#!/bin/bash
# Regenerates everything the site reads from the analysis directory.
#
# The site used to hold hand-copied snapshots of books.json and friends, which drifted:
# the catalogue grew to 163 books while the site still rendered 112. Nothing under
# site/src/data or site/src/content/studies is authored by hand — it is all generated here,
# and this runs as part of `npm run build`.
set -eu
ROOT="$(cd "$(dirname "$0")/.." && pwd)"

python3 "$ROOT/scripts/sync_studies.py"

# Articles are written for a reader, not for the record, and carry their own front matter —
# so they are copied verbatim rather than normalised the way the analyses are.
ART_SRC="$ROOT/analysis/articles"
ART_DST="$ROOT/site/src/content/articles"
# Wipe the destination completely rather than by glob. This tree lives under Documents,
# which is iCloud-synced, and conflict copies appear as "justify-that 2.md" — a file that
# would publish the same article again at a second URL. Only what analysis/articles holds
# may survive here.
mkdir -p "$ART_DST"; find "$ART_DST" -type f -delete
if [ -d "$ART_SRC" ]; then
  for f in "$ART_SRC"/*.md; do
    [ -e "$f" ] || continue
    rev=$(date -r "$f" +%Y-%m-%d)
    awk -v rev="$rev" 'NR==1&&/^---$/{print;print "revised: \"" rev "\"";next}1' "$f" > "$ART_DST/$(basename "$f")"
  done
  n_src=$(ls "$ART_SRC"/*.md 2>/dev/null | wc -l | tr -d ' ')
  n_dst=$(ls "$ART_DST"/*.md 2>/dev/null | wc -l | tr -d ' ')
  echo "  articles: $n_dst"
  if [ "$n_src" != "$n_dst" ]; then
    echo "  ! article count mismatch: $n_src source, $n_dst copied" >&2; exit 1
  fi
  # A duplicated slug publishes the same piece twice; catch it here rather than in a sitemap.
  dupes=$(ls "$ART_DST" | sed -E 's/ [0-9]+\.md$/.md/' | sort | uniq -d)
  if [ -n "$dupes" ]; then echo "  ! duplicate article slugs: $dupes" >&2; exit 1; fi
fi

for f in books comparisons findings summaries library hosting; do
  if [ -f "$ROOT/analysis/$f.json" ]; then
    cp "$ROOT/analysis/$f.json" "$ROOT/site/src/data/$f.json"
    n=$(python3 -c "import json,sys;d=json.load(open(sys.argv[1]));print(len(d) if isinstance(d,list) else len(d.get('books',d)))" "$ROOT/analysis/$f.json")
    echo "  data/$f.json  ($n records)"
  else
    echo "  ! analysis/$f.json missing — site keeps its previous copy"
  fi
done
# Ship the pipeline itself. The method page describes these scripts; a reproducibility
# claim with nothing attached is a claim, not an artefact.
ART="$ROOT/site/public/artefacts"
mkdir -p "$ART"
for f in "$ROOT"/scripts/*.sh "$ROOT"/scripts/*.py; do
  [ -e "$f" ] || continue
  cp "$f" "$ART/$(basename "$f").txt"
done
cp "$ROOT/README.md" "$ART/README.md.txt" 2>/dev/null || true
cp "$ROOT/URL-POLICY.md" "$ART/URL-POLICY.md.txt" 2>/dev/null || true
cp "$ROOT/EDITORIAL-GUIDELINE.md" "$ART/EDITORIAL-GUIDELINE.md.txt" 2>/dev/null || true
echo "  artefacts: $(ls "$ART" | wc -l | tr -d ' ') files"

echo "site data synced"