#!/usr/bin/env python3 """SCPI Data Fixer - Scrapes ideal-investisseur.fr for verified data""" import json, re, time, sys from scrapling.fetchers import Fetcher SHEET_ID = "1dBGv3jIsFDLMJInANzyLvUAF0HYb5iRZ_RQljo-i7XM" # Mapping of SCPI names to their ideal-investisseur URLs SCPI_URLS = { "Eden": "https://www.ideal-investisseur.fr/scpi-avis/eden-10150.html", "Elialys": "https://www.ideal-investisseur.fr/scpi-avis/elialys-advenis-1105.html", "Eurovalys": "https://www.ideal-investisseur.fr/scpi-avis/eurovalys-1043.html", "Aestiam Agora": "https://www.ideal-investisseur.fr/scpi-avis/aestiam-agora-1028.html", "Aestiam Horizon": "https://www.ideal-investisseur.fr/scpi-avis/aestiam-horizon-1071.html", "Linaclub": "https://www.ideal-investisseur.fr/scpi-avis/linaclub-10139.html", "AEW Commerces Europe": None, # Need to find "AEW Diversification Allemagne": None, "AEW Opportunités Europe": None, "AEW Patrimoine Santé": "https://www.ideal-investisseur.fr/scpi-avis/aew-patrimoine-sante-10027.html", "Atout Pierre Diversification": None, "Activimmo": None, "Comète": "https://www.ideal-investisseur.fr/scpi-avis/comete-alderan-10045.html", "Allianz Pierre": "https://www.ideal-investisseur.fr/scpi-avis/allianz-divercity-10008.html", "Alta Convictions": "https://www.ideal-investisseur.fr/scpi-avis/alta-convictions-10048.html", "Altixia Cadence XII": "https://www.ideal-investisseur.fr/scpi-avis/altixia-cadence-XII-12-1101.html", "Altixia Commerces": "https://www.ideal-investisseur.fr/scpi-avis/altixia-commerces-1102.html", "Edissimmo": None, "Genepierre": None, "Rivoli Avenir Patrimoine": None, "MomenTime": "https://www.ideal-investisseur.fr/scpi-avis/MomenTime-10149.html", "Transitions Europe": "https://www.ideal-investisseur.fr/scpi-avis/transitions-europe-10040.html", "New Gen": None, "Epargne Pierre": None, "Epargne Pierre Europe": "https://www.ideal-investisseur.fr/scpi-avis/epargne-pierre-europe-10024.html", "Epargne Pierre Sophia": "https://www.ideal-investisseur.fr/scpi-avis/epargne-pierre-sophia-10144.html", "Atream Hôtels": "https://www.ideal-investisseur.fr/scpi-avis/scpi-atream-hotels-1032.html", "Upeka": "https://www.ideal-investisseur.fr/scpi-avis/upeka-10046.html", "Accès Valeur Pierre": "https://www.ideal-investisseur.fr/scpi-avis/scpi-acces-valeur-pierre-1029.html", "Accimmo Pierre": None, "Imarea Pierre": "https://www.ideal-investisseur.fr/scpi-avis/Imarea-Pierre-10151.html", "Opus Real": None, "Optimale": "https://www.ideal-investisseur.fr/scpi-avis/optimale-consultim-1150.html", "Corum Eurion": "https://www.ideal-investisseur.fr/scpi-avis/corum-eurion-10000.html", "Corum Origin": "https://www.ideal-investisseur.fr/scpi-avis/scpi-corum-1000.html", "Corum USA": None, "Corum XL": "https://www.ideal-investisseur.fr/scpi-avis/corum-xl-1097.html", "Darwin RE01": "https://www.ideal-investisseur.fr/scpi-avis/darwin-re01-10138.html", "Edmond de Rothschild Europa": "https://www.ideal-investisseur.fr/scpi-avis/edr-europa-10133.html", "Elevation Tertiom": "https://www.ideal-investisseur.fr/scpi-avis/elevation-tertiom-10137.html", "Epsicap Explore": "https://www.ideal-investisseur.fr/scpi-avis/epsicap-explore-10140.html", "Epsicap Nano": "https://www.ideal-investisseur.fr/scpi-avis/epsilon-360-10019.html", "Euryale Horizons Santé": None, "Pierval Santé": "https://www.ideal-investisseur.fr/scpi-avis/pierval-sante-1055.html", "Buroboutic