Code/batch_ia/utils/graphs.py
2025-05-28 14:36:30 +02:00

592 lines
28 KiB
Python

import os
import sys
from networkx.drawing.nx_agraph import read_dot
from utils.config import (
REFERENCE_GRAPH_PATH,
determine_threshold_color, get_weight_for_color
)
def parse_graphs(graphe_path):
"""
Charge et analyse les graphes DOT (analyse et référence).
"""
print(graphe_path)
# Charger le graphe à analyser
if not os.path.exists(graphe_path):
print(f"Fichier de graphe à analyser introuvable: {graphe_path}")
sys.exit(1)
# Charger le graphe de référence
reference_path = REFERENCE_GRAPH_PATH
if not os.path.exists(reference_path):
print(f"Fichier de graphe de référence introuvable: {reference_path}")
sys.exit(1)
try:
# Charger les graphes avec NetworkX
graph = read_dot(graphe_path)
ref_graph = read_dot(reference_path)
# Convertir les attributs en types appropriés pour les deux graphes
for g in [graph, ref_graph]:
for node, attrs in g.nodes(data=True):
for key, value in list(attrs.items()):
# Convertir les valeurs numériques
if key in ['niveau', 'ihh_acteurs', 'ihh_pays', 'isg', 'ivc']:
try:
if key in ['isg', 'ivc', 'ihh_acteurs', 'ihh_pays', 'niveau']:
attrs[key] = int(value.strip('"'))
else:
attrs[key] = float(value.strip('"'))
except (ValueError, TypeError):
# Garder la valeur originale si la conversion échoue
pass
elif key == 'label':
# Nettoyer les guillemets des étiquettes
attrs[key] = value.strip('"')
# Convertir les attributs des arêtes
for u, v, attrs in g.edges(data=True):
for key, value in list(attrs.items()):
if key in ['ics', 'cout', 'delai', 'technique']:
try:
attrs[key] = float(value.strip('"'))
except (ValueError, TypeError):
pass
elif key == 'label' and '%' in value:
# Extraire le pourcentage
try:
percentage = value.strip('"').replace('%', '')
attrs['percentage'] = float(percentage)
except (ValueError, TypeError):
pass
return graph, ref_graph
except Exception as e:
print(f"Erreur lors de l'analyse des graphes: {str(e)}")
sys.exit(1)
def extract_data_from_graph(graph, ref_graph):
"""
Extrait toutes les données pertinentes des graphes DOT.
"""
data = {
"products": {}, # Produits finaux (N0)
"components": {}, # Composants (N1)
"minerals": {}, # Minerais (N2)
"operations": {}, # Opérations (N10)
"countries": {}, # Pays (N11)
"geo_countries": {}, # Pays géographiques (N99)
"actors": {} # Acteurs (N12)
}
# Extraire tous les pays géographiques du graphe de référence
for node, attrs in ref_graph.nodes(data=True):
if attrs.get('niveau') == 99:
country_name = attrs.get('label', node)
isg_value = attrs.get('isg', 0)
data["geo_countries"][country_name] = {
"id": node,
"isg": isg_value
}
# Extraire les nœuds du graphe à analyser
for node, attrs in graph.nodes(data=True):
level = attrs.get('niveau', -1)
label = attrs.get('label', node)
if level == 0 or level == 1000: # Produit final
data["products"][node] = {
"label": label,
"components": [],
"assembly": None,
"level": level
}
elif level == 1 or level == 1001: # Composant
data["components"][node] = {
"label": label,
"minerals": [],
"manufacturing": None
}
elif level == 2: # Minerai
data["minerals"][node] = {
"label": label,
"ivc": attrs.get('ivc', 0),
"extraction": None,
"treatment": None,
"ics_values": {}
}
elif level == 10 or level == 1010: # Opération
op_type = label.lower()
data["operations"][node] = {
"label": label,
"type": op_type,
"ihh_acteurs": attrs.get('ihh_acteurs', 0),
"ihh_pays": attrs.get('ihh_pays', 0),
"countries": {}
}
elif level == 11 or level == 1011: # Pays
data["countries"][node] = {
"label": label,
"actors": {},
"geo_country": None,
"market_share": 0
}
elif level == 12 or level == 1012: # Acteur
data["actors"][node] = {
"label": label,
"country": None,
"market_share": 0
}
# Extraire les relations et attributs des arêtes
for source, target, edge_attrs in graph.edges(data=True):
