SNOW-1871392: Quoting changes for column names starting with _

[jochenott]

Please answer these questions before submitting your issue. Thanks!

1. What version of Python are you using? 3.12

2. What operating system and processor architecture are you using? osx-arm64

3. What are the component versions in the environment (pip freeze)? Main component: snowflake-sqlalchemy 1.7.2

4. What did you do?
   We have sqlalchemy code that uses constructs like some_expression.label("_column0"), i.e. it selects columns "as" an identifier that starts with an underscore.

5. What did you expect to see?
   So far (with smaller versions of snowflake-sqlalchemy), no quoting would be used when compiling these identifier names, effectively upper-casing these column names. This is what I expect to see.
   Now, as a result of mraba/underscore_column_id: use _ as column identifier #538, quoting is used. No upper-casing is done and the effective column names used by snowflake are no longer as expected, since snowflake sees a quoted name now and no longer does implicit uppercase conversion.

This breaks many of our queries that worked just fine until 1.7.1 and is a blocker to upgrade. I don't see any clear workaround, I think #538 should be reverted.

To reproduce:

from snowflake.sqlalchemy.snowdialect import SnowflakeDialect
from sqlalchemy.sql.compiler import SQLCompiler
import sqlalchemy as sa

dialect = SnowflakeDialect()

stmt = sa.Column("_a")
print(str(SQLCompiler(dialect, stmt)))

For snowflake-sqlalchemy 1.7.1, the output is:

_a

For snowflake-sqlalchemy 1.7.2, the output is:

"_a"

[sfc-gh-dszmolka]

hi - thanks for raising this with us. We'll take a look.

[sfc-gh-dszmolka]

fixed with #569, awaiting release

[sfc-gh-dszmolka]

released with snowflake-sqlalchemy 1.7.3