Upload rag.py

rag.py

@@ -0,0 +1,71 @@
+import PyPDF2
+import spacy
+from impactscoreTest import classify_actions_rse_impact_score
+from ODD import classify_actions_rse_ODD
+
+# Chargement du modèle spaCy
+nlp = spacy.load("fr_core_news_sm")
+
+# Étape 1 : Prétraitement des documents PDF
+def extract_text_from_pdf(pdf_path):
+    text = ""
+    with open(pdf_path, "rb") as f:
+        pdf_reader = PyPDF2.PdfReader(f)
+        for page in pdf_reader.pages:
+            text += page.extract_text()
+    return text
+
+# Étape 2 : Traitement du langage naturel (NLP)
+def process_text(text):
+    doc = nlp(text)
+    return doc
+
+# Étape 3 : Extraction d'informations clés
+def extract_key_information(doc):
+    key_information = []
+    for entity in doc.ents:
+        if entity.label_ == "ORG":  
+            name = entity.text
+            if name[0].lower() in ['a', 'e', 'i', 'o', 'u', 'y']:  # Vérification si le nom commence par une voyelle
+                key_information.append(f"l'{name}")
+            else:
+                key_information.append(f"le {name}")
+    return key_information
+
+# Étape 4 : Modélisation
+def generate_transition_plan(key_information, impact_score_criteria, odd_criteria):
+    transition_plan = "Voici votre plan de transition : \n\n"
+    
+    transition_plan += "Informations clés :\n"
+    for info in key_information:
+        transition_plan += f"- {info}\n"
+    
+    transition_plan += "\nCritères d'impact score :\n"
+    for criterion, companies in impact_score_criteria.items():
+        transition_plan += f"- {criterion}:\n"
+        for company in companies:
+            transition_plan += f"  * {company['name']} ({company['activity']}, {company['city']})\n"
+
+    transition_plan += "\nCritères ODD :\n"
+    for criterion, companies in odd_criteria.items():
+        transition_plan += f"- {criterion}:\n"
+        for company in companies:
+            transition_plan += f"  * {company['name']} ({company['activity']}, {company['city']})\n"
+    
+    return transition_plan
+
+
+# Étape 5 : Intégration avec le RAG de Hugging Face
+def generate_plan_from_pdf(pdf_path):
+    text = extract_text_from_pdf(pdf_path)
+    doc = process_text(text)
+    key_information = extract_key_information(doc)
+    impact_score_criteria = classify_actions_rse_impact_score(doc)
+    odd_criteria = classify_actions_rse_ODD(doc)  # Utiliser la fonction ODD
+    transition_plan = generate_transition_plan(key_information, impact_score_criteria, odd_criteria)
+    return transition_plan
+
+# Étape 6 : Test et validation
+pdf_path = "trans1.pdf"
+transition_plan = generate_plan_from_pdf(pdf_path)
+print(transition_plan)