if source not in graph.nodes or target not in graph.nodes:
continue
source_level = graph.nodes[source].get('niveau', -1)
target_level = graph.nodes[target].get('niveau', -1)
# Extraire part de marché
market_share = 0
if 'percentage' in edge_attrs:
market_share = edge_attrs['percentage']
elif 'label' in edge_attrs and '%' in edge_attrs['label']:
try:
market_share = float(edge_attrs['label'].strip('"').replace('%', ''))
except (ValueError, TypeError):
pass
# Relations produit → composant
if (source_level == 0 and target_level == 1) or (source_level == 1000 and target_level == 1001):
if target not in data["products"][source]["components"]:
data["products"][source]["components"].append(target)
# Relations produit → opération (assemblage)
elif (source_level == 0 and target_level == 10) or (source_level == 1000 and target_level == 1010):
if graph.nodes[target].get('label', '').lower() == 'assemblage':
data["products"][source]["assembly"] = target
# Relations composant → minerai avec ICS
elif (source_level == 1 or source_level == 1001) and target_level == 2:
if target not in data["components"][source]["minerals"]:
data["components"][source]["minerals"].append(target)
# Stocker l'ICS s'il est présent
if 'ics' in edge_attrs:
ics_value = edge_attrs['ics']
data["minerals"][target]["ics_values"][source] = ics_value
# Relations composant → opération (fabrication)
elif (source_level == 1 or source_level == 1001) and target_level == 10:
if graph.nodes[target].get('label', '').lower() == 'fabrication':
data["components"][source]["manufacturing"] = target
# Relations minerai → opération (extraction/traitement)
elif source_level == 2 and target_level == 10:
op_label = graph.nodes[target].get('label', '').lower()
if 'extraction' in op_label:
data["minerals"][source]["extraction"] = target
elif 'traitement' in op_label:
data["minerals"][source]["treatment"] = target
# Relations opération → pays avec part de marché
elif (source_level == 10 and target_level == 11) or (source_level == 1010 and target_level == 1011):
data["operations"][source]["countries"][target] = market_share
data["countries"][target]["market_share"] = market_share
# Relations pays → acteur avec part de marché
elif (source_level == 11 and target_level == 12) or (source_level == 1011 and target_level == 1012):
data["countries"][source]["actors"][target] = market_share
data["actors"][target]["market_share"] = market_share
data["actors"][target]["country"] = source
# Relations pays → pays géographique
elif (source_level == 11 or source_level == 1011) and target_level == 99:
country_name = graph.nodes[target].get('label', '')
data["countries"][source]["geo_country"] = country_name
# Compléter les opérations manquantes pour les produits et composants
# en les récupérant du graphe de référence si elles existent
# Pour les produits finaux (N0)
for product_id, product_data in data["products"].items():
if product_data["assembly"] is None:
# Chercher l'opération d'assemblage dans le graphe de référence
for source, target, edge_attrs in ref_graph.edges(data=True):
if (source == product_id and
((ref_graph.nodes[source].get('niveau') == 0 and
ref_graph.nodes[target].get('niveau') == 10) or
(ref_graph.nodes[source].get('niveau') == 1000 and
ref_graph.nodes[target].get('niveau') == 1010)) and
ref_graph.nodes[target].get('label', '').lower() == 'assemblage'):
# L'opération existe dans le graphe de référence
assembly_id = target
product_data["assembly"] = assembly_id
# Ajouter l'opération si elle n'existe pas déjà
if assembly_id not in data["operations"]:
data["operations"][assembly_id] = {
"label": ref_graph.nodes[assembly_id].get('label', assembly_id),
"type": "assemblage",
"ihh_acteurs": ref_graph.nodes[assembly_id].get('ihh_acteurs', 0),
"ihh_pays": ref_graph.nodes[assembly_id].get('ihh_pays', 0),
"countries": {}
}
