`merge-work/merge-house.sh`

91 lines
#!/usr/bin/env bash
# merge-house.sh <house_key>
#
# Re-exports each source vendor's catalog with the CURRENT fixed export-catalog.sh
# (scrubbing to the HOUSE name with the COMBINED banned set), concatenates them into
# one deduped combined staging JSONL, and leak-scans the result.
#
# Houses:
#   malibu_house         : malibu + brewster_york            -> "Malibu Wallpaper"
#   phillipe_romano_house: justindavid + phillipe_romano + relativity_textiles + twil_karin -> "Phillipe Romano"
#   hollywood_house      : command54 + hollywood             -> "Hollywood Wallcoverings"
#
# LOCAL ONLY. No deploy.
set -euo pipefail
HK="${1:?usage: merge-house.sh <house_key>}"
REPO="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
cd "$REPO"
EXPORT="$REPO/lib/export-catalog.sh"

case "$HK" in
  malibu_house)
    HOUSE="Malibu Wallpaper"
    SOURCES=("malibu:malibu_catalog" "brewster_york:brewster_catalog")
    BANNED=("Brewster & York" "Brewster" "York" "Chesapeake" "Seabrook" "NextWall" "WallQuest")
    ;;
  phillipe_romano_house)
    HOUSE="Phillipe Romano"
    SOURCES=("justindavid:justindavid_catalog" "phillipe_romano:phillipe_romano_catalog" "relativity_textiles:relativity_textiles_catalog" "twil_karin:twil_karin_catalog")
    BANNED=("York/Brewster (Phillipe Romano source)" "Relativity Textiles" "Justin David" "TWIL Karin" "TWIL" "Command54" "York" "Brewster")
    ;;
  hollywood_house)
    HOUSE="Hollywood Wallcoverings"
    SOURCES=("command54:command54_catalog" "hollywood:hollywood_catalog")
    BANNED=("Command54" "Momentum" "WallQuest" "Chesapeake" "Seabrook" "NextWall" "Brewster" "York")
    ;;
  *) echo "unknown house_key: $HK" >&2; exit 1;;
esac

echo "== merge $HK -> HOUSE='$HOUSE' banned=[${BANNED[*]}] =="

STAGE_FILES=()
for pair in "${SOURCES[@]}"; do
  vcode="${pair%%:*}"; ctable="${pair##*:}"
  out="$REPO/staging/${vcode}.jsonl"
  echo "-- re-export $vcode ($ctable) with fixed scrub -> $out"
  n="$(HOUSE="$HOUSE" bash "$EXPORT" "$ctable" "$out" "${BANNED[@]}")"
  echo "   exported $n rows"
  STAGE_FILES+=("$out")
done

# ---- concatenate + dedup (exact-duplicate lines AND same mfr_sku+color_name) ----
COMBINED="$REPO/staging/${HK}.jsonl"
echo "-- concat + dedup -> $COMBINED"
python3 - "$COMBINED" "${STAGE_FILES[@]}" <<'PY'
import sys, json
out_path = sys.argv[1]
files = sys.argv[2:]
seen_line = set()
seen_key  = set()
kept = 0; dup_line = 0; dup_key = 0
with open(out_path, 'w') as fout:
    for f in files:
        with open(f) as fin:
            for line in fin:
                line = line.rstrip('\n')
                if not line.strip():
                    continue
                # exact-duplicate line
                if line in seen_line:
                    dup_line += 1
                    continue
                seen_line.add(line)
                # same mfr_sku + color_name dedup (only when mfr_sku present)
                try:
                    r = json.loads(line)
                except Exception:
                    r = None
                if r is not None:
                    mk = (r.get('mfr_sku') or r.get('sku') or '')
                    ck = (r.get('color_name') or '')
                    if mk:
                        key = (str(mk).strip().lower(), str(ck).strip().lower())
                        if key in seen_key:
                            dup_key += 1
                            continue
                        seen_key.add(key)
                fout.write(line + '\n')
                kept += 1
print(json.dumps({"kept": kept, "dup_exact_line": dup_line, "dup_mfr_color": dup_key, "total_dedup": dup_line + dup_key}))
PY