Update main.py

main.py

@@ -51,32 +51,43 @@ def health():
 # the llm will say the most similar job to the cv
 @app.route('/compare', methods=['POST'])
 def compare():
-    jobs_skills = request.json.get('jobs_skills')
-    employee_skills = request.json.get('employee_skills')
+    data = request.json
+    jobs_skills = data.get('jobs_skills')
+    employees_skills = data.get('employees_skills')
     
+    # Validate input
     if not isinstance(jobs_skills, list) or not all(isinstance(skill, str) for skill in jobs_skills):
-        raise ValueError("jobs_skills must be a list of strings")
+        return jsonify({"error": "jobs_skills must be a list of strings"}), 400
+    if not isinstance(employees_skills, list) or not all(isinstance(skills, str) for skills in employees_skills):
+        return jsonify({"error": "employees_skills must be a list of strings"}), 400
     
-    # Encode job and employee skills
+    # Encode job skills
     job_embeddings = [model.encode(skill) for skill in jobs_skills]
-    employee_embeddings = model.encode(employee_skills)
+    job_embeddings_tensor = torch.tensor(job_embeddings)
     
-    # Calculate cosine similarity
-    similarities = []
-    employee_embedding_tensor = torch.tensor(employee_embeddings).unsqueeze(0)
-    for job_embedding in job_embeddings:
-        job_embedding_tensor = torch.tensor(job_embedding).unsqueeze(0)
-        similarity = cosine_similarity(employee_embedding_tensor, job_embedding_tensor)
-        similarities.append(similarity.item())
+    # Initialize a dictionary to store similarities for each employee
+    all_similarities = {}
     
-    # Find the job with highest similarity
-    max_similarity_index = similarities.index(max(similarities))
-    max_similarity_job = jobs_skills[max_similarity_index]
+    for idx, employee_skills in enumerate(employees_skills):
+        # Encode employee skills
+        employee_embedding = model.encode(employee_skills)
+        employee_embedding_tensor = torch.tensor(employee_embedding).unsqueeze(0)
+        
+        # Calculate cosine similarity
+        similarities = cosine_similarity(employee_embedding_tensor, job_embeddings_tensor)[0]
+        
+        # Find the job with highest similarity for this employee
+        max_similarity_index = similarities.argmax()
+        max_similarity_job = jobs_skills[max_similarity_index]
+        
+        # Store similarities in the dictionary
+        similarities_dict = {job: similarity for job, similarity in zip(jobs_skills, similarities)}
+        all_similarities[f'employee_{idx+1}'] = {
+            'job': max_similarity_job,
+            'similarities': similarities_dict
+        }
     
-    # Return similarities for all jobs
-    similarities_dict = {job: similarity for job, similarity in zip(jobs_skills, similarities)}
-    
-    return jsonify({'job': max_similarity_job, 'similarities': similarities_dict})
+    return jsonify(all_similarities)
 
 if __name__ == '__main__':
     app.run()