NOM INC.
RE/MAX VISION
Agence Immobilière
_AGENT_BANNER_RE = re.compile(r'class="agency">[^<]*
([^<]+)
', re.S)
_TOUR_RE = re.compile(r'https?://(?:my\.matterport\.com/show/[^"\'\s<>]+'
r'|(?:www\.)?youtube\.com/(?:embed/|watch\?v=)[^"\'\s<>]+'
r'|youtu\.be/[^"\'\s<>]+'
r'|(?:player\.)?vimeo\.com/(?:video/)?\d+[^"\'\s<>]*)', re.I)
# libellés « Bâtiment et intérieur » promus en colonnes du schéma
_INT_FIELDS = {"Chambres": "bedrooms", "Salles de bain": "bathrooms",
"Salles d'eau": "powder_rooms"}
def _clean(s: str) -> str:
return re.sub(r'\s+', ' ', _html.unescape(s)).strip()
def _to_sqft(val: str) -> float | None:
"""« 128.8 Mètres carrés » / « 1 200 Pieds carrés » / « 305.4 » (m² implicite,
unités Centris métriques sur ce portail) -> pi²."""
m = re.search(r'([\d][\d\s ,.]*)', val or "")
if not m:
return None
raw = m.group(1).replace(" ", "").replace(" ", "").replace(" ", "")
if raw.count(",") == 1 and "." not in raw:
raw = raw.replace(",", ".")
else:
raw = raw.replace(",", "")
try:
v = float(raw)
except ValueError:
return None
if v <= 0:
return None
if re.search(r'pied|pi2|pi²|pc\b', val, re.I):
return round(v)
return round(v * 10.7639) # métrique (défaut du portail)
def _parse_auc_detail(html: str) -> dict:
"""Fiche avecuncourtier.com : galerie, description+addendas, GPS,
accordéons (pièces, taxes, bâtiment), courtiers + agence réelle."""
out: dict = {}
details: dict = {}
feats: list[str] = []
# galerie : slider principal seulement (avant les vignettes dupliquées)
main = html.split('gallery-slider__thumbnails')[0]
imgs = list(dict.fromkeys(SITE + u for u in _GAL_RE.findall(main)))
if len(imgs) > 1:
out["images"] = imgs[:80]
mc = _COORD_RE.search(html)
if mc:
lat, lng = float(mc.group(1)), float(mc.group(2))
if 44.5 <= lat <= 63.0 and -80.0 <= lng <= -56.0:
out["lat"], out["lng"] = lat, lng
desc, addendum = "", ""
md = _DESC_RE.search(html)
if md:
desc = _clean(re.sub(r'<[^>]+>', ' ', md.group(1)))
mo = _OPEN_RE.search(html)
if mo:
oh = _clean(re.sub(r'<[^>]+>', ' ', mo.group(1)))
if oh:
details["Visites libres"] = oh[:200]
# accordéons « Détails »
for item in re.split(r'
', html)[1:]:
tm = _ITEM_TITLE_RE.search(item)
title = _clean(tm.group(1)) if tm else ""
if title == "Proximité":
mp = _PROX_RE.search(item)
if mp:
prox = _clean(re.sub(r'<[^>]+>', ' ', mp.group(1)))
if prox:
details["Proximité"] = prox[:300]
feats.append(f"Proximité : {prox}"[:300])
elif title == "Dimension des pièces":
rooms = _ROOM_RE.findall(item)
for name, lvl, dim, floor in rooms:
name, lvl = _clean(name), _clean(lvl)
dim, floor = _clean(dim), _clean(floor)
line = name + (f" — {dim}" if dim else "") + (f", {lvl}" if lvl else "")
if floor:
line += f" ({floor})"
feats.append(line)
elif title == "Addendas":
ma = _ADD_RE.search(item)
if ma:
# les sauts de ligne du texte Centris sont préservés dans le
addendum = _html.unescape(ma.group(1)).strip()
else: # Taxes et dépenses / Bâtiment…
for label, val in _PAIR_RE.findall(item):
label, val = _clean(label).rstrip(":"), _clean(val)
if not label or not val or val in ("0", "-"):
continue
fld = _INT_FIELDS.get(label)
if fld:
mi = re.search(r'\d+', val)
if mi:
out[fld] = int(mi.group())
elif label == "Aire habitable":
a = _to_sqft(val)
if a:
out["area_sqft"] = a
elif label == "Superficie du terrain":
t = _to_sqft(val)
if t:
out["lot_sqft"] = t
val = f"{val} m²" if re.fullmatch(r'[\d\s,.]+', val) else val
elif label == "Année de construction":
my = re.search(r'\b((?:1[6-9]|20)\d{2})\b', val)
if my:
out["year_built"] = int(my.group(1))
if label in details and val not in details[label]:
details[label] = f"{details[label]}, {val}"[:300]
else:
details[label] = val[:300]
if desc or addendum:
out["description"] = "\n\n".join(filter(None, (desc, addendum)))[:6000]
if feats:
out["features"] = feats[:40]
# courtier(s) inscripteur(s) + bannière réelle (encadré agents-wrapper) :
# premier agent = inscripteur principal
i = html.find('class="agents-wrapper"')
if i != -1:
j = html.find('calculator-wrapper', i)
seg = html[i:j if j != -1 else i + 5000]
mn = _AGENT_NAME_RE.search(seg)
if mn:
out["broker_name"] = _clean(mn.group(1))
mt = _AGENT_TEL_RE.search(seg)
if mt:
out["broker_phone"] = _clean(mt.group(1))
mb = _AGENT_BANNER_RE.search(seg)
if mb:
banner = _clean(mb.group(1))
if banner and banner.lower() != "agence immobilière":
details["__agency"] = banner.title() if banner.isupper() else banner
# visite virtuelle / vidéo (hors liens de pied de page /channel/)
mtou = _TOUR_RE.search(html)
if mtou:
details.setdefault("Visite virtuelle / vidéo", mtou.group(0))
if details:
out["details"] = details
return out
|