Allow commas in namespace names

[kalra-mohit]

Namespace values that contain a comma get rejected with IllegalArgumentException. Showed up in practice with SQL Server/JDBC connection-string style namespaces, which use commas to separate options — e.g. sqlserver://host;databaseName=DB1,DB2.

Root cause is boring: NamespaceName's validation regex just never allowed , in the character class:

Pattern.compile(String.format("^[a-zA-Z:;=/0-9_\\-\\.@+]{%d,%d}$", MIN_SIZE, MAX_SIZE));

Before adding it I checked whether , is used as a delimiter anywhere that a newly-permitted comma could break something — namespace lists in query params, NodeId-style token splitting, that sort of thing. NodeId only parses on : and #, and there's no comma-delimited namespace query param anywhere, so there's nothing for this to collide with.

flowchart TD
    In["Namespace value containing a comma
e.g. sqlserver://host;databaseName=DB1,DB2"]
    In --> Old["Old PATTERN: comma not in allowed char class"]
    Old --> OldOut["No match -> IllegalArgumentException"]
    In --> New["New PATTERN: comma added to allowed char class"]
    New --> NewOut["Match -> NamespaceName created"]

Loading

Fix: added , to NamespaceName.PATTERN and updated the validation error message to mention commas as an allowed character.

Testing: added three regression cases to NamespaceNameTest#testValidNamespaceName — a comma-containing name, a bare ,, and a realistic SQL Server connection-string namespace. Confirmed all three throw before the fix and pass after:

./gradlew :api:test --tests "marquez.common.models.NamespaceNameTest"

3 failures before the fix, 15/15 passing after.

Fixes #3110.

[boring-cyborg]

Thanks for opening your first pull request in the Marquez project! Please check out our contributing guidelines (https://github.com/MarquezProject/marquez/blob/main/CONTRIBUTING.md).

[kalra-mohit]

@merobi-hub whenever you get a chance to take a look, happy to address any feedback!