Failed to fetch semantic model

[antfido]

Hey,

I redeployed the extension (I couldn't find a bug in my former deploy), and now I'm getting an error as on the screen below:

Did anyone experience a similar error ? How did you handle it ?

I ran out what to change, my .env file looks good. Maybe I wrongly define example_id in examples ? I defines it as
<my_Looker_model_name>:<my_Looker_explore_name> . I don't add .model and .explore .

Thanks !

[antfido]

hey @LukaFontanilla - do you know what can be a reason for that ? thnx!

[vijayram0690]

@antfido was this issue fixed, i am also running into similar issue.

@LukaFontanilla how can we fix the above issue.

[antfido]

Nope, still stucked here @vijayram0690

[antfido]

@vijayram0690 did you solve it ?

[LukaFontanilla]

I assume this error is returned when the extension first starts up and not after a NL request is submitted? It would be helpful to understand what requests are failing. Can you please check the browser dev tools and both the "Console" and "Network" tabs and share any Looker specific errors you find there? Given the error message I would assume there is a failing Looker API call. It would be helpful to see what that is.

As for the newer Explore Assistant deployment, there are 3 BigQuery tables that are provisioned -> explore_assistant_examples, explore_assistant_refinement_examples, & explore_assistant_samples. Each of these tables should have a row for the given model and explore you want to use with the Explore Assistant (we are no longer reliant on a single set of environment variables for this). Additionally each explore_id value should be in the format you listed, where the model name and explore name are separated by a single ":".

[antfido]

Hey,

so here are the screenshots from console:

For the second part - in BQ and in the config files it's defined with one semicolon, I don't know why error adds one more...

Thanks a lot for your help @LukaFontanilla .

[antfido]

hey @LukaFontanilla - would you be able to look into that one ? Thnx a lot ! :)

[dassowmd]

I'm having a similar issue. I think its related to my exploreKey having '::' in it instead of the expected ':' the code is looking for.

[antfido]

@dassowmd it's just generic output of the error (defined wrongly) so I wouldnt bother about it. is it working for you know ?

[dassowmd]

@antfido I think the issue is that the EXPLORE_ID in the .env file has a single colon, but Looker delimits with a double colon. I didn't catch it, which sent me down a rabbit hole. Once I figured that out it seems to be working fine.

Would it be possible to validate that the EXPLORE_ID set in that file is properly formatted to only have a single colon?

[dassowmd]

@antfido I tried pushing the feature, but can't on this repo.

But I was thinking something like this in explore-assistant-examples/load_examples.py

def validate(args):
    """Validate the given explore_id."""
    if '::' in args.explore_id:
        raise argparse.ArgumentTypeError(f"Invalid explore_id: {args.explore_id}. '::' not allowed")

def main():
    args = parse_arguments()

    # validate before making changes
    validate(args)

    # delete existing rows
    client = get_bigquery_client(args.project_id)
    delete_existing_rows(client, args.project_id, args.dataset_id, args.table_id, args.explore_id)

    # load data from file and insert into BigQuery
    data = load_data_from_file(args.json_file)
    insert_data_into_bigquery(client, args.dataset_id, args.table_id, args.column_name, args.explore_id, data)

Full file:

import argparse
from google.cloud import bigquery
import json

def parse_arguments():
    """Parse command line arguments."""
    parser = argparse.ArgumentParser(description='Load JSON data into BigQuery')
    parser.add_argument('--project_id', type=str, required=True, help='Google Cloud project ID')
    parser.add_argument('--dataset_id', type=str, help='BigQuery dataset ID', default='explore_assistant')
    parser.add_argument('--table_id', type=str, help='BigQuery table ID', default='explore_assistant_examples')
    parser.add_argument('--column_name', type=str, help='Column name, if different than "examples"', default='examples')
    parser.add_argument('--explore_id', type=str, required=True, help='The name of the explore in the model:explore_name format')
    parser.add_argument('--json_file', type=str, help='Path to the JSON file containing the data', default='examples.json')
    return parser.parse_args()

def get_bigquery_client(project_id):
    """Initialize and return a BigQuery client."""
    return bigquery.Client(project=project_id)

def delete_existing_rows(client, project_id, dataset_id, table_id, explore_id):
    """Delete existing rows with the given explore_id."""
    full_table_id = f"{project_id}.{dataset_id}.{table_id}"
    delete_query = f"DELETE FROM `{full_table_id}` WHERE explore_id = @explore_id"
    job_config = bigquery.QueryJobConfig(
        query_parameters=[bigquery.ScalarQueryParameter("explore_id", "STRING", explore_id)]
    )
    delete_job = client.query(delete_query, job_config=job_config)
    delete_job.result()  # Wait for the job to complete
    if delete_job.errors:
        print(f"Failed to delete existing rows for explore_id {explore_id}: {delete_job.errors}")
    else:
        print(f"Successfully deleted rows for explore_id {explore_id}")

def load_data_from_file(json_file_path):
    """Load data from a JSON file."""
    with open(json_file_path, 'r') as file:
        return json.load(file)

def insert_data_into_bigquery(client, dataset_id, table_id, column_name, explore_id, data):
    """Insert data into BigQuery using a SQL INSERT statement."""
    # Convert the data to a JSON string
    data_json = json.dumps(data)

    # Create a BigQuery SQL INSERT statement
    insert_query = f"""
    INSERT INTO `{dataset_id}.{table_id}` (explore_id, `{column_name}`)
    VALUES (@explore_id, @examples)
    """

    # Set up query parameters to prevent SQL injection
    job_config = bigquery.QueryJobConfig(
        query_parameters=[
            bigquery.ScalarQueryParameter("explore_id", "STRING", explore_id),
            bigquery.ScalarQueryParameter("examples", "STRING", data_json)
        ]
    )

    # Execute the query
    query_job = client.query(insert_query, job_config=job_config)
    query_job.result()  # Wait for the query to complete

    # Check if the query resulted in any errors
    if query_job.errors is None:
        print("Data has been successfully inserted.")
    else:
        print(f"Encountered errors while inserting data: {query_job.errors}")

def validate(args):
    """Validate the given explore_id."""
    if '::' in args.explore_id:
        raise argparse.ArgumentTypeError(f"Invalid explore_id: {args.explore_id}. '::' not allowed")

def main():
    args = parse_arguments()

    # validate before making changes
    validate(args)

    # delete existing rows
    client = get_bigquery_client(args.project_id)
    delete_existing_rows(client, args.project_id, args.dataset_id, args.table_id, args.explore_id)

    # load data from file and insert into BigQuery
    data = load_data_from_file(args.json_file)
    insert_data_into_bigquery(client, args.dataset_id, args.table_id, args.column_name, args.explore_id, data)

if __name__ == '__main__':
    main()