# Extraire les relations de l'opération vers les pays
for op_source, op_target, op_edge_attrs in ref_graph.edges(data=True):
if (op_source == assembly_id and
(ref_graph.nodes[op_target].get('niveau') == 11 or ref_graph.nodes[op_target].get('niveau') == 1011)):
country_id = op_target
# Extraire part de marché
market_share = 0
if 'percentage' in op_edge_attrs:
market_share = op_edge_attrs['percentage']
elif 'label' in op_edge_attrs and '%' in op_edge_attrs['label']:
try:
market_share = float(op_edge_attrs['label'].strip('"').replace('%', ''))
except (ValueError, TypeError):
pass
# Ajouter le pays à l'opération
data["operations"][assembly_id]["countries"][country_id] = market_share
# Ajouter le pays s'il n'existe pas déjà
if country_id not in data["countries"]:
data["countries"][country_id] = {
"label": ref_graph.nodes[country_id].get('label', country_id),
"actors": {},
"geo_country": None,
"market_share": market_share
}
else:
data["countries"][country_id]["market_share"] = market_share
# Extraire les relations du pays vers les acteurs
for country_source, country_target, country_edge_attrs in ref_graph.edges(data=True):
if (country_source == country_id and
(ref_graph.nodes[country_target].get('niveau') == 12 or ref_graph.nodes[country_target].get('niveau') == 1012)):
actor_id = country_target
# Extraire part de marché
actor_market_share = 0
if 'percentage' in country_edge_attrs:
actor_market_share = country_edge_attrs['percentage']
elif 'label' in country_edge_attrs and '%' in country_edge_attrs['label']:
try:
actor_market_share = float(country_edge_attrs['label'].strip('"').replace('%', ''))
except (ValueError, TypeError):
pass
# Ajouter l'acteur au pays
data["countries"][country_id]["actors"][actor_id] = actor_market_share
# Ajouter l'acteur s'il n'existe pas déjà
if actor_id not in data["actors"]:
data["actors"][actor_id] = {
"label": ref_graph.nodes[actor_id].get('label', actor_id),
"country": country_id,
"market_share": actor_market_share
}
else:
data["actors"][actor_id]["market_share"] = actor_market_share
data["actors"][actor_id]["country"] = country_id
# Extraire la relation du pays vers le pays géographique
for geo_source, geo_target, geo_edge_attrs in ref_graph.edges(data=True):
if (geo_source == country_id and
ref_graph.nodes[geo_target].get('niveau') == 99):
geo_country_name = ref_graph.nodes[geo_target].get('label', '')
data["countries"][country_id]["geo_country"] = geo_country_name
break # Une seule opération d'assemblage par produit
# Pour les composants (N1)
for component_id, component_data in data["components"].items():
if component_data["manufacturing"] is None:
# Chercher l'opération de fabrication dans le graphe de référence
for source, target, edge_attrs in ref_graph.edges(data=True):
if (source == component_id and
((ref_graph.nodes[source].get('niveau') == 1 and
ref_graph.nodes[target].get('niveau') == 10) or
(ref_graph.nodes[source].get('niveau') == 1001 and
ref_graph.nodes[target].get('niveau') == 1010)) and
ref_graph.nodes[target].get('label', '').lower() == 'fabrication'):
# L'opération existe dans le graphe de référence
manufacturing_id = target
component_data["manufacturing"] = manufacturing_id
# Ajouter l'opération si elle n'existe pas déjà
if manufacturing_id not in data["operations"]:
data["operations"][manufacturing_id] = {
"label": ref_graph.nodes[manufacturing_id].get('label', manufacturing_id),
"type": "fabrication",
"ihh_acteurs": ref_graph.nodes[manufacturing_id].get('ihh_acteurs', 0),
"ihh_pays": ref_graph.nodes[manufacturing_id].get('ihh_pays', 0),
"countries": {}
}
# Extraire les relations de l'opération vers les pays
for op_source, op_target, op_edge_attrs in ref_graph.edges(data=True):
if (op_source == manufacturing_id and
(ref_graph.nodes[op_target].get('niveau') == 11 or ref_graph.nodes[op_target].get('niveau') == 1011)):
