Untitled

def get_document_embedding_to_chunk_mapping(entity_id, embedding_to_chunk_content={}):
    query=f"""
    SELECT cmetadata,embedding,"document"
    FROM fb7d848f95e2.langchain_pg_embedding
    where lower(cmetadata ::text) ilike '%"entity_id"%"{entity_id}"%'
    """
    cursor.execute(query)
    for chunk in cursor.fetchall():
        embedding=tuple(json.loads(chunk[1]))
        chunk_content=chunk[2]
        embedding_to_chunk_content[embedding]=chunk_content
    return embedding_to_chunk_content

def get_document_clusters(entity_ids):
    embedding_to_chunk_content={}
    for entity in entity_ids:
        embedding_to_chunk_content=get_document_embedding_to_chunk_mapping(entity_id, embedding_to_chunk_content)
    
    clustering = DBSCAN(eps=0.3, min_samples=37,metric="cosine").fit(X)
    return clustering.labels_