fix: KeyConditionExpression rejects parenthesized sub-expressions

[AnatolyRugalev]

Summary

The KeyConditionExpression parser rejects parentheses around sub-expressions, returning ValidationException: Expected attribute name, got (. Both DynamoDB and DynamoDB Local accept these expressions. ConditionExpression handles parentheses correctly — the bug is specific to the KeyConditionExpression parser.

Version: dynoxide 0.9.8

Reproduction

dynoxide --port 8000 &

export AWS_ACCESS_KEY_ID=fake AWS_SECRET_ACCESS_KEY=fake \
       AWS_DEFAULT_REGION=us-east-1 EP=http://localhost:8000

# Create table
aws dynamodb create-table --endpoint-url $EP \
  --table-name test --billing-mode PAY_PER_REQUEST \
  --attribute-definitions AttributeName=PK,AttributeType=S AttributeName=SK,AttributeType=S \
  --key-schema AttributeName=PK,KeyType=HASH AttributeName=SK,KeyType=RANGE

# Seed data
aws dynamodb put-item --endpoint-url $EP --table-name test \
  --item '{"PK":{"S":"user#1"},"SK":{"S":"profile"},"name":{"S":"Alice"}}'

# ✅ Works — no parens
aws dynamodb query --endpoint-url $EP --table-name test \
  --key-condition-expression "#pk = :pk AND #sk = :sk" \
  --expression-attribute-names '{"#pk":"PK","#sk":"SK"}' \
  --expression-attribute-values '{":pk":{"S":"user#1"},":sk":{"S":"profile"}}'

# ❌ Fails — parens around each condition
aws dynamodb query --endpoint-url $EP --table-name test \
  --key-condition-expression "(#pk = :pk) AND (#sk = :sk)" \
  --expression-attribute-names '{"#pk":"PK","#sk":"SK"}' \
  --expression-attribute-values '{":pk":{"S":"user#1"},":sk":{"S":"profile"}}'
# => ValidationException: Expected attribute name, got (

Root cause

parse_single_condition in key_condition.rs doesn't handle Token::LParen — it immediately tries to parse an attribute path, which fails on (.

Fix

Two levels of parenthesis support:

1. Outer parens — strip_outer_parens removes balanced parens wrapping the entire expression before parsing (handles (pk = :pk AND sk = :sk) and ((expr)))
2. Per-condition parens — parse_single_condition counts and consumes wrapping parens around individual conditions (handles (pk = :pk) AND (sk = :sk))

Unit tests cover: per-condition parens, outer parens, nested parens ((pk = :pk)), mixed nesting (((pk = :pk)) AND ((begins_with(...)))), and parens with #name references.

Also submitted as a conformance test: nubo-db/dynamodb-conformance#1

Fixes #4

[hicksy]

Nice - the nested and mixed parens coverage looks thorough. Reviewing alongside the other two.

[hicksy]

Two small edits pushed to your branch:

1. Rebased onto current main as our workflows have been updated there recently.
2. cargo fmt (two trailing semicolons in the new consume_close_parens). No logic change.

As you were a first time contributor the auto CI checks needed approval. They should run now. Will merge once green.

[hicksy]

Quick follow-up: the earlier CI run failed on a pre-existing clippy lint on main (new iter_kv_map lint in Rust 1.95, unrelated to this PR). Pushed a fix on main and rebased your branch on top. CI re-running now.