appdev_genai_googlecloud/src/genai-app/main.py [31:94]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
app = Flask(__name__)


async def insert_doc(file_id:int, text:str, metadata:str, user_id:str, embeddings_data:list):
    loop = asyncio.get_running_loop()
    async with Connector(loop=loop) as connector:
        conn: asyncpg.Connection = await connector.connect_async(
            f"{project_id}:{region}:{instance_name}",
            "asyncpg",
            user=f"{database_user}",
            password=f"{database_password}",
            db=f"{database_name}",
        )

        await register_vector(conn)
        
        await conn.execute(
            "INSERT INTO docs_embeddings (document_id, content, metadata, user_id, embedding) VALUES ($1, $2, $3, $4, $5)",
            file_id,
            text,
            json.dumps(metadata),
            user_id,
            np.array(embeddings_data),
        )
        await conn.close()


async def search_doc(embeddings_data:list, user_id:str):
    loop = asyncio.get_running_loop()
    async with Connector(loop=loop) as connector:
        conn: asyncpg.Connection = await connector.connect_async(
            f"{project_id}:{region}:{instance_name}",
            "asyncpg",
            user=f"{database_user}",
            password=f"{database_password}",
            db=f"{database_name}",
        )

        await register_vector(conn)
        similarity_threshold = 0.0 # Now, not set the threshold. You can change the threshold.
        num_matches = 50

        # Find similar products to the query using cosine similarity search
        # over all vector embeddings. This new feature is provided by `pgvector`.
        results = await conn.fetch(
            """SELECT document_id, content, metadata, user_id, 1 - (embedding <=> $1) AS similarity
            FROM docs_embeddings
            WHERE 1 - (embedding <=> $1) > $2 AND user_id = $4
            ORDER BY similarity DESC
            LIMIT $3
            """,
            embeddings_data,
            similarity_threshold,
            num_matches,
            user_id
        )

        await conn.close()
        
        return results

    
def call_Palm(context:str, question:str):
    llm = VertexAI(
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



appdev_with_generative_ai/src/genai-app/main.py [31:94]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
app = Flask(__name__)


async def insert_doc(file_id:int, text:str, metadata:str, user_id:str, embeddings_data:list):
    loop = asyncio.get_running_loop()
    async with Connector(loop=loop) as connector:
        conn: asyncpg.Connection = await connector.connect_async(
            f"{project_id}:{region}:{instance_name}",
            "asyncpg",
            user=f"{database_user}",
            password=f"{database_password}",
            db=f"{database_name}",
        )

        await register_vector(conn)
        
        await conn.execute(
            "INSERT INTO docs_embeddings (document_id, content, metadata, user_id, embedding) VALUES ($1, $2, $3, $4, $5)",
            file_id,
            text,
            json.dumps(metadata),
            user_id,
            np.array(embeddings_data),
        )
        await conn.close()


async def search_doc(embeddings_data:list, user_id:str):
    loop = asyncio.get_running_loop()
    async with Connector(loop=loop) as connector:
        conn: asyncpg.Connection = await connector.connect_async(
            f"{project_id}:{region}:{instance_name}",
            "asyncpg",
            user=f"{database_user}",
            password=f"{database_password}",
            db=f"{database_name}",
        )

        await register_vector(conn)
        similarity_threshold = 0.0 # Now, not set the threshold. You can change the threshold.
        num_matches = 50

        # Find similar products to the query using cosine similarity search
        # over all vector embeddings. This new feature is provided by `pgvector`.
        results = await conn.fetch(
            """SELECT document_id, content, metadata, user_id, 1 - (embedding <=> $1) AS similarity
            FROM docs_embeddings
            WHERE 1 - (embedding <=> $1) > $2 AND user_id = $4
            ORDER BY similarity DESC
            LIMIT $3
            """,
            embeddings_data,
            similarity_threshold,
            num_matches,
            user_id
        )

        await conn.close()
        
        return results

    
def call_Palm(context:str, question:str):
    llm = VertexAI(
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -