-upload CSV

Blog selection → vibe suggestions → topic input → category suggestion → article generation with validation → image generation → preview → publish to WordPress.

See .env.example for the full list. Key groups:

• Per-run Pinterest CSV: tmp/pinterest_engine/<run_id>/pinterest_bulk_upload_<blog_suffix>.csv
• Theme ZIP: artifacts/theme/the-sunday-patio.zip
• Run artifacts: tmp/pinterest_engine/<run_id>/ (manifest, trends, pinclicks, images)

Artifact base paths can be overridden via environment variables (see src/automating_wf/config/paths.py).

Symptom: You stop Streamlit and see Stopping..., then a traceback ending with RuntimeError: Event loop is closed.

This is a benign shutdown-time warning if your expected artifacts were generated. Press Ctrl+C once, then wait for shutdown. Do not press Ctrl+C repeatedly.

If the terminal appears stuck, end the process from Task Manager or taskkill /PID <streamlit_pid> /F.