Essential Artists Dayton — Catalog Tools

Python tools for generating a PDF art catalog and social media images from the Essential Artists Dayton Wix CMS.

What's here

Script / App	Purpose
main.py	Generate the PDF catalog (ArtistCatalog.pdf)
extract_text.py	Pull artist bios and statements out of Artists.csv into text files
generate_summaries.py	Use Claude (or a local Ollama model) to write short social media summaries
social/social.py	Render Instagram and Facebook images for all artists (or one by name)
process_artist_url.py	Generate images for a single artist directly from their website URL
app.py	Web app — artists self-serve their own social images

Setup

Requires Python 3.11 and a virtual environment:

python3.11 -m venv env
source env/bin/activate
pip install -r requirements.txt

Set your Anthropic API key (required for summary generation):

export ANTHROPIC_API_KEY=sk-ant-...

Web app

The simplest way for artists to generate their own images. Run locally:

source env/bin/activate
python app.py

Then open http://localhost:5000. Enter an artist page URL (e.g. https://www.essentialartistsdayton.org/artist/amy-kollar-anderson), review and edit the description, and download Instagram and Facebook images.

For production deployment see Deployment below.

Batch workflow (new artist cohort)

Run these steps after exporting a fresh Artists.csv from Wix CMS:

source env/bin/activate

# 1. Extract bios and statements from CSV into text files
python extract_text.py

# 2. Generate AI summaries for any artists that don't have one yet
python generate_summaries.py

# 3a. Generate social images for all artists
python social/social.py

# 3b. Or generate for a single artist
python social/social.py "Artist Name"

Social images are written to social/output/<Artist_Name>/instagram.png and facebook.png.

PDF catalog

source env/bin/activate
python main.py

Outputs ArtistCatalog.pdf.

Single-artist URL workflow

Process one new artist directly from their website without a CSV export:

source env/bin/activate
python process_artist_url.py https://www.essentialartistsdayton.org/artist/artist-slug

Summary generation options

generate_summaries.py is incremental — it skips artists that already have summaries.

# Use Claude (default)
python generate_summaries.py

# Use a local Ollama model
python generate_summaries.py --provider ollama
python generate_summaries.py --provider ollama --model llama3.2

Deployment

The web app (app.py) is configured for Render via Procfile.

1. Push this repo to GitHub.
2. Create a new Web Service on Render pointed at the repo.
3. Set the environment variable ANTHROPIC_API_KEY in Render's dashboard.
4. Render will run gunicorn app:app automatically.

The app checks text/artist_summaries.json for existing summaries before calling the Claude API, so artists who've already been processed get instant results.

Project structure

lib/
  artists.py      CSV parsing and image URL extraction
  images.py       Image download, caching, EXIF rotation
  text.py         Bio/statement loading, summary loading
  wix.py          Wix image URL → CDN URL
social/
  social.py       Social image renderer
  output/         Generated images (gitignored)
templates/
  index.html      Web app UI
text/
  bios/           Per-artist bio text files
  statements/     Per-artist statement text files
  all_texts.txt   Combined text file for summary generation
  artist_summaries.json  AI-generated summaries (short / medium / long)
image_cache/      Downloaded image cache (gitignored)