country_id = op_target
# Extraire part de marché
market_share = 0
if 'percentage' in op_edge_attrs:
market_share = op_edge_attrs['percentage']
elif 'label' in op_edge_attrs and '%' in op_edge_attrs['label']:
try:
market_share = float(op_edge_attrs['label'].strip('"').replace('%', ''))
except (ValueError, TypeError):
pass
# Ajouter le pays à l'opération
data["operations"][manufacturing_id]["countries"][country_id] = market_share
# Ajouter le pays s'il n'existe pas déjà
if country_id not in data["countries"]:
data["countries"][country_id] = {
"label": ref_graph.nodes[country_id].get('label', country_id),
"actors": {},
"geo_country": None,
"market_share": market_share
}
else:
data["countries"][country_id]["market_share"] = market_share
# Extraire les relations du pays vers les acteurs
for country_source, country_target, country_edge_attrs in ref_graph.edges(data=True):
if (country_source == country_id and
(ref_graph.nodes[country_target].get('niveau') == 12 or ref_graph.nodes[country_target].get('niveau') == 1012)):
actor_id = country_target
# Extraire part de marché
actor_market_share = 0
if 'percentage' in country_edge_attrs:
actor_market_share = country_edge_attrs['percentage']
elif 'label' in country_edge_attrs and '%' in country_edge_attrs['label']:
try:
actor_market_share = float(country_edge_attrs['label'].strip('"').replace('%', ''))
except (ValueError, TypeError):
pass
# Ajouter l'acteur au pays
data["countries"][country_id]["actors"][actor_id] = actor_market_share
# Ajouter l'acteur s'il n'existe pas déjà
if actor_id not in data["actors"]:
data["actors"][actor_id] = {
"label": ref_graph.nodes[actor_id].get('label', actor_id),
"country": country_id,
"market_share": actor_market_share
}
else:
data["actors"][actor_id]["market_share"] = actor_market_share
data["actors"][actor_id]["country"] = country_id
# Extraire la relation du pays vers le pays géographique
for geo_source, geo_target, geo_edge_attrs in ref_graph.edges(data=True):
if (geo_source == country_id and
ref_graph.nodes[geo_target].get('niveau') == 99):
geo_country_name = ref_graph.nodes[geo_target].get('label', '')
data["countries"][country_id]["geo_country"] = geo_country_name
break # Une seule opération de fabrication par composant
return data
def calculate_vulnerabilities(data, config):
"""
Calcule les vulnérabilités combinées pour toutes les opérations et minerais.
"""
thresholds = config.get('thresholds', {})
results = {
"ihh_isg_combined": {}, # Pour chaque opération
"ics_ivc_combined": {}, # Pour chaque minerai
"chains": [] # Pour stocker tous les chemins possibles
}
# 1. Calculer ISG_combiné pour chaque opération
for op_id, operation in data["operations"].items():
isg_weighted_sum = 0
total_share = 0
# Parcourir chaque pays impliqué dans l'opération
for country_id, share in operation["countries"].items():
country = data["countries"][country_id]
geo_country = country.get("geo_country")
if geo_country and geo_country in data["geo_countries"]:
isg_value = data["geo_countries"][geo_country]["isg"]
isg_weighted_sum += isg_value * share
total_share += share
# Calculer la moyenne pondérée
isg_combined = 0
if total_share > 0:
isg_combined = isg_weighted_sum / total_share
# Déterminer couleurs et poids
ihh_value = operation["ihh_pays"]
ihh_color, ihh_suffix = determine_threshold_color(ihh_value, "IHH", thresholds)
isg_color, isg_suffix = determine_threshold_color(isg_combined, "ISG", thresholds)
# Calculer poids combiné
ihh_weight = get_weight_for_color(ihh_color)
isg_weight = get_weight_for_color(isg_color)
combined_weight = ihh_weight * isg_weight
# Déterminer vulnérabilité combinée
if combined_weight in [6, 9]:
vulnerability = "ÉLEVÉE à CRITIQUE"
elif combined_weight in [3, 4]:
vulnerability = "MOYENNE"
else: # 1, 2
vulnerability = "FAIBLE"
# Stocker résultats
results["ihh_isg_combined"][op_id] = {
