← back to Greenland Onboard
scripts/cork-enrich.py
309 lines
#!/usr/bin/env python3
"""Cork enrichment — mirrors full-02-enrich.py for the 213 Greenland CORK SKUs.
Inputs (local, $0):
data/dwgl-map.json mfr <-> Cork-500xxx DW# assignment (canonical scheme)
data/greenland_cork.ndjson specs + raw color per mfr_sku
public/img/Cork-500xxx.jpg pre-downloaded colorway images (PIL median-cut)
Output:
data/enriched-cork.json one record per Cork-500xxx, SAME schema as enriched.json (naturals)
Deterministic PIL median-cut + interior-designer lexicon. NO paid APIs.
"""
import json, os, re, sys
from collections import Counter
from PIL import Image
ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
IMG_DIR = os.path.join(ROOT, "public", "img")
MAP = json.load(open(os.path.join(ROOT, "data", "dwgl-map.json")))
# cork specs keyed by stripped mfr_sku (9 rows have trailing whitespace)
SPECS = {}
for line in open(os.path.join(ROOT, "data", "greenland_cork.ndjson")):
r = json.loads(line)
SPECS[r["mfr_sku"].strip()] = r
# mfr-series prefix -> coastal-city style (canonical, from SKILL)
CITY = {
"S300": "Montauk", "H171": "Nantucket", "G0174": "Newport",
"S176": "Sag Harbor", "G0181": "Cape May", "G0155": "Palm Beach",
"G0111": "Kiawah", "G0110": "Hilton Head", "G0182": "Chatham",
"G0154": "Southampton", "G0148": "Amagansett", "H192": "Martha's Vineyard",
}
def series_of(mfr):
m = re.match(r"([A-Z]0?\d+)", mfr)
return m.group(1) if m else ""
def city_of(mfr):
return CITY.get(series_of(mfr), "Cork")
# ---- Colorway lexicon: name -> RGB anchor (interior-designer vocabulary) ----
LEXICON = {
"Alabaster": (237,234,224), "Bone": (227,218,201), "Oatmeal": (214,201,178),
"Cream": (245,238,220), "Ivory": (240,234,214), "Linen White": (246,241,229),
"Greige": (188,178,162), "Mushroom": (169,157,140), "Fawn": (196,176,148),
"Camel": (176,138,92), "Tan": (200,170,120), "Sand": (210,190,150),
"Wheat": (222,196,140), "Honey": (214,170,96), "Amber": (196,140,60),
"Caramel": (170,120,64), "Toffee": (140,98,54), "Chestnut": (120,80,48),
"Chocolate": (74,52,38), "Espresso": (58,44,34), "Walnut": (92,66,44),
"Umber": (98,72,50), "Terracotta": (188,104,72), "Rust": (160,82,48),
"Clay": (176,120,92), "Brick": (150,72,56), "Sienna": (150,90,56),
"Oatmeal Grey": (198,192,182), "Dove": (196,194,190), "Silver": (196,198,200),
"Pewter": (140,140,142), "Slate": (100,110,120), "Charcoal": (64,66,70),
"Graphite": (54,56,60), "Ink": (32,36,44), "Onyx": (24,24,26),
"Ebony": (30,28,28), "Jet": (18,18,20),
"Fog": (200,204,206), "Mist": (210,214,214), "Ash": (150,150,148),
"Stone": (168,162,152), "Flint": (120,120,116), "Smoke": (128,128,130),
"Celadon": (168,190,168), "Sage": (150,168,138), "Moss": (108,124,90),
"Fern": (96,120,80), "Olive": (120,116,72), "Sagebrush": (140,148,120),
"Eucalyptus": (150,172,150), "Pine": (58,88,64), "Forest": (44,72,52),
"Hunter": (48,80,60), "Verdigris": (100,150,130), "Jade": (80,140,110),
"Wedgwood": (108,144,180), "Slate Blue": (96,120,150), "Denim": (78,104,140),
"Chambray": (140,164,190), "Sky": (168,196,220), "Powder Blue": (188,208,224),
"Cornflower": (120,150,200), "Indigo": (52,64,110), "Navy": (36,48,84),
"Midnight": (24,34,60), "Prussian": (40,60,90), "Teal": (44,120,130),
"Peacock": (34,108,120), "Aegean": (66,110,140), "Marine": (48,80,120),
"Aubergine": (88,58,84), "Plum": (110,72,110), "Mauve": (168,140,160),
"Heather": (150,132,158), "Lilac": (188,170,198), "Wisteria": (176,160,196),
"Amethyst": (120,90,150), "Orchid": (170,120,170), "Wine": (110,44,58),
"Burgundy": (96,40,52), "Merlot": (86,36,48), "Claret": (120,44,56),
"Blush": (224,196,190), "Rose": (206,150,150), "Dusty Rose": (196,158,156),
"Coral": (222,140,120), "Salmon": (228,158,138), "Apricot": (228,180,140),
"Peach": (240,200,168), "Champagne": (232,214,186), "Buff": (222,204,168),
"Gold": (198,158,72), "Brass": (176,140,72), "Bronze": (140,110,66),
"Copper": (168,104,68), "Antique Gold": (168,138,84),
"Snow": (248,248,246), "White": (250,250,248), "Pearl": (238,236,230),
"Taupe": (170,156,140), "Khaki": (176,160,120), "Putty": (196,186,168),
}
def hex_to_rgb(h):
h = h.lstrip("#")
return tuple(int(h[i:i+2], 16) for i in (0, 2, 4))
def rgb_to_hex(rgb):
return "#%02x%02x%02x" % tuple(int(c) for c in rgb)
def nearest_name(rgb):
best, bd = None, 1e9
for name, anchor in LEXICON.items():
d = sum((a - b) ** 2 for a, b in zip(rgb, anchor))
if d < bd:
bd, best = d, name
return best
def dominant_and_top(path):
"""median-cut dominant hex + top-3 colors with %."""
