letters/generate_web_data.py

#!/usr/bin/env python3
"""Generate combined JSON data files for the hicalsoft.github.io web UI.

Reads individual JSON files from letters/ and poetry/ directories and
produces the combined letters.json and poetry.json used by the web pages.

Usage:
    python3 generate_web_data.py              # Generate both
    python3 generate_web_data.py --letters    # Letters only
    python3 generate_web_data.py --poetry     # Poetry only
"""

import argparse
import glob
import json
import os
import sys

LETTERS_SRC = "letters"
POETRY_SRC = "poetry"
LETTERS_OUT = os.path.join("hicalsoft.github.io", "letters", "data", "letters.json")
POETRY_OUT = os.path.join("hicalsoft.github.io", "poetry", "data", "poetry.json")


def generate_letters():
    """Combine individual letter JSON files into one for the web UI."""
    letters = []
    for f in sorted(glob.glob(os.path.join(LETTERS_SRC, "*.json"))):
        with open(f) as fh:
            data = json.load(fh)
        for item in data:
            letters.append({
                "h": item.get("heading", ""),
                "b": item["body"],
                "a": item["author"],
                "r": item.get("recipient", ""),
                "s": item["source"],
                "p": item.get("period", ""),
            })

    authors = {}
    for l in letters:
        authors[l["a"]] = authors.get(l["a"], 0) + 1

    out = {"authors": authors, "letters": letters}

    os.makedirs(os.path.dirname(LETTERS_OUT), exist_ok=True)
    with open(LETTERS_OUT, "w") as f:
        json.dump(out, f, separators=(",", ":"))

    size_mb = os.path.getsize(LETTERS_OUT) / 1024 / 1024
    print(f"Letters: {len(letters)} letters from {len(authors)} authors ({size_mb:.2f} MB)")
    print(f"  → {LETTERS_OUT}")


def generate_poetry():
    """Combine individual poetry JSON files into one for the web UI."""
    poems = []
    for f in sorted(glob.glob(os.path.join(POETRY_SRC, "*.json"))):
        with open(f) as fh:
            data = json.load(fh)
        for item in data:
            poems.append({
                "t": item.get("title", ""),
                "b": item["body"],
                "a": item["author"],
                "s": item["source"],
                "p": item.get("period", ""),
            })

    authors = {}
    for p in poems:
        authors[p["a"]] = authors.get(p["a"], 0) + 1

    out = {"authors": authors, "poems": poems}

    os.makedirs(os.path.dirname(POETRY_OUT), exist_ok=True)
    with open(POETRY_OUT, "w") as f:
        json.dump(out, f, separators=(",", ":"))

    size_mb = os.path.getsize(POETRY_OUT) / 1024 / 1024
    print(f"Poetry: {len(poems)} poems from {len(authors)} poets ({size_mb:.2f} MB)")
    print(f"  → {POETRY_OUT}")


def main():
    parser = argparse.ArgumentParser(
        description="Generate combined JSON data for the hicalsoft.github.io web UI."
    )
    parser.add_argument("--letters", action="store_true", help="Generate letters.json only")
    parser.add_argument("--poetry", action="store_true", help="Generate poetry.json only")
    args = parser.parse_args()

    # If neither flag is set, generate both
    do_letters = args.letters or not (args.letters or args.poetry)
    do_poetry = args.poetry or not (args.letters or args.poetry)

    if do_letters:
        if not os.path.isdir(LETTERS_SRC):
            print(f"Error: {LETTERS_SRC}/ directory not found. Run download_letters.py first.",
                  file=sys.stderr)
            sys.exit(1)
        generate_letters()

    if do_poetry:
        if not os.path.isdir(POETRY_SRC):
            print(f"Error: {POETRY_SRC}/ directory not found. Run download_poetry.py first.",
                  file=sys.stderr)
            sys.exit(1)
        generate_poetry()


if __name__ == "__main__":
    main()