Métropoles": None, "Ficommerce Proximité": None, "Logipierre 3 Résidences Services": "https://www.ideal-investisseur.fr/scpi-avis/logipierre-3-1116.html", "Pierre Expansion Santé": None, "Selectipierre 2 - Paris": "https://www.ideal-investisseur.fr/scpi-avis/selectipierre-2-1072.html", "Cap Foncières et Territoires": "https://www.ideal-investisseur.fr/scpi-avis/cap-foncieres-et-territoire-1045.html", "GMA Essentialis": None, "Affinités Pierre": None, "Attraits Pierre": "https://www.ideal-investisseur.fr/scpi-avis/attraits-pierre-10141.html", "Elysées Grand Large": "https://www.ideal-investisseur.fr/scpi-avis/elysee-grand-large-10145.html", "Elysées Pierre": None, "Cristal Life": "https://www.ideal-investisseur.fr/scpi-avis/cristal-life-10033.html", "Cristal Rente": "https://www.ideal-investisseur.fr/scpi-avis/cristal-rente-intergestion-1104.html", "Iroko Atlas": "https://www.ideal-investisseur.fr/scpi-avis/iroko-atlas-10136.html", "Iroko Zen": "https://www.ideal-investisseur.fr/scpi-avis/iroko-zen-10007.html", "Kyaneos Pierre": None, "Crédit Mutuel Pierre 1": None, "Epargne Foncière": None, "LF Avenir Santé": "https://www.ideal-investisseur.fr/scpi-avis/lf-avenir-sante-10038.html", } def extract_ideal_investisseur(url): """Extract SCPI data from ideal-investisseur.fr page""" page = Fetcher.get(url) data = {} # Extract from table cells cells = page.css('td') for i in range(0, len(cells)-1, 2): label = cells[i].text.strip() value = cells[i+1].text.strip() if i+1 < len(cells) else '' if 'Société de gestion' in label: data['gestionnaire'] = value elif 'Création' in label: data['date_creation'] = value elif 'Taux de distribution brut' in label: data['td'] = value elif 'Performance brute globale annuelle' in label: data['pga'] = value elif 'TRI 10 ans' in label: data['tri_10'] = value elif 'TRI 15 ans' in label: data['tri_15'] = value elif 'Prix de souscription' in label: data['prix_souscription'] = value elif 'Prix de retrait' in label: data['prix_retrait'] = value elif 'Valeur de réalisation' in label and 'Date' not in label: data['val_realisation'] = value elif 'Valeur de reconstitution' in label and 'Date' not in label: data['val_reconstitution'] = value elif 'TOF' in label: data['tof'] = value elif 'Taux d\'endettement' in label or 'endettement' in label.lower(): data['endettement'] = value elif 'Capitalisation' in label: data['capitalisation'] = value elif 'Surface totale' in label: data['surface'] = value elif 'Durée des baux' in label: data['duree_baux'] = value elif 'Nombre d\'immeubles' in label or "Nombre d'immeubles" in label: data['nb_immeubles'] = value elif 'Nombre de locataires' in label: data['nb_locataires'] = value elif 'Nb parts en retrait' in label: data['parts_retrait'] = value elif 'Classification SFDR' in label: data['sfdr'] = value elif 'Label ISR' in label: data['label_isr'] = value return data # Only process rows with errors first PRIORITY_ROWS = { "Eden": True, "Elialys": True, } results = {} for scpi_name, url in SCPI_URLS.items(): if url is None: print(f"SKIP {scpi_name} (no URL)") continue try: print(f"Fetching {scpi_name}...", end=" ", flush=True) data = extract_ideal_investisseur(url) results[scpi_name] = data print(f"OK ({len(data)} fields)") time.sleep(0.5) except Exception as e: print(f"ERROR: {e}") print("\n\n=== RESULTS ===") print(json.dumps(results, indent=2, ensure_ascii=False))