im = Image.open(path).convert("RGB")
im.thumbnail((160, 160))
q = im.quantize(colors=8, method=Image.MEDIANCUT)
pal = q.getpalette()
data = q.get_flattened_data() if hasattr(q, "get_flattened_data") else q.getdata()
counts = Counter(data)
total = sum(counts.values())
ranked = counts.most_common()
tops = []
for idx, cnt in ranked[:3]:
rgb = (pal[idx*3], pal[idx*3+1], pal[idx*3+2])
tops.append({"hex": rgb_to_hex(rgb), "pct": round(100*cnt/total, 1)})
dom = tops[0]["hex"] if tops else "#888888"
return dom, tops
def color_family(hexv):
r, g, b = hex_to_rgb(hexv)
mx, mn = max(r, g, b), min(r, g, b)
light = (r + g + b) / 3
sat = mx - mn
if light > 225 and sat < 24:
return "White"
if light < 55:
return "Black"
if sat < 28:
return "Neutral"
if r >= g and r >= b:
if g > b and abs(r - g) < 40 and g > 110:
return "Warm"
if b > g:
return "Purple" if b > 90 and r - b < 80 else "Red"
return "Red" if r - g > 40 else "Warm"
if g >= r and g >= b:
return "Green"
if b >= r and b >= g:
return "Purple" if r > g and r > 90 else "Blue"
return "Neutral"
TYPO = {
"pruple": "Purple", "dusry": "Dusty", "dusdy": "Dusty", "gery": "Grey",
"gray": "Gray", "beigh": "Beige", "biege": "Beige", "slver": "Silver",
"chacoal": "Charcoal", "chaocoal": "Charcoal", "brnze": "Bronze",
"goldan": "Golden", "bluu": "Blue", "grean": "Green", "greeen": "Green",
"brwon": "Brown", "brownn": "Brown", "natual": "Natural", "naturel": "Natural",
"ivroy": "Ivory", "creme": "Cream", "taupw": "Taupe", "olvie": "Olive",
"steal": "Steel", "sliver": "Silver", "champange": "Champagne",
"wheit": "White", "whte": "White", "blak": "Black", "blck": "Black",
}
DIM_RE = re.compile(r"(width\s*:|cm\b|meter|\d+\s*(cm|m|mm)\b|/\s*\d+|\d+['\"]|colour\s*:.*\d)", re.I)
def is_dimension_junk(c):
if not c or not c.strip():
return True
s = c.strip()
if DIM_RE.search(s):
return True
letters = sum(ch.isalpha() for ch in s)
if letters < 2:
return True
# bare mfr code (letters+digits, no space) is not a color name
if re.fullmatch(r"[A-Z]{1,4}\d{2,}[A-Z0-9-]*", s):
return True
return False
def title_case(s):
small = {"of", "and", "the", "in", "on", "with", "a"}
words = re.split(r"(\s+|-|/)", s.strip())
out = []
for w in words:
if not w or w.isspace() or w in "-/":
out.append(w); continue
lw = w.lower()
out.append(lw if (lw in small and out) else lw.capitalize())
return "".join(out)
def clean_color(raw, dom_hex):
raw = (raw or "").replace("\xa0", " ") # kill non-breaking spaces
raw = re.sub(r"\s+", " ", raw).strip() # collapse doubled spaces
if is_dimension_junk(raw):
return nearest_name(hex_to_rgb(dom_hex)), "derived"
s = raw
toks = re.split(r"(\s+)", s)
fixed = []
for t in toks:
key = re.sub(r"[^a-z]", "", t.lower())
if key in TYPO:
fixed.append(TYPO[key])
else:
fixed.append(t)
s = "".join(fixed)
# drop a trailing/standalone "Cork" so titles don't read "Cork Cork Wallcovering"
s = re.sub(r"\bcork\b", "", s, flags=re.I)
s = re.sub(r"\s+", " ", s).strip()
if not s: # color was literally just "Cork" -> derive from hex
return nearest_name(hex_to_rgb(dom_hex)), "derived"
s = title_case(s)
return s, "cleaned"
def packaging(min_order):
"""Cork = $59.50/yд in 6-yard increments (min 6 / step 6)."""
