depends_on is empty when using backticks around schema and table names in BigQuery queries

[1210yuichi0]

Nice to meet you and thank you for the great OSS.
When using backticks to quote schema and table names in BigQuery queries, the depends_on field in the generated exposures.yml file is empty.

In BigQuery, schema and table names must be quoted with backticks. Documentation

Example 1: Using backticks (fails)

Metabase Query:

SELECT * FROM `work.transactions`  

Generated exposures.yml:

depends_on: []  

Example 2: Without backticks (works but breaks Metabase Field filters)

Metabase Query:

SELECT * FROM work.transactions  

Generated exposures.yml:

depends_on:  
  - ref('payments')  

Expected Behavior

The depends_on field should correctly capture dependencies even when schema and table names are quoted with backticks, as required by BigQuery's syntax.

Environment

• dbt-bigquery: 1.8.0
• dbt-core: 1.8.3
• dbt-metabase: 1.4.1

Steps to Reproduce

1. Create a query in Metabase that uses backticks to quote schema and table names.
2. Run the dbt-metabase sync process.
3. Check the exposures.yml file and observe that depends_on is empty.

Would it be possible to enhance dbt-metabase to correctly parse dependencies from queries that use backticks in BigQuery?

[syou6162]

@yuichi-github-dev The response from the API containing the query looks like this, for example.

• dbt-metabase/tests/fixtures/api/card/29.json

  Lines 70 to 77 in d6a4247

  	"dataset_query": {
  	"database": 2,
  	"type": "native",
  	"native": {
  	"template-tags": {},
  	"query": "select\n count(*)\nfrom STG_payments as p\n left join STG_orders as o on p.order_id = o.order_id\nwhere o.status = 'returned'\n;"
  	}
  	},

The code and regular expression for extracting the schema name and table name from a query like this are defined below, and I think that if you modify it to extend this, you should be able to extract ref correctly even in cases that include backquotes.

• dbt-metabase/dbtmetabase/_exposures.py

  Lines 300 to 307 in d6a4247

  	# Parse SQL for exposures through FROM or JOIN clauses
  	for sql_ref in re.findall(_EXPOSURE_PARSER, native_query):
  	# DATABASE.schema.table -> [database, schema, table]
  	parsed_model_path = [s.strip('"').lower() for s in sql_ref.split(".")]
  	# Scrub CTEs (qualified sql_refs can not reference CTEs)
  	if parsed_model_path[-1] in ctes and "." not in sql_ref:
  	continue

• dbt-metabase/dbtmetabase/_exposures.py

  Line 29 in d6a4247

  _EXPOSURE_PARSER = re.compile(r"[FfJj][RrOo][OoIi][MmNn]\s+([\w.\"`]+)")