import pandas as pd

import numpy as np

from datetime import datetime, timedelta

now = datetime.now()

embeddings = []

for i in range(10):

embeddings.append({

'document_id': f'doc_{i}',

'embedding': np.random.rand(768).astype(np.float32),

'event_timestamp': now - timedelta(days=i),

'created_timestamp': now - timedelta(days=i, hours=1)

})

df_embeddings = pd.DataFrame(embeddings)

df_embeddings.to_parquet('data/document_embeddings.parquet', index=False)

metadata = []

for i in range(10):

metadata.append({

'document_id': f'doc_{i}',

'title': f'Document {i}',

'content': f'This is the content of document {i}',

'source': 'web',

'author': f'author_{i}',

'publish_date': (now - timedelta(days=i*30)).strftime('%Y-%m-%d'),

'event_timestamp': now - timedelta(days=i),

'created_timestamp': now - timedelta(days=i, hours=1)

})

df_metadata = pd.DataFrame(metadata)

df_metadata.to_parquet('data/document_metadata.parquet', index=False)

print('Created RAG data files successfully!')