"ihh_value": ihh_value,
"ihh_color": ihh_color,
"ihh_suffix": ihh_suffix,
"isg_combined": isg_combined,
"isg_color": isg_color,
"isg_suffix": isg_suffix,
"combined_weight": combined_weight,
"vulnerability": vulnerability
}
# 2. Calculer ICS_moyen pour chaque minerai
for mineral_id, mineral in data["minerals"].items():
ics_values = list(mineral["ics_values"].values())
ics_average = 0
if ics_values:
ics_average = sum(ics_values) / len(ics_values)
ivc_value = mineral.get("ivc", 0)
# Déterminer couleurs et poids
ics_color, ics_suffix = determine_threshold_color(ics_average, "ICS", thresholds)
ivc_color, ivc_suffix = determine_threshold_color(ivc_value, "IVC", thresholds)
# Calculer poids combiné
ics_weight = get_weight_for_color(ics_color)
ivc_weight = get_weight_for_color(ivc_color)
combined_weight = ics_weight * ivc_weight
# Déterminer vulnérabilité combinée
if combined_weight in [6, 9]:
vulnerability = "ÉLEVÉE à CRITIQUE"
elif combined_weight in [3, 4]:
vulnerability = "MOYENNE"
else: # 1, 2
vulnerability = "FAIBLE"
# Stocker résultats
results["ics_ivc_combined"][mineral_id] = {
"ics_average": ics_average,
"ics_color": ics_color,
"ics_suffix": ics_suffix,
"ivc_value": ivc_value,
"ivc_color": ivc_color,
"ivc_suffix": ivc_suffix,
"combined_weight": combined_weight,
"vulnerability": vulnerability
}
# 3. Identifier tous les chemins et leurs vulnérabilités
for product_id, product in data["products"].items():
for component_id in product["components"]:
component = data["components"][component_id]
for mineral_id in component["minerals"]:
mineral = data["minerals"][mineral_id]
# Collecter toutes les vulnérabilités dans ce chemin
path_vulnerabilities = []
# Assemblage (si présent)
assembly_id = product["assembly"]
if assembly_id and assembly_id in results["ihh_isg_combined"]:
path_vulnerabilities.append({
"type": "assemblage",
"vulnerability": results["ihh_isg_combined"][assembly_id]["vulnerability"],
"operation_id": assembly_id
})
# Fabrication (si présent)
manufacturing_id = component["manufacturing"]
if manufacturing_id and manufacturing_id in results["ihh_isg_combined"]:
path_vulnerabilities.append({
"type": "fabrication",
"vulnerability": results["ihh_isg_combined"][manufacturing_id]["vulnerability"],
"operation_id": manufacturing_id
})
# Minerai (ICS+IVC)
if mineral_id in results["ics_ivc_combined"]:
path_vulnerabilities.append({
"type": "minerai",
"vulnerability": results["ics_ivc_combined"][mineral_id]["vulnerability"],
"mineral_id": mineral_id
})
# Extraction (si présent)
extraction_id = mineral["extraction"]
if extraction_id and extraction_id in results["ihh_isg_combined"]:
path_vulnerabilities.append({
"type": "extraction",
"vulnerability": results["ihh_isg_combined"][extraction_id]["vulnerability"],
"operation_id": extraction_id
})
# Traitement (si présent)
treatment_id = mineral["treatment"]
if treatment_id and treatment_id in results["ihh_isg_combined"]:
path_vulnerabilities.append({
"type": "traitement",
"vulnerability": results["ihh_isg_combined"][treatment_id]["vulnerability"],
"operation_id": treatment_id
})
# Classifier le chemin
path_info = {
"product": product_id,
"component": component_id,
"mineral": mineral_id,
"vulnerabilities": path_vulnerabilities
}
# Déterminer le niveau de risque du chemin
critical_count = path_vulnerabilities.count({"vulnerability": "ÉLEVÉE à CRITIQUE"})
medium_count = path_vulnerabilities.count({"vulnerability": "MOYENNE"})
if any(v["vulnerability"] == "ÉLEVÉE à CRITIQUE" for v in path_vulnerabilities):
path_info["risk_level"] = "critique"
elif medium_count >= 3:
path_info["risk_level"] = "majeur"
elif any(v["vulnerability"] == "MOYENNE" for v in path_vulnerabilities):
path_info["risk_level"] = "moyen"
else:
path_info["risk_level"] = "faible"
results["chains"].append(path_info)
return results