DynamoDB Enhanced - Change to use regex pattern for the attribute name cleaner

[debora-ito]

Motivation and Context

The attribute name sanitizer checked for a specific list of special characters, but the list was not comprehensive of all the valid special characters for a dynamo attribute name.

Attribute names with a character that was valid but not in the list would get an error like

ExpressionAttributeNames contains invalid key: Syntax error; key: "#AMZN_MAPPED_FirstName+LastName"

Customers submitted issues asking to add a new special character to the list (#3058 | #3139 | #3945 | #5603). This change will help prevent similar issues in the future.

Modifications

Testing

Screenshots (if appropriate)

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)

License

• I confirm that this pull request can be released under the Apache 2 license

[L-Applin]

Running the unit tests, two of them are failing in the codebuild jobs:

• ProjectionExpressionTest.severalNestedAttributes_handledCorrectly
• ProjectionExpressionTest. severalTopLevelAttributes_handledCorrectly

Notably, they check that the firstiteminlist[0] attribute handled correctly, should square brackets be excluded?

some unit test are failing