mo = (min_order or "").strip()
m = re.search(r"(\d+)\s*yard", mo, re.I)
if m:
n = int(m.group(1))
return n, n
return 6, 6
# cork color name lives in pattern_name (leaf) or color_name (child colorway)
def raw_color_of(spec):
c = spec.get("color_name") or spec.get("pattern_name") or ""
return c.replace("\xa0", " ").strip()
def build_title(city, color):
return f"{city}-{color} Cork Wallcovering | Phillipe Romano"
# ---- Description (deterministic local template, $0 — texture line, no LLM needed) ----
FAM_PHRASE = {
"White": "a bright, airy", "Neutral": "a warm, versatile", "Warm": "a golden, sun-warmed",
"Red": "a rich, earthy", "Green": "a soft, organic", "Blue": "a cool, coastal",
"Black": "a deep, dramatic", "Purple": "a moody, refined",
}
def build_description(city, color, fam, width, fire, backing):
tone = FAM_PHRASE.get(fam, "a natural")
w = (width or "").strip()
parts = [
f"The {city} colorway in {color} brings {tone} tone to our Phillipe Romano cork "
f"wallcovering — a 100% natural cork surface with the subtle, tactile grain that only "
f"real cork delivers.",
"Hand-crafted for interiors that want depth and warmth without pattern, cork adds quiet "
"texture to feature walls, studies, and hospitality spaces while softening sound.",
]
specbits = []
if w:
specbits.append(f"{w} wide")
if backing:
specbits.append(f"{backing} backing")
if fire:
specbits.append(f"{fire} fire rating")
if specbits:
parts.append("Specifications: " + ", ".join(specbits) + ".")
parts.append("Sold per yard in 6-yard increments. Request a sample or a trade quote.")
return " ".join(parts)
def main():
out = []
missing_img = 0
for r in MAP:
dwSku = r["dwsku"]
mfr = r["mfr"]
spec = SPECS.get(mfr, {})
path = os.path.join(IMG_DIR, f"{dwSku}.jpg")
try:
dom_hex, tops = dominant_and_top(path)
except Exception:
dom_hex, tops = "#888888", [{"hex": "#888888", "pct": 100.0}]
missing_img += 1
raw = raw_color_of(spec)
color, csrc = clean_color(raw, dom_hex)
fam = color_family(dom_hex)
city = city_of(mfr)
mn, st = packaging(spec.get("minimum_order", ""))
title = build_title(city, color)
alt = f"{title} — {dwSku}"
desc = build_description(city, color, fam, spec.get("width", ""),
spec.get("fire_rating_us", ""), spec.get("backing", ""))
# tags >=2: material(Cork), color-family, city style, brand, quotes
tags = ["Cork", fam, city, "Phillipe Romano", "quotes"]
tags = [t for t in dict.fromkeys(tags) if t]
rec = {
"dwSku": dwSku, "joinKey": mfr, "prefix": "Cork",
"material": "Cork", "city": city,
"rawColor": raw, "cleanColor": color, "colorSource": csrc,
"hex": dom_hex, "topColors": tops, "colorFamily": fam,
"title": title, "alt": alt, "tags": tags, "description": desc,
"min": mn, "step": st, "minimumOrder": spec.get("minimum_order", ""),
"mfrSku": mfr, "mfrModel": mfr,
"collectionCode": series_of(mfr), "collectionName": spec.get("collection", ""),
"width": spec.get("width", ""), "use": spec.get("use", ""),
"repeat": spec.get("pattern_repeat", ""), "fireRating": spec.get("fire_rating_us", ""),
"backing": spec.get("backing", ""), "weight": spec.get("weight", ""),
"installUrl": "", "catalogUrl": "",
"productUrl": spec.get("product_url", ""),
"imageLocal": f"public/img/{dwSku}.jpg",
}
out.append(rec)
json.dump(out, open(os.path.join(ROOT, "data", "enriched-cork.json"), "w"), indent=1)
fams = Counter(x["colorFamily"] for x in out)
csrc = Counter(x["colorSource"] for x in out)
cities = Counter(x["city"] for x in out)
pk = Counter((x["min"], x["step"]) for x in out)
# coverage
cov = {
"records": len(out),
"hex_populated": sum(1 for x in out if x["hex"] and x["hex"] != "#888888"),
"topColors_3": sum(1 for x in out if len(x["topColors"]) == 3),
"color_nonempty": sum(1 for x in out if x["cleanColor"]),
"title_nonempty": sum(1 for x in out if x["title"]),
"alt_nonempty": sum(1 for x in out if x["alt"]),
"tags_ge2": sum(1 for x in out if len(x["tags"]) >= 2),
"description_nonempty": sum(1 for x in out if x.get("description")),
"missing_image": missing_img,
}
print(json.dumps({
"coverage": cov,
"colorFamilies": dict(fams.most_common()),
"colorSource": dict(csrc),
"cities": dict(cities.most_common()),
"packaging": {f"{k[0]}/{k[1]}": v for k, v in pk.items()},
}, indent=2))
if __name__ == "__main__":
main()